Model Selection and Recommendation Report for Central Oahu Watershed Study, Phase 2 Part A NOVEMBER 2008
Prepared for: U.S. Army Corps of Engineers, Honolulu District and
City and County of Honolulu, Department of Environmental Services
Prepared by: PB Americas, Inc. Honolulu, HI and
Northwest Hydraulic Consultants, Inc. Seattle, WA
Central Oahu Watershed Study Phase 2 Part A
EXECUTIVE SUMMARY The U.S. Army Corps of Engineers and the City and County of Honolulu, Department of Environmental Services have partnered to sponsor a study to select and develop a watershed model for the Waikele Stream watershed on the island of Oahu. This work is being accomplished under the Central Oahu Watershed Study – Phase 2 project. The purpose of this report is to document the model selection, screening, and evaluation process to ultimately recommend a suitable watershed model for application to Waikele Stream. The model’s primary intended use is to perform planning-level analyses for management practices, including evaluating the effect of land use changes and implementation of best management practices. In addition, the selected model will be used as a tool to assist in compliance with current and future Total Maximum Daily Load regulations. The selection process was initiated by compiling a broad list of 47 candidate models, each of which was categorized in one of three complexity categories (high complexity, moderate complexity, and low level complexity). These candidate models were then evaluated in each of thirteen specific model capability categories. A systematic ranking procedure was developed to assign a composite numerical rating to each model under consideration. Based on this screening process, seven watershed models were short-listed for further detailed evaluation. These included: HSPF and WARMF (high complexity category), AnnAGNPS and SWAT (moderate complexity category), and N-SPECT, PLOAD, and WTM (low level complexity category). Consideration of additional features/model capabilities, software demonstrations, and continued refinement of the relative rankings led to the selection of N-SPECT, SWAT, WARMF, and HSPF as the top-rated watershed models. These four models were then developed for pilot testing, including additional demonstrations on model development, data manipulation, and output generation. Based on feedback received during the pilot testing demonstrations, three options were presented to the U.S. Army Corps of Engineers and the City and County of Honolulu for consideration. Since each option had its share of advantages and disadvantages associated with them, a summary was prepared to assist in the decision-making process. Based on the needs evaluation and research conducted, our recommendation for Part A is to develop a calibrated WARMF model for application to the Waikele Stream watershed. This option was selected as the most favorable for the following reasons: WARMF is a physically based model that can provide detailed and accurate simulation results, meeting the City and County’s need to develop results that are scientifically defensible, WARMF is capable of producing detailed output which can be easily compared to HSPF output, and knowledge transfer through hands-on training would enable the City and County to verify the model’s suitability for application to additional watersheds on Oahu.
W9128A-05-D-0002
i
Central Oahu Watershed Study Phase 2 Part A
TABLE OF CONTENTS Introduction ...............................................................................................................................1 Project Location and Scope .....................................................................................................1 Model Selection Methodology..................................................................................................2 Identify Model Selection Criteria (Step 1)................................................................................3 Identify Candidate Models, Establish a Rating System, and Develop Screening Matrix (Steps 2-4) .................................................................................................................................4 Identify Seven Short-Listed Models (Step 5)...........................................................................6 Develop Four Top-Rated Models for Pilot Testing (Step 6)....................................................7 N-SPECT ................................................................................................................................9 SWAT ...................................................................................................................................10 WARMF ................................................................................................................................11 HSPF ....................................................................................................................................12 Pilot Model Comparison .........................................................................................................13 Evaluate Alternatives (Step 7)................................................................................................16 Option #1: .............................................................................................................................16 Option #2: .............................................................................................................................17 Option #3: .............................................................................................................................18 Recommend Model(s) (Step 8) ...............................................................................................19 References ..............................................................................................................................21
Appendix A – Field Visit Appendix B – Field Visit Route Map Appendix C – Agency Outreach / Kickoff Meeting Appendix D – Discussion on Short-Listed Models Presentation – June 9, 2008 Appendix E – Meeting Summary Notes Appendix F – Data Assessment Appendix G – Example Output From Uncalibrated SWAT Model Appendix H – Example Output From Uncalibrated WARMF Model Appendix J – Example Output From Uncalibrated HSPF Model
W9128A-05-D-0002
ii
Central Oahu Watershed Study Phase 2 Part A
INTRODUCTION The U.S. Army Corps of Engineers (USACE) and the City and County of Honolulu, Department of Environmental Services (ENV) have partnered to sponsor a study to select and develop a watershed model for the Waikele Stream watershed on the island of Oahu. The primary purpose of this model is to perform planning-level analyses for management practices, including evaluating the effect of land use changes and implementation of best management practices within the watershed. In addition, ENV intends to use the model as a tool to assist in compliance with future Total Maximum Daily Load (TMDL) requirements. ENV is also considering applying the selected watershed model to other basins on Oahu. This work is being accomplished under the Central Oahu Watershed Study – Phase 2 Project (Phase 1 of this study was completed in 2007 [Oceanit, 2007]). The purpose of this report is to document the identification of model criteria and potential candidate models and the subsequent process of model evaluation, ranking, and demonstration to narrow down the model list and provide a recommendation for developing a watershed model for Waikele Stream. PROJECT LOCATION AND SCOPE The focus of this Phase 2 study is on the Waikele watershed (See Figure 1), which is the second largest watershed on Oahu encompassing approximately 46 square miles. The watershed is bound by the Waianae mountain range to the west and the Koolau mountain range to the east and includes the Waikakalaua and Kipapa tributaries. The Phase 2 study is being conducted in two parts; Part A and Part B. This memorandum documents and summarizes the tasks completed in Part A which consist of the following: agency outreach / kickoff meeting, field visit / tour of study location, data inventory / assessment, candidate model evaluation, and model recommendation. Details of the two-day field visit conducted on April 8-9, 2008 are included in Appendix A and Appendix B. The agency outreach effort and kickoff meeting notes and list of attendees are provided in Appendix C. A presentation for discussing the seven short listed models is provided in Appendix D. Appendix E contains summary conference call, meeting, and model demonstration notes. Appendix F contains a summary of the observed data collected from available sources. Part B of this Phase 2 study will consist of detailed development of the watershed model(s) selected in this Part A process.
W9128A-05-D-0002
1
Central Oahu Watershed Study Phase 2 Part A
Figure 1. Location Map MODEL SELECTION METHODOLOGY The model selection process consisted of developing a systematic approach for determining a preferred candidate watershed model for final consideration by USACE and ENV. The multistep process included evaluating various models’ capabilities and synthesizing those capabilities with pre-determined selection criteria. The general methodology is shown in Figure 2. Steps 1 through 8 are documented in greater detail in the following sections. The last step, final selection of a watershed model, will be made by ENV and USACE based on information in this report.
W9128A-05-D-0002
2
Central Oahu Watershed Study Phase 2 Part A
1. Identify Model Selection Criteria
2. Identify Candidate Models
3. Establish a Rating System
4. Develop Screening Matrix
5. Identify Seven Short-Listed Models
6. Develop Four TopRated Models for Pilot Testing
7. Evaluate Alternatives
8. Recommend Model(s)
9. Final Selection by ENV and USACE
Figure 2. Model Selection Flow Chart IDENTIFY MODEL SELECTION CRITERIA (STEP 1) An initial list of model capabilities was developed to provide the basis for assessing and prioritizing USACE/ENV’s needs and evaluating the applicability of various models in meeting those needs. The selection criteria were based on the following model capabilities: Hydrologic modeling capabilities Continuous event modeling (i.e., simulated results every hour vs. results averaged over an event or season) Mixed land use modeling Ability to simulate sediment erosion and deposition Ability to simulate nutrient pollutants (e.g., nitrogen) BMP evaluation Linkage to GIS Graphical user interface for pre-processing (e.g., Windows based input data editors) Post-processing tools (e.g., printing tables and graphs directly from the model software) Well-defined user documentation Widespread use and acceptability of model (e.g., used nationally for TMDL modeling) Model availability (public domain vs. proprietary, software cost, etc.) Requirements for user modeling experience Additional factors that were also considered in the initial screening include: Ability to simulate chemical pollutants (e.g., pesticides) Level of user effort (e.g., few required parameters) Requirements for model maintenance (data updates and input, further calibration as data are available, etc.) Streamflow modeling
W9128A-05-D-0002
3
Central Oahu Watershed Study Phase 2 Part A
IDENTIFY CANDIDATE MODELS, ESTABLISH A RATING SYSTEM, AND DEVELOP SCREENING MATRIX (STEPS 2-4) Based on the model capabilities needed by ENV, a broad list of potential watersheds models was compiled. The list was developed based on publications (AquaTerra, 2002; EPA, 1997; EPA, 2005; Ward and Benaman, 1999), online research, recommendations from agencies, and experience by nhc staff. A ranking system for each model capability was developed to provide a quantifiable base for comparing the initial 47 models by assigning a rating of high, moderate, or low (or none when appropriate) and presenting the results within a matrix (See Figure 3). It should be noted that ratings in Figure 3 are approximate and that models with the same rating for a given criterion may have different levels of competency. A more refined rating system was not considered to be useful for purposes of the screening which was oriented toward creating groups of models with approximately similar capabilities and judging aggregate model capability rather than precise capability with respect to a single criterion. In addition, the matrix evolved during Part A as additional research was conducted (See section on Seven Short-Listed Models). Specific model criteria were evaluated relative to the models within the same complexity grouping (e.g., high, moderate, or low complexity). Thus, a low rating for BMP evaluation in a complex model may equate to a higher rating in a very simple model if the power of the BMP algorithm is on a par with the simple model’s most sophisticated algorithms.
W9128A-05-D-0002
4
Central Oahu Watershed Study Phase 2 Part A Rating of model capabilities:
Low Moderate High None or not evaluated
Use with Minimal Data/Modeler Experience
Public Domain Availability
Widely Used/Accepted
Well Defined User Documentation
Post-Processing
Graphical User Interface
Linkage to GIS
BMP Evaluation
Simulate Nutrients
Simulate Sediment
Mixed Land Use Modeling
Event Modeling (not necessarily continuous)
Hydrologic Modeling
Model ALAWAT AnnAGNPS ANSWERS/ANSWERS2000 APEX CH3D-WES/CH3D-SED CREAMS DR3M-QUAL DWSM EFDC EPIC ETD FHWA GLEAMS GSSHA GSTAR GWLF/BasinSim HEC-6 HEC-HMS HSPF (BASINS)/WinHSPF HUMUS KINEROS (BASINS) LSPC LWWM MIKE SHE (incl. SHE/SHESED) NLEAP NSPECT OWLS PLOAD (BASINS) PRMS QUAL2K/QUAL2E (BASINS) SAM SITEMAP SLAMM/WinSLAMM SLOSS-PHOSPH SLURP SPARROW (incl. USGSREGR) STORM SWAT (BASINS)/AGWA-SWAT SWMM (EPA, PC, XP) WARMF WASP WATERSHED WEPP WMM WMS WTM WWHM/BAHM
Notes
Model Capabilities
1 2
1 2 2 2 1 1
2
2 1
2
3
Notes: 1 Hydraulic, bay, or estuary model and requires flows/runoff be calculated externally from the program. Model capabilities not evaluated. 2 Field scale agricultural model. Model capabilities not evaluated. 3 Pre- and post-processing tool for watershed models. Model capabilities not evaluated.
Figure 3. Initial Model Screening Matrix
W9128A-05-D-0002
5
Central Oahu Watershed Study Phase 2 Part A
IDENTIFY SEVEN SHORT-LISTED MODELS (STEP 5) The short-listing process consisted of identifying a manageable number of top candidate models from the initial screening list of 47 potential models. Each model under consideration was defined in one of three complexity categories: high, moderate, or low level. The models classified as high and moderate are generally physically based models which require extensive input parameters. Models of low complexity are generally simpler, non-physically based and require fewer input parameters. Typically they have little predictive power and are best used in making qualitative or semi-quantitative comparisons of different scenarios. A composite rating for each model was then determined by weighting each model capability in the categories of technical capability, ease of use, and a combination of these two. The top two candidate models in each of the three complexity categories (high, moderate, and low level) were selected. The following two models were selected for the high complexity category: Hydrologic Simulation Program FORTRAN (HSPF) Watershed Analysis Risk Management Framework (WARMF) The following two models were selected for the moderate complexity category: Annualized Agricultural Non-Point Source Pollutant (AnnAGNPS) Soil and Water Assessment Tool (SWAT) The following three models were selected for the low complexity category: NonPoint Source Pollution and Erosion Comparison Tool (N-SPECT) Pollutant Load Application (PLOAD) Watershed Treatment Model (WTM) The selection of WTM was based on a recommendation from Alexandre Remnek of the State Department of Health (DOH) to illustrate a very simple spreadsheet model. Progressing through the selection process, it became evident that USACE/ENV were considering supplemental criteria in addition to those previously discussed in the decision making process. These include: Possible use of the watershed model by other City agencies (primarily DPP), Presence of local technical support, Assurance that the results would be scientifically defensible, and Laying the foundation for future collaboration with the Department of Health. These factors were taken into consideration for the evaluation of alternatives and final recommendation. At the request of ENV, HEC’s Hydrologic Modeling System (HEC-HMS) was included in the Pilot Testing for discussion purposes despite lacking several critical model criteria such as simulating sediment transport and water quality. To aid ENV in gaining an understanding of the level of effort to develop or make modifications to a model, an overview of the seven top-rated candidate models was presented to USACE, ENV, W9128A-05-D-0002
6
Central Oahu Watershed Study Phase 2 Part A
DOT-Highways Division, and Townscape on June 9, 2008. The primary purpose of the presentation was to provide a brief overview of each of the seven short-listed models’ data input requirements, software capabilities (strengths and weaknesses), output format, and level of difficulty for users to operate. The presentation also allowed for side-by-side comparison of the models being considered and provided the basis for discussions on determining the appropriate models to proceed with pilot testing. A copy of the presentation is included in Appendix D. All models considered for the short-list are available in the public domain and have readily available manuals that can be accessed for further model definition. After further research, the assessments for the SWAT and WARMF models were revised. SWAT was initially categorized in a moderately complex category but was reclassified as more complex relative to the WARMF model. This reclassification was primarily due to the accessibility of the model parameters via the model interface rather than the model technical capabilities alone. The revisions to the matrix for the seven short-listed models, with changes for SWAT and WARMF highlighted in yellow, can be seen in Figure 4. Rating of model capabilities:
Low Moderate High None or not evaluated
Use with Minimal Data/Modeler Experience
Public Domain Availability
Widely Used/Accepted
Well Defined User Documentation
Post-Processing
Graphical User Interface
Linkage to GIS
BMP Evaluation
Simulate Nutrients
Simulate Sediment
Mixed Land Use Modeling
Event Modeling (not necessarily continuous)
Hydrologic Modeling
Model AnnAGNPS HSPF (BASINS)/WinHSPF NSPECT PLOAD (BASINS) SWAT (BASINS)/AGWA-SWAT WARMF WTM
Notes
Model Capabilities
Figure 4. Revised Evaluation Matrix for Seven Short-Listed Models With the assistance of Mr. Michael Wong, Hydrologist and Acting Chief, Civil and Technical Works Branch, USACE, the seven short-listed models were narrowed down to four and included at least one model from each complexity category. It was decided that N-SPECT, SWAT, WARMF, and HSPF would be carried forward to the pilot testing phase. A summary of the June 13, 2008 teleconference discussing the seven short-listed models is provided in Appendix E. DEVELOP FOUR TOP-RATED MODELS FOR PILOT TESTING (STEP 6) Each of the four models (N-SPECT, WARMF, SWAT and HSPF) selected for pilot testing was developed for the Waikele Stream watershed using default model coefficients and readily available input data sets. The intent of these pilot models was to demonstrate to ENV the functionality and usability of model interfaces for model development from GIS, parameter refinement, simulation, calibration and access to output. Common data-sources were used to W9128A-05-D-0002
7
Central Oahu Watershed Study Phase 2 Part A
facilitate comparisons of model structures and output format when a source was required by more than one model (i.e. NOAA C-CAP land use, NRCS soils, USGS NED DEM and meteorology; requirements as shown in Table 1 and Table 2, in order of increasing complexity). Table 1. Required Spatial Data Sets for Model Development Parameter
N-SPECT
SWAT
WARMF
HSPF
DEM1
Grid
Grid
Grid
Grid
Land Use
Grid
Grid
Shape
Grid
Soils
Shape
Grid
Manual
Shape or Grid
R-Factor2
Grid
N/A
Grid
N/A
1
Digital Elevation Model Rainfall and runoff factor
2
Table 2. Required Meteorological and Air Quality Datasets For Water Quality Modeling with Maximum Time Step Parameter
N-SPECT
SWAT
WARMF
HSPF
Temperature
N/A
Daily
Daily
Daily
Precipitation
Annual or Event (Grid)
Daily or Hourly
Daily
Hourly
Other Meteorological
N/A
Daily
Daily
Daily
Air Quality
N/A
N/A
Daily
N/A
Output from all four models was generated using 2005 NOAA C-CAP land use and subsequently compared to observed data at the Waikele Stream at Waipahu USGS gage (Site Number 16213000) during March 1991 when precipitation, flow, and water quality data were all available (Appendix F). Model comparisons to gage data were made for demonstration purposes rather than as a means of ranking the accuracy of the models against one another since no attempt had been made to calibrate the models to gage data. As a demonstration of assessing the impacts of a proposed land use change, a future conditions scenario was simulated for each of the four pilot models that would hypothetically change an upper portion of the watershed to impervious. This change is implemented in a slightly different fashion for each of the models, and it was important to demonstrate the process of developing and running such a scenario because land use change analysis is one of the key model functions that ENV wishes to utilize. Notes on development of the four pilot models (N-SPECT, SWAT, WARMF, and HSPF) and some of the talking points from the July 18, 2008 and July 29, 2008 presentations are provided in the following text. This includes a description of how to make a change in the model for future land use conditions. A summary of the July 18, 2008 demonstration and July 29, 2008 and August 6, 2008 discussions on the four pilot models and recommendations for Part B are provided in Appendix E. W9128A-05-D-0002
8
Central Oahu Watershed Study Phase 2 Part A
N-SPECT Of the four pilot models, N-SPECT utilizes the simplest methods for performing calculations and is also the most straightforward to apply. The N-SPECT model is the only one of the pilot models that uses a grid based framework to perform calculations and report results (See Figure 5).
Figure 5. Screenshot of N-SPECT interface within ArcMap As part of the development of the N-SPECT pilot model for Waikele stream, an r-factor and annual precipitation grid were developed. The r-factor data was digitized from 1954 contour maps produced by the USGS for the entire island of Oahu. The annual precipitation data was acquired in a grid format from PRISM at Oregon State University. The future scenario land use change was performed in N-SPECT by drawing a polygon of land use change and running an N-SPECT calculation. The impact of the change was assessed by querying a grid cell at the Waikele Stream at Waipahu USGS gage; the reported value was 91,000,000 kg of sediment as compared to 144,000,000 kg in the un-calibrated existing conditions model and 16,000,000 kg observed by the USGS during the March 1991 event selected for demonstration. The noted reduction in sediment as a result of increased imperviousness in the future conditions scenario was attributed to the fact that N-SPECT does not account for in-channel processes, a shortcoming of the model. It should be noted that these results were generated from an uncalibrated model, and as such, their accuracy, or lack thereof, is of little practical significance. The results are provided to demonstrate the type of results that could be presented using the model rather than actual model results.
W9128A-05-D-0002
9
Central Oahu Watershed Study Phase 2 Part A
SWAT Similar to all of the other non-grid based pilot models, the SWAT model framework is based on linkages of sub-basins and streams. The ArcSWAT toolbar, an add-on for ArcMap GIS, includes functions for sub-basin delineation using a DEM and existing stream data. The existing stream data provided by ENV was cleaned for topological correctness for use in the SWAT pilot model demonstration. Ten sub-basins were delineated for the demonstration, six of which have outlets at USGS gages with historical water-quality data (See Figure 6. Screenshot of ArcSWAT Toolbar interface within ArcMap).
Figure 6. Screenshot of ArcSWAT Toolbar interface within ArcMap The future scenario land use change was performed in SWAT by making a copy of the SWAT project database and redefining the Hydrologic Response Units using a refined land use dataset. The impact of the change was assessed by plotting simulated existing and future scenario suspended sediment and flow for Waikele Stream at Waipahu USGS gage during the March 1991 event selected for demonstration (see Appendix G, Figures G1 – G3). Total event existing and future suspended sediment loads of 1,300,000 kg and 2,000,000 kg were tabulated in EXCEL. It should be noted that these results were generated from an uncalibrated model and as such, their accuracy, or lack thereof, is of little practical significance. The results are provided to demonstrate the type of results that could be presented using the model rather than actual model results. As part of the development of the SWAT pilot model for Waikele stream, the SWAT database was updated with soil attribute data for the entire island of Oahu.
W9128A-05-D-0002
10
Central Oahu Watershed Study Phase 2 Part A
WARMF WARMF was initially categorized in a highly complex category but as discussed earlier, it was reclassified as moderately complex given the accessibility provided by its user interface as compared with SWAT and HSPF. Similar to the other non-grid based pilot models, the WARMF model framework is based on linkages of sub-basins and streams. WARMF does not include functions for performing sub-basin and stream delineations but it does have functions for importing delineations with attributes processed by the ArcSWAT tool or BASINS. For consistency with the SWAT simulations, the same delineations were used for the WARMF model demonstration as those used for SWAT. The imported sub-basin and streams as displayed from within the WARMF interface are shown in Figure 7. Default parameters were assigned to all sub-basins in the pilot WARMF model. A model developed for application would require characterization of each sub-basin from SSURGO soils data. The initial soils data parameters would then be cut and pasted from EXCEL into the ‘Soil Layers’ tab for each subbasin. Some of those parameters, such as hydraulic conductivity, would then be modified during model calibration.
Figure 7. Screenshot of WARMF interface The future scenario land use change was performed in WARMF by editing tabular land cover percentages. A second option of re-importing a new land use dataset was also discussed. Similar to the SWAT demonstration, the impact of the change was assessed by plotting simulated existing and future scenario suspended sediment and flow for Waikele Stream at Waipahu USGS gage during the March 1991 event selected for demonstration (see Appendix H, Figures H1 – H5). Total event existing and future suspended sediment loads of 5,500,000 kg and 19,250,000 kg were tabulated in EXCEL. It should be noted that these results were generated from an uncalibrated model, and as such, their accuracy, or lack thereof, is of little W9128A-05-D-0002
11
Central Oahu Watershed Study Phase 2 Part A
practical significance. The results are provided to demonstrate the type of results that could be presented using the model rather than actual model results. A unique aspect of WARMF is that it implements a unique stakeholder consensus building road map and TMDL calculator. These modules were not demonstrated as part of the pilot model demonstration but would be valuable tools for a TMDL study. HSPF The HSPF model is recognized as being the least accessible of the four models but the model was included as a pilot model due to ENV interest and because of its wide national use for TMDLs and specifically its application to Waikele watershed by DOH. Similar to the other nongrid based pilot models (SWAT and WARMF), the HSPF model framework is based on linkages of sub-basins and streams. HSPF is currently being distributed as part of the EPA BASINS 4.0 package. Unlike prior versions of BASINS which relied on ESRI’s ArcView GIS software, version 4.0 utilizes publicdomain GIS tools based on the open-source MapWindow GIS package (http://www.mapwindow.com/) to pre-process spatial data and develop a basin model. The MapWindow interface was demonstrated with the pilot model, but it was recommended that subbasins and streams be delineated using the ArcSWAT toolbar within ESRI’s ArcMap GIS and only the “import to HSPF” routine within the BASINS MapWindow interface be used for development of an HSPF model. The MapWindow interface has great potential as a stand alone environment for model development, but currently has some bugs and is inferior to other delineation tools such as the one included with ArcSWAT. For consistency with the other pilot model simulations, the same delineations were used for the HSPF model demonstration as those used for SWAT and WARMF. The HSPF model schematic (sub-basin and streams) is displayed from within the WinHSPF interface in Figure 8.
W9128A-05-D-0002
12
Central Oahu Watershed Study Phase 2 Part A
Figure 8. Screenshot showing schematic view within WinHSPF interface The future scenario land use change was performed in HSPF by editing tabular land cover percentages. Like the SWAT and WARMF demonstrations, the impact of the change was assessed by plotting simulated existing and future scenario suspended sediment and flow for Waikele Stream at Waipahu USGS gage during the March 1991 event selected for demonstration (see Appendix J, Figures J1 – J3). Total event existing and future suspended sediment loads of 72,000,000 kg and 88,000,000 kg were tabulated in EXCEL. It should be noted that these results were generated from an uncalibrated model, and as such, their accuracy, or lack thereof, is of little practical significance. The results are provided to demonstrate the type of results that could be presented using the model rather than actual model results. PILOT MODEL COMPARISON The pilot model demonstrations concluded with a discussion of the technical capabilities and user accessibility of the four models. The summary notes from these discussions are included in meeting notes for July 18, 2008 and July 29, 2008 (See Appendix E). Three tables that were introduced as part of these discussions are as follows. Table 3 was referenced to demonstrate the type of output available from each of the models, both by parameter and the time interval of output. As an indicator of model complexity, Table 4 includes a tabulation of pages in each user manual. While it is possible for a model with very few pages of documentation to be a poorly documented complex model, it was our experience that in the case of the four piloted models, the manual page-counts properly reflect their relative complexity and user accessibility. The last table, Table 5, was produced impromptu during the discussion as an indicator of the relative effort for each modeling task for each of the pilot models.
W9128A-05-D-0002
13
Central Oahu Watershed Study Phase 2 Part A
Table 3. Required Spatial Data Sets for Model Development Parameter
N-SPECT
SWAT
WARMF
HSPF
Storm Runoff
Average Annual/Event
Daily/Hourly
Daily/Hourly
Sub-Hourly
Stream Flow
Average Annual/Event
Daily/Hourly
Daily/Hourly
Sub-Hourly
Sub-Surface Flow
N/A
Daily/Hourly
Daily/Hourly
Sub-Hourly
Reservoirs
N/A
Daily/Hourly
Daily/Hourly
Sub-Hourly
Sediment
Average Annual/Event
Daily
Daily/Hourly
Sub-Hourly
N, P, Metals, TSS
Average Annual/Event
Daily
Daily/Hourly
Sub-Hourly
Temperature, DO
N/A
Daily
Daily/Hourly
Sub-Hourly
Algae, Bacteria, pH, Pesticides, etc.
N/A
Daily
Daily
Sub-Hourly
Point Sources
N/A
Daily
Daily
Sub-Hourly
Table 4. Instructional Manuals as an Indicator of Model Complexity Indicator
N-SPECT
SWAT
WARMF
HSPF
Pages of Technical Documentation
70
494
168
N/A1
Pages of User Manual
42
541
185
7501
Pages of Tutorial or Introduction Manual
43
453
110
9712
1
The current HSPF manual and documentation is distributed as a Windows help file database that does not lend itself easily to a page number count. Published user manuals for earlier versions of HSPF were on the order of 750 pages in length. 2
The HSPF tutorial is a series of PDFs and slideshows.
W9128A-05-D-0002
14
Central Oahu Watershed Study Phase 2 Part A
Table 5. Relative Time Commitments Required for Pilot Models Task
N-SPECT
SWAT
WARMF
HSPF
Data Preparation1
few days
1 week
1 week
2 weeks
Calibration
few days
4 weeks
4 weeks
4 weeks
Existing Reporting
2 days
1 week
1 week
1 week
Training (non-recurring)
1 week
1 month
2 months
2 months
1
Meteorological data, GIS pre-processing, etc.
W9128A-05-D-0002
15
Central Oahu Watershed Study Phase 2 Part A
EVALUATE ALTERNATIVES (STEP 7) Based on feedback from the pilot testing demonstration (see Appendix E, meeting notes dated 8/6/08), three options were presented to USACE/ENV for consideration. The advantages and disadvantages of each option were presented to assist in the decision-making process. The underlined items are deemed to have a greater weight on the final decision. OPTION #1: Develop HSPF model for Waikele under this contract; ENV to develop N-SPECT model independently for planning and reporting purposes for other watersheds. PROS
CONS
ENV would have a sound basis for verifying DOH’s TMDLs for the Waikele watershed which would be scientifically defensible
N-SPECT model may not be comprehensive enough to accommodate land use change scenarios or BMP implementation for other watersheds
ENV would have a means of easily applying land use change scenarios and assessing the effects of BMP implementation
N-SPECT model will not provide a defensible basis for contesting TMDLs in other watersheds (only provides order of magnitude estimates)
Good learning tool for ENV’s multi-year, multi-project effort to become proficient in HSPF modeling
There is the possibility that two independent developers could produce significantly different results using the same model (ex. differences in calibration or assumptions for modeling parameters, etc.)
Good technical support available; DOH could also be a local resource for support Consistent model use with DOH (HSPF is being used for all Pearl Harbor basins and Kaukonahua stream)
Due to the complexity of the model, ENV would only be trained to apply the model and obtain output under the current contract, and would not be trained to develop or calibrate HSPF models for other watersheds
HSPF is a well established, widely used model ENV would have sufficient working knowledge of HSPF modeling to collaborate with DOH on future TMDLs for other watersheds
Not cost effective for Waikele in particular since DOH’s calibrated model is expected to be released by the end of the year (2008); it would be more cost-effective to have an independent consultant review the HSPF model DOH has already developed
N-SPECT model would be easily transferable to DPP for future applications
Timing of the Waikele HSPF model completion would be late in terms of being able to verify DOH’s TMDLs (Draft Part B Report is due late-summer 2009) Re-calibration of the HSPF model (as more USGS data becomes available or as warranted by other conditions) would likely require a separate subconsultant contract
W9128A-05-D-0002
16
Central Oahu Watershed Study Phase 2 Part A
OPTION #2: Develop WARMF model for Waikele under this contract; ENV to contract out Quality Control / Independent Technical Review (QC/ITR) of DOH’s TMDL model on an as needed basis. PROS
CONS
ENV would have a sound basis for verifying DOH’s TMDLs for the Waikele watershed which would be scientifically defensible
Additional funds would be required to conduct QC/ITR of DOH’s HSPF model for Waikele and future TMDL watersheds (number of watersheds that would require ITR is unknown)*
ENV would have a means of easily applying land use change scenarios and assessing the effects of BMP implementation
Limited technical support; no known local expertise (although EPA appears committed to providing continuing technical support)
ENV would have the training to be able to develop and apply the WARMF model to other watersheds on Oahu (particularly Kaukonahua stream)
WARMF is not as well established or widely used as HSPF
Future re-calibration of the Waikele WARMF model (as more USGS data becomes available or as warranted by other conditions) can be accomplished in house and would not require a separate subconsultant contract Project schedule would not be dependent on DOH’s schedule *If the WARMF model produces similar flow and pollutant loadings to DOH’s HSPF model and ENV is satisfied with DOH’s application of simulation results to TMDL development and load allocation, then an ITR of the HSPF model may not be needed.
W9128A-05-D-0002
17
Central Oahu Watershed Study Phase 2 Part A
OPTION #3: Use current contract to (1) perform QC/ITR of DOH’s HSPF model for Waikele and (2) develop N-SPECT model for Waikele. PROS
CONS
ENV would have an expert objectively evaluate DOH’s TMDL model for Waikele
N-SPECT model may not be comprehensive enough to accommodate land use change scenarios or BMP implementation for other watersheds
ENV would have a means of easily applying land use change scenarios and assessing the effects of BMP implementation for the Waikele watershed (ENV would use the existing calibrated model developed by DOH with minor adjustments, if any, made by PB-nhc)
N-SPECT model will not provide a defensible basis for contesting TMDLs in other watersheds (only provides order of magnitude estimates) Project schedule may be delayed since deliverable would be dependent on DOH’s timeliness in providing electronic HSPF files for review
Additional funding would not be required, only scope of work change order ENV would have the training to be able to develop and apply the N-SPECT model to other watersheds on Oahu Future re-calibration of the Waikele NSPECT model (as more USGS data becomes available or as warranted by other conditions) can be accomplished in house and would not require a separate subconsultant contract ENV would have sufficient working knowledge of HSPF modeling to collaborate with DOH on future TMDLs for other watersheds N-SPECT model would be easily transferable to DPP for future applications
W9128A-05-D-0002
18
Central Oahu Watershed Study Phase 2 Part A
RECOMMEND MODEL(S) (STEP 8) The culmination of Part A is to make a recommendation to ENV for developing and applying a model to the Waikele Stream watershed. Each of the three options presented previously have their pros and cons, and there is no perfect match for ENV. ENV has recognized this issue in the past, and has responded with the idea that perhaps two models would better meet the range of their needs than a single model. Based on the collected information, budget, and framework of the Phase 2 contract, our recommendation is Option #2 - develop a calibrated WARMF model of the Waikele Stream for Part B. In addition, although not included in the scope-of-work, the N-SPECT model developed as a pilot model in Part A could be refined and provided to ENV, but with only very abbreviated documentation of the NSPECT model development. The N-SPECT model would be adjusted to come into approximate agreement with available Waikele stream data and could subsequently be evaluated with comparisons to a more detailed physical model, such as WARMF. However, ENV would need to evaluate and review the draft N-SPECT model and determine if additional model revisions are necessary before it is applied and used for their purposes. We feel this option provides the best outcome from Part B for ENV for several reasons as listed below: WARMF is a physically based model and can provide detailed and accurate simulation results (as compared to a screening level model). WARMF provides nearly the same level of detailed outputs (spatial, temporal, variety of pollutants) as HSPF, and therefore, WARMF simulation results can be used to compare to DOH’s HSPF results that support the TMDLs. By applying WARMF to Waikele Stream and providing ENV with hands-on training for WARMF, ENV can verify the model’s suitability for additional applications to other Oahu watersheds. Completion of Phase 2, Part B will not be dependent on the DOH schedule for completing and releasing their HSPF model for review. The N-SPECT model can be further evaluated by ENV as a tool to make quick, rough checks on results from more complex models such as WARMF and HSPF. The main disadvantage of the recommended option is that it is “riskier” than the other models being proposed because WARMF is less established than other watershed software programs such as HSPF or SWAT. It is anticipated that ENV will require some level of additional outside help from a public or private entity in developing future WARMF models until ENV becomes proficient at developing the models on their own. ENV would also need to learn N-SPECT from available manuals or seek training from an outside source. However, we feel these concerns are surmountable by ENV. In addition, Option #2 avoids the disadvantages of Option #3 which does not provide a separate physically based model to ENV for comparison to DOH’s HSPF simulations. Option #3 only evaluates the DOH model, but provides no mechanism to develop independent detailed comparisons. Our recommendation also avoids the redundancy and possible confusion inherent in Option #1 which includes developing an independent HSPF model in addition to the W9128A-05-D-0002
19
Central Oahu Watershed Study Phase 2 Part A
DOH HSPF model of the Waikele Stream watershed. Option #2 is the only option that provides a long-term solution to ENV for modeling other streams on Oahu, since N-SPECT (Option #3) may be too simple and HPSF (Option #1) is too complex and not as user-friendly.
W9128A-05-D-0002
20
Central Oahu Watershed Study Phase 2 Part A
REFERENCES Aqua Terra Consultants, Watershed Modeling Needs Assessment, Modeling Strategy, and Model Recommendations for Swamp and Green WQA Projects, Prepared for King County Water and Land Resources Division, Seattle, WA, February 2002. EPA, Compendium of Tools for Watershed Assessment and TMDL Development, Office of Water, Washington D.C., EPA841-B-97-006, May 1997. EPA, TMDL Model Evaluation and Research Needs, Office of Research and Development, Cincinnati, Ohio, EPA/600/R-05/149, November 2005. Oceanit, Central O’ahu Watershed Study Final Report, prepared for Honolulu Board of Water Supply, U.S. Army Corps of Engineers and City and County of Honolulu Department of Environmental Services by Oceanit, Townscape, Inc, and Eugene Dashiell, May 2007. Ward and Benaman, Models for TMDL Application in Texas Watercourses: Screening and Model Review, Online Report CRWR-99-7, submitted to Texas Natural Resource Conservation Commission by George H. Ward Jr. and Jennifer Benaman, December 1999.
W9128A-05-D-0002
21
Central Oahu Watershed Study Phase 2 Part A
APPENDIX A – FIELD VISIT A two-day field visit of the Waikele stream Watershed was conducted on April 8-9, 2008. The purpose of the field visit was to observe typical and atypical land uses within the watershed, as well as gain an understanding of the existing conditions and topography. The field visit consisted of windshield surveys with stops at specific sites and meetings with various land owners and lessees to understand agricultural practices, future development plans, and the general history of the land use. The field visit proved valuable in providing a better understanding of the mechanisms within the watershed that may have an impact on the decision making process for the model selection. The areas of interest that were observed include: USGS monitoring station at Farrington Highway (ID# 16213000) USGS monitoring station at Mililani Storm Drain A (ID# 212604158012700) and Mililani Town Roadway cuts along Kunia Road and Interstate Route H-2 Wheeler Air Force Base (Waikele Stream, Waikakalaua Stream) Schofield Barracks (live fire range, DPW sampling location in Waikele Stream) Del Monte Farms Larry Jefts Farms Castle & Cooke Properties (Koa Ridge Ranch Area) Waipio Acres Subdivision Waikele Ammunition Storage Area (Waikele Stream, Waikele/Kipapa confluence) The following individuals participated in the field visit: Lorayne Shimabuku, USACE Randall Wakumoto, ENV Alexandre Remnek, DOH (4/8/08 only) Greg Takeshima, DOH (4/8/08 only) David Penn, DOH (4/9/08 only) Scot Izuka, USGS (4/9/08 only)
Kevin Wong, PB (4/8/08 only) Todd Bennett, NHC Robert MacArthur, NHC David Hartley, NHC Pamela Nakanishi, PB
The PB-nhc team would also like to acknowledge the following individuals who served as escorts during various portions of the field visit or assisted in the right-of-entry process allowing the team to access locations within private properties and/or restricted areas. James Rice, CWA Program Manager, Army DPW Steve Turnbull, USACE-ERDC Gordon Rezentes, Del Monte Farms Alberto, Del Monte Farms Larry Jefts, Owner, Larry Jefts Farms Ray Kunishige, Site Construction Engineer, Castle & Cooke Homes Hawaii Peter Gibson, James Campbell Estates Beverlee Robinson, Robinson Family Trust Beverly Kaku, Caste & Cooke Homes Hawaii
W9128A-05-D-0002
Central Oahu Watershed Study Phase 2 Part A
APPENDIX C – AGENCY OUTREACH / KICKOFF MEETING An informational kickoff meeting was held on April 10, 2008 at the Frank F. Fasi Municipal Building. The purpose of the meeting was to introduce interested agencies to the project and the management/consultant team, solicit ideas and information, and establish a framework for cooperation to enhance the project. The following agencies were invited: City and County of Honolulu Department of Planning and Permitting Natural Resources Conservation Services U.S. Geological Survey* Department of Health Department of Land and Natural Resources State of Hawaii Department of Transportation USACE Engineer Research and Development Center Directorate of Public Works* U.S. Department of the Navy* City and County of Honolulu Board of Water Supply* *A representative did not attend A copy of the meeting summary and attendance sign-in sheet is attached. The meeting summary was distributed to all invitees and attendees via email.
W9128A-05-D-0002
U.S. ARMY CORPS OF ENGINEERS CITY & COUNTY OF HONOLULU, DEPT. OF ENVIRONMENTAL SERVICES MEETING SUMMARY FOR CENTRAL OAHU WATERSHED STUDY, PHASE 2 WAIKELE STREAM WATERSHED MODEL SELECTION AND APPLICATION PROJECT KICKOFF MEETING APRIL 10, 2008 Frank F. Fasi Municipal Building, 9th Floor Conference Room 9:00 AM to 12:00 PM
Team Representatives: Randall Wakumoto, Gerald Takayesu (ENV) Nani Shimabuku (COE) Todd Bennett, David Hartley, Robert MacArthur (NHC) Pamela Nakanishi (PB) Attendees: See attached sign-in sheet (Attachment A) Introduction COE opened the meeting by introducing the project sponsors and the consultant team. ENV also welcomed the group and described some issues which have led to the necessity of this project (ex. significant land use changes occurring within the agricultural lands and preparation for TMDL compliance). Invitees were asked to introduce themselves, the agency they represent, and their interest in the project. Project Purpose PB-NHC indicated that the Central Oahu Watershed Study (COWS) is in Phase 2 of the project. Phase 1 was completed in May 2007 and identified water resource problems and potential solutions for all of the streams that drain to Pearl Harbor, as well as the Ewa plain (11 watersheds total). Phase 2 will focus on selecting and developing a watershed model for the Waikele basin only. The primary purpose of the model will be (1) to evaluate land use scenarios including BMP implementation and (2) to provide ENV with a tool to assist them with TMDL compliance. The project has been phased as follows: Part A (Model Selection Process) Æ Decision Point Æ Part B (Model Development) Part A is expected to be completed in August 2008; Part B in mid-2009. Overview of Basin Characteristics PB-NHC provided an overview of the physiographic nature of the Waikele watershed with an emphasis on the spatial and temporal variability that is characteristic of the watershed. The erodibility of slopes within the various zones was also discussed. Field photos were shared with the group to illustrate current conditions, as well as historical photos which documented erosion hazards of the past. Page 1 of 3
Data Availability NHC provided a preliminary list of the known sources of precipitation, water quality, and flow data (USGS and DOH TMDL). ENV mentioned sampling at Mililani as part of the original NPDES permit applications. David Penn of DOH provided a handout summarizing the TMDLs for Waikele and Pearl Harbor (see Attachment B). DOH indicated that the most abundant data is available at the mouth of Waikele stream. DOH indicated that there was a National Urban Runoff Program in the early 70s and asked if the current Mililani Storm Drain monitoring station location is the same as one of the old NURP stations. DOH indicated that the Navy performed a sediment study (dredging evaluation) for Pearl Harbor approximately 2 years ago. The findings indicated that there was an accumulation of material at the mouths of the streams and within the estuary. The report is probably available through Navy Region Hawaii and/or the SPAWAR System group based out of San Diego (which also produced other technical reports about Pearl Harbor water quality). DOH also mentioned that SPAWAR was previously interested in a collaborative effort to develop an HSPF model of the entire Pearl Harbor watershed, but Navy funding did not materialize. DOT-Highways indicated that they have some data for sediment removal along State highways. Model Characteristics PB-NHC provided an overview of the model selection process and the issues that need to be considered, including selection criteria (i.e., model capabilities with regard to land use representation, hydrology, water quality parameters represented, routing, and chemical transformation, spatial and temporal detail, data requirements, user support and training requirements, etc.) and the model type (i.e., simple or screening, moderately complex, and highly complex). See Attachment C for summary of discussion topics. Discussion TMDLs for Waikele: DOH indicated that there were regular interagency meetings beginning in 2001 which focused on developing projects and programs to address the TMDLs. There were 7 or 8 impaired waterbodies identified with at least 3 pollutants. It was noted that the TMDL work can be “phased”, meaning the work is on-going and may be modified as additional field data becomes available. DOH’s primary concern is to fulfill their commitment to EPA. DOH indicated that Tetra Tech has developed an HSPF model for the Waikele watershed; however, due to staff rotation and limited funding, application of the model for purposes of Waikele TMDL development is being completed internally. DOH indicated that in the east coast, there are multi-state efforts to reach a consensus regarding model development. DOT-Highways emphasized that they are seeking a fair waste load allocation. ENV indicated that there are approximately 1,000 acres of construction on Oahu at any given time. There are also approximately 25,000 acres of agricultural land. There is concern that while the construction sites on Oahu are permitted and under scrutiny, the agricultural lands have little or no regulatory requirements. ENV also has concerns that alien species are affecting the forested areas. Page 2 of 3
DOH indicated that Oceanit had developed a HEC-HMS model for Waikele watershed. DLNR DOFAW indicated that they are conducting some sampling in the forest reserve/management areas, but are currently working on improving the data quality. Dave Smith will provide a contact name and number for the planner in charge of the project. NRCS indicated that they are developing an AGNPS model for a Hanalei project. DOH indicated that the 1986 Rainfall Atlas will soon be available in digital form. Army DPW indicated that they are monitoring flow and water quality in the Schofield/Wheeler area. The samples are being analyzed by UH. They are also developing a GSSHA model for the area. DOH indicated that the water commission is updating inventories of stream diversions and stream channel modifications. Summary In summary, PB-NHC indicated that following this initial model documentation and recommendation phase, ENV will finalize their model selection, and PB-NHC will proceed with full development of the model. Contact Information The attendees were encouraged to contact the following individuals for additional project information: Project Updates / Information • Nani Shimabuku, COE o Phone: 438-2940 o Email:
[email protected] •
Randall Wakumoto, ENV o Phone: 768-3242 o Email
[email protected]
Technical Questions • Pamela Nakanishi, PB o Phone: 566-2265 o Email:
[email protected] End meeting summary.
Page 3 of 3
ATTACHMENT A
ATTACHMENT A
ATTACHMENT B
Watershed model selection criteria and model types
Watershed model selection criteria and model types (continued)
Watershed Model Selection Criteria and General Model Types (Liberally Interpreted from EPA 841-R-92-002, Shoemaker et al, 1992) - = zero capability, L = Low, M = Medium, H = High
Simple Models
Medium Models
Complex Models
L
M
H
Point Sources
-
M
H
Avg. Annual
H
H
H
Event
L
M
H
Continuous
-
M
H
Storm Runoff
-
H
H
Criteria Land Uses
Mixed Land Uses
Time Scale of Output Hydrology
Base Flow
-
L
Simple Models
Medium Models
Complex Models
Sediment
M
H
H
Nutrients
M
H
H
Others (Temp, Metals, Organics, etc)
L
M
H
Transport, Deposition
-
L
H
Transformation
-
-
- to H
Statistics/Probability
-
M
M to H
Graphics/Maps
L
M to H
M to H
User Formats
L
H
H
Criteria
Pollutant Loading
Pollutant Routing
Model Output
H 29
Watershed model selection criteria and model types (continued) Criteria
Data
Simple Models
Medium Models
30
Watershed model selection criteria and model types (continued)
Complex Models
Simple Models
Medium Models
Complex Models
GUI, GIS
L to H
L to H
L to H
Criteria
Input
L
M
H
Calibration
-
L to M
M to H
Default Data Availability
M to H
M to H
M to H
Required Expertise
L
M
H
Regulatory
L to M
M to H
M to H
Regional/National Applications
L to H
L to H
L to H
Scientific
M
M to H
H
Database Integration
-
L to M
M to H
Evaluation (lumped)
L to M
L to M
M to H
Design/Sizing (specific)
-
L to M
M to H
BMP’s
31
Ease of Use
Acceptance
32
ATTACHMENT C
1
Central Oahu Watershed Study Phase 2 Part A
APPENDIX D – DISCUSSION ON SHORT LISTED MODELS PRESENTATION - JUNE 9, 2008
This appendix contains the presentation given on June 9, 2008 to ENV, USACE, DOTHighways Division, and Townscape. It should be noted that these slides were developed prior to the pilot testing where additional details were obtained on the models. This includes the ability of WARMF and SWAT to simulate at an hourly time step. The SWAT model can perform hydrologic runoff and routing calculations at an hourly time step (when using Green & Ampt but not Curve Number infiltration) but all water quality calculations are performed at daily time steps. Documents reviewed for the June 9th presentation indicated that these models only simulated at a daily time step.
W9128A-05-D-0002
Central Oahu Watershed Study Phase 2 Part A
APPENDIX E - MEETING SUMMARY NOTES This appendix contains the summary notes from a conference call on June 13, 2008, model demonstrations on July 18, 2008, a meeting on July 29, 2008, and a conference call August 6, 2008
W9128A-05-D-0002
MEETING NOTES FOR CENTRAL OAHU WATERSHED STUDY, PHASE 2 TELECONFERENCE JUNE 13, 2008 7:30 AM to 8:30 AM Attendees: Randall Wakumoto, Jerry Takayesu (ENV) Nani Shimabuku, Michael Wong (COE) Todd Bennett, David Hartley, Derek Stuart (nhc) Pamela Nakanishi (PB) Action items are underlined. •
Pam indicated that following Monday’s web conference, Randall inquired about GSSHA since the Army is currently using it for Schofield. PB-nhc followed up with Randall prior to the call and indicated that GSSHA is a highly complex model that didn’t rank as favorably as the other complex models that were evaluated. In addition, it runs a 2D overland flow routine which adds to the complexity of the model and introduces stability issues.
•
Randall mentioned that he had spoken with Russell Leong who indicated that GSSHA was more of a moderate-level model, rather than a complex model as indicated by PB-nhc. Russell also indicated that GSSHA is a CORPS-sponsored model, and therefore user fees would be waived. The Army would also be willing to share data for calibration/comparison purposes.
•
Mike indicated that GSSHA is a gridded model that runs a 2D routine. The benefit of this model is that it can be coupled with a groundwater model. It is usable for TMDL purposes, but there are numerous input parameters which add to the complexity of the model.
•
David indicated that it is important to take into consideration the history of a model’s use. GSSHA is not widely used, it is relatively more complex than all the models evaluated, and the level of detail associated with the 2D aspect is excessive for TMDL studies.
•
Todd indicated that pre-processing for GSSHA would be in WMS. This would require ENV to essentially learn two software programs. In addition, for ENV’s future needs, WMS application to other watersheds under non-CORPS contracts may trigger user fees.
•
Todd commented that nhc feels that WARMF should be one of the models to move forward to the pilot testing phase if ENV is considering selecting one of the more complex, physically based models.
•
Randall indicated that he had concerns with WARMF because the model has not been applied locally, and therefore, there is no framework for comparison. In contrast, there are local entities that are currently using GSSHA. Randall indicated that he will obtain documentation from the Army regarding their selection process on the GSSHA model.
Page 1 of 3
•
Mike indicated that based on the web conference presentation, N-SPECT, SWAT, and HSPF appear like the appropriate models to move forward with for ENV’s purposes.
•
Jerry mentioned that he favored Kevin Wong’s idea from Monday which was to have an independent consultant evaluate DOH’s HSPF model. Jerry mentioned that he would prefer a simpler model to suit their needs. He also indicated that WTM should be dropped from further consideration.
•
Mike emphasized that ENV needs to consider what their goal is and what they are trying to accomplish with this model. He indicated that if ENV is looking to develop only one model for the dual purpose of planning studies and TMDL compliance, then he would suggest SWAT. However, if ENV is willing to develop two models, then he would suggest N-SPECT for the planning level and a more complex model for TMDL evaluation. A model developed for TMDL studies would need to be physically based.
•
Mike indicated that COE will be developing a GSSHA model for the Ala Wai watershed (contact: Cindy Barger). They selected GSSHA to evaluate rain barrel ideas. It is also appropriate to model tidal influences from the canal due to its fully dynamic nature. He was not sure that the same level of detail would be required for the Waikele watershed which is on the order of 3 times larger than Ala Wai. COE is considering using a 15-minute time step for the Ala Wai watershed.
•
David indicated that WARMF is more comparable to SWAT than HSPF. The technical capabilities are similar to SWAT, and in addition, advanced programming makes it more user friendly.
•
Randall indicated that N-SPECT is definitely a candidate because it uses RUSLE, which the City is adopting for its drainage standards. He indicated that PLOAD was not at the top of their list.
•
Jerry questioned if a gridded model is too complex for RUSLE application. Todd indicated that for N-SPECT, some RUSLE parameters are inputted on a grid scale; others are determined internally based on other input data such as the DEM.
•
Jerry asked if a 1D model would work for areas with tidal influence. Todd responded that there would be no backwater calculation component in the models being considered. David commended that backwater calculations are not needed for the main modeling purposes specified by ENV.
•
Mike indicated that N-SPECT was a good option since it was developed by NOAA in Hawaii (contact: Jamie Carter). Nani indicated she has Jamie’s contact info.
•
Jerry indicated that ENV is probably not ready to develop and implement a model as complex as HSPF or GSSHA. He would prefer a consultant analyze/assess DOH’s HSPF work for Waikele.
•
Jerry asked it SWAT was a subset of HSPF. David responded that there are distinct differences between the two models. SWAT was originally an agriculture oriented model developed by USDA/ARS. Only recently was mixed land use and urbanization added. SWAT operates on a daily time step. HSPF, in contrast, has always been capable of Page 2 of 3
representing watersheds with a wide range of land uses. HSPF allows for much smaller time steps which can account for rapid runoff from urban areas. •
Todd asked if DOH wasn’t using HSPF, would it have been on ENV’s short list. Randall indicated that it probably would have not have made the short list since ENV doesn’t require that level of detail. ENV is primarily concerned with monitoring, reporting, tracking BMPs, planning uses (land use changes), and permitting purposes.
•
Todd asked if ENV would use a physically-based model for its monitoring, reporting, tracking BMPs, planning uses, and permitting purposes. Randall indicated that they would use such a model as long as they can learn it and use it effectively. Todd indicated that it is very possible for ENV staff to learn to use one of the more complex models depending on the level of effort they wanted to expend and that the demonstrations from the pilot level studies should give ENV a good idea of what is involved.
•
Todd asked if ENV has received any outside comments regarding the model selection, or if they are basing their decision solely on the information presented. Randall indicated that most of their opinions are based on the matrix provided. Todd cautioned that the matrix has evolved during the short listing process, as nhc continues to conduct additional research. Therefore, some refinement to the matrix may be in order.
•
David indicated that nhc’s familiarity with WARMF has improved throughout this screening process. The fact that the model was previously proprietary, but is now fully in the public domain has been confirmed. David stated that WARMF is easier to learn that HSPF.
•
Randall and Jerry indicated that the three frontrunner models that should be moved forward for benchmarking appear to be SWAT, N-SPECT, and WARMF. Todd confirmed that nhc’s recommendations would be the same (one simpler model and two physically-based models) given that ENV doesn’t want to move HSPF forward for the pilot tests. This will allow for further evaluation of, and comparison between, SWAT and WARMF.
•
Todd will send Randall a link to the GSSHA website for the user’s manual and documentation.
•
Todd indicated that the target date for completion of benchmark testing is mid-July.
End meeting notes
Page 3 of 3
MEETING NOTES FOR CENTRAL OAHU WATERSHED STUDY, PHASE 2 MODEL DEMONSTRATION JULY 18, 2008 PB Conference Room 8:00 AM to 4:00 PM Attendees: See attached sign-in sheet. Items Discussed: A demonstration on building the model, running and modifying scenarios, and producing output for each of the three finalist models (WARMF, SWAT, and N-SPECT) was presented. See attached agenda for specific topics discussed. Post-Demonstration Discussion: (Action items are underlined) •
Jerry’s initial assessment was that all three of the models appeared somewhat more complex to develop and modify than he had expected. Randall indicated that ENV currently uses a spreadsheet model for their reporting purposes.
•
Jerry also mentioned that he was surprised that SWAT was presented as a more complex model than WARMF. Initially, based on the preliminary matrix, WARMF was identified as the more complex model. David replied that through the processes of becoming familiar with WARMF, nhc’s view of the models complexity and usability evolved.
•
Derek noted that the level of effort required for developing a SWAT model is only slightly less than that for an HSPF model. Therefore, if ENV is willing to pursue development of the SWAT model, they should consider going forward with an HSPF model instead because of the added benefit of being able to use the DOH model.
•
David believes that if WARMF were available in the public domain for a longer period of time (and as a result been more commonly known), WARMF would have been a good choice for DOH to apply to developing TMDLs for Waikele and other Hawaii watersheds. WARMF was developed specifically for TMDL modeling.
•
Jerry questioned if other municipalities have used WARMF. Based on the WARMF documentation listed on the EPA website, the model has been applied to over 15 watersheds domestically and internationally. A listing displayed from the Systech Water Resources web-page lists 21 applications of WARMF to watersheds domestically and internationally (www.systechengineering.com/WARMFApplications.html). PB-nhc will attempt to identify and contact municipalities who have applied WARMF for TMDL modeling purposes. PB-nhc will report any feedback received from other WARMF users.
•
Jerry’s primary concern with WARMF was the availability of technical support. Derek mentioned that a WARMF user’s group exists but it is not very active and support for his questions had been provided directly from the model developers (Systech Water
Page 1 of 3
Resources). Limited questions to Systech from nhc have been addressed for free but their website states that they require an hourly support fee. •
Derek noted that SWAT has a large user base and displayed current user group questions (via Google Groups) during the presentation as an example.
•
David mentioned that HSPF would be considered for use in this study because DOH has an existing model and plans to use it this year for TMDL development. PB-nhc will contact Dave Penn to verify that they are planning to use HSPF for development of TMDLs in the Waikele watershed and elsewhere on Oahu and that those models will be available to ENV to use and/or apply.
•
David presented three options for ENV to consider if they want to challenge DOH TMDL values: (1) have an independent consultant review DOH’s HSPF model as needed, (2) have an independent consultant develop an HSPF model to verify DOH’s output, or (3) use another software to verify/confirm the TMDL requirements issued by DOH.
•
Randall mentioned that Steve Turnbull gave ENV a presentation on GSSHA. Steve indicated that GSSHA can be as complex or simple as the user desires.
•
Mike indicated that although COE is planning to develop a GSSHA model for the Ala Wai watershed, they would be willing to consider using WARMF if ENV selects it as their primary modeling tool.
•
David mentioned that N-SPECT is more of a screening-level model, rather than a planninglevel model, and that it provides order of magnitude estimates more than reliable numbers. N-SPECT’s advantage is that it can be used by users with less experience than that required for the more complex models.
•
The following table was developed to provide a general feel for the relative time commitment required to effectively develop and learn how to use each of the watershed models considered: Data Prep (met data, GIS preprocessing, etc.)
Calibration
Existing Reporting
few days
few days
2 days
1 week
WARMF
1 week
4 weeks
1 week
1 month
SWAT
1 week
4 weeks
1 week
2 months
HSPF
2 weeks
4 weeks
1 week
2 months
N-SPECT
•
Scenarios, Interpretation, Reporting
Training (nonrecurring)
Michael Wong noted that learning a model is an ongoing process that evolves after applying the software multiple times.
Page 2 of 3
Post-Meeting Discussion: •
Following the meeting on Friday PB-nhc had discussed quickly constructing and presenting a simple example HSPF model for a short demonstration on Saturday AM. That meeting was delayed to allow a demonstration more equivalent to that of the other three models to be developed. PB-nhc will follow-up with a short demo on the HSPF model via a web conference.
•
An official request for the Waikele HSPF *.uci file and *.wdm database was submitted to DOH on 6/4/08. PB-nhc will contact DOH to inquire about the status of this request. If DOH provides the requested files within a few days, PB-nhc will conduct the demo using the files provided. If DOH is not prepared to release the files, PB-nhc will develop an HSPF model sufficient for demonstration purposes.
•
Target date for the HSPF demo web conference demo: week of July 28th.
End meeting notes
Page 3 of 3
CENTRAL OAHU WATERSHED STUDY, PHASE 2 MODEL DEMONSTRATION AGENDA PB Conference Room July 18, 2008 8:00am
1. Introductions 2. Overview a. Review of Model Selection Process b. Purpose of Model Demonstration c. Key components i. GIS, pre modeling ii. Build watershed model iii. Run watershed model (data input, management and reporting) iv. Scenario building and running 3. WARMF Demonstration a. Brief WARMF Overview b. Preliminary GIS Processing in ArcMAP i. Basin and Stream Delineation ii. Calculate aspect for Sub-Basins iii. Sub-Basin Aspect iv. Land Use c. Watershed Model Development in WARMF i. Import spatial data into engineering module 1. Sub-Basins (Catchments) 2. Streams (Rivers) 3. Land use ii. Enter soils parameters from NRCS data iii. Import time-series into data module 1. Meteorological data 2. Air quality data 3. Observed flow and water quality iv. Edit default parameters d. Baseline model simulation i. Execute model run ii. Compare simulated to observed data iii. Brief comment about calibration e. Example scenario i. Implement land use change ii. Compare output to base line 4. Break for Lunch
5. SWAT Demonstration a. Brief SWAT Overview b. Preliminary GIS processing i. Sub-basins / stream (as in WARMF) ii. Aspect not required or used c. Building SWAT model in ArcMAP i. Import spatial data (Sub-basins, streams, land use, slope, soils) ii. Import time-series 1. Meteorological data/Weather Generator 2. Air quality data is NOT required/used d. Edit default parameters e. Run SWAT for Existing (Baseline) Conditions i. Execute model run ii. Compare to observed iii. Brief Comment on Calibration f. Example scenario i. Implement land use change ii. Compare output to base line 6. Mid Afternoon Break 7. N-SPECT a. Brief N-SPECT Overview b. Preliminary GIS processing (ArcMap) i. Develop r-factor spatial coverage ii. Develop or acquire precipitation iii. Convert land use polygons or aggregate to DEM grid size (i.e. NOAA) c. Building N-SPECT model d. Baseline model simulation i. Execute model run ii. Compare to observed iii. Adjust to calibrated e. Example scenario i. Implement land use change (note scenario management) ii. Compare output to base line 8. Model Summary Comparison a. Data input requirements b. Output parameter types and formats (temporal, spatial, format) 9. Questions and Discussion
MEETING NOTES FOR CENTRAL OAHU WATERSHED STUDY, PHASE 2 HSPF MODEL DEMONSTRATION JULY 29, 2008 PB Conference Room 8:00 AM to 12:00 PM Attendees: See attached sign-in sheet. A demonstration on delineating subbasins, building the model, running and modifying scenarios, and producing output for HSPF was presented. Web Conference - Items Discussed: (Action items are underlined) •
In general, SWAT and HSPF have similar sediment transport and channel process algorithms, but the editors may be more difficult to find in the HSPF interface.
•
Jerry asked how difficult it would be to edit the HSPF file to target a single stream. Derek indicated that it is possible to do so in the WinHSPF GUI interface though not directly from the schematic view; however, nhc typically works directly with the HSPF model text file using text editors since they are familiar with the HSPF file format (this text file is the same file that runs in the background of the WinHSPF GUI).
•
Randall indicated that it appears easy to edit the land use and soil parameters in HSPF. Derek indicated that land use and soil parameters can be imported as a simple copy/paste function from excel. David added that populating the soil type parameters is based on SCS SSURGO dataset. These values can be refined during the calibration process.
•
David indicated that it is very possible for ENV to effectively use a calibrated HSPF model to apply their land use change scenarios. David cautioned that calibration of an HSPF model and critique of DOH’s TMDL model should be done by an experienced model user.
•
David indicated that if ENV is considering maintaining and developing HSPF models, then the best way to learn is to start by applying an existing model, then progress towards developing and calibrating a model.
•
Randall indicated his preference for a model selection is towards WARMF, but his main concern is the availability of technical support. It would be problematic if the only technical support available is on a fee only basis.
•
Todd asked if ENV would like to drop N-SPECT from further consideration. Randall indicated that is not necessarily the case. They may require dual models – something like N-SPECT for their planning purposes and a second model which would be more comparable to DOH’s TMDL model.
•
Randall indicated he would be the primary user of the selected model.
Page 1 of 4
•
Mike indicated that his personal view, independent of ENV, is to recommend WARMF because it is physically-based, yet the interface is easy to understand. His second choice would be HSPF. The drawbacks to WARMF are that soil data needs to be entered manually (although this is the same case for some HSPF parameters) and it is only capable of running a daily time step. If GSSHA were one of the short listed models, he would not necessarily recommend it. Mike also mentioned that in general, more de-bugging is required with physically-based models.
•
ENV indicated that their intent is to have their GIS department perform the initial work required to develop the base files for the model, but ENV would input the data and calibrate the model. It was noted that WARMF data input is accomplished by a simple cut/paste function from excel. HSPF accepts either DSSVue or WDMUtil met data input, but GENSEN (HSPF output viewer) is not compatible with the DSS file format.
•
Derek indicated that the Spatial Analyst extension is required to load the ArcSWAT toolbar and do the GIS pre-processing functions. Another option for editing SWAT input is to load the stand-alone SWAT Editor application which has limited editing commands without the need of GIS. The SWAT Editor application does not allow the user to do any GIS preprocessing (i.e. changing sub-basins, land-use, soil or slope acreages via GIS processing).
•
Mike indicated that re-calibration of an existing model would be required when (1) more data becomes available in a spatially different location or (2) when a significant rainfall event occurs that is larger than the calibrated event and the additional data does not match the observed data. He further recommended programming funds into ENV’s budget to account for re-calibration for an assumed time period, say every 5 to 10 years.
•
Todd asked if ENV is considering implementing N-SPECT and another, more complicated model. Jerry responded that it is likely since there are 32 watersheds on Oahu that need to be addressed.
•
Jerry indicated that some of the smaller watersheds on Oahu are very simple in that there may be only 2 land uses within the entire watershed. In addition, some have no gaging stations. David indicated that if that were the case, parameters from an adjacent, similar watershed could be used.
•
Randall indicated that DPP is not looking into applying a model for their needs. They are currently using RUSLE. It is possible that they may consider using N-SPECT. The model may also be released to developers and other interested parties at some point in the future.
•
Todd indicated that nhc had contacted EPA to get some feedback regarding the availability of technical support for WARMF but they did not yet have a response (currently SYSTECH is providing support for the model). Randall and Jerry indicated that they would not consider moving forward with WARMF if the only source of technical support would be the user’s manual (this answer was in response to Todd hypothetically asking ‘what if SYSTECH were to disappear?’).
•
Pam indicated that PB also contacted other municipalities who have developed WARMF models. Feedback was received for the Truckee River (CA, NV), Catawba River (SC, NC), and San Joaquin River (CA). The feedback received was generally positive.
Page 2 of 4
•
Sherri indicated that Townscape’s current contract is not necessarily to develop a model for the adjacent watershed. It was their intention that the model developed for Waikele would be transferrable to Waialua-Kaiaka basins. Sherri also indicated that based on what she has seen so far, it appears feasible for Townscape to develop a WARMF model.
•
It was decided that additional time would be required by all parties to make a final decision. ENV indicated that delaying the Final Part A report for several weeks would be acceptable. PB-nhc will follow-up with EPA and some of the municipalities contacted to get additional information on WARMF. A follow-up teleconference will be scheduled for the week of 8/4.
Post-Meeting Discussion: After the web conference portion ended, ENV, COE, Townscape, and PB continued discussions. •
Pam indicated that nhc found that the BASINS console functions for developing an HSPF model do not (explicitly) accommodate some of HSPF’s more complex capabilities (e.g. using soils and slope GIS files when developing HRUs). It appears that there is a tradeoff between ease in developing a model (via BASINS) and the full functionality of the options offered. nhc typically does not use the BASINS console in their HSPF modeling.
•
Jerry indicated that ENV would like to collaborate with DOH early in the TMDL process, preferably in the development stage. He emphasized the need for ENV to get involved early in the process, since it is very difficult to change TMDLs once they have been released.
•
A call to Dave Penn (DOH) was made. Dave indicated that DOH is not necessarily applying the HSPF model to watersheds on Oahu islandwide. The model selection is on a projectby-project basis. Dave indicated that besides the Pearl Harbor basins, they are using HSPF for the Kaukonahua watershed. He further indicated that they did not use the BASINS console in developing their HSPF models.
•
Nani asked Dave how the initial selection of the HSPF model was made. Dave indicated that there were discussions between EPA, their contractor (TetraTech), and DOH. After the call, Jerry mentioned that TetraTech had a hand in developing the BASINS platform, which would explain their preference towards HSPF. Dave indicated that WARMF was not one of the models under consideration at the time.
•
Jerry asked when DOH would be prepared to release the HSPF model to ENV. Dave could not pinpoint a date, citing other pressing deadlines.
•
Dave indicated that DOH is completing their scoping efforts for the Kalihi and Nuuanu watersheds.
•
Nani and Pam will check on the official request which was submitted to DOH on 6/4/08.
•
Mike indicated that COE would consider applying the HSPF model to Ala Wai. Currently, Oceanit is using HEC-HMS for the hydrology, HEC-RAS for the streams in the upper watershed, and FLOW2D in the canal. Mike indicated that a 2D model is not required for Waikele.
Page 3 of 4
•
Jerry asked Nani to look into the terms of PB-nhc’s contract to determine if there is a mechanism to allow nhc to review/QC DOH’s HSPF model for Waikele in the near term.
End meeting notes
Page 4 of 4
MEETING NOTES FOR CENTRAL OAHU WATERSHED STUDY, PHASE 2 MODEL SELECTION ALTERNATIVES August 6, 2008 Teleconference 8:00 AM to 9:30 AM Attendees: Randall Wakumoto, Gerald Takayesu (ENV) Sherri Hiraoka (Townscape) Nani Shimabuku (COE) David Hartley, Derek Stuart, Todd Bennett (nhc) Pamela Nakanishi (PB) Handouts: Draft list of alternatives pros/cons (see attached) Items Discussed: •
Jerry indicated that for Option 1, ENV would not continue to apply their existing spreadsheet model to all other watersheds for planning and reporting purposes. Tom Schuler’s simple model was probably not intended for use on larger, more complex watersheds. PB-nhc will revise the table to indicate this revision. It was noted that the training session included in Option 1 would consist of application of the HSPF model and not N-SPECT.
•
Jerry questioned the frequency of re-calibrating an existing model. David indicated that the first calibration parameter is flow. The second component would be sediment, and lowest tier would be other water quality parameters. A situation that would warrant re-calibration of a model may not necessarily be additional flow data at a current location, but new measurements at other locations within the watershed.
•
David indicated that he contacted Tim Wool (EPA Athens Lab, WARMF training coordinator) and relayed information regarding his feedback on continued technical support for WARMF. Tim indicated that WARMF is an adequate TMDL modeling tool, but ultimately it’s up to the local agencies to determine the most appropriate tool for their needs. David feels the most important distinction between WARMF support and HSPF or SWAT is that WARMF is backed by an industry group (EPRI) versus a government agency. David indicated that there is a lot of literature available for download on the EPRI website and is confident that, based on feedback from EPA and EPRI, technical support and software updates will be available for the foreseeable future.
•
Randall asked if Tom Wool still provides training. David indicated that training is available. In 2006, EPA hosted a WARMF training session sponsored by EPRI. Systech also periodically holds training sessions, either in their CA base office or as requested by various agencies.
•
Derek indicated that he also spoke with Joel Herr (Systech). Joel indicated that they do not have an ongoing contract with EPA, but they do provide them with software updates as it
Page 1 of 3
becomes available. Derek also indicated that technical support for software bug-related inquiries are free of charge. •
Todd indicated that overall WARMF is a good option as a middle-of-the-road model in that it provides a good combination of technical capability and usability. HSPF, on the other hand, is time-intensive to learn/apply and is much more complex.
•
David indicated that regardless of the model selected (WARMF or HSPF), ENV can anticipate some need for subconsultant support. The level of support would be less with WARMF than HSPF.
•
Randall indicated that he feels WARMF is a good model and likes that fact that it’s selfcontained, meaning land use changes can be accomplished in the model itself. He reiterated his concern regarding the availability of technical support.
•
Jerry asked if the WARMF and HSPF models were applied to the same watershed, would they provide the same result. David indicated that, in general, the results should be very similar. However, differences in literature values associated with various parameters could conceivably produce significantly different results.
•
Jerry indicated that he does not feel WARMF is an appropriate model for DPP use. Since N-SPECT is fairly simple, it appears much more appropriate for planning purposes. Randall feels that they could convince DPP to use N-SPECT at a planning tool since they are already using RUSLE2 equations. David indicated that N-SPECT is essentially a spreadsheet model with a GIS component.
•
Jerry questioned the need for additional funds indicated under Option 2. Todd responded that an independent technical review (ITR) of DOH’s work is not included in this project’s scope of work. However, if the WARMF model produces similar flow and pollutant loadings to DOH’s HSPF model and ENV is satisfied with DOH’s application of simulation results to TMDL development and load allocation, then an ITR would probably be of less utility to ENV. David added that the current budget would allow nhc to conduct an ITR of DOH’s HSPF model and develop N-SPECT for ENV use in the near-term.
•
Sherri indicated that developing an N-SPECT model is included in their Waialua-Kaiaka contract.
•
Sherri questioned if any of the models can be used to identify hot spots for a particular type of pollutant. David indicated that HSPF and WARMF can provide the desired output at any designated reach outlet. It can also be accomplished, to some degree, in N-SPECT by querying individual cells. Todd added that suspected hot spots can be isolated in WARMF for HSPF by delineating a subbasin for the area in question. Point loads, however, would require a separate designation. Derek added that the model inputs will dictate the output (i.e., mass wasting site will need to be keyed into the model).
•
Jerry asked which models are most prone to crashing. David indicated that N-SPECT is pretty stable in that it will generally execute without crashing. David also believes that the GUI and operation are more stable in WARMF than HSPF and therefore will crash less often. Derek added that HSPF always seems to crash on the first attempt at running a model. Todd believes there are less bugs associated with HSPF than WARMF.
Page 2 of 3
•
Randall asked if it was possible to develop WARMF and N-SPECT models under Option 2. Nani indicated that it was outside the scope of the current contract. Todd indicated that a pilot N-SPECT test model had been developed for demonstration purposes which could be turned over to ENV. However, he cautioned that the model would require refinement before application.
•
Jerry indicated that contrary to his initial feelings, HSPF should be dropped from further consideration. Jerry and Randall generally feel comfortable with the WARMF user interface, but Jerry wants to keep N-SPECT as a candidate.
•
Todd indicated that WARMF is much easier to learn than HSPF and feels that a new model user could pick it up from scratch, even though there is quite a bit to learn.
•
David indicated that Option 1 may appear confrontational to DOH which may hinder future collaboration efforts. Todd added that Option 1 seems like an inefficient way to go about this process. There is also the lingering issue that ENV may have an ITR completed for Waikele, but there are potentially 32 or so other watersheds that may need to be evaluated and/or contested.
•
Randall indicated that Option 1 may actually facilitate collaboration with DOH and does not necessarily have to be construed as confrontational. DOH may view ENV’s attempt in becoming familiar with HSPF modeling as an indication that they want to collaborate on future TMDLs.
•
Randall questioned if N-SPECT can be used as a tool for permit monitoring and tracking BMP performance. David indicated that this can be accomplished at a very approximate level. There is a BMP reduction coefficient that can be applied to the model. David also indicated that EPA (Jim Carlton) is considering adding N-SPECT to the BASINS console, which may indicate some increasing level of interest/use. Todd noted that nhc has identified some technical issues related to N-SPECT.
•
Jerry indicated that ENV would probably require their consultants to use whatever model they select for consistency.
•
Sherri asked if the WARMF data input is similar to the HSPF data input. David indicated that there is some degree of overlap, and being familiar with one model will enhance understanding of the other. Sherri indicated that if this is the case, ENV’s training in WARMF would provide them with a basic understanding of the parameters needed to run the HSPF model, thus allowing them to collaborate/communicate with DOH.
•
Nani questioned if this is also the case for WARMF and N-SPECT. She suggested that ENV re-assess their need for N-SPECT training after they receive the WARMF training.
•
Todd indicated that the goal of Part A is to recommend a model to ENV. PB-nhc will proceed with their recommendation and Preliminary Part A report for government review (date to be determined). ENV will then have the opportunity to evaluate their options before making a final decision.
•
Nani requested that the meeting notes be distributed to Mike Wong for comment.
End meeting notes
Page 3 of 3
CENTRAL OAHU WATERSHED STUDY, PHASE 2 MODEL SELECTION ALTERNATIVES - DRAFT OPTION 1: Develop HSPF model for Waikele under this project; ENV to continue using spreadsheet model for planning and reporting purposes for other watersheds. PROS •
•
•
• •
•
CONS
ENV would have a sound basis for contesting DOH’s TMDLs for the Waikele watershed which would be scientifically defensible ENV would have a means of easily applying land use change scenarios and assessing the effects of BMP implementation Good learning tool for ENV’s multi-year, multiproject effort to become proficient in HSPF modeling Good technical support available; DOH could also be a local resource Consistent model use with DOH (HSPF is being used for all Pearl Harbor basins and Kaukonahua stream) HSPF is a well established model (vs. the relatively new WARMF)
•
•
•
•
•
•
•
Page 1 of 2
Existing spreadsheet model may not be comprehensive enough to accommodate land use change scenarios or BMP implementation for other watersheds (other than Waikele) Existing spreadsheet model will not provide a defensible basis for contesting TMDLs in other watersheds (other than Waikele) There is the possibility that two different developers could get different results using the same model (ex. differences in calibration, etc.) Due to the complexity of the model, ENV would only be trained to apply the model and obtain output and would not be able to develop models for other watersheds Not cost effective for Waikele in particular since DOH’s calibrated model is expected to be released by the end of the year (more costeffective to have an independent consultant review HSPF model DOH has already developed) Timing of the Waikele HSPF model completion would be late in terms of being able to contest DOH’s TMDLs (Draft Part B Report is due Aug 2009) Re-calibration of the model (as more USGS data becomes available) would likely require a subconsultant contract.
CENTRAL OAHU WATERSHED STUDY, PHASE 2 MODEL SELECTION ALTERNATIVES - DRAFT OPTION 2: Develop WARMF model for Waikele under this project; ENV to contract out QC of DOH’s TMDL model on an as needed basis. PROS •
•
•
•
CONS
ENV would have a sound basis for contesting DOH’s TMDLs for the Waikele watershed which would be scientifically defensible ENV would have a means of easily applying land use change scenarios and assessing the effects of BMP implementation ENV would have the training to be able to develop and apply the model to other watersheds on Oahu (particularly Kaukonahua stream) Future re-calibration of the Waikele model can be accomplished in house
•
• •
Additional funds would be required for independent QC of DOH’s HSPF model for Waikele* Limited technical support; no local expertise WARMF is not as established as HSPF
*If the WARMF model developed for Waikele stream is consistent with DOH’s TMDL calculations, then additional funding would not be required.
OPTION 3: Use current contract to have nhc QC DOH’s HSPF model for Waikele and develop N-SPECT model for Waikele. PROS • •
• •
•
CONS
ENV would have an expert objectively evaluate DOH’s TMDL model for Waikele ENV would have a means of easily applying land use change scenarios and assessing the effects of BMP implementation for the Waikele watershed (ENV would use the existing calibrated model developed by DOH with minor adjustments, if any, made by PB-nhc) Additional funding would not be required, only scope of work change order ENV would have the training to be able to develop and apply the N-SPECT model to other watersheds on Oahu Future re-calibration of the Waikele N-SPECT model can be accomplished in house
•
•
•
Page 2 of 2
N-SPECT model may not be comprehensive enough to accommodate land use change scenarios or BMP implementation N-SPECT model will not provide a defensible basis for contesting TMDLs in other watersheds (only provides order of magnitude estimates) Project schedule may be delayed since deliverable would be dependent on DOH’s timeliness in providing electronic files
Central Oahu Watershed Study Phase 2 Part A
APPENDIX F – DATA ASSESSMENT A review of available water quality, flow, meteorological, and air quality data was performed to determine what data are available for calibration of a Waikele Stream watershed water quality model for Part B of Phase 2 - developing a detailed hydrologic model. WATER QUALITY DATA A water quality data inventory for the watershed is tabulated in Table F1, showing the ten USGS stations and an additional four stations where data were collected by Oceanit (Personal Communication, Bob Bourke, Oceanit, August 2008). The stations are also presented in a figure at the end of this appendix - the ten historic USGS stations are shown as red squares, current USGS stations as black hexagons with smaller hexagons inset, and Oceanit stations as purple X’s. Additional data provided by Hawaii DOH was collected by TetraTech at USGS Station Waikele Stream at Waipahu but is not shown separately on the figure. While nearly one hundred water quality parameters are available between these sites only, TSS (Total Suspended Solids), conductivity, water temperature, nitrate/nitrite, total nitrogen, total phosphorus and turbidity data were assessed as part of this review. Water quality samples in the study area are available as instantaneous values, sequential samples (a series of instantaneous observations) and daily mean, max or min. The daily and sequential sample values are preferable for calibration because they are more likely to capture the peak value associated with a storm event than a single instantaneous value. Review of the available data revealed that with only three exceptions, most of the historic data are available as instantaneous values. The exceptions are: Waikele Stream at Waipahu (multiple parameters), Waikakalaua Stream near Wahiawa (water temperature), and Kipapa Stream near Wahiawa (TSS). The records most likely useful for calibration are Waikele Stream at Waipahu daily TSS values between 1972 and 1993 and sequential samples of TSS, Conductivity, Temperature, Nitrate/Nitrite, Total Nitrogen, Total Phosphorus and Turbidity collected by Oceanit in 2002. The 2002 Oceanit samples were also accompanied by concurrent instantaneous observations in tributaries to Waikele. Recent data at the four current USGS stations - Waikele Stream at Waipahu, Waikele Stream at Wheeler Field, Kipapa Stream near Wahiawa and Miliani Storm Drain A - also include sequential TSS observations that should be useful as the record continues in the future. FLOW DATA Eight of the ten USGS stations with water-quality data in the basin also have stream flow and/or stream stage data available. Unfortunately all but three of these stations only report annual peak or instantaneous observations which are not useful for continuous modeling, and one of those three is from a period prior to available water quality data. The two stations with daily data are Waikele Stream at Waipahu and Kipapa Stream near Wahiawa which report values from 1972 and 1968 to current, respectively.
METEOROLOGICAL DATA There are seventeen historical meteorological stations within two miles of the watershed and an additional solar radiation station in Hilo, Hawaii. The meteorological data are provided by
W9128A-05-D-0002
Central Oahu Watershed Study Phase 2 Part A
the National Weather Service (NWS) and the solar station by the Earth System Research Laboratory (ESRL); both are divisions of the National Oceanic and Atmospheric Administration. These stations and their periods of records are tabulated in the Table F2 below and are shown spatially in Figure F1. Looking first at precipitation data, fifteen stations have some precipitation data but the period of record varies. As the water-quality data review noted, the period between 1972 and 1993 is of interest because the availability of TSS, phosphorus, and nitrogen data makes that period likely for calibration. Ten of the NWS stations include data after 1972 and only eight after 1976. For similar reasons, the currently operating stations are of special interest because of new water quality data currently being collected by the USGS. Seven of the meteorological stations remain in operation today (2008). Of the seven stations currently operating, Camp 84, Wahiawa Dam, Mt. Kaala, Upper Wahiawa, Waiahole, and Waipahu are most likely to be used during calibration of current water quality data. For an unknown reason, the Kahana station has significant data gaps in recent years. The appropriateness of the Mt. Kaala, Waikane, and Waiahole stations should also be reviewed given the distance and topographic relief between the stations and the watershed. There is only limited meteorological data at stations other than Wheeler AFB. The original Wheeler AFB 810.1 station was decommissioned in 1999 and replaced with the Wheeler AFB OAHU station so these two records effectively provide one record from 1963 through 2008 with only a short gap between 1970 and 1973. The Wheeler AFB station provides temperature, wind speed, cloud cover, dew point temperature, and air pressure at hourly intervals. The Upper Wahiawa station, which began collecting air temperature data in 1997, is the only station other than Wheeler AFB with any significant meteorological data after 1977. The Waipahu and Pri Wahiawa stations have limited pre-1977 air temperature data but nothing after. Solar radiation is required by many water quality models but the parameter is not collected at standard NWS meteorological stations operated by the NWS. The NOAA ESRL provides solar radiation data at the Mauna Loa Observatory in Hilo, Hawaii, online between 1977 and 2004 and by request from 2004 to current. The post 2004 data is only available in 1-minute averages which would require some post-processing for use in an hourly or daily time step water quality model. AIR QUALITY DATA Air quality values are required for application of the WARMF model. There are air quality data available near Hawaii Volcanoes National Park from the National Atmospheric Deposition Program at the University of Illinois. During the period of 1980 to 1993 the station was called “Mauna Loa” and located at an elevation of 3399 meters. Then during the period of 2000 through 2005 the station was called “Hawaii Volcanoes National Park Thurston” and it was located at an elevation of 1195 meters. An EPA CASTNET air quality monitoring program also operated air quality monitoring equipment between 2000 and 2005 at the location of the Hawaii Volcanoes National Park-Thurston monitor. The influence of recent volcanic activity on air quality should be carefully evaluated before data from these stations are applied to the island of Oahu.
W9128A-05-D-0002
1600000
1610000
1620000
1630000
1640000
1650000
1660000
1670000
1680000
130000
130000
Central Oahu Watershed Study
Meteorology, Flow and Water-Quality Monitoring Stations Scale - 1:90,000 0.5 0.25 0
WAHIAWA MAUKA INTL 882