Environmental Management https://doi.org/10.1007/s00267-018-1041-6
Assessing the Hydrogeomorphic Effects of Environmental Flows using Hydrodynamic Modeling Angela Gregory1 Ryan R. Morrison2 Mark Stone1 ●
●
1234567890();,:
1234567890();,:
Received: 19 September 2017 / Accepted: 30 March 2018 © Springer Science+Business Media, LLC, part of Springer Nature 2018
Abstract Water managers are increasingly using environmental flows (e-flows) as a tool to improve ecological conditions downstream from impoundments. Recent studies have called for e-flow approaches that explicitly consider impacts on hydrogeomorphic processes when developing management alternatives. Process-based approaches are particularly relevant in river systems that have been highly modified and where water supplies are over allocated. One-dimensional (1D) and two-dimensional (2D) hydrodynamic models can be used to resolve hydrogeomorphic processes at different spatial and temporal scales to support the development, testing, and refinement of e-flow hypotheses. Thus, the objective of this paper is to demonstrate the use of hydrodynamic models as a tool for assisting stakeholders in targeting and assessing environmental flows within a decision-making framework. We present a case study of e-flows on the Rio Chama in northern New Mexico, USA, where 1D and 2D hydrodynamic modeling was used within a collaborative process to implement an e-flow experiment. A specific goal of the e-flow process was to improve spawning habitat for brown trout by flushing fine sediments from gravel features. The results revealed that the 2D hydrodynamic model provided much greater insight with respect to hydrodynamic and sediment transport processes, which led to a reduction in the recommended e-flow discharge. The results suggest that 2D hydrodynamic models can be useful tools for improving process understanding, developing e-flow recommendations, and supporting adaptive management even when limited or no data are available for model calibration and validation. Keywords Environmental flows Hydrodynamic modeling Adaptive management Sediment ●
Introduction The integrity of many river ecosystems is threatened or degraded by the regulation and over-allocation of water (Arthington and Pusey 2003; Hauer and Lorang 2004; Propst et al. 2008; Shafroth et al. 2010). Management activities, such as irrigation and municipal withdrawals, hydropower production, and water impoundment, lead to
* Angela Gregory
[email protected] * Mark Stone
[email protected] Ryan R. Morrison
[email protected] 1
Department of Civil Engineering, University of New Mexico, Albuquerque, NM 87131, USA
2
Department of Civil and Environmental Engineering, Colorado State University, Campus Delivery 1372, Fort Collins, CO 805231372, USA
●
●
changes in the natural flow regime of these systems (Poff et al. 1997) and can cause detrimental harm to a wide variety of ecological and geomorphic processes including riparian vegetation recruitment (Braatne et al. 2007; Morrison and Stone 2014, 2015a), fish reproduction (Lucas and Marmulla 2000), macroinvertebrate diversification (Vinson 2001), groundwater recharge (Fleckenstein et al. 2004), and sediment transport (Wohl et al. 2015) (see Poff and Zimmerman (2010) for a comprehensive review of the impacts of altered flow regimes). Environmental flows (henceforth e-flows) can be incorporated into management strategies to repair or prevent further ecological degradation caused by historical management activities. Intended to restore components of the natural flow regime—most notably the magnitude, duration, timing, frequency, or rate of change of flow (Poff et al. 1997)—e-flow implementations have been demonstrated in numerous river systems around the world, including the Palmiet River in South Africa (King et al. 2003), Latrobe River in Australia (Shenton et al. 2011), Colorado River in the United States (Sanderson et al. 2012), and Yellow River
Environmental Management
in China (Yang et al. 2008). More recently, e-flows have been incorporated within adaptive management frameworks to provide learning and improvement opportunities (e.g., Webb et al. 2018; Allan and Watts 2018). The success of eflow projects, however, depends on the specific ecosystem processes targeted by the project and the particular management challenges of each river system. The methods used to design e-flows are quite varied. Common approaches include system-based methods, which consider multiple water demands (such as hydropower, irrigation, and environmental needs) within a basin (e.g., King and Brown 2010; Morrison and Stone 2015b), and optimization approaches, which use operational functions to meet management goals and designs e-flows as either an operational constraint or objective (e.g., Shiau and Wu 2007, 2013; Yin et al. 2012). These methods are useful for exploring how e-flows fit into existing water management operations but are unable to inform managers about the effectiveness of e-flows in meeting ecological goals. Specifically, these methods cannot be used to examine the hydrogeomorphic impacts of flow alternatives, which have largely been ignored in e-flow studies despite their importance to river health (Wohl et al. 2015). The tools that are predominantly used to connect ecology with hydrogeomorphic processes mostly rely on simplified hydraulic models. For example, both the Physical Habitat Simulation System (PHABSIM) (Milhous et al. 1989), and the Functional Flows Model (FFM) (Escobar-Arias and Pasternack 2010) simulate one-dimensional hydrodynamics. These approaches rely upon simplified representations of boundary shear stress, velocity, and water depth. However, the spatial heterogeneity of channel geomorphology and hydrodynamics are important factors in determining the longevity and health of macroinvertebrates, algae, and fish (Bunn and Arthington 2002). Two-dimensional hydrodynamic models can be used to assist managers in planning and evaluating the success of eflow releases, especially when hydrogeomorphic processes are the primary interest. Two-dimensional flow models have the capability to represent spatially and temporally complex elements of flow that are important to ecological and geomorphic processes, such as velocity gradients, transverse flows, and directional shear stresses. The use of two-dimensional hydrodynamic models in ecological studies has steadily increased over the past two decades (e.g., Crowder and Diplas 2000, 2002, Dunbar et al. 2012). Hydrodynamic models have been used to evaluate mechanical river restoration projects (Pasternack et al. 2004), sediment dynamics of aquatic habitat (Gaeuman 2014), the impact of hydropeaking on hydrology as a function of river morphology (Vanzo et al. 2016), alteration of ecosystem function resulting from flow and channel form interactions (Lane et al. 2018), and flow dynamics of river
riffles and rapids (Reinfelds et al. 2010). Recent applications of hydrodynamic models specifically for e-flows include recommending fish spawning flows (Ban et al. 2011) and assessing habitat suitability for aquatic species in the Upper Delaware River, USA (Maloney et al. 2012, 2015). Even with these recent examples, however, hydrodynamic models are seldom used within decision-making frameworks to develop and implement e-flows. The objective of this study was to use two-dimensional hydrodynamic modeling to improve environmental flow recommendations and outcomes regarding hydrogeomorphic processes. We present a case study of e-flows on the Rio Chama in northern New Mexico, USA, where hydrodynamic modeling was used within a collaborative process to implement an e-flow experiment for improving ecological health in the river. A specific goal of the e-flow release was to improve spawning habitat for brown trout (salmo trutta) by flushing fine sediments from fluvial gravel features. The flow experiment was intended to flush fine sediments from gravel features at various locations within the river.
The Study System The Rio Chama originates in southern Colorado’s San Juan and Cumbres Mountains and flows south through northern New Mexico, USA, before entering the Rio Grande (Fig. 1). The hydrology of the Rio Chama is dominated by snowmelt runoff during the late-spring and variable monsoon storms that contribute short bursts of high flows from tributaries during summer monsoon months (Swanson and Meyer 2014). The 8300 km2 watershed includes three dams (Heron, El Vado, and Abiquiu) that are used to manage water for irrigation, recreation, flood control, and municipal needs. The San Juan-Chama Project, a trans-basin water transfer project completed in 1970 and located upstream of Heron Dam, delivers up to 3 × 108 m3 of additional water to the Rio Chama each year from the Colorado River Basin (Flanigan and Hass 2008). The operation of El Vado Dam has increased base flows and reduced peak flow frequencies and magnitudes from background conditions (Fig. 2) (Morrison and Stone 2015b). In 1988, approximately 40 km of the river downstream of El Vado Dam was federally designated as a Wild and Scenic River, prompting instream flow recommendations to maintain flows greater than 11.3 m3 s−1 between October and March for optimum spawning conditions and base flows of 5.2 m3 s−1 to maintain macroinvertebrate food sources for brown trout (Fogg et al. 1992). These recommendations have been met 34 and 72% of the time, respectively, based on our statistical analysis of discharge records from USGS gage 08285500, located immediately downstream of El Vado Dam. Additionally, a peak flow of
Environmental Management Fig. 1 Map of Rio Chama watershed including field sites and location within the Rio Grande basin
142 m3 s−1 was recommended to occur every 5–10 years to maintain channel forming processes that enhance generation of cottonwood trees (Fogg et al. 1992). The geomorphic characteristics of the Rio Chama are complex, with sediment sizes ranging from fine to boulder classifications. Historically, cobble and gravel from above El Vado Dam provided coarse material to the reach (Swanson et al. 2012). Since the installation of the dam, gravel and cobble sources have been limited to materials stored in the banks and delivery of materials from debris flows and tributaries (Swanson et al. 2012). Shale layers in the channel bed are imbricated and act as armoring to the material below in some reaches. In addition, a series of
tributaries contribute erodible shales and sandstone into the Rio Chama during monsoon storms and debris flows (Swanson and Meyer 2014). The reduction in frequency of maximum flows has caused channel narrowing and downcutting (Swanson et al. 2012), reduced floodplain connectivity, decreased the frequency of flushing flows, reduced spawning habitat for brown trout, and decreased the stream power required to rework channel material (Swanson and Meyer 2014). The quality of brown trout spawning habitats are dependent on factors such as water depth and velocity, substrate size, dissolved oxygen concentration and the quantity of fine sediments present (Armstrong et al. 2003;
Environmental Management 200 Discharge (cms)
(A) Rio Chama upstream from El Vado Dam USGS 08284100 150 100 50 0 Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
200 Discharge (cms)
(B) Rio Chama downstream from El Vado Dam USGS 08285500 150 100 50 0 Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Fig. 2 Comparison of discharge records (gray bands encompass 10percentile to 90-percentile values, black lines represent median values) for a USGS gage 08284100 (located upstream of El Vado Reservoir) and b USGS gage 08285500 (located downstream of El Vado Reservoir). Also included are discharge values for 2005, which
highlights the typical changes in hydrologic conditions created by the reservoir. As shown in the two plots, high flows that occur during the snowmelt season have been significantly reduced, and reservoir operations cause small weekly spikes in summer flows to accommodate recreational rafting
Chapman 1988; Crisp 2000). Brown trout prefer gravel sized material for redd development. Kondolf and Wolman (1993) reported that desired median particle sizes for redd development ranges from 5.8 to 50 mm. As the presence of fine sediment increases, trout embryo development is decreased and macroinvertebrate mortality is increased (Olsson and Persson 1988; Soulsby et al. 2001; Turnpenny and Williams 1980). Mechanisms for improving spawning habitat include localized installment of instream structures, flushing flows, and high flow gravel injection. Of the three options available, flushing flows are the least expensive and physically intrusive. Flushing flows have long been recommended as a means for conditioning spawning habitat in coarse-grained rivers through removal of fine sediments, though success rates have varied (Petticrew et al. 2007). However, the peak flow target described above was not intended to flush fine grained materials from sediments for maintenance of brown trout spawning areas.
the 2D hydrodynamic model was developed to guide general e-flow recommendations, here we emphasize the model utility as related to brown trout spawning habitat. The site selection and e-flow recommendations were made collaboratively by a group of geomorphologists, hydrologists, and ecologists (henceforth referred to as the collaborative group). The decision framework for developing e-flow recommendations consisted of the five phases summarized in Fig. 3.
Environmental Flows Process E-flow recommendations were developed for the Rio Chama using an integrated approach that included stakeholder participation, baseline sampling, collaborative workshops, and hydrodynamic modeling (Morrison 2014). The overall purpose of the flow recommendations was to improve aquatic and riparian conditions of the Rio Chama between El Vado Dam and Abiquiu Reservoir. Although
Phase 1: Data Collection The collaborative group selected four project sites between El Vado Dam and Abiquiu Reservoir that were representative of geomorphic and ecological conditions of the overall reach. These sites are referred to as Archuleta, Dark Canyon, Cebolla, and Benson’s Bar (Fig. 1). Access to all four sites is extremely limited due to steep terrain and unmaintained roads. Data were collected at each site to describe channel and floodplain topography, sediment characteristics, floodplain vegetation, benthic macroinvertebrate composition, and fish populations. The data was used to support one-dimensional hydrodynamic model development and to serve as a baseline for observing geomorphic and ecological change over time. Additional topographic and sediment data were collected at the Archuleta and Cebolla sites to develop two-dimensional hydrodynamic models. These sites were chosen because of their locations near major tributaries that contribute fine sediments to the Rio Chama.
Environmental Management
PHASE
1 DATA COLLECTION
2 1D MODELING
3 EXPERT WORKSHOP
4 2D MODELING
5 IMPLEMENT FLOWS
PROCESS: Key steps during each phase De ne project objectives Collect survey data at key study sites Site exploration Initial environmental ows hypotheses Develop 1D hydraulic model based on survey data and site exploration Evaluate model results for various discharge Re ne environmental ow hypotheses according to model results Invite participants to interdisciplinary workshop Develop environmental ow recommendations based on expert opinion and model results Identify data gaps Re ne environmental ow recommendations for speci
ecological
Develop 2D hydrodynamic model to test recommendations Determine target ow release based on model results Field-test environmental ow recommendations Collect eld data before and after ow release Analyze data and evaluate success of ow release Re ne recommendations and 2D hydrodynamic model
ITERATIVE PROCESS
Fig. 3 The key phases of implementing environmental flows on the Rio Chama
Due to space constraints, we have focused this paper on Archuleta site, which is located 11.7 km downstream of El Vado Dam (36°32′N–106°44′W). Detailed site descriptions and results for the remaining sites can be found in Gregory (2013). The Archuleta reach is approximately 875 m long and has an average slope of 0.002. The Rio Nutrias enters immediately upstream of the reach and the Arroyo del Puerto Chiquito (Chiquito) enters the reach 330 m from the top of the reach. The channel primarily consists of gravels with fine sediment deposits located at the confluence of the Rio Chama and Chiquito, and at the heads and tails of vegetated bars. Wolman pebble counts (Wolman 1954) were used to estimate the median particle size within the channel, which was between 32 and 64 mm depending on the location within the channel bed.
conditions in the Rio Chama. The workshop participants consisted of a wide range of disciplines including riparian ecologists, geomorphologists, fish ecologists, and water managers. The workshop was initiated by sharing results from the monitoring activities and 1D hydrodynamic models to inform workshop participants about existing conditions and the potential implications of e-flow alternatives. The workshop participants concluded that e-flows should focus on improving riparian vegetation recruitment, steady base flows, floodplain-channel connectivity, and flushing of fine sediments (Morrison and Stone 2015a). The participants also agreed upon e-flow recommendations to meet these ecological needs (Fig. 4). The e-flow recommendation that was the focus of our 2D hydrodynamic modeling efforts was a release of approximately 70 m3 s−1 from El Vado Dam during the spring season of 2014 to flush fine sediment from gravel bars used by brown trout for spawning. It was recommended that such a flushing flow should be targeted every one to two years and be maintained for a period of 3–5 days.
Phase 4: Two-Dimensional Hydrodynamic Modeling A 2D hydrodynamic model was applied to simulate steady state hydrodynamic conditions at the Cebolla and Archuleta sites. The details of the 2D model development are provided in the methods section below. The use of 2D modeling fits within existing e-flow methodologies, such as Ecological Limitations of Hydrologic Alterations (ELOHA) approach, which advocates for understanding habitat responses to hydraulic conditions (Poff et al. 2010). However, using 2D modeling in e-flow studies moves beyond the regional- to segment-scale habitat-hydraulic relationships supported in the ELOHA methodology and provides an improved approach for relating e-flows to hydrogeomorphic goals.
Phase 2: One-Dimensional Hydrodynamic Modeling Phase 5: E-Flow Implementation One-dimensional hydrodynamic models were developed by the collaborative group for all four project sites using the U. S. Army Corps of Engineers’ Hydrologic Engineering Center River Analysis System (HEC-RAS 2010) software. The flows were assumed to be steady with normal-depth boundary conditions. The models were primarily developed to determine stage-discharge curves for each site and to estimate mean channel and overbank discharges (Morrison and Stone 2015a). The results were a key input for the workshop described in Phase 3.
Phase 3: Expert Workshop A collaborative workshop was held in March 2013 to specify e-flow targets for improving specific ecological
An experimental e-flow release was administered from April 25 to April 27, 2014. The collaborative group designed the release hydrograph with the objective of mobilizing fine sediments while considering other constraints, particularly the limited volume of water available to support the e-flow (Morrison 2014). A peak flow of 58 m3 s −1 was released for a 24-h period before being decreased to 37 m3 s−1 for an additional 24 h and then back to a baseflow of 8 m3 s−1 (Fig. 5). The flow adjustments were blocky in appearance because dam operations were manually adjusted at fixed points in time. The effectiveness of the experimental release was assessed by monitoring changes in fine sediment distributions, and the e-flow recommendations were further investigated and refined using the 2D
Environmental Management Fig. 4 Environmental flow recommendations for improving ecological condition on the Rio Chama below El Vado Dam
Ecological Function
Brown Trout Spawning
Sediment Flushing
provide reliable spawning habitat
from spawning gravels and rework inchannel features
3
3
Riparian Veg Recruitment
O ff-Channel Habitat
Encourage cottonwood recruitment in
Create off-channel habitat and maintain
3
3
Discharge
Less than 3 m /s
70 m /s
140 m /s
170 m /s
River Stage
In-channel
Bankfull to slightly overbank
Overbank
Much higher than overbank
Recurrence Interval
Annually late summer to spring season
Every 1–2 years
Every 3–5 years
Every 10 years or when hydrology conditions allow
Hydrograph Pattern
Consistent during base
Consistent during
Maintained for at least 2 days; recession rate limited to 2-5 cm/day
Maintained for at least 2 days
maintained for 2 days
Fig. 5 Hydrograph of environmental flow release from El Vado Dam in April 2014
hydrodynamic model, as further described in the “Methods” section below.
Methods Two-Dimensional Hydrodynamic Model Development A 2D hydrodynamic model was developed using the United States Bureau of Reclamation’s (USBOR) Sedimentation and River Hydraulics—Two Dimensional (SRH-2D) model. SRH-2D was applied at the Archuleta and Cebolla sites for flows between 14 and 168 m3 s−1 at 14 m3s−1
increments. The details of the Archuleta model are provided here and the Cebolla model and results are described in Gregory (2013). SRH-2D is based on the depth-averaged St. Venant’s equations, which are derived by depth averaging the three-dimensional Navier-Stokes equations (Lai 2009). SRH-2D model development requires the generation of a 2D numerical mesh that is based on topographic data, specification of upstream and downstream boundary conditions, and specification of Manning’s roughness coefficients. The topographic mesh was developed using interpolated point data collected through cross-sectional surveys of the channel and floodplains. The numerical mesh had over 111,000 elements with a mean resolution of approximately 1 m. Following the methodology of
Environmental Management
Bockelmann et al. (2004), the boundary conditions were defined using the results from the HEC-RAS models over the range of modeled flows. Channel and floodplain Manning’s roughness coefficients were selected using a combination of field notes, aerial photographs, Strickler’s method (1923) for calculating roughness values of gravel bars, and Chow’s values (1955) for other surface types. Roughness values ranged from 0.033 to 0.060. SRH-2D includes two options for the estimation of turbulence. The first is the parabolic turbulence equation that can be derived from averaging eddy viscosity over the water depth (Elder 1959). The second turbulence model is the k-ε model (Rastogi and Rodi 1978), which accounts for turbulence mixing lengths and requires multiple parameter inputs. Due to the lack of data available for model calibration and relative simplicity, the parabolic turbulence model was applied with a coefficient value of 0.7, which is the default value (Elder 1959; Fischer et al. 2013; Lai 2009).
Shear Stress and Sediment Incipient Motion Analyses Boundary shear stress (τb) was determined for a range of spatial scales to highlight insights provided through 2D hydrodynamic modeling and to support sediment calculations as described below. Reach scale boundary shear stress was estimated based on the approximation of τb = γRSo where γ is the specific weight of water, R is the hydraulics radius, and So is the bed slope. R was determined as the average value determined with HEC-RAS. Cross-sectional bed shear stress was determined directly from the HECRAS results. Bed shear stress was also investigated for the 2D model domain and was provided as output from the SRH-2D model. Incipient motion conditions were estimated for every 2D grid cell using the Shields–Rouse equation (Guo 2002). For each of the simulated flows, the Shields–Rouse equation was applied to determine if the critical shear stress (τ*c) for incipient motion was exceeded for the target particle size of 5 mm. Thus, a threshold discharge for mobilizing fine sediments was identified for each grid cell. We refer to this threshold flow as the critical discharge. The critical discharge values where compiled in a map that summarizes the estimated discharge required to flush fine sediments for the full the model domain.
Monitoring Process Sediment samples were collected along the margins of five gravel bars (1 at Archuleta, 3 at Benson’s Bar, and 1 at Dark Canyon) without replication prior to the e-flow release. The sites were selected based on their accessibility during preexperimental flow conditions and because fine sediment
deposits were visually observed at these locations before the experimental release. Samples were collected from the top 2 to 3 cm of sediment over a 330 cm2 area at each location. Samples were collected at the same five sites following the e-flow event, again without replication. The percent fines contained in each sample before and after the e-flow event was determined using ASTM D6913-04 (2009) and ASTM D2487-11 (2011) for classification of particle size. Fine grained materials were classified as all sediments that passed a number 4 sieve (