Geographical Information Systems, Decision

0 downloads 0 Views 6MB Size Report
information from spatial databases and prepare an input file to HEC-HMS. ..... HEC-1 and HEC-2, have been updated and renamed HEC-HMS and HEC-RAS, ... the GIS is incomplete, inaccurate, or both) different levels of manual operation.
1

Geographical Information Systems, Decision Support Systems, and Urban Stormwater Management

James P. Heaney, David Sample, and Leonard Wright University of Colorado Boulder, Colorado

Final Report to the US Environmental Protection Agency Edison, NJ

1

This report was prepared by the University of Colorado under Cooperative Agreement No. CZ826256-01-0 with the EPA. The information presented does not necessarily reflect the views of the Agency, and no official endorsement should be inferred. The mention of trade names or commercial products does not imply endorsement by the United States government.

Table of Contents Abstract .............................................................................................................................................................................................v 1.0 Introduction................................................................................................................................................................................ 1 2.0 Literature Review...................................................................................................................................................................... 2 2.1 Overview of Sources of Reviewed Literature .................................................................................................................. 2 2.2 GIS as a Spatial Database for Urban Stormwater Modeling......................................................................................... 2 2.2.1 GIS as a pre-processor for urban stormwater models ............................................................................................. 3 2.2.2 GIS as a post-processor for urban stormwater models ........................................................................................... 4 2.2.3 GIS used to estimate spatial input parameters ......................................................................................................... 4 2.2.4 GIS used to estimate non-point source pollutant loads .......................................................................................... 5 2.3 Integration of GIS and Hydrologic Time Series.............................................................................................................. 5 2.4 Integration of GIS and Urban Stormwater Models ......................................................................................................... 6 2.5 Management Evaluation Using Integrated GIS and Urban Stormwater Models ....................................................... 7 2.6 Trends in the Integration of GIS and Urban Stormwater Modeling............................................................................. 8 3.0 Summary of Available GIS Urban Stormwater Modeling Software ................................................................................ 9 3.1 SWMM and EPA Windows SWMM.............................................................................................................................. 11 3.2 PCSWMM ’98 and PCSWMM GIS................................................................................................................................ 11 3.3 XP-SWMM by XP Software (Also Available as Visual Hydro by CAiCE)............................................................ 12 3.4 SWMM-DUET.................................................................................................................................................................... 13 3.5 DHI Software ....................................................................................................................................................................... 13 3.5.1 MIKE SWMM............................................................................................................................................................. 13 3.5.2 MOUSE and MOUSE GIS........................................................................................................................................ 13 3.6 Wallingford Software-HydroWorks and InfoWorks .................................................................................................... 16 3.7 Summary............................................................................................................................................................................... 17 4.0 Future Urban Stormwater Modeling in a DSS Environment........................................................................................... 19 4.1 State Information................................................................................................................................................................. 21 4.1.1 GIS................................................................................................................................................................................. 21 4.1.2 Time series ................................................................................................................................................................... 22 4.1.3 Relational database..................................................................................................................................................... 23 4.2 Process Information-Simulation Tools ............................................................................................................................ 28 4.3 Evaluation Tools ................................................................................................................................................................. 28 4.4 Overall DSS for Water Management............................................................................................................................... 28 5.0 Application of GIS and DSS to Micro Storm Analysis .................................................................................................... 31 5.1 Spatial Scale and GIS-Stormwater Modeling ................................................................................................................ 32 5.2 Description of Happy Acres Case Study GIS................................................................................................................ 37 5.3 Simulation Tools for Hydraulic Design.......................................................................................................................... 44 5.4 Simulation Tools for Hydrologic Analysis .................................................................................................................... 49 5.4.1 Hydrologically functional landscaping................................................................................................................... 49 5.4.2 Determination of runoff volumes using NRCS method....................................................................................... 52 5.4.3 Breakdown of calculated volumes per function .................................................................................................... 52 5.5 Simulation Tools for Cost Analysis ................................................................................................................................ 55 5.6 Optimization of Control Options for Happy Acres....................................................................................................... 59 5.7 Decision Support Systems and the Happy Acres Case Study..................................................................................... 61 6.0 Summary and Conclusions.................................................................................................................................................... 62 6.1 Summary............................................................................................................................................................................... 62 6.2 Conclusions ......................................................................................................................................................................... 62 7.0 References ................................................................................................................................................................................ 65 Appendix: Happy Acres Database............................................................................................................................................. 72

ii

List of Figures Figure 3.1: PCSWMM output.................................................................................................................................................... 12 Figure 3.2: Visual Hydro ............................................................................................................................................................. 14 Figure 3.3: Mouse GIS user action............................................................................................................................................ 15 Figure 3.4: System response to user action, Mouse GIS........................................................................................................ 15 Figure 3.5: InfoWorks from Wallingford Software ................................................................................................................ 16 Figure 4.1: DSS structure and components .............................................................................................................................. 20 Figure 4.2: Relational database query example in ArcView using water use data ........................................................... 25 Figure 4.3: Spatial results for example query from figure 4.2. ............................................................................................. 26 Figure 4.4: Query results output to Excel using Avenue script tool and Microsoft DDE................................................ 27 Figure 4.5: CU-CADSWES DSS............................................................................................................................................... 29 Figure 4.6: Danish Hydraulic Institute DSS, based on integrated water resources modeling......................................... 30 Figure 5.1: Proposed DSS for microstorm analysis ................................................................................................................ 32 Figure 5.2: BASINS dataset for Boulder, Colorado ............................................................................................................... 34 Figure 5.3: ArcView coverage of Boulder, Colorado............................................................................................................. 35 Figure 5.4: City of Boulder ArcView GIS coverage for University Hill neighborhood, Boulder, Colorado............... 36 Figure 5.5: AutoCAD file for University Hills neighborhood, Boulder, Colorado........................................................... 38 Figure 5.6: AutoCAD coverage for study area........................................................................................................................ 39 Figure 5.7: Study area topography ............................................................................................................................................ 40 Figure 5.8: Study area land use.................................................................................................................................................. 41 Figure 5.9: Study area soils ......................................................................................................................................................... 42 Figure 5.10: Study area sewer network..................................................................................................................................... 45 Figure 5.11: Conventional storm drainage............................................................................................................................... 50 Figure 5.12: Illustration of hydrologically functional landscape ......................................................................................... 51 Figure 5.13: Allocation of available storage for initial abstraction and land use.............................................................. 55

iii

List of Tables Table 3.1: Summary of available urban stormwater modeling software with GIS linkages ........................................... 10 Table 3.2: Characteristics of urban storm stormwater models .............................................................................................. 18 Table 5.1: Available BASINS data attributes.......................................................................................................................... 33 Table 5.2: Minimum horizontal accuracy and example features for various map scales in urban areas....................... 34 Table 5.3: Mix of land uses in Happy Acres ........................................................................................................................... 37 Table 5.4: AutoCAD layers for study area............................................................................................................................... 43 Table 5.5: Right of way characteristics..................................................................................................................................... 43 Table 5.6: Lot characteristics for residential parcels .............................................................................................................. 44 Table 5.7: Aggregate characteristics for commercial, apartments, and schools ................................................................ 44 Table 5.8: Sewer network design hydrology ........................................................................................................................... 46 Table 5.9: Sewer network design hydraulics ........................................................................................................................... 47 Table 5.10: Sewer network design cost.................................................................................................................................... 48 Table 5.11: Initial abstraction as a function of curve numbers, CN..................................................................................... 49 Table 5.12: SCS hydrologic classifications, and calculation of unit storage values, 1/99$............................................. 53 Table 5.13: Calculation of developed and predevelopment stormwater volumes for Happy Acres .............................. 54 Table 5.14: Land valuation for medium density lot, 1/99$.................................................................................................... 56 Table 5.15: Cost analysis of landscaping for medium density lot, 1/99$............................................................................ 57 Table 5.16: Calculation of unit costs for controls, including opportunity costs for land, 1/99$..................................... 58 Table 5.17: Results of LP optimization-land use allocation by function (includes opportunity costs)......................... 60 Table 5.18: Least-cost LP solutions for land Use/BMP options (including land costs) for Happy Acres.................... 61 Table A-1: Parcel attributes ........................................................................................................................................................ 73 Table A-2: Right of way attributes ............................................................................................................................................ 78

iv

Abstract This report reviews the application of Geographic Information System (GIS) technology to the field of urban stormwater modeling. The GIS literature is reviewed in the context of its use as a spatial database for urban stormwater modeling, integration of GIS and hydrologic time series, and integration of GIS and urban stormwater models (from both a software and management perspective). The available urban stormwater modeling software is reviewed and discussed with respect to their GIS integration capabilities. Decision Support Systems (DSS) are reviewed with respect to their integration with GIS, and their applicability to urban stormwater management problems. A simplified neighborhood scale DSS is presented that includes a GIS, a database, a stormwater system design template, and an optimization capability for screening alternatives. The area and soil based NRCS method is used for calculating runoff from GIS information. Using economic analysis that compares the costs of controls, including the opportunity cost of land for land intensive controls, the optimal selection of Best Management Practice (BMP) controls was accomplished by use of a linear programming (LP) method. The intent of this presentation is to provide an example of the types of problems that become possible to explore with the application of DSS and GIS technology on a small scale. This field is evolving rapidly, and warrants carefully targetted research efforts, particularly at developing nonspecific software tools that aid in integrating existing models.

1.0 Introduction A mathematical model of an urban hydrologic response to precipitation usually requires extensive data due to the complexity of surfaces, flow paths and conduits found in developed locales. Many of these data are geographic in nature; e.g., geographic boundaries of the hydrologic basin provide boundary conditions of the mathematical model. Therefore the marriage of mathematical stormwater models and geographic information systems (GIS) is a natural development of simulation and database technology. The relationship between urban stormwater models and GIS may take many forms. This is apparent from the nearly 50 journal articles, conference proceedings and internet reports surveyed for this review of recent literature. The relationship between GIS and urban stormwater models may be distinct, where the GIS functions as a separate pre- and post-processor; or the distinction may be blurred, where the model is seamlessly integrated to the GIS. The purpose of this report is to accomplish several tasks. In chapter 2 a review of technical literature is performed to determine how GIS is being used in the field of urban storm stormwater modeling. Next, in chapter 3, the predominant urban stormwater models are reviewed within the context of the taxonomy developed in chapter 2. Then, in chapter 4, looking at the future directions of urban stormwater models, Decision Support Systems (DSSs) are described. DSS is now being used extensively for river basin modeling, particularly in the hydropower context. This type of system lends itself to unstructured problems where data integration is a key to evaluation of the problem. The various components of DSS including models, database structure, GIS, optimization, and time series management are discussed. A process level DSS is developed for a textbook subdivision in chapter 5. This DSS contains a GIS, including graphic features and a relational database, a system simulator, and an optimizer. Stormwater design templates were created using Excel spreadsheets, paralleling the design problem from the textbook. Next, GIS data were utilized in a simple hydrologic model using the NRCS (National Resources Conservation Service) method. This data was combined with unit cost data into a linear programming model (LP) in order to develop the least costly mix of BMP controls that maintain the same initial abstraction after development as before. Suggestions for further improvement of the DSS are made by comparison of the DSS structure with those found in chapter 4. Finally conclusion are presented in chapter 6.

1

2.0 Literature Review 2.1 Overview of Sources of Reviewed Literature The GIS literature is broad, due to the wide variety of areas that utilize geographic data. Likewise, the literature describing GIS applications in water resources is itself very broad. However, much of this work in water resources has been in the area of natural hydrology and large-scale, river-basin hydrology. GIS has a long history in this area due in large part to the early availability of remotely sensed spatial data suited for this purpose. A good overview of the concepts of GIS and database technology and their application within the field of natural systems hydrology is found in Singh and Fiorentino (1996). The use of GIS in modeling urban stormwater systems has been more limited due to the need for large, expensive and detailed spatial and temporal databases, along with the fact that many computer tools used in urban stormwater modeling are not easily amenable to integration with GIS. However, as local data gathering efforts have increased and software integration has evolved, the use of GIS in urban stormwater is now widespread. Shamsi et al. (1995) estimate that more than 70% of the information used by local governments is georeferenced. Much of this information has been, or will be, transferred to a digital format, usually a GIS. Recent literature was found in several distinct fields. From the water resources field, recent conferences focusing on urban stormwater have several papers on GIS. Proceedings from two European conferences on urban stormwater by Butler and Maksimovic (eds. 1998), and Seiker and Verworn (eds. 1996), have a wealth of current information on GIS. The American Water Resources Association (AWRA) has sponsored conferences specific to the use of GIS in water resources, such as Harlin and Lanfear (1993) and Hallam et al. (1996). These reports have sections devoted to urban stormwater, of which modeling is a recurring theme. Significant literature in this area was also found on the internet. The Center for Research in Water Resources at The University of Texas at Austin has a large online library of reports and papers on the use of GIS for hydrologic research, some of which concerns the modeling of urban areas (University of Texas, 1998). Other resources were found in the GIS field. One software provider, Environmental Systems Research Institute (ESRI), hosts a large annual international user conference. The proceedings for these conferences are located on the internet at http://www.esri.com (ESRI 1998). The International Association of Hydrological Sciences (IAHS) publishes the proceedings from its many conferences, some of which have dealt specifically with the integration and application of GIS and water resources management (e.g. Kovar and Nachtnebel 1996). Other IAHS conferences have focused on applications, which usually have several papers on using GIS for that application. For example Simonovic et al. (1995) edited “Modeling and Management of Sustainable Basin-Scale Water Resource Systems”, proceedings from a 1995 conference in Boulder, CO. which contained several papers on GIS and model integration. 2.2 GIS as a Spatial Database for Urban Stormwater Modeling The most basic role a GIS can play in the modeling of urban stormwater is that of a simple preprocessor of spatial data. As a pre-processor, GIS may simply store geographic information in a database, or it may be used to calculate model-input parameters from stored geographic data. 2

Frequently data are exported from the GIS in a file format consistent with a model-input file. As a post-processor, GIS may be used to map water surface elevations, concentrations, etc., or to derive spatial statistics based on model output. Shamsi (1998) describes the batch transfer of data from a GIS to SWMM as the interchange data. The GIS and SWMM are operated separately, with no direct interlink. The GIS is used to extract data required by SWMM from the spatial database into a file compatible with a SWMM input file. A recurring theme in recent literature focuses on the ability to get the most out of data by assuring that information tools are consistent. This idea has been termed “hydroinformatics” and is especially prominent in the recent European literature (Fuchs and Scheffer 1996). 2.2.1 GIS as a pre-processor for urban stormwater models Many municipalities store general spatial information in a GIS, and the information is used for a wide variety of purposes and functions within the institutional framework. VanGelder and Miller (1996) describe a typical use of GIS as a spatial database for modeling stormwater from a municipal airport. Detailed georeferenced data were used in conjunction with maintenance data to develop an operation and management schedule as well as to link node information needed to create a SWMM EXTRAN model. Pryl et al. (1998) use a GIS to export details of the urban stormwater network to a hydraulic simulator for Prague in the Czech Republic. The Danish Hydraulic Institute (DHI) program Model Of Urban SEwers (MOUSE) was used to simulate various scenarios for development of an urban stormwater master plan. Rodriguez et al. (1998) used a GIS to study stormwater characteristics of an urban area in Nantes, France. This study used the urban land parcel as the base hydrologic unit of a detailed hydrologic model, as opposed to the more typical basin defined by topography and the layout of the stormwater network. A detailed water budget was performed around the owner-defined parcel. This physically based hydrologic model was then used with the stormwater network to analyze the behavior of urban catchments under a wide variety of storm events. The idea of using small hydrologic units based on land ownership for urban stormwater modeling is ideally suited for GIS applications and is useful when simulating the effect of management decisions made at the parcel level. Sotic et al. (1998) began a preliminary design of CSO facilities in Kumodraz, Yugoslavia with paper maps. Existing paper maps and other data were used to create a GIS, which in turn was used to aid in the design and analysis of the CSO system. This “hydroinformatic” approach consists of developing a set of tools to collect and process data in a consistent manner. The attention to consistency in data transferability is to assure that the greatest value is achieved from the dataset. In this case, the GIS was used to integrate a Digital Elevation Model (DEM), the street network, and the sewer network; then this information was transferred to the BEAMUS hydraulic simulation model (Sotic et al. 1998). A similar hydroinformatic approach is described for the town of Pilsen in the Czech Republic by Hora et al. (1998). Beginning with paper maps, a GIS was built from the ground-up. The complete process is described, ending with an information tool that was used to create a hydrodynamic model of the sewer system, store monitored flow and rain data, evaluate current hydraulic sewer capacity and evaluate the feasibility of alternative sewer developments. Barbe et al. (1993) integrate data transfer from a GIS and a SCADA system to a SWMM model of the Jefferson Parish stormwater stormwater system in Louisiana. The SWMM RUNOFF block was used to simulate the hydrologic runoff characteristics of the area. Geospatial data were transferred from the GIS to the SWMM RUNOFF data file. Similarly, the EXTRAN block 3

was used to simulate the pipe network, and the network connectivity was transferred from the GIS to the SWMM EXTRAN data file. Time series data from 150 monitoring sites were transferred from a SCADA system to the SWMM model for calibration purposes. 2.2.2 GIS as a post-processor for urban stormwater models GIS may also be used to accept model output. Xu et al. (1998) describe a mixed land use hydrologic model that uses GIS as a pre- and post-processor of model information. For this application, the model output of time series of simulated flows may be depicted dynamically through an ArcView interface. Sorensen (1996) describes a typical use of GIS to present model output, that of depicting flood inundation maps from the GIS. MIKE GIS is a modeling tool from DHI that interfaces between ArcInfo or ArcView and MIKE, a flood assessment model. First developed to study flood management in Bangladesh, MIKE GIS uses both the maximum flood extent and the time series of flooding to analyze expected damages from peak inundation and the duration of inundation (Sorensen 1996). A key element to this work is that the GIS is used for more than mapping model output, but that spatial analysis is done with the GIS that adds to the information gained from the model output alone. Shamsi (1998) discusses the difference between transferring data files between ArcView and SWMM and creating an interface that uses SWMM output as a spatial coverage layer in a GIS. This “interface method” (as opposed to the interchange method described above) involves creating a SWMM menu within ArcView. Pre- and post-processors of SWMM input and output files create input files, read output files, and join and unjoin data files (Shamsi 1998). These options are made available in ArcView; however SWMM is run separately from ArcView (Shamsi 1998). 2.2.3 GIS used to estimate spatial input parameters One of the most important hydrographic features of an urban surface is impervious area. Fankhauser (1998) describes a method to estimate impervious area from color infrared aerial photographs and orthophotos. These images have a ground resolution of 25 to 75 centimeters. A raster based GIS, IDRISI, was used to estimate imperviousness to within 10% of the value determined manually for an entire basin. However, the deviation for individual catchments was much higher. For this reason, this method was recommended only for large project areas where the high costs of parameter estimation could be justified. Olivera et al. (1996) use GIS to calculate hydrographic properties of terrain for non-point load estimation. Flow paths calculated from paths of steepest descent are used to calculate flow properties of basins. Cluis et al. (1996) use topographic data and GIS functions to derive important hydrographic characteristics of the terrain such as overland flow paths in a raster based format. Mercado (1996) describes the use of detailed spatial information in the creation of a stormwater model in Tallahassee, FL using XPSWMM software. Scanned and georectified black and white aerial photography was used as a background with other GIS based data, including two foot contour elevations, streams, buildings, roads, etc. A DEM was created in ArcInfo, and the Triangulated Irregular Network (TIN) and Grid functions were used to define areas of high slope and erosion potential, flow gradients and very accurate subbasin delineation (Mercado 1996). 4

Herath et al. (1996) used high-resolution raster data sets to develop a distributed GIS-based urban hydrologic model. Data sets included 50 m x 50 m and 20 m x 20 m land use grids; 1:25,000 plans were used to develop imperviousness by land use, a 50 m x 50 m DEM, population density, water supply data, and rainfall. Herath et al. (1996) integrated the hydrologic model with the GIS, by writing the numerical simulation codes within the GIS, thus reducing problems of data transfer. However, the computational time was felt to be too high for practical use due to inefficiencies of performing the hydrologic simulation within the GIS (Herath et al. 1996). Olivera et al. (1998) developed a GIS-based preprocessor for the new HEC-HMS model developed by the Army Corps of Engineers’ Hydrologic Engineering Center. HEC-HMS is an updated version of the popular HEC-1 hydrologic model. Olivera et al. (1998) describe HECPrePro as a system of ArcView scripting programs and controls to extract hydrographic information from spatial databases and prepare an input file to HEC-HMS. Using SCS curve numbers and a DEM, HEC-PrePro delineates streams and basin boundaries, determines their interconnectivity, and calculates parameters for each stream and basin (Olivera 1998). A benefit to automating the calculation of hydrologic parameters that were traditionally estimated manually is that results are reproducible, i.e., they are not dependent on the bias or experience of the modeler. 2.2.4 GIS used to estimate non-point source pollutant loads Using land use as a predictor of non-point source loads is a common use of GIS and hydrologic models. Hauber and Joeres (1996) describe how a GIS was used to preprocess urban pollutant loads for the Source Loading and Management Model (SLAMM). Similarly, Wright et al. (1995) estimated nutrient loads from developed areas in the Onondaga Lake stormwater basin in upstate NY with the GRASS GIS. These preprocessed loads were then routed from the developed basins using the SWMM RUNOFF model. Battin et al. (1998) describe the EPA’s BASINS (Better Assessment Science Integrating Point and Non-Point Sources) software, which integrates watershed point and non-point source load data, the watershed hydrology program HSPF and the receiving water quality simulation program QUAL2E. Olivera et al. (1996) describe the use of GIS to account for the spatial variability of terrain in pollutant loading from a variety of land uses. The authors review the strength of GIS in quantifying spatially distributing loads, and point out that this is a distinct advantage over lumped models. Scarborough and Yetter (1998) evaluated the Non-Point Source (NPS) module in BASINS 2.0 and found it to be a useful tool for evaluating NPS pollution. However, several problems were found when evaluating a small watershed with the GIS data included with the program. The most critical problem was that of coverage alignment (Scarborough and Yetter 1998). Boundaries of land use and watershed boundaries did not match for the test case study, the St. Jones watershed in Delaware. 2.3 Integration of GIS and Hydrologic Time Series For the purposes of urban stormwater modeling, spatial data may usually be viewed as static. Changes in geographic data are typically modeled in a scenario manner, e.g., a model run may be 5

done for an undeveloped watershed, and then a developed scenario is performed using the same hydrologic conditions. Hydrologic and meteorological data are commonly a time series of discrete values. Therefore some attention must be paid to the integration of spatial and time series data. This idea of consistency among data is key to the concept of hydroinformatics. Pryl et al. (1998) describe the integration of time series with GIS to accomplish urban stormwater master planning in the Czech Republic. Similarly, Rodriguez et al. (1998) use time series in their analysis of the water budget based on parcel-level urban spatial data. Time series integration was a key element in the work reported by Barbe (1996) in Louisiana. A large network of 150 monitoring locations fed a SCADA system with many time series data that were integrated with GIS data and the SWMM model. An Oracle database was used to manage non-spatial data for this project (Barbe 1993). Da Costa et al. (1995, 1996) examined this problem in developing the Portuguese Water Resources Information System. The integration of GIS with temporal data is described as one of the great challenges of developing this system (da Costa et al. 1996). To accomplish this integration, a database was developed using Oracle software to underlie the information system. A special processing module was developed to interface time series data with the GIS. The GIS portion used the ESRI ArcView software. Sorensen et al. (1996) describe the use of time series in an application of MIKE GIS in Bangladesh. Sotic et al. (1998) describe the integration of rainfall and flow time series with geographic data in a hydroinformatic manner in Yugoslavia. Wolf-Schumann and Vaillant (1996) describe in detail the need for integrated time series with georeferenced data. The development of TimeView, a time series management tool, is described as adding a whole dimension (time) to spatial data. TimeView is integrated with ArcView, so that a user can select a geographic feature in ArcView (e.g. a monitored manhole), and TimeView returns a time series of measured data in graphical format. 2.4 Integration of GIS and Urban Stormwater Models The linking of GIS and several hydrologic process models (beyond creating pre-processed data files within the GIS) is examined by Charnock et al. (1996) and DeVantier and Feldman (1993). Issues of differing scale properties and error propagation are addressed. The use of GIS as a central hub of information, which is fed to several satellite process models, is favored over coupling all the processes in one large program. Kopp (1996) addresses these same issues and argues for more data standards to streamline hydroinformatics. Sponemann et al. (1996) explain how a GIS can be shared among many varied users, e.g. gas utilities, water utilities, stormwater, etc. thus maximizing the benefits derived from data collection and management. Greene and Cruise (1995) developed an urban watershed modeling system using the SCS rainfall-runoff methodology and GIS parcel attributes. Meyer et al. (19993) developed a raster based GIS for an urban subdivision in Ft. Collins, Colorado and found that the results compared favorably with non-GIS hydrologic studies of the same area. Shamsi (1998) distinguishes three forms of information exchange between ArcView and SWMM. The interchange and interface methods are described above, and involve the transfer of information between ArcView and SWMM, which are run independently. Shamsi (1998) defines the third method, integration, as the most advanced of the methods. SWMM is used as the hydrologic and hydraulic simulator and is executed from within ArcView. This form of integration includes performing all program tasks within ArcView: creating SWMM input data, 6

editing data files, executing SWMM, and displaying output results (Shamsi 1998). Integration as defined by Shamsi (1998) combines a SWMM Graphical User Interface (GUI) with a GIS to provide a complete data environment. The advantages of a GUI were advanced by Shamsi (1997), who provided a summary of software features and needs for SWMM interfaces. Feinberg and Uhrick (1997) discuss integrating an infrastructure database in Broward County, FL with a GIS and water distribution and wastewater models. The HydroWorks model is used to simulate the wastewater collection system, with close integration with the database of infrastructure characteristics and the GIS. Refsgaard et al. (1995) describe the evolution of DHI’s land process hydrologic model, SHE, and its extensive use of GIS. Ribeiro (1996) describes the use of a raster-based GIS to interface with HSPF to analyze the effects of basin urbanization. Hellweger (1996) developed an ArcView application using the Avenue scripting language to perform the model calculations of USDA’s hydrologic model TR-55. Mark et al. (1997) use the MOUSE program from DHI to evaluate stormwater in Dhaka, along the banks of the Ganges and Bramaputra rivers in Bangladesh. Integration of GIS, time series, and the hydraulic model were accomplished to better understand flooding characteristics. Maximum inundation and duration of inundation were mapped using MOUSE and GIS. Shamsi and Fletcher (1996) describe in detail the linkage of ArcView and SWMM for the City of Huntington, WV. ArcView is shown to be a user-friendly environment to perform stormwater modeling. Bellal et al. (1996) studied partly urbanized basins using a linked GIS and hydrologic model. The hydrologic model was based on a non-urban water budget, with modifications to account for urbanization. The GIS was based on a DEM and raster-based land use data. 2.5 Management Evaluation Using Integrated GIS and Urban Stormwater Models The integration of GIS, time series data, and an urban stormwater model is usually done to evaluate management options. These options may be watershed-based, which would likely include non-urban areas, or they may be local to the urban area. Rodriguez et al. (1998) describe an integrated GIS and urban hydrologic model to evaluate small storm hydrology for parcel level management decisions. Tskhai et al. (1995) use a GIS linked with an optimization model to evaluate ecological and economic alternatives for the Upper Ob River in the Altai region of Russia. While not strictly an urban runoff model in the traditional sense, this project does link urban management decisions with an economic optimization model. Makropoulos et al. (1998) focus on urban sustainability to evaluate stormwater systems. Beginning with the idea that low energy solutions that control impacts at the source are more sustainable, Makropoulos et al. (1998) demonstrate how a raster-based GIS (IDRISI) can be used to integrate theoretical concepts and site specific spatial characteristics. The strength of GIS can be used as a common ground between specialists and non-specialists to help them communicate effectively. Bellal et al. (1996) studied the effect of urbanization on a watershed using a linked hydrologic model based on a DEM and a GIS. A water budget approach was used around each raster unit to account for changes due to urbanization. Mark et al. (1997) describe a detailed evaluation of flood management techniques in Dhaka, Bangladesh, using MOUSE GIS. Xue et al. (1996) and Xue and Bechtel (1997) describe the development of a model designed to evaluate the effectiveness of Best Management Practices 7

(BMP’s). This model, called the Best Management Practices Assessment Model (BMPAM), was linked with ArcView to create an integrated management tool. This integrated model was used to evaluate the pollutant load reduction potential of a hypothetical wet pond in Okeechobee, Florida. Kim et al. (1998) used ArcView with an economic evaluation model and a hydraulic simulator to evaluate storm sewer design alternatives. The hydraulic simulator was used to generate initial design alternatives, which where in turn evaluated with an economic model. The GIS was used to store spatial information, generate model input, and present alternative solutions. The complete package of GIS, economic evaluation model, and hydraulic simulator was termed a Planning Support System (Kim et al. 1998). 2.6 Trends in the Integration of GIS and Urban Stormwater Modeling The trend towards a data-centric suite of evaluation tools is clear. The central idea behind the European concept of hydroinformatics is that a consistent database is used for a variety of purposes. The model is no longer the central unit driving the decision process. Neither, however, has the GIS become the central data tool, due in large part to its inability to handle temporal information effectively. Researchers who have paid equal attention to the model (the processes), the GIS (the spatial data), and the temporal information (time series of hydrologic processes) seem to have had considerable success. The integration of GIS and urban stormwater models should therefore include integration with a database structure equipped to handle time series. Several advanced applications have used a non-graphic database (e.g. dBase, Oracle, Access) that is queried by both the GIS and the hydrologic/hydraulic models. While clearly an evolving area, this approach seems to hold the most promise for the purpose of urban stormwater decision support systems.

8

3.0 Summary of Available GIS Urban Stormwater Modeling Software As described in section 2, a useful taxonomy to define the different ways a GIS is used in urban hydrologic and hydraulic modeling is presented by Shamsi (1998). The three methods defined by Shamsi (1998) are data interchange, program interface, and program integration (Shamsi 1998). A fourth grouping was added for this report, the “intermediate program”. Several commercial modeling products feature a data management program to facilitate data transfer between the GIS and a model. A short description is given below in order of increasing sophistication. Data Interchange: a batch process is used to transfer data to and from the model data set. For example, the GIS may be used to calculate model input parameters e.g., catchment slope, or to query an existing spatial coverage, such as land use. Then portions of the GIS query file can be copied into a model-input file with no direct link between the GIS and the model. The model is executed independently from the GIS, and portions of the output files may be copied back into the GIS as a new spatial coverage for presentation purposes. Intermediate Program: a data management program is used to transfer information between a GIS and a model. This data management program is written specifically to import data from a variety of common third party GIS software, and export to a model data set. Under certain conditions this intermediate program could be defined as an interface, but generally it is not. Program Interface: a direct link consisting of a pre- and post-processor is used to transfer information between the GIS and the model. This process automates the data interchange method. Model-specific menu options are added to the GIS. The model is executed independently from the GIS, however the input file is created in the GIS. For example, in the data interchange method, the user finds a portion of a file and copies it. An interface automates this process, so that the pre- and post-processor finds the appropriate portion of the file automatically. Program Integration: while the interface method can’t launch the model from the GIS, under the integration method, the model and the GIS are together within one Graphical User Interface (GUI). This represents the closest relationship between GIS and model, though “closest” does not necessarily mean “best”. It may be more efficient for a model to be independent from a GIS in certain situations. As noted elsewhere in this report, the development of a GIS for use in urban hydrologic and hydraulic modeling is an expensive investment. Typically the most advanced tools are created for advanced applications, where a full GIS is in place. For some applications, a DOS-based model may still be the most appropriate. However, as more urban areas create GIS coverages, the integration of modeling software and GIS software will become more useful and more prevalent. The Storm Water Management Model (SWMM) is the most widely used urban hydrologic/hydraulic model in the US. In addition to SWMM, numerous other hydrologic models were created in the US during the 70s including the US Army Corps of Engineers Hydrologic Engineering Center “HEC” series of models (HEC-1 through 6). Two of the most popular models, HEC-1 and HEC-2, have been updated and renamed HEC-HMS and HEC-RAS, 9

respectively. These two models have been updated from the original DOS model with a MS Windows based GUI. HSPF, and ILLUDAS are other models developed in the 70’s, which are still used today. The original SWMM model, available at no charge from the US EPA (at the following website: http://ftp.epa.gov/epa_ceam/wwwhtml/ceamhome.htm) was written in Fortran-77 for mainframe computers (Huber and Dickinson 1988). The model was originally written during the 70s, with several major improvements made in the early 80s. It has continued to evolve since being ported to personal computers. Version 4.31 is the latest release; however numerous other modifications exist to the program (e.g. UD-SWMM, a modification of SWMM by the Urban Stormwater and Flood Control District of Denver, Colorado). SWMM runs in MS-DOS in a text-based environment, which is not the user-friendly windows and graphical user interface (GUI) based environment that is expected today. Despite these shortcomings, it has an active user community within the United States. Lack of funding support for SWMM during the 80s and 90s meant that the model had to be selfsustaining. Interested parties such as local governments, consultants, and third party developers added their own refinements to the model, with very little support from the federal government. Because these refinements added value to the original program code, the developers started to charge for these improvements. XP-SWMM (XP-Software 1998) and PCSWMM (CHI 1998) are examples of this type of refinement. The SWMM user’s listserver has developed into a selfsustaining community of users. Information on accessing the listserver can be found at http://www.chi.on.ca/swmmusers.html During the 1980’s, several models started to evolve in Europe. Two of them are HydroWorks, from Wallingford Software in Great Britain, and MOUSE from the Danish Hydraulic Institute, DHI, in Denmark. Unlike EPA SWMM, these models are proprietary. These models are listed in table 3.1, with the addition of MikeSWMM, which is the result of a recent collaborative effort between DHI and Camp, Dresser, and McKee (CDM). This product uses the latest SWMM model engine available from the US EPA, and adds the MIKE GUI and MOUSE GIS from DHI. Table 3.1: Summary of available urban stormwater modeling software with GIS linkages Product Model

Interface Company/Source

HydroWorks/ InfoWorks Mouse GIS

Hydroworks Hydroworks Mouse

Mike

MikeSWMM

SWMM

Mike

PCSWMM/GIS SWMM

PCSWMM

XPSWMM

XPSWMM

SWMM

Website

HR Wallingford/ Montgomery Watson Danish Hydraulic Institute/

www.wallingford-software.co.uk

Danish Hydraulic Institute/ Camp Dresser and McKee Computational Hydraulics International CAiCHE

www.mikeswmm.com

www.dhi.dk

www.chi.on.ca www.xpsoftware.com

The following sections describe commercial and public domain products that are currently available for urban hydrologic and hydraulic modeling. The above taxonomy is used to define how each one handles information transfer between a GIS and the model. However, the reader is 10

cautioned that while integration may be the most advanced method of using a GIS and model together, it is not necessarily the best method for every application. For some applications (especially when the GIS is incomplete, inaccurate, or both) different levels of manual operation may be more appropriate. For example, a limited GIS may exist for an urban watershed, along with very detailed and accurate CAD files. Certain commercial products (e.g. Visual Hydro by CAiCE) can handle CAD drawings better than a product designed to run a pre-existing GIS. If resources were not available to create a GIS, it would be appropriate to use a product suited to the available data. 3.1 SWMM and EPA Windows SWMM As stated previously, SWMM is a DOS based program developed under US EPA funding during the late 1970’s and early 1980’s. There is no provision to link directly or indirectly with a GIS other than through standard input text files. This is the most basic version of SWMM available. This version of SWMM is important because it is in the public domain, and the source code is readily available. The latest version of the DOS based SWMM can be found at http://www.ccee.orst.edu/swmm/ In 1994, the US EPA produced a Windows-based GUI for SWMM. This program (also available at http://www.epa.gov/ost/SWMM_WINDOWS/) runs on Windows version 3.1, and is therefore somewhat outdated. This program is also limited by the fact that the DOS based SWMM engine is in a constant state of improvement by developers and users because the Fortran source code is available. Unfortunately, the Windows SWMM program used the SWMM engine available circa 1994, and the newer versions of the SWMM engine cannot easily be substituted. Therefore the program has quickly become outdated, and has few users. Windows SWMM could not be linked directly to a GIS program. To use either of these programs with a GIS, the data-interchange method must be deployed to transfer information from a GIS to an input file. The GIS may be used to store and estimate model input parameters. The GIS could be queried for the needed values, and the values could then be transferred to the input file. The level of automation to perform this task depends on the user. It could be as simple as copying the needed values onto a Windows clipboard and pasting them into the input file, or developing special queries from the GIS to create an input file automatically. 3.2 PCSWMM ’98 and PCSWMM GIS PCSWMM-98 is a set of 32 bit applications designed to facilitate running SWMM. These tools include an ASCII text editor, an animated hydraulic grade line plot, a chart wizard, an Internet wizard, a batch file control, a rainfall analysis package, a bibliographic database, a sensitivity analysis wizard, and a calibration wizard. The GUI allows files from many sources to be linked, including those accessed across Intranets and Internets. PC-SWMM GIS is an optional tool that works directly with CAD or GIS files in constructing a link-node database for running the model from the existing data sources. After importing the data from a CAD or GIS file, an aggregation tool allows semiautomatic construction of a simplified link-node model. This reduces model complexity, and provides a direct analog to the aggregated catchment concept in the original SWMM. An example of output from a PC-SWMM example run is found in figure 3.1. 11

Figure 3.1: PCSWMM output (CHI, 1999) PCSWMM GIS is an intermediate data management program designed to accept data from a GIS package and transfer it to a SWMM input file. Because it is a more sophisticated method of transferring information from a GIS to a model than the data-interchange method, but it is not an interface as defined by Shamsi (1998), a fourth category was added to the taxonomy, that of the intermediate program. PCSWMM GIS and PCSWMM’98 were developed by CHI in Guelph, Ontario. According to the CHI website, (www.chi.on.ca), PCSWMMGIS does not perform any parameter estimation calculations. It accepts geographic data from an external GIS, within which the parameter estimation calculations and queries are performed. However, it does perform tasks specific to SWMM modeling, such as performing geographic and hydrologic aggregation calculations that are commonly done to simplify a SWMM model. 3.3 XP-SWMM by XP Software (Also Available as Visual Hydro by CAiCE) XP-SWMM32 by XP Software (also included in Visual Hydro, by CAiCE) is a full 32-bit MS Windows application. The program has been enhanced by the addition of a graphics database, and an adaptive dynamic wave solution algorithm that is more stable than the matrix method used in the original SWMM. The program is divided into a stormwater layer, which includes hydrology and water quality, a wastewater layer, which includes storage treatment and water 12

quality routing for BMP analysis, and a hydro-dynamic/hydraulics layer for simulation of open or closed conduits. The user-friendly GUI is based upon a graphical representation of the modeled system using a link-node architecture. An example of input and output processing in Visual Hydro is found in figure 3.2. Because the links and nodes are set up on a coordinate system basis, files can be translated between most CAD and GIS software systems. CAD or GIS files can also be used as a backdrop for the system being modeled. However, since there is no interface with a GIS, data interchange method must be used to transfer parameters (e.g., slope, width, percent imperviousness, etc.) from a GIS to the model. However, the program can import and export files from and to a GIS. 3.4 SWMM-DUET SWMM-DUET is the only fully integrated application of a model into a GIS. It was developed using ArcInfo and the native ArcInfo development language AML (Shamsi 1998). SWMM DUET uses relational databases, both pre- and post-processors, and expert system logic to integrate the SWMM environment and the graphical paradigm of ArcInfo (Shamsi 1998). Future plans include an ArcView version of this product (Shamsi 1998). 3.5 DHI Software 3.5.1 MIKE SWMM MikeSWMM is a proprietary GUI for SWMM from the Danish Hydraulic Institute and Camp, Dresser and McKee, Inc. Mike SWMM can be integrated with a GIS system using Mouse GIS, also available from DHI. Mike SWMM is a classified as an ArcView interface due to its ability to link with the Mouse GIS program, which is described in the follow section. 3.5.2 MOUSE and MOUSE GIS Mouse GIS is a module for MikeSWMM and Mouse users that also allows tight integration between the GIS and the model database. Mouse GIS is an ArcView GIS application. Files do not need to be translated and converted from the GIS to the model format. The DHI product for stormwater modeling, Mouse, uses the Mike GUI within the MS Windows environment. Mouse is a dynamic 32-bit model running in MS Windows that is capable of modeling any type or combination of open or closed conduits and pressurized or gravity flows. An example of the result of a simple query that illustrates the operating environment of Mouse GIS can be seen in figures 3.4 and 3.5. Each object within Mouse GIS has database attributes that can be queried. Mouse GIS is an interface between ArcView and the hydraulic pipe simulator, MOUSE. Mouse is a sophisticated proprietary hydraulic model that is commonly compared to SWMM.

13

Figure 3.2: Visual Hydro (CAiCHE, 1998) 14

Figure 3.3: Mouse GIS user action (www.dhi.dk/mouse/)

Figure 3.4: System response to user action, Mouse GIS (www.dhi.dk/mouse/). 15

3.6 Wallingford Software-HydroWorks and InfoWorks HydroWorks and InfoWorks are companion products produced by Wallingford, Inc. of the UK. Wallingford has taken a different approach to managing geospatial data. InfoWorks is designed to import relational and geospatial data from third party software (e.g. Access and ArcView). Once transferred to InfoWorks, the data is then used to create and run a HydroWorks model. Hydroworks is an urban stormwater modeling system with a user friendly GUI. HydroWorks uses a fully dynamic solution technique that solves backwater and unsteady open or closed conduit situations. InfoWorks performs GIS-type operations, and is designed to operate with HydroWorks, the hydrologic and hydraulic simulator produced by Wallingford, Inc. While the relationship between InfoWorks and HydroWorks may be defined as an interface or even fully integrated, InfoWorks is not a GIS interface. An example of InfoWorks is shown in figure 3.5. Data from a general use GIS product like ArcView would need to be imported into InfoWorks, much like the PCSWMM GIS program from CHI.

Figure 3.5: InfoWorks from Wallingford Software (HR-Wallingford, 1999)

16

3.7 Summary A summary of model and GIS features is presented in table 3.2. As described above, and summarized in table 3.2, the problem of transferring geographic and hydrographic data between a GIS and a simulation model has been handled several different ways by various software developers. It may appear self evident that a tight integration between the hydraulic model and the GIS is desirable. However, the question should be raised; how integrated should these two types of software be? For example, should a GIS include a hydraulic model as part of a toolbox within the GIS? This may, or may not, be desirable. Therefore it should not be assumed that because SWMM DUET has integrated SWMM within ArcInfo that it is the best modeling tool. For example, the expert GUI of XP SWMM may be more useful for a given application, despite the fact that it does not interface directly with a GIS, nor does it have an intermediate data management program. What is common among the recent software developments is a transferability of fundamental database information. This theme is formerly known as a Decision Support System (DSS). Under a DSS framework, neither the GIS nor the model are “central” to the process. Both GIS and model serve satellite functions to a central master database. A more fundamental look at this question is given in chapter 5. The question “which model works best with GIS?” is impossible to answer. Depending on the problem at hand, several products are designed to work with an existing GIS. The answer largely depends on the state of information available. If an existing ArcInfo database is in place, SWMMDUET would work well. Other products have used an information management approach over GIS integration. This may be best suited for applications with disparate data sources. Differences amongst hydraulic models may be more important. The DHI suite of models may be appealing for certain applications. The organization of the HydroInfo/ HydroWorks or PCSWMM’98/PCSWMM GIS software may be best suited for other applications. Each has unique and valuable features, and no recommendation is made in this report for a specific software package. The future evolution of both GIS and urban stormwater modeling, and their possible convergence, appears to be centering upon object intelligence and smaller, programmable component tools. For example, ESRI’s stated goal of its next generation of programs (possibly ArcView 4.0) is to rewrite and enhance its programs to use standard MS Windows routines that can be called via dynamic link libraries (DLLs). An early example is the product called MapObjects, which allows a programmer to insert a GIS-like application within a Visual Basic or Visual C++ program, and make queries and ArcView-like functions upon GIS databases without the ArcView program itself. Existing tools like Evolver, for nonlinear optimization, and @Risk for Monte Carlo simulation are also available as DLLs (Palisade Corporation, 1998). Urban stormwater modeling tools appear to be evolving into using similar tools as they take advantage of existing libraries such as spreadsheet and graphic add-ons, (e.g., Visual Hydro, PCSWMM), and are rewritten in object-oriented programs such as Visual C++, Visual Basic, or Java. The future convergence of GIS and urban stormwater modeling will probably utilize these common sets of tools to take advantage of the easier interoperability. Such tools make integration of these disparate components possible into an integrated Decision Support System, the subject of the next chapter.

17

Table 3.2: Characteristics of urban storm stormwater models Software SWMM Products: EPA SWMM Windows SWMM PCSWMM’98/ PCSWMM GIS Visual Hydro/XP-SWMM SWMMDUET MIKE SWMM/ DHI Products MOUSE, Mike-11 MOUSE GIS HydroWorks/ InfoWorks

Data Interchange

Intermediate Program

GIS/Model Interface

GIS/Model Integration

X X X X X X X

X

Advantages/Disadvantages DOS based Based on SWMM circa 1994 PCSWMM GIS is a data management program Imports CAD, GIS files ArcInfo based ArcView based (via MOUSE GIS) ArcView

InfoWorks is a data management program for geographic and relational databases.

18

4.0 Future Urban Stormwater Modeling in a DSS Environment Much of the data used in distributed (and lumped-distributed) hydrologic modeling requires some level of spatially referenced information. Conversely, purely lumped hydrologic models by definition do not require data to be spatially referenced. This report is focused on lumpeddistributed models and the type of information required to use them. Lumped-distributed models are typically defined by sub-catchments within a stormwater basin. The hydrologic parameters are lumped within each sub-catchment. On the basin scale, however, the discretization among sub-catchments provides spatial distribution. Some of the data used in these distributed models may be more efficiently stored in forms other than GIS spatial database structures (Reitsma et al. 1996). For example, relational data models may be more efficient in storing certain attribute information. Time series are another form of data commonly used in hydrologic modeling. These data are frequently stored in a relational form, despite some shortcomings of this structure for time series (Reitsma et al. 1996). Besides model input, decision-makers frequently require analysis of model output, and the analysis may not necessarily be spatially referenced. For these reasons, future model development should not only focus on the role of GIS in modeling, but on how all information is stored and used. Due to the complexity of tools required to fully support a complex hydrologic decision, a system made up of more than a GIS and simulation model is needed. An integrated suite of tools is required to manage information. These tools are referred to as Decision Support Systems (DSS). Although the model is important, much of the focus has shifted to the related needs of relational database management, developing geographical information systems, and a sophisticated user friendly interface, all combined in DSS. Figure 4.1 describes these necessary components of a DSS (Reitsma et al., 1996). The evolution of DSS may be seen as a natural extension of simulation models (e.g. SWMM, MOUSE, HydroWorks), GIS (e.g. ArcView, IDRISI, ArcInfo), relational databases (e.g. Dbase, Oracle, Access) and evaluation tools (e.g. optimization software). Reitsma (1996) define a DSS for water resources applications: “Decision support systems are computer-based systems which integrate state information, dynamic or process information, and plan evaluation tools into a single software implementation.” In this definition, state information refers to data which represents the system’s state at any point in time, process information represents the first principles governing resource behavior, and evaluation tools refer to software used for transforming raw data into information useful for decision making. A simple representation of DSS components is shown in figure 4.1. The GIS and the simulation model are only components of the DSS in figure 4.1. Future model development should focus not only on GIS interfaces and integration with models, but should include integration with a more complete management information system.. The view for future model development should be broader than only GIS integration, because hydrologic decision making requires more than just spatial information. In a DSS, the GIS only handles spatial data. Spatially referenced information is only one form of state data that is relevant to hydrologic and

19

hydraulic modeling. Time series and attribute data are also crucial to the analysis, and may be handled poorly in a GIS database format designed to manage spatially referenced data. A thorough background on DSSs and their application to reservoir decisions can be found in Jamieson and Fedra (1996a), Fedra and Jamieson (1996), and Jamieson and Fedra (1996b). These series of articles describe the conceptual design, planning capability, and example application of the Water Ware DSS, a complex river basin DSS that combines a “GIS, a georeferenced database, groundwater flow, surface water flow, hydrologic processes, demand forecasting, and water-resources planning” (Jamieson and Fedra 1996a). Reservoir operation and management was one of the first areas of civil engineering in which DSSs were applied. Because of the complicated decision criteria governing urban stormwater management, Davis et al. (1991) studied a prototype DSS developed to analyze the impact of different catchment policies. Driscoll (1993) developed a DSS to assist highway engineers in determining which construction sites would contribute to a receiving water quality problem. Azzout et al. (1995) discuss a DSS under development that would assist in determining the feasibility of alternative techniques in urban stormwater management.

DSS Evaluation Tools -Multi Criteria Evaluation -Visualization -Status Checking

State Information -Databases -Geographic Information

Process Information -(Simulation) Models

Figure 4.1: DSS structure and components (Reitsma et al. 1996) 20

The theme of the following sections is that the parts of a DSS are separate but complementary. They should be able to transfer information to needed process models and evaluation tools without complications. There is no need to house everything under one umbrella, i.e. to perform all modeling tasks in an integrated GIS/hydraulic model. 4.1 State Information In one form of DSS, state information drives the system. This is a “data-centric” view, and it differs from the more traditional model-based analysis commonly used in urban water resources modeling. This fundamental change in perspective may be more important to the future of stormwater modeling than efficient program interfacing. The modeler will need to have tools that handle spatial and temporal data for purposes of modeling, rather than spending resources manually transforming data into the format needed for the model. While this is the idea behind much of the discussion in section 4, a fully integrated GIS/model like SWMMDUET may not be the best modeling tool for the future. It may be that an intermediate database manager (e.g. HydroInfo, PCSWMM GIS, etc) may be closer to a DSS than full GIS integration. State information is stored in relational databases or spatial databases in a modern DSS. Instead of integrating all data forms into one database model, the relational and the spatial information are kept separate, and are linked together to form a geo-relational database structure. 4.1.1 GIS The focus of this report has been on spatial data for modeling purposes. GIS is a critical part of the DSS for systems that are spatially distributed. Since some spatial discretization is needed to model urban hydrologic systems, much effort has been placed on smoothly transferring spatial data to the model and vice versa. Under the DSS data-centered framework, the GIS is one part of the central database of state information. Due to the popularity of GIS software, there has been some interest in housing the entire DSS within the GIS framework. For example, Walsh (1993) investigate spatial DSS, a GIS driven DSS. Reitsma et al. (1996) describe some of the problems associated with a GIS-based DSS: “Recent developments in modeling in GIS (NCGIA 1991; 1993) suggest that GIS can be extended even further into other domains of modeling, e.g., water resources. This type of architecture does offer certain advantages in that it makes use of sophisticated software for management and evaluation of spatial data. A distinct problem, however, is that although rapid improvements are being made in the integration of GIS and modeling (NCGIA 1991; 1993), the full integration of all three components of DSS in GIS is, to say the least, problematic.” To facilitate a non-GIS-based DSS framework, i.e. GIS as a component but not central to the DSS, there are several considerations for GIS. First, the spatial database in the GIS must communicate with other DSS components. This means that much of the interfacing/integration of models and GIS discussed by Shamsi (1998) and reviewed in Chapter 3 must be extended to include other DSS components. Second, spatial tools should be available for modification by the modeler. The GUI should include a dynamic toolbox. For example, if the GIS performs an

21

aggregation calculation in one way, the modeler may wish to modify the algorithm without having to re-write a lot of computer code. The spatial analysis of topographic and hydrographic data may be efficiently carried out in a GIS. GIS software, e.g. ArcView, contain tools that take basic geographic input parameters, e.g. a DEM, and create stormwater boundaries, do slope analysis, etc. Land use and soil coverages are commonly used to estimate hydrologic parameters. Shamsi (1998) discusses several ways that SWMM input parameters may be estimated using GIS. Subarea characteristics such as area, width of overland flow, percent imperviousness and slope may be estimated for the RUNOFF block of SWMM. Parameters used for water quality simulation with the TRANSPORT block of SWMM such as curb length may be estimated from road characteristics in a GIS. Similarly, land use data may be used from a GIS to create SWMM TRANSPORT input files for water quality simulation. Hellweger and Maidment (1999) discuss the details of the spatial analysis required to create an input file for the HEC-HMS model. While not specifically an urban model, it may be useful to review the procedures used. A method to define sub-basin boundaries and stream network connectivity was developed using GIS data layers derived from digital terrain data. Intersecting the sub-basin and stream network layers results in a node-arc representation of the watershed. This information is used to develop an input file for the HEC-HMS model. In this example, an underlying assumption was that streams flowed perpendicular to topographic contour lines. While many of the tools and methods described by Hellweger and Maidment (1999) are useful for modeling natural hydrologic systems, the effect of managed systems in urban areas significantly complicates the analysis. For example, gravity sewers and engineered open channels may have slopes that are independent from the ground surface slope, possibly crossing natural stormwater boundaries and otherwise defying a general physics-based analysis that is used when describing natural systems. Managed or altered hydrologic systems may also be operated based on logic other than the processes that drive a natural system. For example, flow may be diverted from a stream only during dry weather for irrigation purposes, thereby exaggerating the apparent peaking ratio of a stream gauging station. The problems associated with a “pure” GIS analysis of an urban, managed system highlight the advantages of integrating GIS, simulation tools, and relational databases into a DSS. The DSS framework addresses many of the problems associated with using a GIS for urban analysis because of the ability to access and manage related, auxiliary information. 4.1.2 Time series The analysis of time series data is equally important to modeling as the analysis of spatial data. Temporal data includes flow and rain time series, water quality data, etc., as well as dynamic model output. The DSS could include a time series toolbox. Statistical tests and statistical models could reside in this portion of the DSS, for comparison with process models and for analyzing model output. An example of some of this type of pre-processing is that which is currently done in outside statistical packages, or even using Microsoft Excel. Continuous simulation modeling usually will require large amounts of time series data for input

22

purposes. Urban stormwater models that have the capability of continuous simulation usually are capable of reading several different formats of rainfall data. For example, SWMM reads the following formats (Gregory and James, 1996): 1. 2. 3. 4.

National Weather Service Hourly Rainfall Data (in two formats). Pre-1980 National Weather Service Hourly Rainfall Data User Defined Hourly Rainfall Data Canadian Atmospheric Environment Service Hourly Rainfall Data

In SWMM, the standard modules RUNOFF, TRANSPORT, EXTRAN, and STORAGE can import the above formats of time series data. In addition, the modules RAIN, TEMP, STATS, and COMBINE can be used to preprocess time series data. HSP-F, the Hydrologic Simulation Program FORTRAN, includes several time series facilities (Gregory and James, 1996). Several single purpose time series data management programs are available. The HEC-DSS, or the Hydrologic Engineering Center Data Storage System (not Decision Support System), was developed to link time series data with the various HEC watershed management programs. ANNIE, developed by the US Geological Survey, uses watershed data management (WDM) files, and can import WATSTORE files (Gregory and James, 1996). Both ANNIE and HECDSS are non-proprietary FORTRAN models. Due to the multitude of file formats it is difficult to import and export datasets between different modeling environments. For this reason, the CASCADE2 time series management program was developed (Wang and James, 1997). This program, written in Visual Basic, runs under MS Windows and bridges the gap between SWMM and HEC-DSS formats. To be used within a relational database, the time value must be stored, which creates a redundancy of information. This is because a time series is defined by the start time, the time interval, and either the length of the interval or the end time (Reitsma et al. 1996). Another disadvantage of the relational approach is that the DSS must store the criteria for searching the time series (Reitsma et al. 1996). The importance of this redundancy becomes more evident in the case of real time control, which utilizes signal processing and control theory. Lavallee et al. (1996) describe a real time control system developed for the Quebec urban area to manage a stormwater system to minimize CSOs. The unique data needs and system architecture of the RTC system support many of the concepts of DSS due to the demand for timely decisions and vast amounts of data available.. 4.1.3 Relational database An example of a relational database query and its results is presented in this subsection. This example is presented within the context of a relational database contained within a GIS. The same queries can be made in a non-graphic relational database. The linked tabular structure of a relational database allows for extremely complex and powerful queries to be constructed, thus relevant information is made available to the user. The City of Aurora, Colorado has developed a very good base system for GIS. A subcatchment was chosen from the Shop Creek watershed of Aurora, Colorado, a pilot area for GIS development for the City of Aurora. The available themes from this area are as follows: 1. Water lines 23

2. Digital elevation models 3. Rain gages 4. Stream gages 5. Parcels 6. Sewer lines 7. Sewer manholes 8. Digital orthophotos 9. Streets centerlines 10. Sewer tap locations 11. Water meter locations 12. Impervious areas (created by tracing the digital orthophotos) Many tables are associated with each of these themes. An important feature of ArcView is the use of the relational database structure. Tables are linked to graphical features, or themes (analogous to layers in AutoCAD) through the use of spatial geocoding. The user links or joins the tables by choosing a common column, or field between them. The three main types of relationships among tables are: 1. One to One 2. One to Many or Many to One 3. Many to Many All of the records in the one to one table could be placed in the same table. However, good database practice suggests organizing the tables around their functions, instead of the other way around. For example, many attributes are associated with your name, but only your address and phone number are listed in a telephone directory. The first two of these types of relationships is shown in figure 4.2. The two tables nearest the bottom, “Attributes of Theme1.shp” and “Attributes of Parcel” are joined by a one to one relationship, with the fields “Parcel-ID” being the common column. This is again the relationship between “Attributes of Parcel” and “Attributes of Address”, using the fields “Parcel-Id” and “Address-Id” as the common columns (it is not necessary that they have the same name). Lastly, a one to many relationship is shown by the indexing of “Attributes of Address” and “all_9295.dbf” with the fields “Gistag” and “Gisno”. The function of this linking is essentially the following. The theme1.shp table contains the parcels that are located within the small subcatchment. The Attributes of Parcel table contains data on all parcels. Attributes of Address contain address information, including the GIS tag number needed within the Water Use database. This database lists monthly water use data within entire Shop Creek basin, so many records are associated with each parcel. The query shown in figure 4.2 illustrates the power of this tool. The query asks for all linked records in which the water use in a month is over 10,000 gallons. The results of the query are highlighted within the tables. These queries can be moved to the top of their respective tables for further visual analysis. Alternatively, by clicking on the view with the current theme set to Theme1.shp, the visual results of the query can be seen by highlighting parcels that used at least 10,000 gallons a month as shown in figure 4.3.

24

Figure 4.2: Relational database query example in ArcView using water use data

25

Figure 4.3: Spatial results for example query from figure 4.2.

26

The results of the query can also be output to an Excel spreadsheet by using ArcView’s Avenue script language and Microsoft’s Dynamic Data Exchange (DDE). This capability is incorporated as a toolbar shown as an “X” in figure 4.3. The results of this query, output to MS Excel, can be found in figure 4.4.

Figure 4.4: Query results output to Excel using Avenue script tool and Microsoft DDE An example of a relational database within a DSS can be found in Reitsma et al. (1996). In a review of the TERRA DSS system, the authors explain that the data were divided into seven main groupings: 1. 2. 3. 4. 5. 6. 7.

Time Series Data Historical Data Physical Attribute Data Operational Constraints Model Data Security Data Meta Data

Meta data, the last group, is data about data; and allows the Data Management Interface (DMI), a program component of the DSS, to refer only to the meta data, which keeps track of the data structure and where and how the data is stored. This allows the DSS program to be relatively

27

free of data constraints (Reitsma et al., 1996). Although the relational database model has some shortcomings, particularly for time series, it remains the database structure of choice for DSS, as it is the prevailing database model at present. 4.2 Process Information-Simulation Tools In the DSS framework, the process information is contained in simulation models. Process models simulate transitions of the state of the system, as described by the geo-relational database. The simulation model must therefore communicate in some fashion with the rest of the system. For stormwater management models, this may occur in much the same way as described in chapter 3. Data must be transferred to the model from spatial and relational databases. This may occur in a variety of ways, from the rudimentary (but effective) data interchange methods to full-fledged integration in a DSS, running along with the other tools that make up a DSS. The difference from the methods described in chapter 3 is that the communication is not only with the GIS, but also with all elements of the DSS. 4.3 Evaluation Tools Evaluation tools assist the decision-maker by presenting the output from the process and state information in a manner consistent with resource or policy appraisal (Reitsma et al. 1996). Evaluation tools may be of many forms. While much of the above discussion is framed around the excellent review of DSS by Reitsma et al. (1996), the discussion of optimization deviates somewhat from their discourse. Reitsma et al. (1996) do not consider optimization tools to be strictly an evaluation tool, nor do they feel that optimization has been accepted by the user community. While perhaps true for classical optimization techniques, the development of new Intelligent Search Techniques (IST) is proving to be useful for many realistic problems that are unsuitable for traditional methods. 4.4 Overall DSS for Water Management An overall DSS for water management of hydropower and river operations is shown in figure 4.5. This DSS combines the concepts of a centralized database, including hydrologic as well as spatial information, and utilizes two different models that access that data; the Modular Modeling System (MMS) which is a watershed and general environmental model, and RiverWare, which models rivers and reservoirs. Evaluation tools are included within each of the model components. The DSS includes a GIS as a tool for the user to query the common spatial database. This DSS was developed by the Center for Advanced Decision Support in Water and Environmental Systems (CADSWES) at the University of Colorado at Boulder, with support from the Tennessee Valley Authority and the US Bureau of Reclamation. This DSS focuses on large watersheds with complex reservoir and hydropower operations.

28

Figure 4.5: CU-CADSWES DSS (Fulp et al., 1994) A DSS framework for the urban stormwater field is presented in figure 4.6. This DSS is an amalgamation of the different components of the Mike series of software produced by the Danish Hydraulic Institute (DHI), emphasizing their interoperability and common database, Mike Info. The database (relational and spatial) is the common link between separate functions and applications of the DSS. The peripheral models include Mike-11 for urban drainage, Mike SHE for distributed watershed modeling, WUS for river basin planning, and NAM for statistical analysis of streamflow/unguaged catchments.

29

Figure 4.6: Danish Hydraulic Institute DSS, based on integrated water resources modeling (DHI, 1998)

30

5.0 Application of GIS and DSS to Micro Storm Analysis This chapter focuses upon the application of GIS, database management, and DSS to the urban stormwater management problem. A textbook case study from Tchobanoglous (1981) is used to develop a GIS and an accompanying relational database. The database is used with hydrologic and hydraulic models, and a cost analysis module. The combination of these components represents a systematic urban stormwater design tool. The tool is then interfaced with an optimization software package to develop optimal designs of the proposed network. The costs of these designs can then be compared with a decentralized approach to controlling runoff. This was done by using the GIS in conjunction with the NRCS analysis, which computes the initial abstraction storage volume that is lost as a result of development. Using unit costs developed in Heaney et al. (1999a), the optimal suite of controls can be selected using linear programming (LP). A diagram of the process used in the chapter is found in figure 5.1. The reader may notice similarities between some of the components of a DSS and figure 5.1. In particular, the problem consists of a database, simulation tools, and evaluation tools, similar in concept to that of a DSS presented by Reitsma et al. (1996). The database includes GIS and its inherent spatial database, but also a cost database, and a hydrologic database. The simulation tools consist of the NRCS curve number method for computation of initial abstraction, the hydrologic model spreadsheet template, the hydraulic model spreadsheet template, and the costing module. The evaluation tool consists of a genetic algorithm to optimize the stormwater network, and a linear programming model to evaluate proposed controls based upon unit costs developed in Heaney et al. (1999a). Although not integrated into a single software program, the process shown here closely parallels that of a DSS. The utility of GIS (to the urban stormwater field) is enhanced by its close integration with the database, models, and analysis tools used in the problem. Because of the large investment in time and resources necessary to construct an urban GIS, there is a natural tendency for the GIS system to move to center stage. However, the value of the GIS is when it is fully integrated within a DSS which is then used to address complex processes that cannot be easily solved by other means. Key considerations are the concepts of accuracy and scale as they apply to GIS data. Since the datasets presented here vary substantially in terms of their level of detail and scale, a discussion of spatial scale becomes necessary.

31

DSS

Evaluation Tools Optimization Linear Programming (LP) Genetic Algorithms (GA)

Simulation Tools Database Relational (nongraphic) addresses billing unit costs time series input data GIS/Spatial Database Themes Topography Soils Land use Streets Right of way Pipe network Parcels

NRCS CN Hydrologic Method Rational Method Hydraulic Design Template Cost Template

Figure 5.1: Proposed DSS for microstorm analysis 5.1 Spatial Scale and GIS-Stormwater Modeling A recent software development, BASINS 2.0, developed by TetraTech for the US Environmental Protection Agency, has created interest in the development of model-graphical user interfaceGIS linkages within the water community. BASINS 2.0 runs within ArcView 3.0 and includes a national dataset on the attributes listed in Table 5.1 (Battin, et al. 1998).

32

Table 5.1: Available BASINS data attributes (Battin et al. 1998) Spatially Distributed Data Land use/land cover (GIRAS) Urbanized areas Populated place location Reach File, version 1 (RF1) Reach File, version 3 (RF3) Soils (STATSGO) Elevation (DEM) Major roads Environmental Monitoring Data Water quality monitoring station summaries Water quality observation data Bacteria monitoring station summaries Weather Station Sites (477) Clean Water Needs Survey Point Source Data Permit Compliance System Industrial Facilities Discharge (IFD) sites Toxic Release Inventory (TRI) sites

USGS Hydrologic unit boundaries Drinking water supplies Dam sites EPA region boundaries State boundaries County boundaries Federal and Indian Lands Ecoregions USGS gaging stations Fish and wildlife advisories National Sediment Inventory (NSI) Shellfish Contamination Inventory

Resource Conservation & Recovery Act (RCRA) sites Mineral availability system/mineral industry location Superfund national priority list sites

BASINS 2.0 includes tools for automatic watershed delineation and handling of digital elevation models (DEM). Its main data handling routines include: Target, which is a regional, or state level broad-based watershed water quality or point source assessment tool; Assess, which operates a smaller scale of one or a few watersheds and enclosed discharge points or water quality stations; and Data Mining, which dynamically links water discharge stations and geographic location information. Modeling tools include a nonpoint source model (later to be enhanced by the addition of SWAT, the MS Windows based nonpoint source model developed by the USDA), HSPF, Qual-2E, and Toxiroute. Model post processors include graphs (Battin, Kinerson, and Lahlou 1998). EPA SWMM may be linked with BASINS in the future.

33

The accepted accuracy levels of mapping work are listed in Table 5.2 (Shamsi et al. 1995). Most of the BASINS work and modeling have been on a watershed or regional level scale. An example is shown in figure 5.2. The size of this file relative to the area it represents reflects a scale of about 1:2000. Table 5.2: Minimum horizontal accuracy and example features for various map scales in urban areas (Shamsi et al. 1995) Map Scale

1”=50’ 1’=100’ 1”=200’ 1”=2000’

Minimum Horizontal Accuracy, per National Map Accuracy Standards ± 1.25’ ± 2.50’ ± 5.00’ ± 40’

Examples of Smallest Features Depicted Manholes, catch basins Utility poles, fence lines Buildings, edge of pavement Transportation, developed areas, watersheds

Figure 5.2: BASINS dataset for Boulder, Colorado Automatic watershed delineation of undeveloped areas may be appropriate at this scale. However, urban systems have extremely altered topography. The topography in these types of catchments can be represented by a dense DEM; however, development of watersheds based

34

upon triangular irregular networks (TINs) from this information is not presently reliable. This is not to say that the database information presented from a watershed level scale has no value. Actually, having the information presented in figure 5.2 can provide the modeler with possible alternative sources of data, possibly structures that may not have been considered, etc. However, a key disadvantage of using GIS information from different scales of accuracy is that a vector GIS cannot show any uncertainty. An assumption of the GIS model is that the points are known to 100% accuracy. This leaves it up to the reader to verify locations and discrepancies, particularly when the scales, and the resultant accuracy, differ widely. In addition, the memory requirements for regional level stormwater-GIS modeling are staggering. For example, the City of Boulder has an ongoing GIS project, a broad view of which is shown in figure 5.3 (Brown and Caldwell and Camp, Dresser, and McKee, 1997).

Figure 5.3: ArcView coverage of Boulder, Colorado (Brown and Caldwell and Camp, Dresser, and McKee, 1997) Minor roads are outlined in light green, major roads are outlined in thick maroon; creeks are shown in light blue, lakes in shaded blue, and sub-basins boundaries in black. Not shown for better clarity, but available, are parcels, zoning, topography, watershed boundaries, and several other miscellaneous themes. Also not shown is the database describing each graphic entity (for example, the parcel database). Even at this finer resolution, urban stormwater modeling is at too aggregate a scale to evaluate sets of alternatives that include micro-topographical changes to implement BMPs.

35

In order to evaluate the effects of source and neighborhood-level BMPs, the coverage as depicted in figure 5.4 is needed. This area is a block in the University Hill neighborhood of Boulder. The parcel theme is shown in red, the street centerline is shown in green, and the streams are shown in blue. Topography is not shown, but exists in this database at the 40 foot contour interval, reflecting a scale of about 1:200.

Figure 5.4: City of Boulder ArcView GIS coverage for University Hill neighborhood, Boulder, Colorado.

36

Moving towards a finer dataset, another parallel project at the City of Boulder, in the Public Works/Public Utilities group, is an Automated Mapping/Facilities Management (AM/FM) project in which the city’s infrastructure is being mapped by street surveys and aerial photography. The end product at the present time is a tiled set of AutoCAD maps representing portions of the city. The representation of this project for the same block in the University Hill neighborhood is shown in figure 5.5. The scale of this information is approximately 1:100. The green layer signifies building rooflines, yellow is the street centerline and parking areas/driveways, red is sidewalks, and black is the curblines. This file has been edited extensively to eliminate extraneous lines and close polygons. Since the end product of the project was a set of AutoCAD maps, manual and automatic processes on the digital photography result in multiple lines whose ends may not match and polygons that do not close. Although acceptable for graphic presentation, this information is of limited value for extracting data for stormwater evaluations. Extensive cleanup is necessary for this information prior to inputting it into a GIS. Topography for this information is available for an additional cost at a 2-foot contour interval. At the present time, conversion of this data to ArcInfo and ArcView coverages is underway. 5.2 Description of Happy Acres Case Study GIS A textbook study area, nicknamed “Happy Acres”, was selected from Tchobanoglous (1981). A GIS coverage for this case study was developed. The study area was first digitized in AutoCAD, then edited for geometric consistency, i.e., parallel lines were kept parallel, polygons were joined from separated lines, to make the transition to GIS easier. The mix of land uses for the area is laid out in table 5.3. The reconstructed AutoCAD drawing of the area is shown in figure 5.6. The topography of the study area and the layout of the storm sewer system are shown in figure 5.7 (Tchobanoglous 1981). Land use is shown in figure 5.8. Soils data is shown in figure 5.9. The entire study area is divided into 54 sub-areas that range in size from 0.8 to 5.4 acres in size. A description of the attribute information in figure 5.6 is found in table 5.4. Table 5.3: Mix of land uses in Happy Acres Land Use

Acres

Residential, low density Residential, medium density Apartments School Commercial Total

37

20.8 51.7 10.0 5.7 18.4 106.6

Dwelling units/acre 2-3 6-8 10 N/A N/A

Figure 5.5: AutoCAD file for University Hills neighborhood, Boulder, Colorado.

38

Figure 5.6: AutoCAD coverage for study area (adapted from Tchobanoglous, 1981)

39

N

Sewer2.sh p Manhol e2 .sh p Con tour s o f Tch ob an_po ints 4_point.shp Tchoban_parce ls2 _regio n.shp Tchoban_roads 2_r egi on .s hp

100

0

100

200 Meters

Figure 5.7: Study area topography (adapted from Tchobanoglous, 1981)

40

Tchoban_roads2_region.shp Nwgrd3 Apartment Commercial LD Residential MD Residential School No Data

N W 300

0

E

300 600 900 1200 1500 1800 2100 2400 Feet

S

Figure 5.8: Study area land use (adapted from Tchobanoglous, 1981)

41

Tchoban_roads2_region.shp Tchoban_drainage2_region.shp Soilgrid Clay Rock Silt No Data

N W 300

0

E

300 600 900 1200 1500 1800 2100 2400 Feet

S

Figure 5.9: Study area soils (adapted from Tchobanoglous, 1981)

42

Table 5.4: AutoCAD layers for study area Layer/Object Category Streets Manholes Sewer lines Land use boundary Hydrologic boundary Parcel Rooflines Driveways Soils

Color Not shown (for clarity) Blue Red Aqua Blue Green Magenta Orange Not shown (for clarity)

The AutoCAD layers shown in table 5.4 become the following ArcView themes: 1. 2. 3. 4. 5. 6. 7. 8. 9.

Streets Manholes Sewer lines Land use boundary Hydrologic boundary Parcel Rooflines Driveways Soils

A relational database is associated with each graphic object, grouped according to type. Attributes associated with parcels are address and land area; and with streets are right of way width, length, land area, and street name. Soils and land use exist in separate tables, and this information was combined with the parcel and street databases by performing an intersection query on the two themes. The results of the query can also be output to an Excel spreadsheet by using ArcView’s Avenue script language and Microsoft’s Dynamic Data Exchange (DDE). This procedure was used to extract the relevant attribute information for parcels and streets. The rights of way identified in figures 5.6 through 5.9 were assigned widths based upon the following criteria. Minor streets within the development have a 50 foot right of way, a minor arterial is given a 60 foot right of way, and a major arterial a 70 foot right of way. The profile of each right of way is given in table 5.4. The reader is referred to Heaney et al. (1999a) for further details on the database. Table 5.5: Right of way characteristics R/W ft

Length, ft 50 28,680 60 1,124 70 2,741

Curb ft

Parking Landscaping Sidewalk Traffic ft strip, ft ft Lanes, ft 4 8 10 8 20 4 16 10 8 22 4 16 18 8 24

43

Note: Some of the parameters are summed from both sides of the street. Lot characteristics for the two single lot residential land use classifications are presented in table 5.6. Lots were aggregated in this manner for the optimization; however the GIS contains the full heterogeneity of each parcel. Table 5.6: Lot characteristics for residential parcels Land Use

No. of Parcels

MD Residential (6-8 DU/AC) LD Residential (2-5 DU/AC)

Roof Patio Driveway Landscap- Total Area SF SF ing Area SF SF SF 255 1,600 200 600 3,600 6,000 51 2,000 400 800 9,800 13,000

For the apartments, commercial, and school land uses, an aggregate analysis was used because these land uses exhibited multi-parcel characteristics, such as for parking. A summary of these characteristics is found in table 5.7 Table 5.7: Aggregate characteristics for commercial, apartments, and schools Land Use

Apartments Commercial School

No. of parcels 2 6 3

Parking LandscapRoof Parcel ing Area Area Area SF SF SF SF 2 162,680 46,927 75,083 40,670 1 481,070 152,839 304,678 23,553 1 149,407 69,080 51,807 28,521

Stories

5.3 Simulation Tools for Hydraulic Design The storm sewer network for the Happy Acres subdivision is diagrammed in figure 5.10. A spreadsheet template has been developed to simulate and optimize storm sewer design for the Happy Acres neighborhood-see tables 5.8 to 5.10. The value of better data obtained using GIS can be estimated by evaluating the designs with and without this better information. The following columns in table 5.7 represent data that can be obtained partially or totally with a GIS system for this example. Column Description 5 Sewer length 6 Stormwater area 7 Dwelling units per acre The output from table 5.8 is the design peak discharge leaving each subcatchment. This information is input to the sewer design table 5.9 that finds feasible combinations of pipe diameters and slopes. The constraints on the design are: Minimum depth of cover for the sewer, and

44

Minimum velocity in the pipe. The decision variables are pipe diameter (column 8) and slope (column 6). Trial and error procedures are used to find a feasible solution to the design problem. In more sophisticated analysis, the costs of the alternative systems are evaluated as shown in table 5.10. The background for development of the cost relationships found in this table can be found in Heaney et al. (1999a), and is based upon data obtained from R.S. Means (1996a). Additional GIS data are helpful for the cost analysis. Specifically, soil conditions (column 8) affect the side slopes of the sewer excavations, and the bedding costs. 18

14E

17

17A

17B

16

16A

16B

15

15A

15B

12A

12B

9A

9B

7A

7B

14F 14

14D

14C

14B

14A

13I

13H 13

13G

13F

13E

13D

13C

13B

13A

12

11G

11F

11E

11D

11C

11B

11A

11

10E

10D

10C

10B

10A

10

9

8D

8C

8B

8A

8

7

1

2

3

4

Figure 5.10: Study area sewer network (adapted from Tchobanoglous, 1981)

45

5

6

16C

Table 5.8: Sewer network design hydrology (Heaney et al. 1999)

46

Table 5.9: Sewer network design hydraulics (Heaney et al. 1999)

47

Table 5.10: Sewer network design cost (Heaney et al. 1999)

48

Using a new intelligent search technique called genetic algorithms (GAs), the optimal design was found by having Evolver (Palisade Corp., 1998), a commercially available GA, evaluate different combinations of pipe diameters and slopes until the least cost design is found. 5.4 Simulation Tools for Hydrologic Analysis Heaney, Wright, and Sample (1999) describe a method for using the NRCS curve number (CN) approach for evaluating micro storms. The fundamental principle is that development should not reduce the initial soil moisture storage that existed prior to development. This initial soil moisture storage is equivalent to the initial abstraction as calculated using the Natural Resources Conservation Service (NRCS) curve number (CN) method. The initial abstraction is a good measure of the ability of the soil system to filter the stormwater. The initial abstraction, as a function of CN, is shown in table 5.11. Inspection of table 5.11 reveals the importance of CN. A low CN of 30 corresponds to an initial abstraction of 4.67 inches. Even at a CN of 80, the initial abstraction is still 0.5 inches. If the original CN is fairly low, then a significant amount of soil moisture storage is lost if this area is rendered impervious by development. Table 5.11: Initial abstraction as a function of curve numbers, CN CN 20 30 40 50 60

Ia, inches 8 4.67 3 2 1.33

CN 70 80 90 100

Ia, inches 0.86 0.5 0.22 0.02

This method uses the concept of modifying the CNs for the developed condition so that the modified CN is the same as the natural CN. The more cost-effective controls tend to focus on utilizing the pervious area for more intensive infiltration. Alternatively, we seek to design hydrologically functional landscapes as described in the next section. 5.4.1 Hydrologically functional landscaping Traditional landscaping relies on covering most, if not all, of the pervious area with grass. The lot is graded so that stormwater drains to the street and/or the rear of the lot as shown in figure 5.11 (Dewberry and Davis 1996). An example of a hydrologically functional landscape is shown in figure 5.12 (Prince Georges County 1997). The general idea is to maximize the infiltration of stormwater by providing depressions, draining runoff from impervious areas to pervious areas, providing more circuitous routes for the stormwater to increase the time of concentration, etc.

49

Figure 5.11: Conventional storm drainage (Dewberry and Davis 1996).

50

Figure 5.12: Illustration of hydrologically functional landscape (Prince Georges County 1997).

51

5.4.2 Determination of runoff volumes using NRCS method Each developed land use is assigned a curve number (CN) based upon work done by the Soil Conservation Service (1986). The initial abstraction, or available storage, is estimated by the following equation: 200 Ia = −2 5.1 CN The final list of 10 permeable and 16 impermeable candidate land uses with their expected effectiveness as measured by their curve number (CN) and the associated initial abstraction in inches, calculated using equation 5.1, are shown in table 5.12. The CNs range from 25 to 98. The initial abstraction associated with a CN of 25 is 6.00 inches of precipitation. Making this land impervious increases the CN to 98 with an associated initial abstraction of only 0.04 inches, a major loss of infiltration capacity. Using unit costs in $/square feet, which are developed in section 5.5 (and detailed in Heaney et al. 1999a) and having determined the appropriate abstraction, it is possible to convert the control option costs to $/gallon, which is done in the last four columns of table 5.12. Several different functional land uses are given in table 5.12. These include two kinds of aspens, fair, and good (referring to the health and density of the stand), two kinds of driveways, permeable and impermeable, three types of grass cover, good, fair, and poor (again referring to health and density), four types of parking, a traditional impervious surface, and three of gradually increasing porosity, two types of patios, permeable and impermeable, two kinds of roofs, with retention and without, two kinds of sidewalks, permeable and impermeable, storage (detention pond), four types of streets, a traditional street profile with curb and gutter, a street with curb and gutter and porous pavement, an impervious street with swales, and a street with porous pavement and swales, two types of swales of progressively greater infiltration capacity (and greater area), and two kinds of wooded areas, fair and poor, again referring to health and density of the trees. These values are unique to the soil type heading the column. The NRCS method aggregates clay and silt together as soil type "B", and rock as soil type "D". Unit costs expressed as $/gallon are useful for comparative purposes, as will be seen later. 5.4.3 Breakdown of calculated volumes per function A functional analysis within each land use and soil classification was performed by adding the total areas for the functions of roof, lawns, driveways, and parking (for non-right of way uses), and streets, curbs, parking, sidewalks, and lawns for right of way areas. Volumes of developed runoff can then be calculated by multiplying the initial abstraction by the appropriate area. Predevelopment runoff can be calculated by using the composite curve number for Happy Acres prior to development of 63.07, determining an initial abstraction for each soil group, and multiplying this again by the area as done for the developed volumes. The result of this analysis is found in table 5.13. This provides a snapshot of the increase in runoff volume for each land use generated by development. Because the NRCS method is unique to soil characteristics, this is further broken down by soil group.

52

Table 5.12: SCS hydrologic classifications, and calculation of unit storage values, 1/99$ Curve Number

28

48

57

63

Unit Unit Costs in $/gallons cost B C D $/sf A B C D 5.14 2.17 1.51 1.17 $2.00 $0.62 $1.48 $2.13 $2.73

Aspen G

25

30

41

48

6.00

4.67

2.88

2.17

$3.00

$0.80 $1.03 $1.67 $2.22

Driveway 1 Driveway 2 Grass F

98 70 49

98 80 69

98 85 79

98 87 84

0.04 0.86 2.08

0.04 0.50 0.90

0.04 0.35 0.53

0.04 0.30 0.38

$0.23 $0.25 $0.81

$9.21 $9.21 $9.21 $9.21 $0.47 $0.80 $1.13 $1.34 $0.63 $1.45 $2.45 $3.42

Grass G

39

61

74

80

3.13

1.28

0.70

0.50

$1.03

$0.53 $1.29 $2.35 $3.30

Grass P

68

79

86

89

0.94

0.53

0.33

0.25

$0.70

$1.19 $2.12 $3.45 $4.55

Parking 1 Parking 2 Parking 3 Parking 4 Patio 1 Patio 2 Roof 1 Roof 2 Sidewalk 1 Sidewalk 2 Storage

98 61 46 36 95 76 95 85 98 70 15

98 75 65 55 95 85 95 85 98 80 20

98 83 77 67 95 89 95 85 98 85 35

98 87 82 72 95 91 95 85 98 87 40

0.04 1.28 2.35 3.56 0.11 0.63 0.11 0.35 0.04 0.86 11.33

0.04 0.67 1.08 1.64 0.11 0.35 0.11 0.35 0.04 0.50 8.00

0.04 0.41 0.60 0.99 0.11 0.25 0.11 0.35 0.04 0.35 3.71

0.04 0.30 0.44 0.78 0.11 0.20 0.11 0.35 0.04 0.30 3.00

$0.23 $0.25 $0.26 $0.28 $0.19 $0.19 $0.00 $1.50 $0.19 $0.19 $5.00

$9.21 $0.31 $0.18 $0.13 $2.89 $0.49 $0.00 $6.82 $7.44 $0.36 $0.71

Street 1 Street 2

98 70

98 80

98 85

98 87

0.04 0.86

0.04 0.50

0.04 0.35

0.04 0.30

$0.25 $0.26

$9.77 $9.77 $9.77 $9.77 $0.49 $0.84 $1.19 $1.41

Street 3 Street 4 Swales 1

76 61 46

85 75 65

89 83 77

91 87 82

0.63 1.28 2.35

0.35 0.67 1.08

0.25 0.41 0.60

0.20 0.30 0.44

$0.27 $0.28 $3.00

Swales 2 Swales 2 Woods:Fair: Woods are grazed but not Woods F burned, and some forest litter Woods:Good: Woods without grazing, and Woods G adequate litter and brush

29 36

50 60

62 73

67 79

4.90 3.56

2.00 1.33

1.23 0.74

0.99 0.53

$6.00 $0.80

$0.68 $1.22 $1.74 $2.17 $0.35 $0.67 $1.09 $1.49 $2.05 $4.47 $8.06 $10.9 6 $1.97 $4.81 $7.85 $9.77 $0.36 $0.96 $1.73 $2.41

25

55

70

77

6.00

1.64

0.86

0.60

$1.40

$0.37 $1.37 $2.62 $3.76

Cover Description No.

Type 1 Permeable 2 Permeable 1 Impervious 2 Impervious 3 Permeable 4 Permeable 5 Permeable 6 4 5 6 7 8 9 10 11 12 13

Impervious Impervious Impervious Impervious Impervious Impervious Impervious Impervious Impervious Impervious Permeable

14 Impervious 15 Impervious 16 Impervious 17 Impervious 18 Permeable 19 Permeable 20 Permeable 21 Permeable

Cover type and hydrologic condition Aspen-mountain brush mixture: Fair:3070% ground cover Aspen-mountain brush mixture: Good: >70% ground cover Driveway Driveway-porous pavement Lawns, pasture, grassland: Fair condition (grass cover 50-75%) Lawns, pasture, grassland: Good condition (grass cover >75%) Lawns, pasture, grassland: Poor condition (grass cover < 50%) Parking Porous parking 1 Porous parking 2 Porous parking 3 Patio Porous patio Roof Roof with detention Sidewalks Sidewalks with porous materials Storage-off-site in infiltration/detention basins Street with curb and gutter Street with curb and gutter and porous pavement Street with swales Street with swales and porous pavement Swales 1

ID Aspen F

A

B

Initial Abstraction in inches C

D

Source: adapted from SCS, 1986

53

A

$9.21 $0.60 $0.39 $0.27 $2.89 $0.88 $0.00 $6.82 $7.44 $0.62 $1.00

$9.21 $0.98 $0.71 $0.46 $2.89 $1.25 $0.00 $6.82 $7.44 $0.88 $2.16

$9.21 $1.34 $0.97 $0.58 $2.89 $1.57 $0.00 $6.82 $7.44 $1.04 $2.67

Table 5.13: Calculation of developed and predevelopment stormwater volumes for Happy Acres Soil Types B

Soil Types D, Total

sf

sf

Volume

Volume

Total Vol.

Roof

46927

0

46927

412

0

412

4580

0

4580

Parking

75083

0

75083

255

0

255

7327

0

7327

0

0

0

0

0

0

0

0

0

Lawns

40670

0

40670

4334

0

4334

3969

0

3969

Roof

95132

57707

152839

834

506

1341

9284

49

9333

Parking

44810

259868

304678

152

884

1036

4373

86

4459

0

0

0

0

0

0

0

0

0

6839

16714

23553

729

696

1425

667

68

735

140800

267200

408000

1235

2344

3579

13741

229

13969

0

0

0

0

0

0

0

0

0

52800

100200

153000

180

341

520

5153

33

5186

34514

2191

36705

9954

0

9954

Lawns MD Residential

Roof Parking Driveway Lawns

LD Residential

cf

cf

353666

538755

892420

37686

22448

60134

Patio

17600

33400

51000

154

293

447

Roof

102000

0

102000

895

0

895

Parking Driveway Lawns School

cf

Undev., Undev. D cf cf

Apartments

Driveway

sf

Undev., B cf

Function

Commercial

Developed, B Developed, D Developed

Volume Tot. Volume

Land Use

Driveway

Area, Total

Volume

0

0

0

0

0

0

0

0

0

0

40800

0

40800

139

0

139

3982

0

3982

47939

0

47939

491233

0

491233

52344

0

52344

Patio

20400

0

20400

179

0

179

Roof

69080

0

69080

606

0

606

6742

0

6742

Parking

51806

0

51806

176

0

176

5056

0

5056

0

0

0

0

0

0

0

0

0

28521

0

28521

3039

0

3039

2783

0

2783

Driveway Lawns

0

Streets 50 ROW

659728

774288

1434016

Street with curb and gutter Parking

105556

123886

229443

359

421

780

10301

41

10342

105556

123886

229443

359

421

780

10301

41

10342

Sidewalks

105556

123886

229443

359

421

780

10301

41

10342

curb

52778

61943

114721

180

211

390

5151

21

5171

Lawns

52778

61943

114721

3952

1966

5918

5151

192

5343

87540

0

87540

Street with curb and gutter Parking

11672

0

11672

40

0

40

1139

0

1139

23344

0

23344

79

0

79

2278

0

2278

Sidewalks

11672

0

11672

40

0

40

1139

0

1139

curb

5836

0

5836

20

0

20

570

0

570

Lawns

5836

0

5836

437

0

437

570

0

570

13195

189531

202726

Street with curb and gutter Parking

1508

21661

23169

5

74

79

147

7

154

3016

43321

46337

10

147

158

294

14

309

Sidewalks

1508

21661

23169

5

74

79

147

7

154

curb

754

10830

11584

3

37

39

74

4

77

Lawns

754

10830

11584

56

344

400

74

34

107

60 ROW

70 ROW

Total

1724282

54

140882

210758

The functions were then compared across land uses by computing the difference between the sum of the function’s pre-development and post-development storage volumes. The result is plotted as a bar chart in figure 5.13. The greatest impact is from streets and roofs, with roughly equal values of storage volume reduction. Patios are insignificant in this analysis. Lawns actually add a great deal of storage, offsetting somewhat the drastic reductions from roofs and streets. Driveways and parking lots result in smaller reductions in volume, however, the local impact may be significant. 140000

120000

100000

Volume, post development, (CF) Volume, predevelopment (CF) Difference

Volume in Cubic Feet

80000

60000

40000

20000

0 Roof

Parking

Driveway

Lawns

Patio

Streets

-20000

-40000 Function

Figure 5.13: Allocation of available storage for initial abstraction and land use. 5.5 Simulation Tools for Cost Analysis If the cost of modifying the CNs can be determined, then cost-effective strategies can be developed for maintaining the undeveloped CN for each parcel or combination of parcels. Since most BMPs are land intensive, a careful evaluation of their costs must include land valuation. The costs used in the analysis were developed in Heaney et al. (1999), for each control and each land use. The procedure for calculation of the land component of controls within one land use, medium density residential, is outlined in table 5.14.

55

Table 5.14: Land valuation for medium density lot, 1/99$ Component

SF

Roof-house Roof-garage Driveway Yard Patio Total

1200 400 600 3600 200 6000

% of $/sf total 20.0% $56.25 6.7% $34.00 10.0% $4.00 60.0% $1.00 3.3% $4.00 100.0%

Construction Total Land $ Unimproved Cost, $ Land, $ $67,500 $8,790 $5,860 $13,600 $2,930 $1,953 $2,400 $4,395 $2,930 $3,600 $26,370 $17,580 $800 $1,465 $977 $87,900 $43,950 $29,300

An estimate of the cost in $/sf is found in column 4 of table 5.14. Next, the construction cost (column 5) is obtained by multiplying column 2 by column 4. Next, the percentage in column 3 is multiplied by the total of column 5 to obtain an estimate of the land cost, in column 6. Column 7, the unimproved land cost, is obtained by multiplying the values in column 6 by 2/3. The value of the 3,600 square feet of land for the yard function is $26,370. Next, opportunity costs must be calculated. This procedure is illustrated in table 5.15. The value of $26,370 is annualized, using an interest rate of 6%, and an infinite term (as in equation 6.2), to obtain $1,582/year. Then, this value is spread over 25 years at 6%, to obtain $20,226. Dividing this value by 3,600 square feet gives $5.62/square feet. This value is used for all grass types as the underlying value of the land is assumed to be constant irrespective of the type of grass. Landscaping costs were developed from RS Means (1996b), and updated to January 1999, and are presented in table 5.15 (for a medium density residential lot). The initial capital investment consists of the cost of soil preparation including sod, topsoil, and soil conditioners, and an irrigation system. For a good lawn, the present value of the initial landscaping investment is $2.22 per square foot. Costs for lesser quality lawns drop to $1.71/sf and $.95/sf for fair and poor quality lawns. For the good lawn system, operation and maintenance costs add an additional $2.45 per square foot bringing the total to $10.29 per square foot. An estimated 10 percent of this total cost is allocated to stormwater management. Similar estimates were made for fair and poor lawns. The resulting total costs per square foot vary from $0.70 to $1.03 per square foot. Better lawns have a lower CN and are thereby preferable from the viewpoint of being able to store more water. Similar estimates were made for the land valuation for low-density residential lots, commercial, apartments, and schools. A similar procedure was followed for these uses, except that the commercial, apartments, and schools are aggregated as one lot. However, they also cost more. The cost for each control was then estimated using these land valuations. The matrix of controls and land uses is presented in table 5.16. A linear programming model is used to find the least costly mix for each land use. See Heaney et al. (1999b) for a more detailed explanation of this method.

56

Table 5.15: Cost analysis of landscaping for medium density lot, 1/99$

Item A. Initial Capital Investment 1. Soil preparation Initial cost of sod Initial cost of topsoil, 6" Spreading topsoil, 6" Soil conditioners Sprinkler system

Input Data

2. Opportunity Cost of Land Land Investment Cost Opportunity cost investment rate Annual cost, $/yr. Interest rate per year Present worth over 25 years Cost in $/ft2 Total of initial capital investment B. Operation & Maintenance Costs, $ Lawn watering Inches per year % of pervious area that is irrigated Cost of water, $/1,000 gallons Present worth factor Present worth, $/ft2 Lawn maintenance Weeks per year $/week Maintenance area, ft2 Present worth, $/ft2 Sprinkler system maintenance Total operation and maintenance costs, $ C. Total Cost, $/ft2 Portion attributable to stormwater Assumed % D. Cost for Stormwater

Good $/ft2

Fair $/ft2

Poor $/ft2

$0.43 $0.50 $0.64 $0.03 $0.62 $2.22

$0.34 $0.40 $0.51 $0.02 $0.44 $1.71

$0.26 $0.30 $0.38 $0.01 $0.00 $0.95

$5.62 $7.84

$5.62 $7.33

$5.62 $6.57

$0.24

$0.15

$0.09

$0.98 $0.25 $1.46 $9.31

$0.50 $0.15 $0.80 $8.13

$0.35 $0.00 $0.44 $7.01

$0.93

$0.81

$0.70

$26,370 6% $1,582 0.06 $20,226

20 80% $1.50 12.78

26 $8.46 2880

10%

57

Table 5.16: Calculation of unit costs for controls, including opportunity costs for land, 1/99$ ID Aspen F Aspen G Driveway 1 Driveway 2 Grass F Grass G Grass P Parking 1 Parking 2 Parking 3 Parking 4 Patio 1 Patio 2 Roof 1 Roof 2 Sidewalk 1 Sidewalk 2 Storage Street 1 Street 2 Street 3 Street 4 Swales 1 Swales 2 Woods F Woods G

LD Res MD Res Commercial School Apartments RW50 RW60 RW70 $/sf $/sf $/sf $/sf $/sf $/sf $/sf $/sf $2.00 $2.00 $2.00 $2.00 $2.00 $2.00 $2.00 $2.00 $3.00 $3.00 $3.00 $3.00 $3.00 $3.00 $3.00 $3.00 $0.23 $0.23 $0.23 $0.23 $0.23 $0.23 $0.23 $0.23 $0.25 $0.25 $0.25 $0.25 $0.25 $0.25 $0.25 $0.25 $0.60 $0.60 $2.12 $2.49 $1.22 $0.60 $0.60 $0.60 $0.69 $0.69 $2.18 $2.56 $1.29 $0.69 $0.69 $0.69 $0.49 $0.49 $2.01 $2.38 $1.11 $0.49 $0.49 $0.49 $0.23 $0.23 $0.23 $0.23 $0.23 $0.23 $0.23 $0.23 $0.25 $0.25 $0.25 $0.25 $0.25 $0.25 $0.25 $0.25 $0.26 $0.26 $0.26 $0.26 $0.26 $0.26 $0.26 $0.26 $0.28 $0.28 $0.28 $0.28 $0.28 $0.28 $0.28 $0.28 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.00 $0.00 $0.00 $0.00 $0.00 $0.00 $0.00 $0.00 $1.50 $1.50 $1.50 $1.50 $1.50 $1.50 $1.50 $1.50 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $5.00 $5.00 $5.00 $5.00 $5.00 $5.00 $5.00 $5.00 $0.25 $0.25 $0.25 $0.25 $0.25 $0.25 $0.25 $0.24 $0.26 $0.26 $0.26 $0.26 $0.26 $0.26 $0.26 $0.26 $0.27 $0.27 $0.27 $0.27 $0.27 $0.27 $0.27 $0.27 $0.29 $0.28 $0.28 $0.28 $0.28 $0.28 $0.29 $0.28 $3.00 $3.00 $3.00 $3.00 $3.00 $3.00 $3.00 $3.00 $6.00 $6.00 $6.00 $6.00 $6.00 $6.00 $6.00 $6.00 $0.80 $0.80 $0.80 $0.80 $0.80 $0.80 $0.80 $0.80 $1.40 $1.40 $1.40 $1.40 $1.40 $1.40 $1.40 $1.40

58

5.6 Optimization of Control Options for Happy Acres The results of the LP optimizations are summarized in tables 5.17 and 5.18. The results are allocated along functional grouping within each soil class in table 5.17, and aggregated for each land use type in table 5.18. The least cost design allocates the appropriate control option to the appropriate soil type and land use (soil is reflected in its predevelopment CN, land use is reflected in the influence of land valuation on the cost of the control). The changes in control options affect the appearance of the neighborhood, and this is evident by inspection of table 5.17. For example, porous pavements were selected (with curb and gutter) for the street design in the rocky soil. In the clay and silt soils where more percolation can take place, the LP model selected a street design with porous pavement and swales instead of curb and gutter. A similar allocation took place with parking areas; both were porous, however, the more permeable soils resulted in a design that had a higher infiltration capacity. The more permeable driveway, patio, and sidewalk choices were chosen in both soil types. Good grass was selected over the other options for all soil areas, except in commercial areas where poor was selected in silts and clays, and fair was selected in rock. This is due to the relatively small amount of landscaped area in commercial areas. There may be other aesthetic factors with commercial areas that would put a higher premium on a higher quality grass other than for a stormwater quality function. Aspens were chosen, but in small amounts, so it would not look significantly different than a typical subdivision. The roof choice remained the standard, rather the roof with detention, due to its relatively high unit cost. Storage was chosen when no other controls were feasible, the highest values, as expected, were in commercial areas with rocky soils, which would not have much infiltration capacity. The cost of the optimal solution for each soil class and land use is found in table 5.18. The total cost for the controls would be $5.2 million, some of which overlaps with money that would be spent for landscaping anyway. About half of this amount is used to attempt to control runoff from transportation related functions. What differs from a traditional subdivision development is the allocation of use. A traditional subdivision would have allocated everything in ground cover to the high quality grass, (particularly for commercial areas) and neglected the woods and aspens (although some exceptions to this exist, mainly for aesthetics). In commercial areas, the detention storage, would have been utilized. For sidewalks, patios, streets, and parking lots, nonporous pavement would have been chosen. Curb and gutter would have replaced swales along street rights of way. An important note here is that this DSS cannot dynamically change land uses. For example, the net amount of area used for rights of way, 39 acres out of the 106 total (see table 5.18), must remain the same. Likewise, the amounts and locations for medium density and low density, as well as the other land uses, must remain the same. What has been done here, however, is to attempt to allocate storage optimally throughout each of these land uses. A more general problem exists which would allow tradeoffs between the land uses. This problem is extremely complex because it involves re-creation of the GIS for each iteration.

59

Table 5.17: Results of LP optimization-land use allocation by function (includes opportunity costs) Land Use Area in Soil Group B in acres 50 60 70 LD MD HD Comm Sch Street 1 Street 2 Street 3 Street 4 Sidewalk 1 Sidewalk 2 Grass P Grass F Grass G Swales 1 Swales 2 Storage Parking 1 Parking 2 Parking 3 Parking 4 Roof 1 Roof 2 Driveway 1 Driveway 2 Patio 1 Patio 2 Woods F Woods G Aspen F Aspen G

Land Use Area in Soil Group D in acres 50 60 70 LD MD HD Comm Sch 11.38 0.00 2.74

9.69 1.08 0.03 2.42 0.21 0.01

0.50 0.00 0.50 0.16 0.38

3.03 0.26 0.01 11.23 6.49 0.57

0.65

1.12 0.00 1.12 0.00 4.60 0.00

1.00 0.06 0.00

0.24

0.83 0.00 0.83 0.00 0.35 0.00

0.04 1.72

0.14

1.19

0.00

2.34 3.23 1.08

2.04 1.03

1.59

0.00 6.13 0.00

0.94 1.21

0.00 2.30

0.47 0.40

0.00 0.77

0.25 0.79 0.36

0.00 9.20 0.00

60

0.00

2.15

0.00 0.00

1.32 5.97

0.00

0.00

Table 5.18: Least-cost LP solutions for land Use/BMP options (including land costs) for Happy Acres.

Land Use 50 ft ROW 60 ft ROW 70 ft ROW Low Density Residential Medium Density Residential Apartments Commercial School SUM Land Use 50 ft ROW 60 ft ROW 70 ft ROW Low Density Residential Medium Density Residential Apartments Commercial School

Soil B Area (acres)

Total (acres)

Soil D Area (acres) 15.15 1.55 0.05 15.02 12.97 3.73 3.37 3.43 55.27

17.78 0.00 4.35 0.00 21.57 0.00 7.67 0.00 51.37

32.92 1.55 4.41 15.02 34.54 3.73 11.04 3.43 106.64

Cost in Soil B, $ $443,554 $36,463 $1,058 $376,677 $361,197 $98,633 $39,267 $106,305 $1,463,153

Cost in Soil D, $ $1,484,917 $$247,981 $$1,509,515 $$517,237 $$3,759,650 TOTAL

Sum, $ $1,928,471 $36,463 $249,039 $376,677 $1,870,712 $98,633 $556,503 $106,305 $5,222,803 $5,220,000

5.7 Decision Support Systems and the Happy Acres Case Study The previous sections have illustrated how a simple hydrologic model can be constructed with basic GIS information. The methods presented in this report allow hydrologic and economic analysis to be performed on micro scales not traditionally used in urban analysis. These micro scales, although unfamiliar, must be used to properly evaluate BMPs for the control of locally generated stormwater runoff. This same information can be used as building blocks for SWMM. SWMM aggregates information in a manner controlled by the user, into an equivalent rectangular catchment. Several methods of aggregation are available within SWMM add-on packages (such as PCSWMM). Unfortunately, this method homogenizes the parcels within each subcatchment, i.e., they lose their unique hydrologic characteristics. The aggregation was typically done so that the user was not overwhelmed by data, as most had to be handled manually. However, within the context of a DSS, appropriate tools can be used to process the data, so smaller scales may be evaluated. A disadvantage of the DSS process used in this case study and outlined in figure 5.1 is that most of the analysis is one way, i.e., there is not a true interchange of information between the modules. The most obvious example is the GIS. It would be desirable to optimize land use in a general form of a land allocation model considering the effects of land valuation, soils, and control options. In order to do this efficiently, the spatial database underlying the parcel delineation must be re-created for each iteration of the model. Of course, this level of integration is also the most difficult and expensive.

61

6.0 Summary and Conclusions 6.1 Summary In summary, GIS has transformed our approach to the urban stormwater management problem. Not only are input parameters in the model itself becoming more easily obtainable, but also the scale of possible evaluations has decreased to a point that it is now possible to effectively evaluate source controls. The case study process shown in figure 5.1 provides a preliminary evaluation of the complex urban stormwater problem and the linked problem of allocation of land use. Several models exist that utilize GIS information; the degree of integration that is desirable remains debatable. Due to the widely disparate spatial scales involved, and the detailed amount of information available in a GIS, it is quite possible for the analyst to be drowned in data that may not be needed in evaluating the problem. The urban stormwater problem needs to be of primary concern to the analyst; rather than the micro maintenance of the GIS. The problem should be the primary focus, even more so than the model, or the database used. As the models evolve into more general Decision Support Systems, they will tend to become more data centered, and computational engines more interchangeable. The GIS data will become more available and standardized, and will be an important tool. One lesson to be learned from the 90s and the computer software explosion that has transformed the working world is that too much reliance on any one technology can lead to obsolescence. DSS promises to be the technology that links many of these tools together to enable the analyst to explore new challenging problems in old contexts. 6.2 Conclusions Advances in development of computer software have produced two key linked technologies: relational databases and geographic information systems. The combination of these two has affected the development of another technology, decision support systems, that has been applied to complex unstructured water resources and environmental problems. Most DSSs include these two technologies, with the addition of simulation models, an evaluation tool (can include optimization), and a graphical user interface. The graphical user interface, mainly the MSWindows interface, is another advance that has both transformed software as well changed the standard of model development. Construction of programs within this environment tends to be more difficult due to its object oriented architecture, however, it is also inherently more dynamic than constructing programs within older environments such as FORTRAN-77. This is primarily due to the advent of structured programming techniques that tend to keep data handling processes out of the main program files, which tends to advance a more data centric approach to modeling. The structured techniques also avoid the use of “spaghetti code” in which it is difficult to debug code due to vague loops and “GOTO” statements that branch the program in many different directions. New types of solvers are now available that can serve as better evaluation tools for a DSS. These include genetic algorithms (GA), simulated annealing (SA), and the relative ease with which linear programming (LP) solvers are used. These optimization tools allow rapid evaluation of both linear and nonlinear problems, which can assist the designer in finding the better or best solution.

62

Urban stormwater models have been created according to specific needs and available funding. The predominant US model, SWMM, was created in the late 60s and early 70s. There is an active user community for this largely public domain model. Several enhancements to the model, namely PC SWMM, Visual Hydro (XP SWMM), and MikeSWMM, are now available in the private domain as well. These enhancements contain facilities that include graphical user interfaces for ease of program use, GIS and CAD interfaces for construction of models based upon the best available system mapping, and external links to available databases to enhance the use of available system data. European models, in particular the DHI and the HR-Wallingford series, have been significantly ahead of the US modeling community in the use of GUIs and GIS. The reason for this gap is primarily the result of funding. Funding for urban stormwater modeling in the US ceased in the early 80s. Meanwhile, the European models were developed and enjoyed significant funding during the 80s and early 90s from both national governments as well as the European Union. These models may have become self-supporting by the creation of companies that sell the licensed product. This enables future enhancements in the models to be made, as well as user support from a centralized source. The US should focus its efforts on the use of linked technologies to take advantage of significant savings that can be realized by avoiding the re-creation of common tools currently available. For example, spreadsheet technology in the US has been effectively standardized upon MS Excel (even if you don’t use it, you use a program that can read these files). Input and output processing within new models could make use of this application, which would allow the user greater flexibility in terms of pre- and post-processing of model output. Visual Hydro provides a good example of the use of spreadsheet tools for data input and output. The US has been a leader in the GIS and database software development field; available links to these programs will continue to evolve and interfaces with GIS should become easier to construct than those at present. A significant portion of this effort is the development of both the graphic features of the GIS and the associated system attributes as well. The case study outlined in this report, although using a simplified hydrologic model, provides a possible outline of the use of this data for problems that have remained intractable to this point, for example, the selection of the appropriate BMP control technology for each parcel. Further work needs to be done to enhance the development of DSS technology to the urban stormwater field. The funding resources should carefully target the development of models and DSSs that link available tools rather than recreate them, and provide a common set of technologies that the user may combine with other available software. The funding should also seek to complement or prod the development of existing commercial software, rather than supplant the market by the introduction of competing products. A possible model could also be the European model community, in which the government funds the initial development of the model, then licenses it to a nonprofit company that markets and sells the model at a self-sustaining price. Care should be taken in that as the model interfaces become easier to run, they may be used inappropriately. A stated goal within the DSS community is to bring the computing power to the level of the decision-maker, rather than an intermediary. This works well if the decision-maker, or their assistant, is trained in the field of urban stormwater. The field of urban stormwater modeling involves the use of complex boundary conditions. Using GIS involves the use of wildly different scales where the uncertainty in the information may not be immediately evident

63

to the user. Such complex problems require a technically competent professional to carefully use and evaluate the information the DSS presents. Rather than simply using a sophisticated set of tools to solve the same problem more efficiently than we can at present, the problems evaluated will become more complex as well as the possible array of solutions to them. The advent of DSS and its inherent technologies, relational databases and GIS, have transformed the field of urban stormwater modeling and allow the evaluation of previously intractable problems.

64

Abstract This report reviews the application of Geographic Information System (GIS) technology to the field of urban stormwater modeling. The GIS literature is reviewed in the context of its use as a spatial database for urban stormwater modeling, integration of GIS and hydrologic time series, and integration of GIS and urban stormwater models (from both a software and management perspective). The available urban stormwater modeling software is reviewed and discussed with respect to their GIS integration capabilities. Decision Support Systems (DSS) are reviewed with respect to their integration with GIS, and their applicability to urban stormwater management problems. A simplified neighborhood scale DSS is presented that includes a GIS, a database, a stormwater system design template, and an optimization capability for screening alternatives. The area and soil based NRCS method is used for calculating runoff from GIS information. Using economic analysis that compares the costs of controls, including the opportunity cost of land for land intensive controls, the optimal selection of Best Management Practice (BMP) controls was accomplished by use of a linear programming (LP) method. The intent of this presentation is to provide an example of the types of problems that become possible to explore with the application of DSS and GIS technology on a small scale. This field is evolving rapidly, and warrants carefully targetted research efforts, particularly at developing nonspecific software tools that aid in integrating existing models.

1.0 Introduction A mathematical model of an urban hydrologic response to precipitation usually requires extensive data due to the complexity of surfaces, flow paths and conduits found in developed locales. Many of these data are geographic in nature; e.g., geographic boundaries of the hydrologic basin provide boundary conditions of the mathematical model. Therefore the marriage of mathematical stormwater models and geographic information systems (GIS) is a natural development of simulation and database technology. The relationship between urban stormwater models and GIS may take many forms. This is apparent from the nearly 50 journal articles, conference proceedings and internet reports surveyed for this review of recent literature. The relationship between GIS and urban stormwater models may be distinct, where the GIS functions as a separate pre- and post-processor; or the distinction may be blurred, where the model is seamlessly integrated to the GIS. The purpose of this report is to accomplish several tasks. In chapter 2 a review of technical literature is performed to determine how GIS is being used in the field of urban storm stormwater modeling. Next, in chapter 3, the predominant urban stormwater models are reviewed within the context of the taxonomy developed in chapter 2. Then, in chapter 4, looking at the future directions of urban stormwater models, Decision Support Systems (DSSs) are described. DSS is now being used extensively for river basin modeling, particularly in the hydropower context. This type of system lends itself to unstructured problems where data integration is a key to evaluation of the problem. The various components of DSS including models, database structure, GIS, optimization, and time series management are discussed. A process level DSS is developed for a textbook subdivision in chapter 5. This DSS contains a GIS, including graphic features and a relational database, a system simulator, and an optimizer. Stormwater design templates were created using Excel spreadsheets, paralleling the design problem from the textbook. Next, GIS data were utilized in a simple hydrologic model using the NRCS (National Resources Conservation Service) method. This data was combined with unit cost data into a linear programming model (LP) in order to develop the least costly mix of BMP controls that maintain the same initial abstraction after development as before. Suggestions for further improvement of the DSS are made by comparison of the DSS structure with those found in chapter 4. Finally conclusion are presented in chapter 6.

1

2.0 Literature Review 2.1 Overview of Sources of Reviewed Literature The GIS literature is broad, due to the wide variety of areas that utilize geographic data. Likewise, the literature describing GIS applications in water resources is itself very broad. However, much of this work in water resources has been in the area of natural hydrology and large-scale, river-basin hydrology. GIS has a long history in this area due in large part to the early availability of remotely sensed spatial data suited for this purpose. A good overview of the concepts of GIS and database technology and their application within the field of natural systems hydrology is found in Singh and Fiorentino (1996). The use of GIS in modeling urban stormwater systems has been more limited due to the need for large, expensive and detailed spatial and temporal databases, along with the fact that many computer tools used in urban stormwater modeling are not easily amenable to integration with GIS. However, as local data gathering efforts have increased and software integration has evolved, the use of GIS in urban stormwater is now widespread. Shamsi et al. (1995) estimate that more than 70% of the information used by local governments is georeferenced. Much of this information has been, or will be, transferred to a digital format, usually a GIS. Recent literature was found in several distinct fields. From the water resources field, recent conferences focusing on urban stormwater have several papers on GIS. Proceedings from two European conferences on urban stormwater by Butler and Maksimovic (eds. 1998), and Seiker and Verworn (eds. 1996), have a wealth of current information on GIS. The American Water Resources Association (AWRA) has sponsored conferences specific to the use of GIS in water resources, such as Harlin and Lanfear (1993) and Hallam et al. (1996). These reports have sections devoted to urban stormwater, of which modeling is a recurring theme. Significant literature in this area was also found on the internet. The Center for Research in Water Resources at The University of Texas at Austin has a large online library of reports and papers on the use of GIS for hydrologic research, some of which concerns the modeling of urban areas (University of Texas, 1998). Other resources were found in the GIS field. One software provider, Environmental Systems Research Institute (ESRI), hosts a large annual international user conference. The proceedings for these conferences are located on the internet at http://www.esri.com (ESRI 1998). The International Association of Hydrological Sciences (IAHS) publishes the proceedings from its many conferences, some of which have dealt specifically with the integration and application of GIS and water resources management (e.g. Kovar and Nachtnebel 1996). Other IAHS conferences have focused on applications, which usually have several papers on using GIS for that application. For example Simonovic et al. (1995) edited “Modeling and Management of Sustainable Basin-Scale Water Resource Systems”, proceedings from a 1995 conference in Boulder, CO. which contained several papers on GIS and model integration. 2.2 GIS as a Spatial Database for Urban Stormwater Modeling The most basic role a GIS can play in the modeling of urban stormwater is that of a simple preprocessor of spatial data. As a pre-processor, GIS may simply store geographic information in a database, or it may be used to calculate model-input parameters from stored geographic data. 2

Frequently data are exported from the GIS in a file format consistent with a model-input file. As a post-processor, GIS may be used to map water surface elevations, concentrations, etc., or to derive spatial statistics based on model output. Shamsi (1998) describes the batch transfer of data from a GIS to SWMM as the interchange data. The GIS and SWMM are operated separately, with no direct interlink. The GIS is used to extract data required by SWMM from the spatial database into a file compatible with a SWMM input file. A recurring theme in recent literature focuses on the ability to get the most out of data by assuring that information tools are consistent. This idea has been termed “hydroinformatics” and is especially prominent in the recent European literature (Fuchs and Scheffer 1996). 2.2.1 GIS as a pre-processor for urban stormwater models Many municipalities store general spatial information in a GIS, and the information is used for a wide variety of purposes and functions within the institutional framework. VanGelder and Miller (1996) describe a typical use of GIS as a spatial database for modeling stormwater from a municipal airport. Detailed georeferenced data were used in conjunction with maintenance data to develop an operation and management schedule as well as to link node information needed to create a SWMM EXTRAN model. Pryl et al. (1998) use a GIS to export details of the urban stormwater network to a hydraulic simulator for Prague in the Czech Republic. The Danish Hydraulic Institute (DHI) program Model Of Urban SEwers (MOUSE) was used to simulate various scenarios for development of an urban stormwater master plan. Rodriguez et al. (1998) used a GIS to study stormwater characteristics of an urban area in Nantes, France. This study used the urban land parcel as the base hydrologic unit of a detailed hydrologic model, as opposed to the more typical basin defined by topography and the layout of the stormwater network. A detailed water budget was performed around the owner-defined parcel. This physically based hydrologic model was then used with the stormwater network to analyze the behavior of urban catchments under a wide variety of storm events. The idea of using small hydrologic units based on land ownership for urban stormwater modeling is ideally suited for GIS applications and is useful when simulating the effect of management decisions made at the parcel level. Sotic et al. (1998) began a preliminary design of CSO facilities in Kumodraz, Yugoslavia with paper maps. Existing paper maps and other data were used to create a GIS, which in turn was used to aid in the design and analysis of the CSO system. This “hydroinformatic” approach consists of developing a set of tools to collect and process data in a consistent manner. The attention to consistency in data transferability is to assure that the greatest value is achieved from the dataset. In this case, the GIS was used to integrate a Digital Elevation Model (DEM), the street network, and the sewer network; then this information was transferred to the BEAMUS hydraulic simulation model (Sotic et al. 1998). A similar hydroinformatic approach is described for the town of Pilsen in the Czech Republic by Hora et al. (1998). Beginning with paper maps, a GIS was built from the ground-up. The complete process is described, ending with an information tool that was used to create a hydrodynamic model of the sewer system, store monitored flow and rain data, evaluate current hydraulic sewer capacity and evaluate the feasibility of alternative sewer developments. Barbe et al. (1993) integrate data transfer from a GIS and a SCADA system to a SWMM model of the Jefferson Parish stormwater stormwater system in Louisiana. The SWMM RUNOFF block was used to simulate the hydrologic runoff characteristics of the area. Geospatial data were transferred from the GIS to the SWMM RUNOFF data file. Similarly, the EXTRAN block 3

was used to simulate the pipe network, and the network connectivity was transferred from the GIS to the SWMM EXTRAN data file. Time series data from 150 monitoring sites were transferred from a SCADA system to the SWMM model for calibration purposes. 2.2.2 GIS as a post-processor for urban stormwater models GIS may also be used to accept model output. Xu et al. (1998) describe a mixed land use hydrologic model that uses GIS as a pre- and post-processor of model information. For this application, the model output of time series of simulated flows may be depicted dynamically through an ArcView interface. Sorensen (1996) describes a typical use of GIS to present model output, that of depicting flood inundation maps from the GIS. MIKE GIS is a modeling tool from DHI that interfaces between ArcInfo or ArcView and MIKE, a flood assessment model. First developed to study flood management in Bangladesh, MIKE GIS uses both the maximum flood extent and the time series of flooding to analyze expected damages from peak inundation and the duration of inundation (Sorensen 1996). A key element to this work is that the GIS is used for more than mapping model output, but that spatial analysis is done with the GIS that adds to the information gained from the model output alone. Shamsi (1998) discusses the difference between transferring data files between ArcView and SWMM and creating an interface that uses SWMM output as a spatial coverage layer in a GIS. This “interface method” (as opposed to the interchange method described above) involves creating a SWMM menu within ArcView. Pre- and post-processors of SWMM input and output files create input files, read output files, and join and unjoin data files (Shamsi 1998). These options are made available in ArcView; however SWMM is run separately from ArcView (Shamsi 1998). 2.2.3 GIS used to estimate spatial input parameters One of the most important hydrographic features of an urban surface is impervious area. Fankhauser (1998) describes a method to estimate impervious area from color infrared aerial photographs and orthophotos. These images have a ground resolution of 25 to 75 centimeters. A raster based GIS, IDRISI, was used to estimate imperviousness to within 10% of the value determined manually for an entire basin. However, the deviation for individual catchments was much higher. For this reason, this method was recommended only for large project areas where the high costs of parameter estimation could be justified. Olivera et al. (1996) use GIS to calculate hydrographic properties of terrain for non-point load estimation. Flow paths calculated from paths of steepest descent are used to calculate flow properties of basins. Cluis et al. (1996) use topographic data and GIS functions to derive important hydrographic characteristics of the terrain such as overland flow paths in a raster based format. Mercado (1996) describes the use of detailed spatial information in the creation of a stormwater model in Tallahassee, FL using XPSWMM software. Scanned and georectified black and white aerial photography was used as a background with other GIS based data, including two foot contour elevations, streams, buildings, roads, etc. A DEM was created in ArcInfo, and the Triangulated Irregular Network (TIN) and Grid functions were used to define areas of high slope and erosion potential, flow gradients and very accurate subbasin delineation (Mercado 1996). 4

Herath et al. (1996) used high-resolution raster data sets to develop a distributed GIS-based urban hydrologic model. Data sets included 50 m x 50 m and 20 m x 20 m land use grids; 1:25,000 plans were used to develop imperviousness by land use, a 50 m x 50 m DEM, population density, water supply data, and rainfall. Herath et al. (1996) integrated the hydrologic model with the GIS, by writing the numerical simulation codes within the GIS, thus reducing problems of data transfer. However, the computational time was felt to be too high for practical use due to inefficiencies of performing the hydrologic simulation within the GIS (Herath et al. 1996). Olivera et al. (1998) developed a GIS-based preprocessor for the new HEC-HMS model developed by the Army Corps of Engineers’ Hydrologic Engineering Center. HEC-HMS is an updated version of the popular HEC-1 hydrologic model. Olivera et al. (1998) describe HECPrePro as a system of ArcView scripting programs and controls to extract hydrographic information from spatial databases and prepare an input file to HEC-HMS. Using SCS curve numbers and a DEM, HEC-PrePro delineates streams and basin boundaries, determines their interconnectivity, and calculates parameters for each stream and basin (Olivera 1998). A benefit to automating the calculation of hydrologic parameters that were traditionally estimated manually is that results are reproducible, i.e., they are not dependent on the bias or experience of the modeler. 2.2.4 GIS used to estimate non-point source pollutant loads Using land use as a predictor of non-point source loads is a common use of GIS and hydrologic models. Hauber and Joeres (1996) describe how a GIS was used to preprocess urban pollutant loads for the Source Loading and Management Model (SLAMM). Similarly, Wright et al. (1995) estimated nutrient loads from developed areas in the Onondaga Lake stormwater basin in upstate NY with the GRASS GIS. These preprocessed loads were then routed from the developed basins using the SWMM RUNOFF model. Battin et al. (1998) describe the EPA’s BASINS (Better Assessment Science Integrating Point and Non-Point Sources) software, which integrates watershed point and non-point source load data, the watershed hydrology program HSPF and the receiving water quality simulation program QUAL2E. Olivera et al. (1996) describe the use of GIS to account for the spatial variability of terrain in pollutant loading from a variety of land uses. The authors review the strength of GIS in quantifying spatially distributing loads, and point out that this is a distinct advantage over lumped models. Scarborough and Yetter (1998) evaluated the Non-Point Source (NPS) module in BASINS 2.0 and found it to be a useful tool for evaluating NPS pollution. However, several problems were found when evaluating a small watershed with the GIS data included with the program. The most critical problem was that of coverage alignment (Scarborough and Yetter 1998). Boundaries of land use and watershed boundaries did not match for the test case study, the St. Jones watershed in Delaware. 2.3 Integration of GIS and Hydrologic Time Series For the purposes of urban stormwater modeling, spatial data may usually be viewed as static. Changes in geographic data are typically modeled in a scenario manner, e.g., a model run may be 5

done for an undeveloped watershed, and then a developed scenario is performed using the same hydrologic conditions. Hydrologic and meteorological data are commonly a time series of discrete values. Therefore some attention must be paid to the integration of spatial and time series data. This idea of consistency among data is key to the concept of hydroinformatics. Pryl et al. (1998) describe the integration of time series with GIS to accomplish urban stormwater master planning in the Czech Republic. Similarly, Rodriguez et al. (1998) use time series in their analysis of the water budget based on parcel-level urban spatial data. Time series integration was a key element in the work reported by Barbe (1996) in Louisiana. A large network of 150 monitoring locations fed a SCADA system with many time series data that were integrated with GIS data and the SWMM model. An Oracle database was used to manage non-spatial data for this project (Barbe 1993). Da Costa et al. (1995, 1996) examined this problem in developing the Portuguese Water Resources Information System. The integration of GIS with temporal data is described as one of the great challenges of developing this system (da Costa et al. 1996). To accomplish this integration, a database was developed using Oracle software to underlie the information system. A special processing module was developed to interface time series data with the GIS. The GIS portion used the ESRI ArcView software. Sorensen et al. (1996) describe the use of time series in an application of MIKE GIS in Bangladesh. Sotic et al. (1998) describe the integration of rainfall and flow time series with geographic data in a hydroinformatic manner in Yugoslavia. Wolf-Schumann and Vaillant (1996) describe in detail the need for integrated time series with georeferenced data. The development of TimeView, a time series management tool, is described as adding a whole dimension (time) to spatial data. TimeView is integrated with ArcView, so that a user can select a geographic feature in ArcView (e.g. a monitored manhole), and TimeView returns a time series of measured data in graphical format. 2.4 Integration of GIS and Urban Stormwater Models The linking of GIS and several hydrologic process models (beyond creating pre-processed data files within the GIS) is examined by Charnock et al. (1996) and DeVantier and Feldman (1993). Issues of differing scale properties and error propagation are addressed. The use of GIS as a central hub of information, which is fed to several satellite process models, is favored over coupling all the processes in one large program. Kopp (1996) addresses these same issues and argues for more data standards to streamline hydroinformatics. Sponemann et al. (1996) explain how a GIS can be shared among many varied users, e.g. gas utilities, water utilities, stormwater, etc. thus maximizing the benefits derived from data collection and management. Greene and Cruise (1995) developed an urban watershed modeling system using the SCS rainfall-runoff methodology and GIS parcel attributes. Meyer et al. (19993) developed a raster based GIS for an urban subdivision in Ft. Collins, Colorado and found that the results compared favorably with non-GIS hydrologic studies of the same area. Shamsi (1998) distinguishes three forms of information exchange between ArcView and SWMM. The interchange and interface methods are described above, and involve the transfer of information between ArcView and SWMM, which are run independently. Shamsi (1998) defines the third method, integration, as the most advanced of the methods. SWMM is used as the hydrologic and hydraulic simulator and is executed from within ArcView. This form of integration includes performing all program tasks within ArcView: creating SWMM input data, 6

editing data files, executing SWMM, and displaying output results (Shamsi 1998). Integration as defined by Shamsi (1998) combines a SWMM Graphical User Interface (GUI) with a GIS to provide a complete data environment. The advantages of a GUI were advanced by Shamsi (1997), who provided a summary of software features and needs for SWMM interfaces. Feinberg and Uhrick (1997) discuss integrating an infrastructure database in Broward County, FL with a GIS and water distribution and wastewater models. The HydroWorks model is used to simulate the wastewater collection system, with close integration with the database of infrastructure characteristics and the GIS. Refsgaard et al. (1995) describe the evolution of DHI’s land process hydrologic model, SHE, and its extensive use of GIS. Ribeiro (1996) describes the use of a raster-based GIS to interface with HSPF to analyze the effects of basin urbanization. Hellweger (1996) developed an ArcView application using the Avenue scripting language to perform the model calculations of USDA’s hydrologic model TR-55. Mark et al. (1997) use the MOUSE program from DHI to evaluate stormwater in Dhaka, along the banks of the Ganges and Bramaputra rivers in Bangladesh. Integration of GIS, time series, and the hydraulic model were accomplished to better understand flooding characteristics. Maximum inundation and duration of inundation were mapped using MOUSE and GIS. Shamsi and Fletcher (1996) describe in detail the linkage of ArcView and SWMM for the City of Huntington, WV. ArcView is shown to be a user-friendly environment to perform stormwater modeling. Bellal et al. (1996) studied partly urbanized basins using a linked GIS and hydrologic model. The hydrologic model was based on a non-urban water budget, with modifications to account for urbanization. The GIS was based on a DEM and raster-based land use data. 2.5 Management Evaluation Using Integrated GIS and Urban Stormwater Models The integration of GIS, time series data, and an urban stormwater model is usually done to evaluate management options. These options may be watershed-based, which would likely include non-urban areas, or they may be local to the urban area. Rodriguez et al. (1998) describe an integrated GIS and urban hydrologic model to evaluate small storm hydrology for parcel level management decisions. Tskhai et al. (1995) use a GIS linked with an optimization model to evaluate ecological and economic alternatives for the Upper Ob River in the Altai region of Russia. While not strictly an urban runoff model in the traditional sense, this project does link urban management decisions with an economic optimization model. Makropoulos et al. (1998) focus on urban sustainability to evaluate stormwater systems. Beginning with the idea that low energy solutions that control impacts at the source are more sustainable, Makropoulos et al. (1998) demonstrate how a raster-based GIS (IDRISI) can be used to integrate theoretical concepts and site specific spatial characteristics. The strength of GIS can be used as a common ground between specialists and non-specialists to help them communicate effectively. Bellal et al. (1996) studied the effect of urbanization on a watershed using a linked hydrologic model based on a DEM and a GIS. A water budget approach was used around each raster unit to account for changes due to urbanization. Mark et al. (1997) describe a detailed evaluation of flood management techniques in Dhaka, Bangladesh, using MOUSE GIS. Xue et al. (1996) and Xue and Bechtel (1997) describe the development of a model designed to evaluate the effectiveness of Best Management Practices 7

(BMP’s). This model, called the Best Management Practices Assessment Model (BMPAM), was linked with ArcView to create an integrated management tool. This integrated model was used to evaluate the pollutant load reduction potential of a hypothetical wet pond in Okeechobee, Florida. Kim et al. (1998) used ArcView with an economic evaluation model and a hydraulic simulator to evaluate storm sewer design alternatives. The hydraulic simulator was used to generate initial design alternatives, which where in turn evaluated with an economic model. The GIS was used to store spatial information, generate model input, and present alternative solutions. The complete package of GIS, economic evaluation model, and hydraulic simulator was termed a Planning Support System (Kim et al. 1998). 2.6 Trends in the Integration of GIS and Urban Stormwater Modeling The trend towards a data-centric suite of evaluation tools is clear. The central idea behind the European concept of hydroinformatics is that a consistent database is used for a variety of purposes. The model is no longer the central unit driving the decision process. Neither, however, has the GIS become the central data tool, due in large part to its inability to handle temporal information effectively. Researchers who have paid equal attention to the model (the processes), the GIS (the spatial data), and the temporal information (time series of hydrologic processes) seem to have had considerable success. The integration of GIS and urban stormwater models should therefore include integration with a database structure equipped to handle time series. Several advanced applications have used a non-graphic database (e.g. dBase, Oracle, Access) that is queried by both the GIS and the hydrologic/hydraulic models. While clearly an evolving area, this approach seems to hold the most promise for the purpose of urban stormwater decision support systems.

8

3.0 Summary of Available GIS Urban Stormwater Modeling Software As described in section 2, a useful taxonomy to define the different ways a GIS is used in urban hydrologic and hydraulic modeling is presented by Shamsi (1998). The three methods defined by Shamsi (1998) are data interchange, program interface, and program integration (Shamsi 1998). A fourth grouping was added for this report, the “intermediate program”. Several commercial modeling products feature a data management program to facilitate data transfer between the GIS and a model. A short description is given below in order of increasing sophistication. Data Interchange: a batch process is used to transfer data to and from the model data set. For example, the GIS may be used to calculate model input parameters e.g., catchment slope, or to query an existing spatial coverage, such as land use. Then portions of the GIS query file can be copied into a model-input file with no direct link between the GIS and the model. The model is executed independently from the GIS, and portions of the output files may be copied back into the GIS as a new spatial coverage for presentation purposes. Intermediate Program: a data management program is used to transfer information between a GIS and a model. This data management program is written specifically to import data from a variety of common third party GIS software, and export to a model data set. Under certain conditions this intermediate program could be defined as an interface, but generally it is not. Program Interface: a direct link consisting of a pre- and post-processor is used to transfer information between the GIS and the model. This process automates the data interchange method. Model-specific menu options are added to the GIS. The model is executed independently from the GIS, however the input file is created in the GIS. For example, in the data interchange method, the user finds a portion of a file and copies it. An interface automates this process, so that the pre- and post-processor finds the appropriate portion of the file automatically. Program Integration: while the interface method can’t launch the model from the GIS, under the integration method, the model and the GIS are together within one Graphical User Interface (GUI). This represents the closest relationship between GIS and model, though “closest” does not necessarily mean “best”. It may be more efficient for a model to be independent from a GIS in certain situations. As noted elsewhere in this report, the development of a GIS for use in urban hydrologic and hydraulic modeling is an expensive investment. Typically the most advanced tools are created for advanced applications, where a full GIS is in place. For some applications, a DOS-based model may still be the most appropriate. However, as more urban areas create GIS coverages, the integration of modeling software and GIS software will become more useful and more prevalent. The Storm Water Management Model (SWMM) is the most widely used urban hydrologic/hydraulic model in the US. In addition to SWMM, numerous other hydrologic models were created in the US during the 70s including the US Army Corps of Engineers Hydrologic Engineering Center “HEC” series of models (HEC-1 through 6). Two of the most popular models, HEC-1 and HEC-2, have been updated and renamed HEC-HMS and HEC-RAS, 9

respectively. These two models have been updated from the original DOS model with a MS Windows based GUI. HSPF, and ILLUDAS are other models developed in the 70’s, which are still used today. The original SWMM model, available at no charge from the US EPA (at the following website: http://ftp.epa.gov/epa_ceam/wwwhtml/ceamhome.htm) was written in Fortran-77 for mainframe computers (Huber and Dickinson 1988). The model was originally written during the 70s, with several major improvements made in the early 80s. It has continued to evolve since being ported to personal computers. Version 4.31 is the latest release; however numerous other modifications exist to the program (e.g. UD-SWMM, a modification of SWMM by the Urban Stormwater and Flood Control District of Denver, Colorado). SWMM runs in MS-DOS in a text-based environment, which is not the user-friendly windows and graphical user interface (GUI) based environment that is expected today. Despite these shortcomings, it has an active user community within the United States. Lack of funding support for SWMM during the 80s and 90s meant that the model had to be selfsustaining. Interested parties such as local governments, consultants, and third party developers added their own refinements to the model, with very little support from the federal government. Because these refinements added value to the original program code, the developers started to charge for these improvements. XP-SWMM (XP-Software 1998) and PCSWMM (CHI 1998) are examples of this type of refinement. The SWMM user’s listserver has developed into a selfsustaining community of users. Information on accessing the listserver can be found at http://www.chi.on.ca/swmmusers.html During the 1980’s, several models started to evolve in Europe. Two of them are HydroWorks, from Wallingford Software in Great Britain, and MOUSE from the Danish Hydraulic Institute, DHI, in Denmark. Unlike EPA SWMM, these models are proprietary. These models are listed in table 3.1, with the addition of MikeSWMM, which is the result of a recent collaborative effort between DHI and Camp, Dresser, and McKee (CDM). This product uses the latest SWMM model engine available from the US EPA, and adds the MIKE GUI and MOUSE GIS from DHI. Table 3.1: Summary of available urban stormwater modeling software with GIS linkages Product Model

Interface Company/Source

HydroWorks/ InfoWorks Mouse GIS

Hydroworks Hydroworks Mouse

Mike

MikeSWMM

SWMM

Mike

PCSWMM/GIS SWMM

PCSWMM

XPSWMM

XPSWMM

SWMM

Website

HR Wallingford/ Montgomery Watson Danish Hydraulic Institute/

www.wallingford-software.co.uk

Danish Hydraulic Institute/ Camp Dresser and McKee Computational Hydraulics International CAiCHE

www.mikeswmm.com

www.dhi.dk

www.chi.on.ca www.xpsoftware.com

The following sections describe commercial and public domain products that are currently available for urban hydrologic and hydraulic modeling. The above taxonomy is used to define how each one handles information transfer between a GIS and the model. However, the reader is 10

cautioned that while integration may be the most advanced method of using a GIS and model together, it is not necessarily the best method for every application. For some applications (especially when the GIS is incomplete, inaccurate, or both) different levels of manual operation may be more appropriate. For example, a limited GIS may exist for an urban watershed, along with very detailed and accurate CAD files. Certain commercial products (e.g. Visual Hydro by CAiCE) can handle CAD drawings better than a product designed to run a pre-existing GIS. If resources were not available to create a GIS, it would be appropriate to use a product suited to the available data. 3.1 SWMM and EPA Windows SWMM As stated previously, SWMM is a DOS based program developed under US EPA funding during the late 1970’s and early 1980’s. There is no provision to link directly or indirectly with a GIS other than through standard input text files. This is the most basic version of SWMM available. This version of SWMM is important because it is in the public domain, and the source code is readily available. The latest version of the DOS based SWMM can be found at http://www.ccee.orst.edu/swmm/ In 1994, the US EPA produced a Windows-based GUI for SWMM. This program (also available at http://www.epa.gov/ost/SWMM_WINDOWS/) runs on Windows version 3.1, and is therefore somewhat outdated. This program is also limited by the fact that the DOS based SWMM engine is in a constant state of improvement by developers and users because the Fortran source code is available. Unfortunately, the Windows SWMM program used the SWMM engine available circa 1994, and the newer versions of the SWMM engine cannot easily be substituted. Therefore the program has quickly become outdated, and has few users. Windows SWMM could not be linked directly to a GIS program. To use either of these programs with a GIS, the data-interchange method must be deployed to transfer information from a GIS to an input file. The GIS may be used to store and estimate model input parameters. The GIS could be queried for the needed values, and the values could then be transferred to the input file. The level of automation to perform this task depends on the user. It could be as simple as copying the needed values onto a Windows clipboard and pasting them into the input file, or developing special queries from the GIS to create an input file automatically. 3.2 PCSWMM ’98 and PCSWMM GIS PCSWMM-98 is a set of 32 bit applications designed to facilitate running SWMM. These tools include an ASCII text editor, an animated hydraulic grade line plot, a chart wizard, an Internet wizard, a batch file control, a rainfall analysis package, a bibliographic database, a sensitivity analysis wizard, and a calibration wizard. The GUI allows files from many sources to be linked, including those accessed across Intranets and Internets. PC-SWMM GIS is an optional tool that works directly with CAD or GIS files in constructing a link-node database for running the model from the existing data sources. After importing the data from a CAD or GIS file, an aggregation tool allows semiautomatic construction of a simplified link-node model. This reduces model complexity, and provides a direct analog to the aggregated catchment concept in the original SWMM. An example of output from a PC-SWMM example run is found in figure 3.1. 11

Figure 3.1: PCSWMM output (CHI, 1999) PCSWMM GIS is an intermediate data management program designed to accept data from a GIS package and transfer it to a SWMM input file. Because it is a more sophisticated method of transferring information from a GIS to a model than the data-interchange method, but it is not an interface as defined by Shamsi (1998), a fourth category was added to the taxonomy, that of the intermediate program. PCSWMM GIS and PCSWMM’98 were developed by CHI in Guelph, Ontario. According to the CHI website, (www.chi.on.ca), PCSWMMGIS does not perform any parameter estimation calculations. It accepts geographic data from an external GIS, within which the parameter estimation calculations and queries are performed. However, it does perform tasks specific to SWMM modeling, such as performing geographic and hydrologic aggregation calculations that are commonly done to simplify a SWMM model. 3.3 XP-SWMM by XP Software (Also Available as Visual Hydro by CAiCE) XP-SWMM32 by XP Software (also included in Visual Hydro, by CAiCE) is a full 32-bit MS Windows application. The program has been enhanced by the addition of a graphics database, and an adaptive dynamic wave solution algorithm that is more stable than the matrix method used in the original SWMM. The program is divided into a stormwater layer, which includes hydrology and water quality, a wastewater layer, which includes storage treatment and water 12

quality routing for BMP analysis, and a hydro-dynamic/hydraulics layer for simulation of open or closed conduits. The user-friendly GUI is based upon a graphical representation of the modeled system using a link-node architecture. An example of input and output processing in Visual Hydro is found in figure 3.2. Because the links and nodes are set up on a coordinate system basis, files can be translated between most CAD and GIS software systems. CAD or GIS files can also be used as a backdrop for the system being modeled. However, since there is no interface with a GIS, data interchange method must be used to transfer parameters (e.g., slope, width, percent imperviousness, etc.) from a GIS to the model. However, the program can import and export files from and to a GIS. 3.4 SWMM-DUET SWMM-DUET is the only fully integrated application of a model into a GIS. It was developed using ArcInfo and the native ArcInfo development language AML (Shamsi 1998). SWMM DUET uses relational databases, both pre- and post-processors, and expert system logic to integrate the SWMM environment and the graphical paradigm of ArcInfo (Shamsi 1998). Future plans include an ArcView version of this product (Shamsi 1998). 3.5 DHI Software 3.5.1 MIKE SWMM MikeSWMM is a proprietary GUI for SWMM from the Danish Hydraulic Institute and Camp, Dresser and McKee, Inc. Mike SWMM can be integrated with a GIS system using Mouse GIS, also available from DHI. Mike SWMM is a classified as an ArcView interface due to its ability to link with the Mouse GIS program, which is described in the follow section. 3.5.2 MOUSE and MOUSE GIS Mouse GIS is a module for MikeSWMM and Mouse users that also allows tight integration between the GIS and the model database. Mouse GIS is an ArcView GIS application. Files do not need to be translated and converted from the GIS to the model format. The DHI product for stormwater modeling, Mouse, uses the Mike GUI within the MS Windows environment. Mouse is a dynamic 32-bit model running in MS Windows that is capable of modeling any type or combination of open or closed conduits and pressurized or gravity flows. An example of the result of a simple query that illustrates the operating environment of Mouse GIS can be seen in figures 3.4 and 3.5. Each object within Mouse GIS has database attributes that can be queried. Mouse GIS is an interface between ArcView and the hydraulic pipe simulator, MOUSE. Mouse is a sophisticated proprietary hydraulic model that is commonly compared to SWMM.

13

Figure 3.2: Visual Hydro (CAiCHE, 1998) 14

Figure 3.3: Mouse GIS user action (www.dhi.dk/mouse/)

Figure 3.4: System response to user action, Mouse GIS (www.dhi.dk/mouse/). 15

3.6 Wallingford Software-HydroWorks and InfoWorks HydroWorks and InfoWorks are companion products produced by Wallingford, Inc. of the UK. Wallingford has taken a different approach to managing geospatial data. InfoWorks is designed to import relational and geospatial data from third party software (e.g. Access and ArcView). Once transferred to InfoWorks, the data is then used to create and run a HydroWorks model. Hydroworks is an urban stormwater modeling system with a user friendly GUI. HydroWorks uses a fully dynamic solution technique that solves backwater and unsteady open or closed conduit situations. InfoWorks performs GIS-type operations, and is designed to operate with HydroWorks, the hydrologic and hydraulic simulator produced by Wallingford, Inc. While the relationship between InfoWorks and HydroWorks may be defined as an interface or even fully integrated, InfoWorks is not a GIS interface. An example of InfoWorks is shown in figure 3.5. Data from a general use GIS product like ArcView would need to be imported into InfoWorks, much like the PCSWMM GIS program from CHI.

Figure 3.5: InfoWorks from Wallingford Software (HR-Wallingford, 1999)

16

3.7 Summary A summary of model and GIS features is presented in table 3.2. As described above, and summarized in table 3.2, the problem of transferring geographic and hydrographic data between a GIS and a simulation model has been handled several different ways by various software developers. It may appear self evident that a tight integration between the hydraulic model and the GIS is desirable. However, the question should be raised; how integrated should these two types of software be? For example, should a GIS include a hydraulic model as part of a toolbox within the GIS? This may, or may not, be desirable. Therefore it should not be assumed that because SWMM DUET has integrated SWMM within ArcInfo that it is the best modeling tool. For example, the expert GUI of XP SWMM may be more useful for a given application, despite the fact that it does not interface directly with a GIS, nor does it have an intermediate data management program. What is common among the recent software developments is a transferability of fundamental database information. This theme is formerly known as a Decision Support System (DSS). Under a DSS framework, neither the GIS nor the model are “central” to the process. Both GIS and model serve satellite functions to a central master database. A more fundamental look at this question is given in chapter 5. The question “which model works best with GIS?” is impossible to answer. Depending on the problem at hand, several products are designed to work with an existing GIS. The answer largely depends on the state of information available. If an existing ArcInfo database is in place, SWMMDUET would work well. Other products have used an information management approach over GIS integration. This may be best suited for applications with disparate data sources. Differences amongst hydraulic models may be more important. The DHI suite of models may be appealing for certain applications. The organization of the HydroInfo/ HydroWorks or PCSWMM’98/PCSWMM GIS software may be best suited for other applications. Each has unique and valuable features, and no recommendation is made in this report for a specific software package. The future evolution of both GIS and urban stormwater modeling, and their possible convergence, appears to be centering upon object intelligence and smaller, programmable component tools. For example, ESRI’s stated goal of its next generation of programs (possibly ArcView 4.0) is to rewrite and enhance its programs to use standard MS Windows routines that can be called via dynamic link libraries (DLLs). An early example is the product called MapObjects, which allows a programmer to insert a GIS-like application within a Visual Basic or Visual C++ program, and make queries and ArcView-like functions upon GIS databases without the ArcView program itself. Existing tools like Evolver, for nonlinear optimization, and @Risk for Monte Carlo simulation are also available as DLLs (Palisade Corporation, 1998). Urban stormwater modeling tools appear to be evolving into using similar tools as they take advantage of existing libraries such as spreadsheet and graphic add-ons, (e.g., Visual Hydro, PCSWMM), and are rewritten in object-oriented programs such as Visual C++, Visual Basic, or Java. The future convergence of GIS and urban stormwater modeling will probably utilize these common sets of tools to take advantage of the easier interoperability. Such tools make integration of these disparate components possible into an integrated Decision Support System, the subject of the next chapter.

17

Table 3.2: Characteristics of urban storm stormwater models Software SWMM Products: EPA SWMM Windows SWMM PCSWMM’98/ PCSWMM GIS Visual Hydro/XP-SWMM SWMMDUET MIKE SWMM/ DHI Products MOUSE, Mike-11 MOUSE GIS HydroWorks/ InfoWorks

Data Interchange

Intermediate Program

GIS/Model Interface

GIS/Model Integration

X X X X X X X

X

Advantages/Disadvantages DOS based Based on SWMM circa 1994 PCSWMM GIS is a data management program Imports CAD, GIS files ArcInfo based ArcView based (via MOUSE GIS) ArcView

InfoWorks is a data management program for geographic and relational databases.

18

4.0 Future Urban Stormwater Modeling in a DSS Environment Much of the data used in distributed (and lumped-distributed) hydrologic modeling requires some level of spatially referenced information. Conversely, purely lumped hydrologic models by definition do not require data to be spatially referenced. This report is focused on lumpeddistributed models and the type of information required to use them. Lumped-distributed models are typically defined by sub-catchments within a stormwater basin. The hydrologic parameters are lumped within each sub-catchment. On the basin scale, however, the discretization among sub-catchments provides spatial distribution. Some of the data used in these distributed models may be more efficiently stored in forms other than GIS spatial database structures (Reitsma et al. 1996). For example, relational data models may be more efficient in storing certain attribute information. Time series are another form of data commonly used in hydrologic modeling. These data are frequently stored in a relational form, despite some shortcomings of this structure for time series (Reitsma et al. 1996). Besides model input, decision-makers frequently require analysis of model output, and the analysis may not necessarily be spatially referenced. For these reasons, future model development should not only focus on the role of GIS in modeling, but on how all information is stored and used. Due to the complexity of tools required to fully support a complex hydrologic decision, a system made up of more than a GIS and simulation model is needed. An integrated suite of tools is required to manage information. These tools are referred to as Decision Support Systems (DSS). Although the model is important, much of the focus has shifted to the related needs of relational database management, developing geographical information systems, and a sophisticated user friendly interface, all combined in DSS. Figure 4.1 describes these necessary components of a DSS (Reitsma et al., 1996). The evolution of DSS may be seen as a natural extension of simulation models (e.g. SWMM, MOUSE, HydroWorks), GIS (e.g. ArcView, IDRISI, ArcInfo), relational databases (e.g. Dbase, Oracle, Access) and evaluation tools (e.g. optimization software). Reitsma (1996) define a DSS for water resources applications: “Decision support systems are computer-based systems which integrate state information, dynamic or process information, and plan evaluation tools into a single software implementation.” In this definition, state information refers to data which represents the system’s state at any point in time, process information represents the first principles governing resource behavior, and evaluation tools refer to software used for transforming raw data into information useful for decision making. A simple representation of DSS components is shown in figure 4.1. The GIS and the simulation model are only components of the DSS in figure 4.1. Future model development should focus not only on GIS interfaces and integration with models, but should include integration with a more complete management information system.. The view for future model development should be broader than only GIS integration, because hydrologic decision making requires more than just spatial information. In a DSS, the GIS only handles spatial data. Spatially referenced information is only one form of state data that is relevant to hydrologic and

19

hydraulic modeling. Time series and attribute data are also crucial to the analysis, and may be handled poorly in a GIS database format designed to manage spatially referenced data. A thorough background on DSSs and their application to reservoir decisions can be found in Jamieson and Fedra (1996a), Fedra and Jamieson (1996), and Jamieson and Fedra (1996b). These series of articles describe the conceptual design, planning capability, and example application of the Water Ware DSS, a complex river basin DSS that combines a “GIS, a georeferenced database, groundwater flow, surface water flow, hydrologic processes, demand forecasting, and water-resources planning” (Jamieson and Fedra 1996a). Reservoir operation and management was one of the first areas of civil engineering in which DSSs were applied. Because of the complicated decision criteria governing urban stormwater management, Davis et al. (1991) studied a prototype DSS developed to analyze the impact of different catchment policies. Driscoll (1993) developed a DSS to assist highway engineers in determining which construction sites would contribute to a receiving water quality problem. Azzout et al. (1995) discuss a DSS under development that would assist in determining the feasibility of alternative techniques in urban stormwater management.

DSS Evaluation Tools -Multi Criteria Evaluation -Visualization -Status Checking

State Information -Databases -Geographic Information

Process Information -(Simulation) Models

Figure 4.1: DSS structure and components (Reitsma et al. 1996) 20

The theme of the following sections is that the parts of a DSS are separate but complementary. They should be able to transfer information to needed process models and evaluation tools without complications. There is no need to house everything under one umbrella, i.e. to perform all modeling tasks in an integrated GIS/hydraulic model. 4.1 State Information In one form of DSS, state information drives the system. This is a “data-centric” view, and it differs from the more traditional model-based analysis commonly used in urban water resources modeling. This fundamental change in perspective may be more important to the future of stormwater modeling than efficient program interfacing. The modeler will need to have tools that handle spatial and temporal data for purposes of modeling, rather than spending resources manually transforming data into the format needed for the model. While this is the idea behind much of the discussion in section 4, a fully integrated GIS/model like SWMMDUET may not be the best modeling tool for the future. It may be that an intermediate database manager (e.g. HydroInfo, PCSWMM GIS, etc) may be closer to a DSS than full GIS integration. State information is stored in relational databases or spatial databases in a modern DSS. Instead of integrating all data forms into one database model, the relational and the spatial information are kept separate, and are linked together to form a geo-relational database structure. 4.1.1 GIS The focus of this report has been on spatial data for modeling purposes. GIS is a critical part of the DSS for systems that are spatially distributed. Since some spatial discretization is needed to model urban hydrologic systems, much effort has been placed on smoothly transferring spatial data to the model and vice versa. Under the DSS data-centered framework, the GIS is one part of the central database of state information. Due to the popularity of GIS software, there has been some interest in housing the entire DSS within the GIS framework. For example, Walsh (1993) investigate spatial DSS, a GIS driven DSS. Reitsma et al. (1996) describe some of the problems associated with a GIS-based DSS: “Recent developments in modeling in GIS (NCGIA 1991; 1993) suggest that GIS can be extended even further into other domains of modeling, e.g., water resources. This type of architecture does offer certain advantages in that it makes use of sophisticated software for management and evaluation of spatial data. A distinct problem, however, is that although rapid improvements are being made in the integration of GIS and modeling (NCGIA 1991; 1993), the full integration of all three components of DSS in GIS is, to say the least, problematic.” To facilitate a non-GIS-based DSS framework, i.e. GIS as a component but not central to the DSS, there are several considerations for GIS. First, the spatial database in the GIS must communicate with other DSS components. This means that much of the interfacing/integration of models and GIS discussed by Shamsi (1998) and reviewed in Chapter 3 must be extended to include other DSS components. Second, spatial tools should be available for modification by the modeler. The GUI should include a dynamic toolbox. For example, if the GIS performs an

21

aggregation calculation in one way, the modeler may wish to modify the algorithm without having to re-write a lot of computer code. The spatial analysis of topographic and hydrographic data may be efficiently carried out in a GIS. GIS software, e.g. ArcView, contain tools that take basic geographic input parameters, e.g. a DEM, and create stormwater boundaries, do slope analysis, etc. Land use and soil coverages are commonly used to estimate hydrologic parameters. Shamsi (1998) discusses several ways that SWMM input parameters may be estimated using GIS. Subarea characteristics such as area, width of overland flow, percent imperviousness and slope may be estimated for the RUNOFF block of SWMM. Parameters used for water quality simulation with the TRANSPORT block of SWMM such as curb length may be estimated from road characteristics in a GIS. Similarly, land use data may be used from a GIS to create SWMM TRANSPORT input files for water quality simulation. Hellweger and Maidment (1999) discuss the details of the spatial analysis required to create an input file for the HEC-HMS model. While not specifically an urban model, it may be useful to review the procedures used. A method to define sub-basin boundaries and stream network connectivity was developed using GIS data layers derived from digital terrain data. Intersecting the sub-basin and stream network layers results in a node-arc representation of the watershed. This information is used to develop an input file for the HEC-HMS model. In this example, an underlying assumption was that streams flowed perpendicular to topographic contour lines. While many of the tools and methods described by Hellweger and Maidment (1999) are useful for modeling natural hydrologic systems, the effect of managed systems in urban areas significantly complicates the analysis. For example, gravity sewers and engineered open channels may have slopes that are independent from the ground surface slope, possibly crossing natural stormwater boundaries and otherwise defying a general physics-based analysis that is used when describing natural systems. Managed or altered hydrologic systems may also be operated based on logic other than the processes that drive a natural system. For example, flow may be diverted from a stream only during dry weather for irrigation purposes, thereby exaggerating the apparent peaking ratio of a stream gauging station. The problems associated with a “pure” GIS analysis of an urban, managed system highlight the advantages of integrating GIS, simulation tools, and relational databases into a DSS. The DSS framework addresses many of the problems associated with using a GIS for urban analysis because of the ability to access and manage related, auxiliary information. 4.1.2 Time series The analysis of time series data is equally important to modeling as the analysis of spatial data. Temporal data includes flow and rain time series, water quality data, etc., as well as dynamic model output. The DSS could include a time series toolbox. Statistical tests and statistical models could reside in this portion of the DSS, for comparison with process models and for analyzing model output. An example of some of this type of pre-processing is that which is currently done in outside statistical packages, or even using Microsoft Excel. Continuous simulation modeling usually will require large amounts of time series data for input

22

purposes. Urban stormwater models that have the capability of continuous simulation usually are capable of reading several different formats of rainfall data. For example, SWMM reads the following formats (Gregory and James, 1996): 1. 2. 3. 4.

National Weather Service Hourly Rainfall Data (in two formats). Pre-1980 National Weather Service Hourly Rainfall Data User Defined Hourly Rainfall Data Canadian Atmospheric Environment Service Hourly Rainfall Data

In SWMM, the standard modules RUNOFF, TRANSPORT, EXTRAN, and STORAGE can import the above formats of time series data. In addition, the modules RAIN, TEMP, STATS, and COMBINE can be used to preprocess time series data. HSP-F, the Hydrologic Simulation Program FORTRAN, includes several time series facilities (Gregory and James, 1996). Several single purpose time series data management programs are available. The HEC-DSS, or the Hydrologic Engineering Center Data Storage System (not Decision Support System), was developed to link time series data with the various HEC watershed management programs. ANNIE, developed by the US Geological Survey, uses watershed data management (WDM) files, and can import WATSTORE files (Gregory and James, 1996). Both ANNIE and HECDSS are non-proprietary FORTRAN models. Due to the multitude of file formats it is difficult to import and export datasets between different modeling environments. For this reason, the CASCADE2 time series management program was developed (Wang and James, 1997). This program, written in Visual Basic, runs under MS Windows and bridges the gap between SWMM and HEC-DSS formats. To be used within a relational database, the time value must be stored, which creates a redundancy of information. This is because a time series is defined by the start time, the time interval, and either the length of the interval or the end time (Reitsma et al. 1996). Another disadvantage of the relational approach is that the DSS must store the criteria for searching the time series (Reitsma et al. 1996). The importance of this redundancy becomes more evident in the case of real time control, which utilizes signal processing and control theory. Lavallee et al. (1996) describe a real time control system developed for the Quebec urban area to manage a stormwater system to minimize CSOs. The unique data needs and system architecture of the RTC system support many of the concepts of DSS due to the demand for timely decisions and vast amounts of data available.. 4.1.3 Relational database An example of a relational database query and its results is presented in this subsection. This example is presented within the context of a relational database contained within a GIS. The same queries can be made in a non-graphic relational database. The linked tabular structure of a relational database allows for extremely complex and powerful queries to be constructed, thus relevant information is made available to the user. The City of Aurora, Colorado has developed a very good base system for GIS. A subcatchment was chosen from the Shop Creek watershed of Aurora, Colorado, a pilot area for GIS development for the City of Aurora. The available themes from this area are as follows: 1. Water lines 23

2. Digital elevation models 3. Rain gages 4. Stream gages 5. Parcels 6. Sewer lines 7. Sewer manholes 8. Digital orthophotos 9. Streets centerlines 10. Sewer tap locations 11. Water meter locations 12. Impervious areas (created by tracing the digital orthophotos) Many tables are associated with each of these themes. An important feature of ArcView is the use of the relational database structure. Tables are linked to graphical features, or themes (analogous to layers in AutoCAD) through the use of spatial geocoding. The user links or joins the tables by choosing a common column, or field between them. The three main types of relationships among tables are: 1. One to One 2. One to Many or Many to One 3. Many to Many All of the records in the one to one table could be placed in the same table. However, good database practice suggests organizing the tables around their functions, instead of the other way around. For example, many attributes are associated with your name, but only your address and phone number are listed in a telephone directory. The first two of these types of relationships is shown in figure 4.2. The two tables nearest the bottom, “Attributes of Theme1.shp” and “Attributes of Parcel” are joined by a one to one relationship, with the fields “Parcel-ID” being the common column. This is again the relationship between “Attributes of Parcel” and “Attributes of Address”, using the fields “Parcel-Id” and “Address-Id” as the common columns (it is not necessary that they have the same name). Lastly, a one to many relationship is shown by the indexing of “Attributes of Address” and “all_9295.dbf” with the fields “Gistag” and “Gisno”. The function of this linking is essentially the following. The theme1.shp table contains the parcels that are located within the small subcatchment. The Attributes of Parcel table contains data on all parcels. Attributes of Address contain address information, including the GIS tag number needed within the Water Use database. This database lists monthly water use data within entire Shop Creek basin, so many records are associated with each parcel. The query shown in figure 4.2 illustrates the power of this tool. The query asks for all linked records in which the water use in a month is over 10,000 gallons. The results of the query are highlighted within the tables. These queries can be moved to the top of their respective tables for further visual analysis. Alternatively, by clicking on the view with the current theme set to Theme1.shp, the visual results of the query can be seen by highlighting parcels that used at least 10,000 gallons a month as shown in figure 4.3.

24

Figure 4.2: Relational database query example in ArcView using water use data

25

Figure 4.3: Spatial results for example query from figure 4.2.

26

The results of the query can also be output to an Excel spreadsheet by using ArcView’s Avenue script language and Microsoft’s Dynamic Data Exchange (DDE). This capability is incorporated as a toolbar shown as an “X” in figure 4.3. The results of this query, output to MS Excel, can be found in figure 4.4.

Figure 4.4: Query results output to Excel using Avenue script tool and Microsoft DDE An example of a relational database within a DSS can be found in Reitsma et al. (1996). In a review of the TERRA DSS system, the authors explain that the data were divided into seven main groupings: 1. 2. 3. 4. 5. 6. 7.

Time Series Data Historical Data Physical Attribute Data Operational Constraints Model Data Security Data Meta Data

Meta data, the last group, is data about data; and allows the Data Management Interface (DMI), a program component of the DSS, to refer only to the meta data, which keeps track of the data structure and where and how the data is stored. This allows the DSS program to be relatively

27

free of data constraints (Reitsma et al., 1996). Although the relational database model has some shortcomings, particularly for time series, it remains the database structure of choice for DSS, as it is the prevailing database model at present. 4.2 Process Information-Simulation Tools In the DSS framework, the process information is contained in simulation models. Process models simulate transitions of the state of the system, as described by the geo-relational database. The simulation model must therefore communicate in some fashion with the rest of the system. For stormwater management models, this may occur in much the same way as described in chapter 3. Data must be transferred to the model from spatial and relational databases. This may occur in a variety of ways, from the rudimentary (but effective) data interchange methods to full-fledged integration in a DSS, running along with the other tools that make up a DSS. The difference from the methods described in chapter 3 is that the communication is not only with the GIS, but also with all elements of the DSS. 4.3 Evaluation Tools Evaluation tools assist the decision-maker by presenting the output from the process and state information in a manner consistent with resource or policy appraisal (Reitsma et al. 1996). Evaluation tools may be of many forms. While much of the above discussion is framed around the excellent review of DSS by Reitsma et al. (1996), the discussion of optimization deviates somewhat from their discourse. Reitsma et al. (1996) do not consider optimization tools to be strictly an evaluation tool, nor do they feel that optimization has been accepted by the user community. While perhaps true for classical optimization techniques, the development of new Intelligent Search Techniques (IST) is proving to be useful for many realistic problems that are unsuitable for traditional methods. 4.4 Overall DSS for Water Management An overall DSS for water management of hydropower and river operations is shown in figure 4.5. This DSS combines the concepts of a centralized database, including hydrologic as well as spatial information, and utilizes two different models that access that data; the Modular Modeling System (MMS) which is a watershed and general environmental model, and RiverWare, which models rivers and reservoirs. Evaluation tools are included within each of the model components. The DSS includes a GIS as a tool for the user to query the common spatial database. This DSS was developed by the Center for Advanced Decision Support in Water and Environmental Systems (CADSWES) at the University of Colorado at Boulder, with support from the Tennessee Valley Authority and the US Bureau of Reclamation. This DSS focuses on large watersheds with complex reservoir and hydropower operations.

28

Figure 4.5: CU-CADSWES DSS (Fulp et al., 1994) A DSS framework for the urban stormwater field is presented in figure 4.6. This DSS is an amalgamation of the different components of the Mike series of software produced by the Danish Hydraulic Institute (DHI), emphasizing their interoperability and common database, Mike Info. The database (relational and spatial) is the common link between separate functions and applications of the DSS. The peripheral models include Mike-11 for urban drainage, Mike SHE for distributed watershed modeling, WUS for river basin planning, and NAM for statistical analysis of streamflow/unguaged catchments.

29

Figure 4.6: Danish Hydraulic Institute DSS, based on integrated water resources modeling (DHI, 1998)

30

5.0 Application of GIS and DSS to Micro Storm Analysis This chapter focuses upon the application of GIS, database management, and DSS to the urban stormwater management problem. A textbook case study from Tchobanoglous (1981) is used to develop a GIS and an accompanying relational database. The database is used with hydrologic and hydraulic models, and a cost analysis module. The combination of these components represents a systematic urban stormwater design tool. The tool is then interfaced with an optimization software package to develop optimal designs of the proposed network. The costs of these designs can then be compared with a decentralized approach to controlling runoff. This was done by using the GIS in conjunction with the NRCS analysis, which computes the initial abstraction storage volume that is lost as a result of development. Using unit costs developed in Heaney et al. (1999a), the optimal suite of controls can be selected using linear programming (LP). A diagram of the process used in the chapter is found in figure 5.1. The reader may notice similarities between some of the components of a DSS and figure 5.1. In particular, the problem consists of a database, simulation tools, and evaluation tools, similar in concept to that of a DSS presented by Reitsma et al. (1996). The database includes GIS and its inherent spatial database, but also a cost database, and a hydrologic database. The simulation tools consist of the NRCS curve number method for computation of initial abstraction, the hydrologic model spreadsheet template, the hydraulic model spreadsheet template, and the costing module. The evaluation tool consists of a genetic algorithm to optimize the stormwater network, and a linear programming model to evaluate proposed controls based upon unit costs developed in Heaney et al. (1999a). Although not integrated into a single software program, the process shown here closely parallels that of a DSS. The utility of GIS (to the urban stormwater field) is enhanced by its close integration with the database, models, and analysis tools used in the problem. Because of the large investment in time and resources necessary to construct an urban GIS, there is a natural tendency for the GIS system to move to center stage. However, the value of the GIS is when it is fully integrated within a DSS which is then used to address complex processes that cannot be easily solved by other means. Key considerations are the concepts of accuracy and scale as they apply to GIS data. Since the datasets presented here vary substantially in terms of their level of detail and scale, a discussion of spatial scale becomes necessary.

31

DSS

Evaluation Tools Optimization Linear Programming (LP) Genetic Algorithms (GA)

Simulation Tools Database Relational (nongraphic) addresses billing unit costs time series input data GIS/Spatial Database Themes Topography Soils Land use Streets Right of way Pipe network Parcels

NRCS CN Hydrologic Method Rational Method Hydraulic Design Template Cost Template

Figure 5.1: Proposed DSS for microstorm analysis 5.1 Spatial Scale and GIS-Stormwater Modeling A recent software development, BASINS 2.0, developed by TetraTech for the US Environmental Protection Agency, has created interest in the development of model-graphical user interfaceGIS linkages within the water community. BASINS 2.0 runs within ArcView 3.0 and includes a national dataset on the attributes listed in Table 5.1 (Battin, et al. 1998).

32

Table 5.1: Available BASINS data attributes (Battin et al. 1998) Spatially Distributed Data Land use/land cover (GIRAS) Urbanized areas Populated place location Reach File, version 1 (RF1) Reach File, version 3 (RF3) Soils (STATSGO) Elevation (DEM) Major roads Environmental Monitoring Data Water quality monitoring station summaries Water quality observation data Bacteria monitoring station summaries Weather Station Sites (477) Clean Water Needs Survey Point Source Data Permit Compliance System Industrial Facilities Discharge (IFD) sites Toxic Release Inventory (TRI) sites

USGS Hydrologic unit boundaries Drinking water supplies Dam sites EPA region boundaries State boundaries County boundaries Federal and Indian Lands Ecoregions USGS gaging stations Fish and wildlife advisories National Sediment Inventory (NSI) Shellfish Contamination Inventory

Resource Conservation & Recovery Act (RCRA) sites Mineral availability system/mineral industry location Superfund national priority list sites

BASINS 2.0 includes tools for automatic watershed delineation and handling of digital elevation models (DEM). Its main data handling routines include: Target, which is a regional, or state level broad-based watershed water quality or point source assessment tool; Assess, which operates a smaller scale of one or a few watersheds and enclosed discharge points or water quality stations; and Data Mining, which dynamically links water discharge stations and geographic location information. Modeling tools include a nonpoint source model (later to be enhanced by the addition of SWAT, the MS Windows based nonpoint source model developed by the USDA), HSPF, Qual-2E, and Toxiroute. Model post processors include graphs (Battin, Kinerson, and Lahlou 1998). EPA SWMM may be linked with BASINS in the future.

33

The accepted accuracy levels of mapping work are listed in Table 5.2 (Shamsi et al. 1995). Most of the BASINS work and modeling have been on a watershed or regional level scale. An example is shown in figure 5.2. The size of this file relative to the area it represents reflects a scale of about 1:2000. Table 5.2: Minimum horizontal accuracy and example features for various map scales in urban areas (Shamsi et al. 1995) Map Scale

1”=50’ 1’=100’ 1”=200’ 1”=2000’

Minimum Horizontal Accuracy, per National Map Accuracy Standards ± 1.25’ ± 2.50’ ± 5.00’ ± 40’

Examples of Smallest Features Depicted Manholes, catch basins Utility poles, fence lines Buildings, edge of pavement Transportation, developed areas, watersheds

Figure 5.2: BASINS dataset for Boulder, Colorado Automatic watershed delineation of undeveloped areas may be appropriate at this scale. However, urban systems have extremely altered topography. The topography in these types of catchments can be represented by a dense DEM; however, development of watersheds based

34

upon triangular irregular networks (TINs) from this information is not presently reliable. This is not to say that the database information presented from a watershed level scale has no value. Actually, having the information presented in figure 5.2 can provide the modeler with possible alternative sources of data, possibly structures that may not have been considered, etc. However, a key disadvantage of using GIS information from different scales of accuracy is that a vector GIS cannot show any uncertainty. An assumption of the GIS model is that the points are known to 100% accuracy. This leaves it up to the reader to verify locations and discrepancies, particularly when the scales, and the resultant accuracy, differ widely. In addition, the memory requirements for regional level stormwater-GIS modeling are staggering. For example, the City of Boulder has an ongoing GIS project, a broad view of which is shown in figure 5.3 (Brown and Caldwell and Camp, Dresser, and McKee, 1997).

Figure 5.3: ArcView coverage of Boulder, Colorado (Brown and Caldwell and Camp, Dresser, and McKee, 1997) Minor roads are outlined in light green, major roads are outlined in thick maroon; creeks are shown in light blue, lakes in shaded blue, and sub-basins boundaries in black. Not shown for better clarity, but available, are parcels, zoning, topography, watershed boundaries, and several other miscellaneous themes. Also not shown is the database describing each graphic entity (for example, the parcel database). Even at this finer resolution, urban stormwater modeling is at too aggregate a scale to evaluate sets of alternatives that include micro-topographical changes to implement BMPs.

35

In order to evaluate the effects of source and neighborhood-level BMPs, the coverage as depicted in figure 5.4 is needed. This area is a block in the University Hill neighborhood of Boulder. The parcel theme is shown in red, the street centerline is shown in green, and the streams are shown in blue. Topography is not shown, but exists in this database at the 40 foot contour interval, reflecting a scale of about 1:200.

Figure 5.4: City of Boulder ArcView GIS coverage for University Hill neighborhood, Boulder, Colorado.

36

Moving towards a finer dataset, another parallel project at the City of Boulder, in the Public Works/Public Utilities group, is an Automated Mapping/Facilities Management (AM/FM) project in which the city’s infrastructure is being mapped by street surveys and aerial photography. The end product at the present time is a tiled set of AutoCAD maps representing portions of the city. The representation of this project for the same block in the University Hill neighborhood is shown in figure 5.5. The scale of this information is approximately 1:100. The green layer signifies building rooflines, yellow is the street centerline and parking areas/driveways, red is sidewalks, and black is the curblines. This file has been edited extensively to eliminate extraneous lines and close polygons. Since the end product of the project was a set of AutoCAD maps, manual and automatic processes on the digital photography result in multiple lines whose ends may not match and polygons that do not close. Although acceptable for graphic presentation, this information is of limited value for extracting data for stormwater evaluations. Extensive cleanup is necessary for this information prior to inputting it into a GIS. Topography for this information is available for an additional cost at a 2-foot contour interval. At the present time, conversion of this data to ArcInfo and ArcView coverages is underway. 5.2 Description of Happy Acres Case Study GIS A textbook study area, nicknamed “Happy Acres”, was selected from Tchobanoglous (1981). A GIS coverage for this case study was developed. The study area was first digitized in AutoCAD, then edited for geometric consistency, i.e., parallel lines were kept parallel, polygons were joined from separated lines, to make the transition to GIS easier. The mix of land uses for the area is laid out in table 5.3. The reconstructed AutoCAD drawing of the area is shown in figure 5.6. The topography of the study area and the layout of the storm sewer system are shown in figure 5.7 (Tchobanoglous 1981). Land use is shown in figure 5.8. Soils data is shown in figure 5.9. The entire study area is divided into 54 sub-areas that range in size from 0.8 to 5.4 acres in size. A description of the attribute information in figure 5.6 is found in table 5.4. Table 5.3: Mix of land uses in Happy Acres Land Use

Acres

Residential, low density Residential, medium density Apartments School Commercial Total

37

20.8 51.7 10.0 5.7 18.4 106.6

Dwelling units/acre 2-3 6-8 10 N/A N/A

Figure 5.5: AutoCAD file for University Hills neighborhood, Boulder, Colorado.

38

Figure 5.6: AutoCAD coverage for study area (adapted from Tchobanoglous, 1981)

39

N

Sewer2.sh p Manhol e2 .sh p Con tour s o f Tch ob an_po ints 4_point.shp Tchoban_parce ls2 _regio n.shp Tchoban_roads 2_r egi on .s hp

100

0

100

200 Meters

Figure 5.7: Study area topography (adapted from Tchobanoglous, 1981)

40

Tchoban_roads2_region.shp Nwgrd3 Apartment Commercial LD Residential MD Residential School No Data

N W 300

0

E

300 600 900 1200 1500 1800 2100 2400 Feet

S

Figure 5.8: Study area land use (adapted from Tchobanoglous, 1981)

41

Tchoban_roads2_region.shp Tchoban_drainage2_region.shp Soilgrid Clay Rock Silt No Data

N W 300

0

E

300 600 900 1200 1500 1800 2100 2400 Feet

S

Figure 5.9: Study area soils (adapted from Tchobanoglous, 1981)

42

Table 5.4: AutoCAD layers for study area Layer/Object Category Streets Manholes Sewer lines Land use boundary Hydrologic boundary Parcel Rooflines Driveways Soils

Color Not shown (for clarity) Blue Red Aqua Blue Green Magenta Orange Not shown (for clarity)

The AutoCAD layers shown in table 5.4 become the following ArcView themes: 1. 2. 3. 4. 5. 6. 7. 8. 9.

Streets Manholes Sewer lines Land use boundary Hydrologic boundary Parcel Rooflines Driveways Soils

A relational database is associated with each graphic object, grouped according to type. Attributes associated with parcels are address and land area; and with streets are right of way width, length, land area, and street name. Soils and land use exist in separate tables, and this information was combined with the parcel and street databases by performing an intersection query on the two themes. The results of the query can also be output to an Excel spreadsheet by using ArcView’s Avenue script language and Microsoft’s Dynamic Data Exchange (DDE). This procedure was used to extract the relevant attribute information for parcels and streets. The rights of way identified in figures 5.6 through 5.9 were assigned widths based upon the following criteria. Minor streets within the development have a 50 foot right of way, a minor arterial is given a 60 foot right of way, and a major arterial a 70 foot right of way. The profile of each right of way is given in table 5.4. The reader is referred to Heaney et al. (1999a) for further details on the database. Table 5.5: Right of way characteristics R/W ft

Length, ft 50 28,680 60 1,124 70 2,741

Curb ft

Parking Landscaping Sidewalk Traffic ft strip, ft ft Lanes, ft 4 8 10 8 20 4 16 10 8 22 4 16 18 8 24

43

Note: Some of the parameters are summed from both sides of the street. Lot characteristics for the two single lot residential land use classifications are presented in table 5.6. Lots were aggregated in this manner for the optimization; however the GIS contains the full heterogeneity of each parcel. Table 5.6: Lot characteristics for residential parcels Land Use

No. of Parcels

MD Residential (6-8 DU/AC) LD Residential (2-5 DU/AC)

Roof Patio Driveway Landscap- Total Area SF SF ing Area SF SF SF 255 1,600 200 600 3,600 6,000 51 2,000 400 800 9,800 13,000

For the apartments, commercial, and school land uses, an aggregate analysis was used because these land uses exhibited multi-parcel characteristics, such as for parking. A summary of these characteristics is found in table 5.7 Table 5.7: Aggregate characteristics for commercial, apartments, and schools Land Use

Apartments Commercial School

No. of parcels 2 6 3

Parking LandscapRoof Parcel ing Area Area Area SF SF SF SF 2 162,680 46,927 75,083 40,670 1 481,070 152,839 304,678 23,553 1 149,407 69,080 51,807 28,521

Stories

5.3 Simulation Tools for Hydraulic Design The storm sewer network for the Happy Acres subdivision is diagrammed in figure 5.10. A spreadsheet template has been developed to simulate and optimize storm sewer design for the Happy Acres neighborhood-see tables 5.8 to 5.10. The value of better data obtained using GIS can be estimated by evaluating the designs with and without this better information. The following columns in table 5.7 represent data that can be obtained partially or totally with a GIS system for this example. Column Description 5 Sewer length 6 Stormwater area 7 Dwelling units per acre The output from table 5.8 is the design peak discharge leaving each subcatchment. This information is input to the sewer design table 5.9 that finds feasible combinations of pipe diameters and slopes. The constraints on the design are: Minimum depth of cover for the sewer, and

44

Minimum velocity in the pipe. The decision variables are pipe diameter (column 8) and slope (column 6). Trial and error procedures are used to find a feasible solution to the design problem. In more sophisticated analysis, the costs of the alternative systems are evaluated as shown in table 5.10. The background for development of the cost relationships found in this table can be found in Heaney et al. (1999a), and is based upon data obtained from R.S. Means (1996a). Additional GIS data are helpful for the cost analysis. Specifically, soil conditions (column 8) affect the side slopes of the sewer excavations, and the bedding costs. 18

14E

17

17A

17B

16

16A

16B

15

15A

15B

12A

12B

9A

9B

7A

7B

14F 14

14D

14C

14B

14A

13I

13H 13

13G

13F

13E

13D

13C

13B

13A

12

11G

11F

11E

11D

11C

11B

11A

11

10E

10D

10C

10B

10A

10

9

8D

8C

8B

8A

8

7

1

2

3

4

Figure 5.10: Study area sewer network (adapted from Tchobanoglous, 1981)

45

5

6

16C

Table 5.8: Sewer network design hydrology (Heaney et al. 1999)

46

Table 5.9: Sewer network design hydraulics (Heaney et al. 1999)

47

Table 5.10: Sewer network design cost (Heaney et al. 1999)

48

Using a new intelligent search technique called genetic algorithms (GAs), the optimal design was found by having Evolver (Palisade Corp., 1998), a commercially available GA, evaluate different combinations of pipe diameters and slopes until the least cost design is found. 5.4 Simulation Tools for Hydrologic Analysis Heaney, Wright, and Sample (1999) describe a method for using the NRCS curve number (CN) approach for evaluating micro storms. The fundamental principle is that development should not reduce the initial soil moisture storage that existed prior to development. This initial soil moisture storage is equivalent to the initial abstraction as calculated using the Natural Resources Conservation Service (NRCS) curve number (CN) method. The initial abstraction is a good measure of the ability of the soil system to filter the stormwater. The initial abstraction, as a function of CN, is shown in table 5.11. Inspection of table 5.11 reveals the importance of CN. A low CN of 30 corresponds to an initial abstraction of 4.67 inches. Even at a CN of 80, the initial abstraction is still 0.5 inches. If the original CN is fairly low, then a significant amount of soil moisture storage is lost if this area is rendered impervious by development. Table 5.11: Initial abstraction as a function of curve numbers, CN CN 20 30 40 50 60

Ia, inches 8 4.67 3 2 1.33

CN 70 80 90 100

Ia, inches 0.86 0.5 0.22 0.02

This method uses the concept of modifying the CNs for the developed condition so that the modified CN is the same as the natural CN. The more cost-effective controls tend to focus on utilizing the pervious area for more intensive infiltration. Alternatively, we seek to design hydrologically functional landscapes as described in the next section. 5.4.1 Hydrologically functional landscaping Traditional landscaping relies on covering most, if not all, of the pervious area with grass. The lot is graded so that stormwater drains to the street and/or the rear of the lot as shown in figure 5.11 (Dewberry and Davis 1996). An example of a hydrologically functional landscape is shown in figure 5.12 (Prince Georges County 1997). The general idea is to maximize the infiltration of stormwater by providing depressions, draining runoff from impervious areas to pervious areas, providing more circuitous routes for the stormwater to increase the time of concentration, etc.

49

Figure 5.11: Conventional storm drainage (Dewberry and Davis 1996).

50

Figure 5.12: Illustration of hydrologically functional landscape (Prince Georges County 1997).

51

5.4.2 Determination of runoff volumes using NRCS method Each developed land use is assigned a curve number (CN) based upon work done by the Soil Conservation Service (1986). The initial abstraction, or available storage, is estimated by the following equation: 200 Ia = −2 5.1 CN The final list of 10 permeable and 16 impermeable candidate land uses with their expected effectiveness as measured by their curve number (CN) and the associated initial abstraction in inches, calculated using equation 5.1, are shown in table 5.12. The CNs range from 25 to 98. The initial abstraction associated with a CN of 25 is 6.00 inches of precipitation. Making this land impervious increases the CN to 98 with an associated initial abstraction of only 0.04 inches, a major loss of infiltration capacity. Using unit costs in $/square feet, which are developed in section 5.5 (and detailed in Heaney et al. 1999a) and having determined the appropriate abstraction, it is possible to convert the control option costs to $/gallon, which is done in the last four columns of table 5.12. Several different functional land uses are given in table 5.12. These include two kinds of aspens, fair, and good (referring to the health and density of the stand), two kinds of driveways, permeable and impermeable, three types of grass cover, good, fair, and poor (again referring to health and density), four types of parking, a traditional impervious surface, and three of gradually increasing porosity, two types of patios, permeable and impermeable, two kinds of roofs, with retention and without, two kinds of sidewalks, permeable and impermeable, storage (detention pond), four types of streets, a traditional street profile with curb and gutter, a street with curb and gutter and porous pavement, an impervious street with swales, and a street with porous pavement and swales, two types of swales of progressively greater infiltration capacity (and greater area), and two kinds of wooded areas, fair and poor, again referring to health and density of the trees. These values are unique to the soil type heading the column. The NRCS method aggregates clay and silt together as soil type "B", and rock as soil type "D". Unit costs expressed as $/gallon are useful for comparative purposes, as will be seen later. 5.4.3 Breakdown of calculated volumes per function A functional analysis within each land use and soil classification was performed by adding the total areas for the functions of roof, lawns, driveways, and parking (for non-right of way uses), and streets, curbs, parking, sidewalks, and lawns for right of way areas. Volumes of developed runoff can then be calculated by multiplying the initial abstraction by the appropriate area. Predevelopment runoff can be calculated by using the composite curve number for Happy Acres prior to development of 63.07, determining an initial abstraction for each soil group, and multiplying this again by the area as done for the developed volumes. The result of this analysis is found in table 5.13. This provides a snapshot of the increase in runoff volume for each land use generated by development. Because the NRCS method is unique to soil characteristics, this is further broken down by soil group.

52

Table 5.12: SCS hydrologic classifications, and calculation of unit storage values, 1/99$ Curve Number

28

48

57

63

Unit Unit Costs in $/gallons cost B C D $/sf A B C D 5.14 2.17 1.51 1.17 $2.00 $0.62 $1.48 $2.13 $2.73

Aspen G

25

30

41

48

6.00

4.67

2.88

2.17

$3.00

$0.80 $1.03 $1.67 $2.22

Driveway 1 Driveway 2 Grass F

98 70 49

98 80 69

98 85 79

98 87 84

0.04 0.86 2.08

0.04 0.50 0.90

0.04 0.35 0.53

0.04 0.30 0.38

$0.23 $0.25 $0.81

$9.21 $9.21 $9.21 $9.21 $0.47 $0.80 $1.13 $1.34 $0.63 $1.45 $2.45 $3.42

Grass G

39

61

74

80

3.13

1.28

0.70

0.50

$1.03

$0.53 $1.29 $2.35 $3.30

Grass P

68

79

86

89

0.94

0.53

0.33

0.25

$0.70

$1.19 $2.12 $3.45 $4.55

Parking 1 Parking 2 Parking 3 Parking 4 Patio 1 Patio 2 Roof 1 Roof 2 Sidewalk 1 Sidewalk 2 Storage

98 61 46 36 95 76 95 85 98 70 15

98 75 65 55 95 85 95 85 98 80 20

98 83 77 67 95 89 95 85 98 85 35

98 87 82 72 95 91 95 85 98 87 40

0.04 1.28 2.35 3.56 0.11 0.63 0.11 0.35 0.04 0.86 11.33

0.04 0.67 1.08 1.64 0.11 0.35 0.11 0.35 0.04 0.50 8.00

0.04 0.41 0.60 0.99 0.11 0.25 0.11 0.35 0.04 0.35 3.71

0.04 0.30 0.44 0.78 0.11 0.20 0.11 0.35 0.04 0.30 3.00

$0.23 $0.25 $0.26 $0.28 $0.19 $0.19 $0.00 $1.50 $0.19 $0.19 $5.00

$9.21 $0.31 $0.18 $0.13 $2.89 $0.49 $0.00 $6.82 $7.44 $0.36 $0.71

Street 1 Street 2

98 70

98 80

98 85

98 87

0.04 0.86

0.04 0.50

0.04 0.35

0.04 0.30

$0.25 $0.26

$9.77 $9.77 $9.77 $9.77 $0.49 $0.84 $1.19 $1.41

Street 3 Street 4 Swales 1

76 61 46

85 75 65

89 83 77

91 87 82

0.63 1.28 2.35

0.35 0.67 1.08

0.25 0.41 0.60

0.20 0.30 0.44

$0.27 $0.28 $3.00

Swales 2 Swales 2 Woods:Fair: Woods are grazed but not Woods F burned, and some forest litter Woods:Good: Woods without grazing, and Woods G adequate litter and brush

29 36

50 60

62 73

67 79

4.90 3.56

2.00 1.33

1.23 0.74

0.99 0.53

$6.00 $0.80

$0.68 $1.22 $1.74 $2.17 $0.35 $0.67 $1.09 $1.49 $2.05 $4.47 $8.06 $10.9 6 $1.97 $4.81 $7.85 $9.77 $0.36 $0.96 $1.73 $2.41

25

55

70

77

6.00

1.64

0.86

0.60

$1.40

$0.37 $1.37 $2.62 $3.76

Cover Description No.

Type 1 Permeable 2 Permeable 1 Impervious 2 Impervious 3 Permeable 4 Permeable 5 Permeable 6 4 5 6 7 8 9 10 11 12 13

Impervious Impervious Impervious Impervious Impervious Impervious Impervious Impervious Impervious Impervious Permeable

14 Impervious 15 Impervious 16 Impervious 17 Impervious 18 Permeable 19 Permeable 20 Permeable 21 Permeable

Cover type and hydrologic condition Aspen-mountain brush mixture: Fair:3070% ground cover Aspen-mountain brush mixture: Good: >70% ground cover Driveway Driveway-porous pavement Lawns, pasture, grassland: Fair condition (grass cover 50-75%) Lawns, pasture, grassland: Good condition (grass cover >75%) Lawns, pasture, grassland: Poor condition (grass cover < 50%) Parking Porous parking 1 Porous parking 2 Porous parking 3 Patio Porous patio Roof Roof with detention Sidewalks Sidewalks with porous materials Storage-off-site in infiltration/detention basins Street with curb and gutter Street with curb and gutter and porous pavement Street with swales Street with swales and porous pavement Swales 1

ID Aspen F

A

B

Initial Abstraction in inches C

D

Source: adapted from SCS, 1986

53

A

$9.21 $0.60 $0.39 $0.27 $2.89 $0.88 $0.00 $6.82 $7.44 $0.62 $1.00

$9.21 $0.98 $0.71 $0.46 $2.89 $1.25 $0.00 $6.82 $7.44 $0.88 $2.16

$9.21 $1.34 $0.97 $0.58 $2.89 $1.57 $0.00 $6.82 $7.44 $1.04 $2.67

Table 5.13: Calculation of developed and predevelopment stormwater volumes for Happy Acres Soil Types B

Soil Types D, Total

sf

sf

Volume

Volume

Total Vol.

Roof

46927

0

46927

412

0

412

4580

0

4580

Parking

75083

0

75083

255

0

255

7327

0

7327

0

0

0

0

0

0

0

0

0

Lawns

40670

0

40670

4334

0

4334

3969

0

3969

Roof

95132

57707

152839

834

506

1341

9284

49

9333

Parking

44810

259868

304678

152

884

1036

4373

86

4459

0

0

0

0

0

0

0

0

0

6839

16714

23553

729

696

1425

667

68

735

140800

267200

408000

1235

2344

3579

13741

229

13969

0

0

0

0

0

0

0

0

0

52800

100200

153000

180

341

520

5153

33

5186

34514

2191

36705

9954

0

9954

Lawns MD Residential

Roof Parking Driveway Lawns

LD Residential

cf

cf

353666

538755

892420

37686

22448

60134

Patio

17600

33400

51000

154

293

447

Roof

102000

0

102000

895

0

895

Parking Driveway Lawns School

cf

Undev., Undev. D cf cf

Apartments

Driveway

sf

Undev., B cf

Function

Commercial

Developed, B Developed, D Developed

Volume Tot. Volume

Land Use

Driveway

Area, Total

Volume

0

0

0

0

0

0

0

0

0

0

40800

0

40800

139

0

139

3982

0

3982

47939

0

47939

491233

0

491233

52344

0

52344

Patio

20400

0

20400

179

0

179

Roof

69080

0

69080

606

0

606

6742

0

6742

Parking

51806

0

51806

176

0

176

5056

0

5056

0

0

0

0

0

0

0

0

0

28521

0

28521

3039

0

3039

2783

0

2783

Driveway Lawns

0

Streets 50 ROW

659728

774288

1434016

Street with curb and gutter Parking

105556

123886

229443

359

421

780

10301

41

10342

105556

123886

229443

359

421

780

10301

41

10342

Sidewalks

105556

123886

229443

359

421

780

10301

41

10342

curb

52778

61943

114721

180

211

390

5151

21

5171

Lawns

52778

61943

114721

3952

1966

5918

5151

192

5343

87540

0

87540

Street with curb and gutter Parking

11672

0

11672

40

0

40

1139

0

1139

23344

0

23344

79

0

79

2278

0

2278

Sidewalks

11672

0

11672

40

0

40

1139

0

1139

curb

5836

0

5836

20

0

20

570

0

570

Lawns

5836

0

5836

437

0

437

570

0

570

13195

189531

202726

Street with curb and gutter Parking

1508

21661

23169

5

74

79

147

7

154

3016

43321

46337

10

147

158

294

14

309

Sidewalks

1508

21661

23169

5

74

79

147

7

154

curb

754

10830

11584

3

37

39

74

4

77

Lawns

754

10830

11584

56

344

400

74

34

107

60 ROW

70 ROW

Total

1724282

54

140882

210758

The functions were then compared across land uses by computing the difference between the sum of the function’s pre-development and post-development storage volumes. The result is plotted as a bar chart in figure 5.13. The greatest impact is from streets and roofs, with roughly equal values of storage volume reduction. Patios are insignificant in this analysis. Lawns actually add a great deal of storage, offsetting somewhat the drastic reductions from roofs and streets. Driveways and parking lots result in smaller reductions in volume, however, the local impact may be significant. 140000

120000

100000

Volume, post development, (CF) Volume, predevelopment (CF) Difference

Volume in Cubic Feet

80000

60000

40000

20000

0 Roof

Parking

Driveway

Lawns

Patio

Streets

-20000

-40000 Function

Figure 5.13: Allocation of available storage for initial abstraction and land use. 5.5 Simulation Tools for Cost Analysis If the cost of modifying the CNs can be determined, then cost-effective strategies can be developed for maintaining the undeveloped CN for each parcel or combination of parcels. Since most BMPs are land intensive, a careful evaluation of their costs must include land valuation. The costs used in the analysis were developed in Heaney et al. (1999), for each control and each land use. The procedure for calculation of the land component of controls within one land use, medium density residential, is outlined in table 5.14.

55

Table 5.14: Land valuation for medium density lot, 1/99$ Component

SF

Roof-house Roof-garage Driveway Yard Patio Total

1200 400 600 3600 200 6000

% of $/sf total 20.0% $56.25 6.7% $34.00 10.0% $4.00 60.0% $1.00 3.3% $4.00 100.0%

Construction Total Land $ Unimproved Cost, $ Land, $ $67,500 $8,790 $5,860 $13,600 $2,930 $1,953 $2,400 $4,395 $2,930 $3,600 $26,370 $17,580 $800 $1,465 $977 $87,900 $43,950 $29,300

An estimate of the cost in $/sf is found in column 4 of table 5.14. Next, the construction cost (column 5) is obtained by multiplying column 2 by column 4. Next, the percentage in column 3 is multiplied by the total of column 5 to obtain an estimate of the land cost, in column 6. Column 7, the unimproved land cost, is obtained by multiplying the values in column 6 by 2/3. The value of the 3,600 square feet of land for the yard function is $26,370. Next, opportunity costs must be calculated. This procedure is illustrated in table 5.15. The value of $26,370 is annualized, using an interest rate of 6%, and an infinite term (as in equation 6.2), to obtain $1,582/year. Then, this value is spread over 25 years at 6%, to obtain $20,226. Dividing this value by 3,600 square feet gives $5.62/square feet. This value is used for all grass types as the underlying value of the land is assumed to be constant irrespective of the type of grass. Landscaping costs were developed from RS Means (1996b), and updated to January 1999, and are presented in table 5.15 (for a medium density residential lot). The initial capital investment consists of the cost of soil preparation including sod, topsoil, and soil conditioners, and an irrigation system. For a good lawn, the present value of the initial landscaping investment is $2.22 per square foot. Costs for lesser quality lawns drop to $1.71/sf and $.95/sf for fair and poor quality lawns. For the good lawn system, operation and maintenance costs add an additional $2.45 per square foot bringing the total to $10.29 per square foot. An estimated 10 percent of this total cost is allocated to stormwater management. Similar estimates were made for fair and poor lawns. The resulting total costs per square foot vary from $0.70 to $1.03 per square foot. Better lawns have a lower CN and are thereby preferable from the viewpoint of being able to store more water. Similar estimates were made for the land valuation for low-density residential lots, commercial, apartments, and schools. A similar procedure was followed for these uses, except that the commercial, apartments, and schools are aggregated as one lot. However, they also cost more. The cost for each control was then estimated using these land valuations. The matrix of controls and land uses is presented in table 5.16. A linear programming model is used to find the least costly mix for each land use. See Heaney et al. (1999b) for a more detailed explanation of this method.

56

Table 5.15: Cost analysis of landscaping for medium density lot, 1/99$

Item A. Initial Capital Investment 1. Soil preparation Initial cost of sod Initial cost of topsoil, 6" Spreading topsoil, 6" Soil conditioners Sprinkler system

Input Data

2. Opportunity Cost of Land Land Investment Cost Opportunity cost investment rate Annual cost, $/yr. Interest rate per year Present worth over 25 years Cost in $/ft2 Total of initial capital investment B. Operation & Maintenance Costs, $ Lawn watering Inches per year % of pervious area that is irrigated Cost of water, $/1,000 gallons Present worth factor Present worth, $/ft2 Lawn maintenance Weeks per year $/week Maintenance area, ft2 Present worth, $/ft2 Sprinkler system maintenance Total operation and maintenance costs, $ C. Total Cost, $/ft2 Portion attributable to stormwater Assumed % D. Cost for Stormwater

Good $/ft2

Fair $/ft2

Poor $/ft2

$0.43 $0.50 $0.64 $0.03 $0.62 $2.22

$0.34 $0.40 $0.51 $0.02 $0.44 $1.71

$0.26 $0.30 $0.38 $0.01 $0.00 $0.95

$5.62 $7.84

$5.62 $7.33

$5.62 $6.57

$0.24

$0.15

$0.09

$0.98 $0.25 $1.46 $9.31

$0.50 $0.15 $0.80 $8.13

$0.35 $0.00 $0.44 $7.01

$0.93

$0.81

$0.70

$26,370 6% $1,582 0.06 $20,226

20 80% $1.50 12.78

26 $8.46 2880

10%

57

Table 5.16: Calculation of unit costs for controls, including opportunity costs for land, 1/99$ ID Aspen F Aspen G Driveway 1 Driveway 2 Grass F Grass G Grass P Parking 1 Parking 2 Parking 3 Parking 4 Patio 1 Patio 2 Roof 1 Roof 2 Sidewalk 1 Sidewalk 2 Storage Street 1 Street 2 Street 3 Street 4 Swales 1 Swales 2 Woods F Woods G

LD Res MD Res Commercial School Apartments RW50 RW60 RW70 $/sf $/sf $/sf $/sf $/sf $/sf $/sf $/sf $2.00 $2.00 $2.00 $2.00 $2.00 $2.00 $2.00 $2.00 $3.00 $3.00 $3.00 $3.00 $3.00 $3.00 $3.00 $3.00 $0.23 $0.23 $0.23 $0.23 $0.23 $0.23 $0.23 $0.23 $0.25 $0.25 $0.25 $0.25 $0.25 $0.25 $0.25 $0.25 $0.60 $0.60 $2.12 $2.49 $1.22 $0.60 $0.60 $0.60 $0.69 $0.69 $2.18 $2.56 $1.29 $0.69 $0.69 $0.69 $0.49 $0.49 $2.01 $2.38 $1.11 $0.49 $0.49 $0.49 $0.23 $0.23 $0.23 $0.23 $0.23 $0.23 $0.23 $0.23 $0.25 $0.25 $0.25 $0.25 $0.25 $0.25 $0.25 $0.25 $0.26 $0.26 $0.26 $0.26 $0.26 $0.26 $0.26 $0.26 $0.28 $0.28 $0.28 $0.28 $0.28 $0.28 $0.28 $0.28 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.00 $0.00 $0.00 $0.00 $0.00 $0.00 $0.00 $0.00 $1.50 $1.50 $1.50 $1.50 $1.50 $1.50 $1.50 $1.50 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $0.19 $5.00 $5.00 $5.00 $5.00 $5.00 $5.00 $5.00 $5.00 $0.25 $0.25 $0.25 $0.25 $0.25 $0.25 $0.25 $0.24 $0.26 $0.26 $0.26 $0.26 $0.26 $0.26 $0.26 $0.26 $0.27 $0.27 $0.27 $0.27 $0.27 $0.27 $0.27 $0.27 $0.29 $0.28 $0.28 $0.28 $0.28 $0.28 $0.29 $0.28 $3.00 $3.00 $3.00 $3.00 $3.00 $3.00 $3.00 $3.00 $6.00 $6.00 $6.00 $6.00 $6.00 $6.00 $6.00 $6.00 $0.80 $0.80 $0.80 $0.80 $0.80 $0.80 $0.80 $0.80 $1.40 $1.40 $1.40 $1.40 $1.40 $1.40 $1.40 $1.40

58

5.6 Optimization of Control Options for Happy Acres The results of the LP optimizations are summarized in tables 5.17 and 5.18. The results are allocated along functional grouping within each soil class in table 5.17, and aggregated for each land use type in table 5.18. The least cost design allocates the appropriate control option to the appropriate soil type and land use (soil is reflected in its predevelopment CN, land use is reflected in the influence of land valuation on the cost of the control). The changes in control options affect the appearance of the neighborhood, and this is evident by inspection of table 5.17. For example, porous pavements were selected (with curb and gutter) for the street design in the rocky soil. In the clay and silt soils where more percolation can take place, the LP model selected a street design with porous pavement and swales instead of curb and gutter. A similar allocation took place with parking areas; both were porous, however, the more permeable soils resulted in a design that had a higher infiltration capacity. The more permeable driveway, patio, and sidewalk choices were chosen in both soil types. Good grass was selected over the other options for all soil areas, except in commercial areas where poor was selected in silts and clays, and fair was selected in rock. This is due to the relatively small amount of landscaped area in commercial areas. There may be other aesthetic factors with commercial areas that would put a higher premium on a higher quality grass other than for a stormwater quality function. Aspens were chosen, but in small amounts, so it would not look significantly different than a typical subdivision. The roof choice remained the standard, rather the roof with detention, due to its relatively high unit cost. Storage was chosen when no other controls were feasible, the highest values, as expected, were in commercial areas with rocky soils, which would not have much infiltration capacity. The cost of the optimal solution for each soil class and land use is found in table 5.18. The total cost for the controls would be $5.2 million, some of which overlaps with money that would be spent for landscaping anyway. About half of this amount is used to attempt to control runoff from transportation related functions. What differs from a traditional subdivision development is the allocation of use. A traditional subdivision would have allocated everything in ground cover to the high quality grass, (particularly for commercial areas) and neglected the woods and aspens (although some exceptions to this exist, mainly for aesthetics). In commercial areas, the detention storage, would have been utilized. For sidewalks, patios, streets, and parking lots, nonporous pavement would have been chosen. Curb and gutter would have replaced swales along street rights of way. An important note here is that this DSS cannot dynamically change land uses. For example, the net amount of area used for rights of way, 39 acres out of the 106 total (see table 5.18), must remain the same. Likewise, the amounts and locations for medium density and low density, as well as the other land uses, must remain the same. What has been done here, however, is to attempt to allocate storage optimally throughout each of these land uses. A more general problem exists which would allow tradeoffs between the land uses. This problem is extremely complex because it involves re-creation of the GIS for each iteration.

59

Table 5.17: Results of LP optimization-land use allocation by function (includes opportunity costs) Land Use Area in Soil Group B in acres 50 60 70 LD MD HD Comm Sch Street 1 Street 2 Street 3 Street 4 Sidewalk 1 Sidewalk 2 Grass P Grass F Grass G Swales 1 Swales 2 Storage Parking 1 Parking 2 Parking 3 Parking 4 Roof 1 Roof 2 Driveway 1 Driveway 2 Patio 1 Patio 2 Woods F Woods G Aspen F Aspen G

Land Use Area in Soil Group D in acres 50 60 70 LD MD HD Comm Sch 11.38 0.00 2.74

9.69 1.08 0.03 2.42 0.21 0.01

0.50 0.00 0.50 0.16 0.38

3.03 0.26 0.01 11.23 6.49 0.57

0.65

1.12 0.00 1.12 0.00 4.60 0.00

1.00 0.06 0.00

0.24

0.83 0.00 0.83 0.00 0.35 0.00

0.04 1.72

0.14

1.19

0.00

2.34 3.23 1.08

2.04 1.03

1.59

0.00 6.13 0.00

0.94 1.21

0.00 2.30

0.47 0.40

0.00 0.77

0.25 0.79 0.36

0.00 9.20 0.00

60

0.00

2.15

0.00 0.00

1.32 5.97

0.00

0.00

Table 5.18: Least-cost LP solutions for land Use/BMP options (including land costs) for Happy Acres.

Land Use 50 ft ROW 60 ft ROW 70 ft ROW Low Density Residential Medium Density Residential Apartments Commercial School SUM Land Use 50 ft ROW 60 ft ROW 70 ft ROW Low Density Residential Medium Density Residential Apartments Commercial School

Soil B Area (acres)

Total (acres)

Soil D Area (acres) 15.15 1.55 0.05 15.02 12.97 3.73 3.37 3.43 55.27

17.78 0.00 4.35 0.00 21.57 0.00 7.67 0.00 51.37

32.92 1.55 4.41 15.02 34.54 3.73 11.04 3.43 106.64

Cost in Soil B, $ $443,554 $36,463 $1,058 $376,677 $361,197 $98,633 $39,267 $106,305 $1,463,153

Cost in Soil D, $ $1,484,917 $$247,981 $$1,509,515 $$517,237 $$3,759,650 TOTAL

Sum, $ $1,928,471 $36,463 $249,039 $376,677 $1,870,712 $98,633 $556,503 $106,305 $5,222,803 $5,220,000

5.7 Decision Support Systems and the Happy Acres Case Study The previous sections have illustrated how a simple hydrologic model can be constructed with basic GIS information. The methods presented in this report allow hydrologic and economic analysis to be performed on micro scales not traditionally used in urban analysis. These micro scales, although unfamiliar, must be used to properly evaluate BMPs for the control of locally generated stormwater runoff. This same information can be used as building blocks for SWMM. SWMM aggregates information in a manner controlled by the user, into an equivalent rectangular catchment. Several methods of aggregation are available within SWMM add-on packages (such as PCSWMM). Unfortunately, this method homogenizes the parcels within each subcatchment, i.e., they lose their unique hydrologic characteristics. The aggregation was typically done so that the user was not overwhelmed by data, as most had to be handled manually. However, within the context of a DSS, appropriate tools can be used to process the data, so smaller scales may be evaluated. A disadvantage of the DSS process used in this case study and outlined in figure 5.1 is that most of the analysis is one way, i.e., there is not a true interchange of information between the modules. The most obvious example is the GIS. It would be desirable to optimize land use in a general form of a land allocation model considering the effects of land valuation, soils, and control options. In order to do this efficiently, the spatial database underlying the parcel delineation must be re-created for each iteration of the model. Of course, this level of integration is also the most difficult and expensive.

61

6.0 Summary and Conclusions 6.1 Summary In summary, GIS has transformed our approach to the urban stormwater management problem. Not only are input parameters in the model itself becoming more easily obtainable, but also the scale of possible evaluations has decreased to a point that it is now possible to effectively evaluate source controls. The case study process shown in figure 5.1 provides a preliminary evaluation of the complex urban stormwater problem and the linked problem of allocation of land use. Several models exist that utilize GIS information; the degree of integration that is desirable remains debatable. Due to the widely disparate spatial scales involved, and the detailed amount of information available in a GIS, it is quite possible for the analyst to be drowned in data that may not be needed in evaluating the problem. The urban stormwater problem needs to be of primary concern to the analyst; rather than the micro maintenance of the GIS. The problem should be the primary focus, even more so than the model, or the database used. As the models evolve into more general Decision Support Systems, they will tend to become more data centered, and computational engines more interchangeable. The GIS data will become more available and standardized, and will be an important tool. One lesson to be learned from the 90s and the computer software explosion that has transformed the working world is that too much reliance on any one technology can lead to obsolescence. DSS promises to be the technology that links many of these tools together to enable the analyst to explore new challenging problems in old contexts. 6.2 Conclusions Advances in development of computer software have produced two key linked technologies: relational databases and geographic information systems. The combination of these two has affected the development of another technology, decision support systems, that has been applied to complex unstructured water resources and environmental problems. Most DSSs include these two technologies, with the addition of simulation models, an evaluation tool (can include optimization), and a graphical user interface. The graphical user interface, mainly the MSWindows interface, is another advance that has both transformed software as well changed the standard of model development. Construction of programs within this environment tends to be more difficult due to its object oriented architecture, however, it is also inherently more dynamic than constructing programs within older environments such as FORTRAN-77. This is primarily due to the advent of structured programming techniques that tend to keep data handling processes out of the main program files, which tends to advance a more data centric approach to modeling. The structured techniques also avoid the use of “spaghetti code” in which it is difficult to debug code due to vague loops and “GOTO” statements that branch the program in many different directions. New types of solvers are now available that can serve as better evaluation tools for a DSS. These include genetic algorithms (GA), simulated annealing (SA), and the relative ease with which linear programming (LP) solvers are used. These optimization tools allow rapid evaluation of both linear and nonlinear problems, which can assist the designer in finding the better or best solution.

62

Urban stormwater models have been created according to specific needs and available funding. The predominant US model, SWMM, was created in the late 60s and early 70s. There is an active user community for this largely public domain model. Several enhancements to the model, namely PC SWMM, Visual Hydro (XP SWMM), and MikeSWMM, are now available in the private domain as well. These enhancements contain facilities that include graphical user interfaces for ease of program use, GIS and CAD interfaces for construction of models based upon the best available system mapping, and external links to available databases to enhance the use of available system data. European models, in particular the DHI and the HR-Wallingford series, have been significantly ahead of the US modeling community in the use of GUIs and GIS. The reason for this gap is primarily the result of funding. Funding for urban stormwater modeling in the US ceased in the early 80s. Meanwhile, the European models were developed and enjoyed significant funding during the 80s and early 90s from both national governments as well as the European Union. These models may have become self-supporting by the creation of companies that sell the licensed product. This enables future enhancements in the models to be made, as well as user support from a centralized source. The US should focus its efforts on the use of linked technologies to take advantage of significant savings that can be realized by avoiding the re-creation of common tools currently available. For example, spreadsheet technology in the US has been effectively standardized upon MS Excel (even if you don’t use it, you use a program that can read these files). Input and output processing within new models could make use of this application, which would allow the user greater flexibility in terms of pre- and post-processing of model output. Visual Hydro provides a good example of the use of spreadsheet tools for data input and output. The US has been a leader in the GIS and database software development field; available links to these programs will continue to evolve and interfaces with GIS should become easier to construct than those at present. A significant portion of this effort is the development of both the graphic features of the GIS and the associated system attributes as well. The case study outlined in this report, although using a simplified hydrologic model, provides a possible outline of the use of this data for problems that have remained intractable to this point, for example, the selection of the appropriate BMP control technology for each parcel. Further work needs to be done to enhance the development of DSS technology to the urban stormwater field. The funding resources should carefully target the development of models and DSSs that link available tools rather than recreate them, and provide a common set of technologies that the user may combine with other available software. The funding should also seek to complement or prod the development of existing commercial software, rather than supplant the market by the introduction of competing products. A possible model could also be the European model community, in which the government funds the initial development of the model, then licenses it to a nonprofit company that markets and sells the model at a self-sustaining price. Care should be taken in that as the model interfaces become easier to run, they may be used inappropriately. A stated goal within the DSS community is to bring the computing power to the level of the decision-maker, rather than an intermediary. This works well if the decision-maker, or their assistant, is trained in the field of urban stormwater. The field of urban stormwater modeling involves the use of complex boundary conditions. Using GIS involves the use of wildly different scales where the uncertainty in the information may not be immediately evident

63

to the user. Such complex problems require a technically competent professional to carefully use and evaluate the information the DSS presents. Rather than simply using a sophisticated set of tools to solve the same problem more efficiently than we can at present, the problems evaluated will become more complex as well as the possible array of solutions to them. The advent of DSS and its inherent technologies, relational databases and GIS, have transformed the field of urban stormwater modeling and allow the evaluation of previously intractable problems.

64

7.0 References Azzout, Y., Barraud, S., Cres, F. N., and Alfakih, E. (1995) Decision Aids for Alternative Techniques in Urban Storm Management, Water Science and Technology, 32 (1): 41-48. Barbe, D.E., Miller, H., and Jalla, S. (1993) Development of a Computer Interface among GDS, SCADA and SWMM for Use in Urban Runoff Simulation. In Harlin, J.M and Lanfear, K.J. (eds.) Proc. of the Symposium on Geographic Information Systems and Water Resources. American Water Resources Association, Bethesda, MD. p. 113-120. Battin, A., Kinerson, R., and Lahlou, M. (1998) EPA's Better Assessment Science Integrating Point and Nonpoint Sources (BASINS)-A Powerful Tool for Managing Watersheds. Internet file retrieved 11/6/98 from The Center for Research in Water Resources, The University of Texas at Austin. http://www.crwr.utexas.edu/gis/gishyd98/epa/ battin/p447.htm. Bellal, M., Sillen, X., and Zech, Y. (1996) Coupling GIS with a Distributed Hydrological Model for Studying the Effect of Various Urban Planning Options on RainfallRunoff Relationships in Urbanized Watersheds. In Kovar, K., and Nachtnebel, H.P. (eds.) HydroGIS ’96: Application of Geographic Information Systems in Hydrology and Water Resources Management. International Association of Hydrologic Sciences Publication No. 235. IAHS Press, Wallingford, UK. p. 99106. Brown and Caldwell and Camp, Dresser and McKee (1997) Boulder Creek Watershed Study, Phase I, November, 1997. Prepared for the City of Boulder. Butler, D. and Maksimovic, C. (eds.) (1998) UDM ’98 Fourth International Conf. on Developments in Urban Drainage Modeling. Imperial College of Science, Technology & Medicine, London, UK. CAiCHE (1998) Visual Hydro Software, Tampa, FL. Charnock, T.W., Hedges, P.D., and Elgy, J. (1996) Linking multiple process models with GIS. In Kovar, K., and Nachtnebel, H.P. (eds.) HydroGIS ’96: Application of Geographic Information Systems in Hydrology and Water Resources Management. International Association of Hydrologic Sciences Publication No. 235. IAHS Press, Wallingford, UK. p. 29-36. Cluis, D., Martz, L., Quentin, E., and Rechatin, C. (1996) Coupling GIS and DEM to Classify the Hortonian Pathways of Non-point Sources to the Hydrologic Network. In Kovar, K., and Nachtnebel, H.P. (eds.) HydroGIS ’96: Application of Geographic Information Systems in Hydrology and Water Resources Management. International Association of Hydrologic Sciences Publication No. 235. IAHS Press, Wallingford, UK. p. 37-44. Computational Hydraulics International (CHI) (1998) PCSWMM, PCSWMM GIS Software, Guelph, Ontario, Canada. da Costa, J.R, Lacerda, M., and Jesus, H.B. (1995) The Portuguese Water Resources

Information System: Using OOP to Integrate Time Series and GIS. Proc. of 1995 ESRI User Conference. Internet file retrieved from http://www.esri.com/library/userconf/ proc95/to300/p296.html da Costa, J.R., Jesus, H.B., and Lacerida, M. (1996) Integrating GIS and Time Series Analysis for Water Resources Management in Portugal. In Kovar, K., and Nachtnebel, H.P. (eds.) HydroGIS ’96: Application of Geographic Information Systems in Hydrology and Water Resources Management. International Association of Hydrologic Sciences Publication No. 235. IAHS Press, Wallingford, UK. p. 289-297. Davis, J. R., Nanninga, P. M., Biggins, J., and Laut, P. (1991) Prototype Decision Support System for Analyzing Impact of Catchment Policies, Journal of Water Resources Planning and Management, 117 (4): 399-414. DeVantier, B.A. and Feldman, A.D. (1993) Review of GIS Applications in Hydrologic Modeling, Journal of Water Resources Planning and Management, 119 (2): 246261. Dewberry and Davis (1996) Land Development Handbook. McGraw-Hill, New York. DHI, Inc. (1998) Mouse, Mike, MikeSHE, NAM, WUS, MikeInfo, MikeSWMM Software, Copenhagen, Denmark. Dion, T.R. (1993) Land Development for Civil Engineers. Wiley-Interscience, New York. Driscoll, E. D. (1993) A Decision Support System for Highway Runoff Monograph: in: James, W., (Ed.) New Techniques for Modeling of Stormwater Quality Impacts, p. 101-122, Guelph, ON. ESRI (1998) Internet page: http://www.esri.com., Redlands, CA. Fankhauser, R. (1998) Automatic Determination of Imperviousness in Urban Areas from Digital Orthophotos. In UDM ’98, the Fourth International Conference on Developments in Urban Drainage Modeling, Butler, D. and Maksimovic, C. (Eds.) 21-24 September, 1998, London, UK, IAWQ/IAHR, pages 321-326. Federal Highway Administration (1996) Urban Drainage Design Manual, HEC No. 22, FHWA-SA-96-078, Washington, D.C. Fedra, K. and Jamieson, D. G. (1996) The ‘Water Ware’ Decision-Support System for River-Basin Planning. 2. Planning Capability, Journal of Hydrology, 177 (3/4), p. 177-198. Feinberg, D., and Uhrick., S.W. (1997) Integrating GIS with Water and Wastewater Hydraulic Models. Proc. of 1997 ESRI User Conference. Internet file retrieved from http://www.esri.com/library/userconf/proc97/PROC97/TO200/ PAP199/P199.HTM Fuchs, L., and Scheffer, C. (1996) Hydroinformatic Tools in Urban Drainage. In Seiker, F., and Verworn, H.R. (eds.). IAHR/IAWQ Proc. of the 7th Annual Conf. on Urban Storm Drainage. Hannover, Germany. Vols. III, p. 1587-1592. Fulp, T., Harkins, J., Williams, B., Vickers, B., King, D., Martin, K., Shiao, L (1994)

Decision Support for Water Resources Management in the Colorado River, United States Bureau of Reclamation; Boulder City, NV. Greene, R.G. and Cruise, J.F. (1995) Urban Watershed Modeling Using Geographic Information System, Journal of Water Resources Planning and Management, 121 (4): 318-325. Gregory, M. and James, W. (1995) Management of Time-Series Data for Long-Term, Continuous Stormwater Modeling. Chapter 8 in James, W. (Ed.) Advances in Modeling the Management of Stormwater Impacts, Computational Hydraulics International, Guelph, Ontario, Canada. Hallam, C.A., Salisbury, J.M., Lanfear, K.J., and Battaglin, W.A. (eds.) (1996) Proc. of the AWRA Annual Symposium: GIS and Water Resources. American Water Resources Association, Herndon, VA. Harlin, J.M and Lanfear, K.J. (eds.) (1993) Proc. of the Symposium on Geographic Information Systems and Water Resources. American Water Resources Association, Bethesda, MD. Hauber, S.M. and Joeres, E.F. (1996) Using a GIS for Estimating Input Parameters in Urban Stormwater Quality Modeling, Water Resources Bulletin 32(6), p. 13411352. Heaney, J. P., Wright, L., and Sample, D. (1998) Research Needs in Wet Weather Flows, Final Report of Project 96-IRM-1, Water Environment Federation, Alexandria, VA. Heaney, J.P., Sample, D., and Wright, L. (1999a) Costs of Urban Stormwater Systems, Report to the US EPA, Edison, NJ. Heaney, J.P., Wright, L., and Sample, D. (1999b) Innovative Methods for Optimization of Urban Stormwater Systems, Report to the US EPA, Edison, NJ. Hellweger, F. (1996) TABHYD-TR55 Tabular Hydrograph Method in ArcView. Internet file retrieved from The Center for Research in Water Resources, The University of Texas at Austin. http://www.crwr.utexas.edu/gis/gishyd98/runoff/webfiles/TABHYD/ TABHYD.HTM Hellweger, F. and Maidment, D. (1999) Definition and Connection of Hydrologic Elements using Geographic Data, Journal of Hydrologic Engineering, 4 (1): 1018. Herath, S., Musiake, K., and Hironaka, S. (1996) Development and Application of a GIS Based Distributed Catchment Model for Urban Areas. In Seiker, F., and Verworn, H.R. (eds.) IAHR/IAWQ Proc. of the 7th Annual Conf. on Urban Storm Drainage. Hannover, Germany. Vols. III, p. 1695-1700. Hora, J., Kuby, R., and Suchanek, M. (1998) Information Technologies and Hydroinformatic Tools in the Town of Pilsen. In Butler, D. and Maksimovic, C. (eds.) UDM ’98 Fourth International Conf. on Developments in Urban Drainage Modeling. Imperial College of Science, Technology & Medicine, London, UK.

p. 481- 486. HR-Wallingford, Inc. (1998) Hydroworks, Infoworks Software, Wallingford, Oxfordshire, UK. Huber, W.C. and Dickinson, R.E. (1988) Storm Water Management Model User's Manual, Version 4, EPA/600/3-88/001a (NTIS PB88-236641/AS), Environmental Protection Agency, Athens, GA, 1988. Jamieson, D. G. and Fedra, K. (1996a) The ‘Water Ware’ Decision-Support System for River-Basin Planning. 1. Conceptual Design, Journal of Hydrology, 177 (3/4), p. 163-175. Jamieson, D. G. and Fedra, K. (1996b) The ‘Water Ware’ Decision-Support System for River-Basin Planning. 3. Example Applications, Journal of Hydrology, 177 (3/4), p. 199-211. Kim, H.-B., Kim, K.-M., and Lee, J.-C. (1998) Sewer Alternative Generation Using GIS and Simulation Models in a Planning Support System. Internet file retrieved from the ESRI 1998 International User Conference, http://www.esri.com/library/ userconf/proc98/PROCEED/ Kopp, S.M. (1996) Linking GIS and Hydrologic Models: Where Have We Been, Where Are We Going? In Kovar, K., and Nachtnebel, H.P. (eds.) HydroGIS ’96: Application of Geographic Information Systems in Hydrology and Water Resources Management. International Association of Hydrologic Sciences Publication No. 235. IAHS Press, Wallingford, UK. p. 133-139. Kovar, K., and Nachtnebel, H.P. (eds.) (1996) HydroGIS ’96: Application of Geographic Information Systems in Hydrology and Water Resources Management. International Association of Hydrologic Sciences Publication No. 235. IAHS Press, Wallingford, UK. Lavallee, P., Marcoux, C., Bonin, R., (1996) Performance of an Integrated Real Time Control System: Application to CSOs Control, in WEF Urban Wet Weather Pollution, Controlling Sewer Overflows and Stormwater Runoff, p. 12.13-12-22, Alexandria, VA. Litman, T. (1998) Transportation Cost Analysis: Techniques, Estimates and Implications. Victoria Transport Policy Institute, Victoria, British Columbia. Makropoulos, C., Butler, D., and Maksimovic, C. (1998) A GIS Based Methodology for the Evaluation of Suitability of Urban Areas for Source Control Application. In Butler, D. and Maksimovic, C. (eds.) UDM ’98 Fourth International Conf. on Developments in Urban Drainage Modeling. Imperial College of Science, Technology & Medicine, London, UK. p. 59-66. Mark, O., van Kalken, T., Rabbi, K., and Kjelds, J. (1997) A Mouse GIS Study of the Drainage in Dhaka City. Proc. of the 1997 ESRI User Conference. Internet file retrieved from http://www.esri.com/library/userconf/proc97/ PROC97/TO500/PAP487/P487.HTM Mercado, R.M. (1996) Geographic Information Systems (GIS) and Image Processing for

Stormwater Management Modeling Using XPSWMM in Tallahassee, Florida. In Hallam, C.A., Salisbury, J.M., Lanfear, K.J., and Battaglin, W.A. (eds.) Proc. of the AWRA Annual Symposium: GIS and Water Resources. American Water Resources Association, Herndon, VA. p. 305-313. Meyer, S.P., Salem, T.H., and Labadie, J.W. (1993) Geographic Information Systems in Urban Storm-Water Management, Journal of Water Resources Planning and Management, 119 (2): 206-228. Miles, S.W. and Heaney, J.P. (1988) Better than “Optimal” Method for Designing Drainage Systems. Journal of Water Resources Planning and Management, 114(5): 477-499. Olivera, F, Reed, S., and Maidment, D. (1998) HEC-PrePro v.2.0: An ArcView PreProcessor for HEC's Hydrologic Modeling System. 1998 ESRI User's Conference, San Diego, CA. http://www.esri.com/library/userconf/proc98/PROCEED. HTM Olivera, F., Maidment, D.R., and Charbeneau, R.J. (1996) Spatially Distributed Modeling of Storm Runoff and Non-Point Source Pollution Using Geographic Information Systems (GIS). CRWR Online Report 96-4. Internet file retrieved from The Center for Research in Water Resources, The University of Texas at Austin. http://www.crwr.utexas.edu/gis/ gishyd98/library/olivera/ header.htm Palisade Corporation (1998) @Risk, Evolver Software, Newfield, NY. Prince Georges County (1997) Low-Impact Development Design Manual, Prince Georges County, Largo, MD, November, 1997. Pryl, K., Vanecek, S., and Vasek, P. (1998) Data Processing and Manipulation Tools Used for Urban Drainage Systems. In Butler, D. and Maksimovic, C. (eds.) UDM ’98 Fourth International Conf. on Developments in Urban Drainage Modeling. Imperial College of Science, Technology & Medicine, London, UK. p. 327- 332. R.S. Means (1996a) Heavy Construction Cost Data, 10th Annual Edition, R.S. Means Company, Inc., Kingston, MA. R.S. Means (1996b) Landscaping Unit Cost Data, R.S. Means Company, Inc., Kingston, MA. Refsgaard, J.C., Storm, B. and Refsgaard, A. (1995) Recent Developments of the Systeme Hydrologique Europeen (SHE) Towards the MIKE SHE. In Simonovic, S., Kundzewicz, Z., Rosbjerg, D., and Takeuchi, K. (eds.) Modeling and Management of Sustainable Basin-Scale Water Resource Systems. International Hydrologic Sciences Publication No. 231. IAHS Press, Wallingford, UK. p. 427-434. Reitsma, R. F. (1996) Structure and Support of Water-Resource Management and Decision-Making, Journal of Hydrology, 177 (3/4), p. 253-268. Reitsma, R. F., Zagona, E.A., Chapra, S.C., and Strzepek, K.M. (1996) Decision Support Systems (DSS) for Water Resources Management, chapter 33 in Mays, L.W.

(Ed.) Water Resources Handbook, McGraw-Hill, New York, NY. Ribeiro, C.T. (1996) Impact of Land Use on Water Resources: Integrating HSPF and a Raster-Vector GIS. In Kovar, K., and Nachtnebel, H.P. (eds.) HydroGIS ’96: Application of Geographic Information Systems in Hydrology and Water Resources Management. International Association of Hydrologic Sciences Publication No. 235. IAHS Press, Wallingford, UK. p. 349-356. Riley, A. (1998) Restoring Streams in Cities, A Guide for Planners, Policymakers, and Citizens. Island Press, Rodriguez, F., Andrieu, H., Creutin, J.D., and Raimbault, G. (1998) Relevance of Geographic Information Systems for Urban Hydrological Analysis. In Butler, D. and Maksimovic, C. (eds.) UDM ’98 Fourth International Conf. on Developments in Urban Drainage Modeling. Imperial College of Science, Technology & Medicine, London, UK. p. 333-340. Scarborough, R.W., and Yetter, C.H. (1998) A Case Study Evaluation of the ArcView/Non-point Source Module Interface in EPA’s BASINS 2.0 Watershed Model. Internet file retrieved from the 1998 ESRI International User’s Conference, http://www.esri.com/library/ userconf/proc98/PROCEED/ABSTRACT/A247.HTM Seiker, F., and Verworn, H.R. (eds.) (1996) IAHR/IAWQ Proc. of the 7th Annual Conf. on Urban Storm Drainage. Hannover, Germany. Vols. I, II, and III. Shamsi, U. M., Benner, S. P., and Fletcher, B. A. (1995) A Computer Mapping Program for Sewer Systems, chapter 7 in Advances in Modeling the Management of Stormwater Impacts, CHI: Guelph, Canada, W. James, Editor, 1995. Shamsi, U.M. (1997) SWMM Graphics. Chapter 7 in James, W. (Ed.), Advances in Modeling the Management of Stormwater Impacts, Volume 5. Computational Hydraulics International, Guelph, Ontario. p. 129-153. Shamsi, U.M. (1998) ArcView Applications in SWMM Modeling. Chapter 11 in James, W. (ed.) Advances in Modeling the Management of Stormwater Impacts, Volume 6. Computational Hydraulics International, Guelph, Ontario, Canada. p. 219233. Shamsi, U.M., and Fletcher, B.A. (1996) Arcview Applications in Stormwater and Wastewater Management. In Hallam, C.A., Salisbury, J.M., Lanfear, K.J., and Battaglin, W.A. (eds.) Proc. of the AWRA Annual Symposium: GIS and Water Resources. American Water Resources Association, Herndon, VA. p. 259-268. Simonovic, S., Kundzewicz, Z., Rosbjerg, D., and Takeuchi, K. (eds.) (1995) Modeling and Management of Sustainable Basin-Scale Water Resource Systems. International Hydrologic Sciences Publication No. 231. IAHS Press, Wallingford, UK. Singh, V. P. and Fiorentino, M. (1996) Geographical Information Systems in Hydrology, Kluwer Academic Publishers, Boston, MA. Sorensen, H.R., Kjelds, J.T., Deckers, F., and Waardenburg, F. (1996) Application of

GIS in Hydrological and Hydraulic Modeling: DLIS and MIKE11-GIS. In Kovar, K., and Nachtnebel, H.P. (eds.) HydroGIS ’96: Application of Geographic Information Systems in Hydrology and Water Resources Management. International Association of Hydrologic Sciences Publication No. 235. IAHS Press, Wallingford, UK. p. 149-156. Sotic, A., Despotovic, J., Petrovic, J., Babic, B., Djukie, A., and Prodanovioc, D. (1998) Hydroinformatic Approach in Sewer System Design-Kumodraz case study. In Butler, D. and Maksimovic, C. (eds.) UDM ’98 Fourth International Conf. on Developments in Urban Drainage Modeling. Imperial College of Science, Technology & Medicine, London, UK. p. 341-346. Sponemann, P., Beeneken, L., Fuchs, L., Prodanovic, D., and Schneider, S. (1996) Criteria for Geographic Information Systems Used in Urban Drainage. In Seiker, F., and Verworn, H.R. (eds.) (1996) IAHR/IAWQ Proc. of the 7th Annual Conf. on Urban Storm Drainage. Hannover, Germany. Vols. III. p. 1689-1694. Tchobanoglous, G. (1981) Wastewater Engineering: Collection and Pumping of Wastewater, McGraw-Hill, New York, New York. Tskhai, A., Shirokova, S., Konev, D., Koshelev, K., and Tskhai, T. (1995) GIS “Hydromonitoring” and Optimization Model of Enterprise Water Protection Activity. In Simonovic, S., Kundzewicz, Z., Rosbjerg, D., and Takeuchi, K. (eds.) Modeling and Management of Sustainable Basin-Scale Water Resource Systems. International Hydrologic Sciences Publication No. 231. IAHS Press, Wallingford, UK. p. 263-270. University of Texas (1998) Internet page: http://www. crwr.utexas.edu/ The Center for Research in Water Resources. Urban Land Institute (1989) Project Infrastructure Development Handbook. Urban Land Institute, Washington, D.C. VanGelder, P., and Miller, M. (1996) GIS as an Aid in the Evaluation of Drainage Facilities at the Albany County Airport, New York. In Hallam, C.A., Salisbury, J.M., Lanfear, K.J., and Battaglin, W.A. (eds.) Proc. of the AWRA Annual Symposium: GIS and Water Resources. American Water Resources Association, Herndon, VA. p. 295-303. Walsh, M.R. (1993) Toward Spatial Decision Support Systems in Water Resources, Journal of Water Resources Planning and Management, 119 (2): 158-169. Wang, Y, and James, W. (1997) Integration of US Army Corps of Engineers’ TimeSeries Data Management System with Continuous SWMM Modeling. Chapter 2 in James, W. (Ed.) Advances in Modeling the Management of Stormwater Impacts, volume 5, Computational Hydraulics International, Guelph, Ontario, Canada. Wolf-Schumann, U., and Vaillant, S. (1996) Time View: A Times Series Management System for GIS and Hydrologic Systems. In Kovar, K., and Nachtnebel, H.P. (eds.) HydroGIS ’96: Application of Geographic Information Systems in Hydrology and Water Resources Management. International Association of

Hydrologic Sciences Publication No. 235. IAHS Press, Wallingford, UK. p. 7987. Wright, L.T., Nix, S.J., Hassett, J.M. and Moffa, P.E. (1995) A Preliminary Urban Nonpoint Source Management Plan: A Modeling Approach. Chapter 4 in James, W. (ed.), Modern Methods for Modeling the Management of Stormwater Impacts. Computational Hydraulics International, Guelph, Ontario. p. 51-62. Xu, Z.X., Schultz, G.A., and Ito, K. (1998) GIS Application in a Watershed-Based Water Resources Management. In Butler, D. and Maksimovic, C. (eds.) UDM ’98 Fourth International Conf. on Developments in Urban Drainage Modeling. Imperial College of Science, Technology & Medicine, London, UK. p. 487-494. Xue, R.Z., and Bechtel, T.J. (1997) Integration of Stormwater Runoff and Pollutant Model with BMP Assessment Model Using ArcView GIS. Proc. of 1997 ESRI User Conference. Internet file retrieved 11/6/98 from http://www.esri.com/library/userconf/proc97/ PROC97/TO700/PAP656/P656.HTM Xue, R.Z., Bechtel, T.J., and Chen, Z. (1996) Developing a User-Friendly Tool for BMP Assessment Model Using a GIS. In Hallam, C.A., Salisbury, J.M., Lanfear, K.J., and Battaglin, W.A. (eds.) Proc. of the AWRA Annual Symposium: GIS and Water Resources. American Water Resources Association, Herndon, VA. p. 285294.

Appendix: Happy Acres Database

Table A-1: Parcel attributes Address 100 101 200 200 201 100 200 201 105 110 120 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 123 125 127 129 100 101 102 106 108 110 120 121 130 131 140 141 150 151 160 161 170 171 180 181 190 191 151 160 161 165

Street

Soil

Land Use

Alpine Street Alpine Street Cedar Street Ashmount Street Ashmount Street Highland Street Birch Avenue Birch Avenue Center Street Center Street Center Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Maple Street Oak Street Oak Street Oak Street Oak Street Oak Street Oak Street Oak Street Oak Street Oak Street Oak Street Oak Street Oak Street Oak Street Oak Street Oak Street Oak Street Oak Street Oak Street Oak Street Oak Street Oak Street Oak Street Acorn Street Acorn Street Acorn Street Acorn Street

Silt Silt Clay Rock Rock Rock Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Clay Clay Clay Clay

Apartments Apartments Commercial Commercial Commercial Commercial Commercial Commercial LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential LD Residential MD Residential MD Residential MD Residential MD Residential

Area SF 50320 112360 25957 154915 72968 80450 100139 46642 14235 18488 6844 15082 9927 11751 9742 11025 8744 11441 7667 12942 11518 11728 7707 12053 14291 17653 8015 13857 13778 11207 18674 15565 13029 14017 16758 19500 22449 14049 10172 11049 11131 11239 11681 11993 12611 12127 12680 12646 12749 13048 12818 12950 12886 13016 12955 13412 13618 14363 11552 6019 5286 3926 3853

Roof SF 0 46927 0 57707 0 0 95132 0 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 1600 1600 1600 1600

Parking Drive-ways, SF SF 37740 0 37343 0 24659 0 89462 0 69319 0 76427 0 0 0 44810 0 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 800 600 600 600 600

Patios SF 0 0 0 0 0 0 0 0 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 400 200 200 200 200

Impervious, SF 37740 84270 24659 147169 69319 76427 95132 44810 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 3200 2400 2400 2400 2400

Pervious SF 12580 28090 1298 7746 3648 4022 5007 1832 11035 15288 3644 11882 6727 8551 6542 7825 5544 8241 4467 9742 8318 8528 4507 8853 11091 14453 4815 10657 10578 8007 15474 12365 9829 10817 13558 16300 19249 10849 6972 7849 7931 8039 8481 8793 9411 8927 9480 9446 9549 9848 9618 9750 9686 9816 9755 10212 10418 11163 8352 3619 2886 1526 1453

Address 170 171 176 179 180 181 182 100 101 110 111 120 121 131 135 139 141 150 151 160 161 170 171 180 181 190 191 100 101 111 121 131 141 150 151 154 155 161 165 166 170 171 180 181 190 191 100 101 110 111 112 116 120 121 131 141 151 161 180 190 101 111 121 131 141 181 191 100 110 120 130

Street

Soil

Land Use

Acorn Street Acorn Street Acorn Street Acorn Street Acorn Street Acorn Street Acorn Street Ash Street Ash Street Ash Street Ash Street Ash Street Ash Street Ash Street Ash Street Ash Street Ash Street Ash Street Ash Street Ash Street Ash Street Ash Street Ash Street Ash Street Ash Street Ash Street Ash Street Ash-Acorn Connec Ash-Acorn Connec Ash-Acorn Connec Ash-Acorn Connec Ash-Acorn Connec Ash-Acorn Connec Ash-Acorn Connec Ash-Acorn Connec Ash-Acorn Connec Ash-Acorn Connec Ash-Acorn Connec Ash-Acorn Connec Ash-Acorn Connec Ash-Acorn Connec Ash-Acorn Connec Ash-Acorn Connec Ash-Acorn Connec Ash-Acorn Connec Ash-Acorn Connec Birch Avenue Birch Avenue Birch Avenue Birch Avenue Birch Avenue Birch Avenue Birch Avenue Birch Avenue Birch Avenue Birch Avenue Birch Avenue Birch Avenue Birch Avenue Birch Avenue Cedar Street Cedar Street Cedar Street Cedar Street Cedar Street Cedar Street Cedar Street Elm Street Elm Street Elm Street Elm Street

Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay

MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD

Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential

Area SF 5543 3926 5800 3926 4788 3926 4783 5750 6785 6600 6765 6620 6744 6724 6703 6683 6662 3919 6642 4481 6621 4763 6601 4878 6581 4326 6560 3127 3180 3039 3157 2994 3086 4739 3157 5648 3109 3089 3149 5648 4630 3349 4818 2948 4551 2686 6469 6554 6477 6522 6484 6492 6499 6490 6457 6425 6360 6328 6560 6568 6572 6580 6588 6595 6603 6663 6671 6481 6448 6416 6384

Roof SF 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600

Parking Drive-ways, SF SF 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600

Patios SF 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200

Impervious, SF 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400

Pervious SF 3143 1526 3400 1526 2388 1526 2383 3350 4385 4200 4365 4220 4344 4324 4303 4283 4262 1519 4242 2081 4221 2363 4201 2478 4181 1926 4160 727 780 639 757 594 686 2339 757 3248 709 689 749 3248 2230 949 2418 548 2151 286 4069 4154 4077 4122 4084 4092 4099 4090 4057 4025 3960 3928 4160 4168 4172 4180 4188 4195 4203 4263 4271 4081 4048 4016 3984

Address 140 150 160 170 106 101 111 120 140 141 150 151 100 101 120 121 141 161 100 101 121 140 100 101 120 141 100 101 120 141 101 100 101 110 111 120 121 130 131 140 141 150 151 156 160 161 165 166 170 171 180 181 190 191 193 101 110 120 130 140 150 156 158 160 170 180 190 161 130 170 190

Street

Soil

Land Use

Elm Street Elm Street Elm Street Elm Street Forest Avenue Main Street Main Street Main Street Main Street Main Street Main Street Main Street Street A Street A Street A Street A Street A Street A Street B Street B Street B Street B Street C Street C Street C Street C Street D Street D Street D Street D Street E Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Sycamore Street Ashmount Street Ashmount Street Ashmount Street Ashmount Street Ashmount Street Ashmount Street Ashmount Street Ashmount Street Ashmount Street Ashmount Street Ashmount Street Ashmount Street Main Street Street A Street A Street A

Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Clay Rock Rock Rock Rock Rock Rock Rock Rock Rock Rock Rock Rock Rock Rock Rock Rock

MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD

Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential

Area SF 6351 6319 6286 6254 6428 4993 5154 6770 6636 6323 4939 6323 5072 4644 5072 4789 4934 5079 4787 4953 4953 4787 5609 4737 5609 4888 5254 5461 5254 5461 5192 6480 6511 6460 6712 6439 6470 6419 6492 6399 6514 6378 6536 6358 6337 6558 6580 6317 6296 5931 6276 5744 6255 6274 5919 6649 5611 5524 6461 6805 6624 6875 6554 6693 6533 6461 5691 6323 5072 5072 5072

Roof SF 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600

Parking Drive-ways, SF SF 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600

Patios SF 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200

Impervious, SF 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400

Pervious SF 3951 3919 3886 3854 4028 2593 2754 4370 4236 3923 2539 3923 2672 2244 2672 2389 2534 2679 2387 2553 2553 2387 3209 2337 3209 2488 2854 3061 2854 3061 2792 4080 4111 4060 4312 4039 4070 4019 4092 3999 4114 3978 4136 3958 3937 4158 4180 3917 3896 3531 3876 3344 3855 3874 3519 4249 3211 3124 4061 4405 4224 4475 4154 4293 4133 4061 3291 3923 2672 2672 2672

Address 141 160 180 181 190 191 160 161 171 190 191 180 181 190 191 100 120 151 171 190 191 126 130 136 140 150 160 170 171 181 191 193 151 155 161 165 171 175 179 101 111 121 131 141 151 176 180 181 190 191 193 195 201 221 231 241 244 250 251 254 260 261 270 274 280 281 290 291 100 101 110

Street

Soil

Land Use

Street B Street B Street B Street B Street B Street B Street C Street C Street C Street C Street C Street D Street D Street D Street D Street E Street E Street E Street E Street E Street E Birch Avenue Birch Avenue Birch Avenue Birch Avenue Birch Avenue Birch Avenue Birch Avenue Birch Avenue Birch Avenue Birch Avenue Birch Avenue Cedar Street Cedar Street Cedar Street Cedar Street Cedar Street Cedar Street Cedar Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Elm Street Forest Avenue Forest Avenue Forest Avenue

Rock Rock Rock Rock Rock Rock Rock Rock Rock Rock Rock Rock Rock Rock Rock Rock Rock Rock Rock Rock Rock Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt

MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD MD

Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential

Area SF 4953 4787 4787 4953 4787 4953 5609 5039 5189 5609 5340 5254 5461 5254 5461 6520 6520 5363 5533 6520 5704 6507 6515 6522 6530 6537 6545 6552 6345 6939 7911 5095 6610 6618 6625 6633 6641 6648 6656 6663 6667 6671 6676 6680 6684 6070 6675 6688 6941 6693 4843 4131 6416 6106 6452 6627 6706 6894 6665 6256 6865 6682 6463 6886 6909 6699 6765 6716 6312 7572 6424

Roof SF 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600

Parking Drive-ways, SF SF 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600

Patios SF 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200

Impervious, SF 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400

Pervious SF 2553 2387 2387 2553 2387 2553 3209 2639 2789 3209 2940 2854 3061 2854 3061 4120 4120 2963 3133 4120 3304 4107 4115 4122 4130 4137 4145 4152 3945 4539 5511 2695 4210 4218 4225 4233 4241 4248 4256 4263 4267 4271 4276 4280 4284 3670 4275 4288 4541 4293 2443 1731 4016 3706 4052 4227 4306 4494 4265 3856 4465 4282 4063 4486 4509 4299 4365 4316 3912 5172 4024

Address 111 120 130 140 141 150 151 160 161 170 171 180 181 186 190 191 200 201 205 210 211 220 221 230 231 240 241 250 251 261 270 271 280 281 290 291 293 121 125 100 101

Street

Soil

Land Use

Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Forest Avenue Main Street Center Street Walnut Street Walnut Street

Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt Silt

MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential MD Residential School School School

Area SF 6971 6294 6313 6353 6998 6333 6875 6372 6694 6392 6619 8120 6724 6312 6079 6599 6558 6500 6389 6562 6266 6566 6326 6570 6133 6575 6025 6579 6193 6379 6583 6169 6587 5411 3196 5894 3230 5200 8600 97601 43206

Roof SF 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 1600 0 69080 0

Parking Drive-ways, SF SF 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 8600 0 0 0 43206 0

Patios SF 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 0 0 0

Impervious, SF 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 2400 8600 69080 43206

Pervious SF 4571 3894 3913 3953 4598 3933 4475 3972 4294 3992 4219 5720 4324 3912 3679 4199 4158 4100 3989 4162 3866 4166 3926 4170 3733 4175 3625 4179 3793 3979 4183 3769 4187 3011 796 3494 830 2800 0 28521 0

Table A-2: Right of way attributes Street Name

Acorn Street Alpine Street Ash Street Ash-Acorn Connector Ashmount Street Ashmount Street ext. Aspen Street Birch Avenue Cedar Street Center Street Elm Street Forest Avenue Highland Street Main Street Maple Street Oak Street Street A Street B Street C Street D Street E stub between Elm and Forest Sycamore Street Walnut Street Total

RW width, ft 50 50 50 50 50 50 50 50 50 60 50 50 50 70 50 50 50 50 50 50 50 50 50 50

RW Area, length, sf ft 1640 81990 1125 56272 1205 60251 844 42214 870 43492 1620 80981 851 42537 2574 128701 2899 144940 1124 67445 2639 131944 2622 131119 831 41568 2741 191895 2153 107667 1751 87540 490 24491 465 23267 517 25829 415 20756 397 19875 519 25951 1086 1167

54281 58349 1693357