Sample input/output files for SWAT simulations
Funded By:
National Academies of Sciences Gulf Research Program
Funding Cycle:
Healthy Ecosystems 4
Research Group:
Development of Gulf Coast Resiliency Management Plan Using Sentinel Species and Natural Infrastructure
Gioia Kennedy
Environmental Defense Fund
gkennedy@edf.org
chemical transport, hydrology, hydrology model, watershed, climate
Abstract:
Soil and Water Assessment Tool (SWAT) is a widely-used semi-distributed watershed hydrology model, freely available and actively supported by a global community of modelers and developers. SWAT has been validated in the prediction of flashflood discharge and has been applied to the Galveston Bay region for estimation of fresh water and sediment inflows and the impact of precipitation and land use changes on nearby Mission-Aransas National Estuarine Research Reserve SWAT physical processes include: a rainfall/runoff curve number-based approach to surface hydrology, vertical and lateral groundwater transport and storage, reservoir dynamics, eroded soil and sediment processes, chemical fate and transport dynamics including partitioning, transformation, and transport in different media (soil, runoff, surface water, sediment), and land use management. Complete theoretical documentation may be found at: https://swat.tamu.edu/docs/. We ran SWAT under different weather and hypothetical chemical release scenarios. This dataset includes the base input files to SWAT from which the different simulations can be recreated by changing the weather (.pcp, .tmp), selected chemical (basins.bsn) and time period (file.cio) inputs accordingly as well as corresponding sample output files. SWAT provided estimates of chemical transport in stormwater, eroded soil, and rivers, accounting for drainage from the entire Galveston Bay watershed, as boundary conditions and inputs into the Delft3D coastal model used to estimate flooding. See the project website (https://createnbs.org/toxic-flooding/toxic-flood-modeling) for more information.
Suggested Citation:
Lauren Padilla. Sample input/output files for SWAT simulations. Distributed by: GRIIDC, Harte Research Institute, Texas A&M University–Corpus Christi. doi:10.7266/d0pxt0ds
Purpose:
In our study, we were concerned about chemicals that are persistent – present in the environment long enough to expose humans and ecosystems; mobile – readily transported from their industrial petrochemical sources in flood waters and eroded soil and sediment that are caused by stormwater and storm surge; and toxic – harm the health of humans, plants and animals who come into contact with them. Specific chemical classes of concern that carry these traits include: polycyclic aromatic hydrocarbons (PAHs), per- and polyfluoroalkyl substances (PFAS), and metals. Soil and Water Assessment Tool (SWAT) provided estimates of chemical transport in stormwater, eroded soil, and rivers, accounting for drainage from the entire Galveston Bay watershed, as boundary conditions and inputs into the Delft3D coastal model used to estimate flooding. In conjunction with Delft3D, the modeling results show where and how stormwater and flooding may affect facilities and move contaminants into vulnerable communities and ecosystems. In addition, we used the coupled model to evaluate the benefit of community master plans that include nature-based solutions in two case studies: Galena Park, TX and Texas City, TX.
Data Parameters and Units:
Full list of parameters can be found in the "swat-io-documentation-2012.pdf" document.
Methods:
SWAT requires input data defining weather, elevation/topography, soil, land cover, crop and land management practices, and lakes and reservoirs.We ran SWAT under different weather and hypothetical chemical release scenarios. This dataset includes the base input files to SWAT from which the different simulations can be recreated by changing the weather (.pcp, .tmp), selected chemical (basins.bsn) and time period (file.cio) inputs accordingly as well as corresponding sample output files. To account for the effects of variable weather and environmental conditions, we ran year-long simulations between 2005 and 2020. Without information on the amount and timing of past chemical releases, we modeled hypothetical releases and resulting chemical amounts and concentrations in waterways assuming that all industrial areas started with the same potential chemical on site per unit area at the beginning of each year. The result was a range of chemical transport estimates reflecting local differences in landscape characteristics (topography, soil, imperviousness) in and around industrial facilities and differences in the volume and flow rate of receiving waterbodies. To achieve more realistic chemical transport simulations, we customized the SWAT source code for chemical partitioning in streams. In the EDF custom version, we updated the sediment phase processes of deposition, resuspension, burial, and unburial so that they are directly coupled to modeled sediment fluxes and thus made dynamic and dependent on predicted sediment inflows, deposits and bank erosion that are a function of stream flows and conditions. Previously sorbed-phase chemical fluxes associated with these processes were constant, determined by user-provided settling and resuspension velocities. In addition, chemical buried by sediment was considered permanently removed from the stream. In EDF’s new version, buried chemical is stored and may resurface if the bed sediment in excess of the active surface sediment layer becomes resuspended in the stream. Aqueous phase processes are largely unchanged in the EDF version with the exception of a correction to the diffusion between pore water and stream water. Previously diffusion occurred proportional to the difference between total chemical mass in stream and sediment. Diffusion has been corrected to be proportional to the difference between chemical concentrations in stream water and pore water. The modified code is open source and publicly available on Github (https://github.com/edf-org/edf_custom_swat). The spatial domain of the SWAT model encompassed the drainage area of the San Jacinto River and other major waterways draining into Galveston Bay with two exceptions. We excluded the Trinity River basin in this model because of its smaller number of petrochemical facilities, large drainage area and outlet into a region of Galveston Bay distinct from the San Jacinto. We also excluded the drainage area around Texas City because the flat topography there made watershed delineation too uncertain given available elevation data. Using a digital elevation model, we divided the domain into 332 subbasins, balancing the need for higher spatial resolution against computational complexity. We accomplished this by delineating larger subbasins farther upland in rural and forested headwaters and finer resolution subbasins in the dense industrial areas that were potential chemical sources. We also increased resolution at the mouth of streams entering Galveston Bay, at the sites of stream monitoring gages and at the outfalls of Addicks and Barker Reservoirs, Lake Houston and Lake Conroe. Within each subbasin, we modeled unique combinations of dominant land use, land slope, and soil, also known as hydrologic response units (HRUs). Excluding some minor combinations limited to small areas, the model included 2,169 HRUs domain-wide. We ran year-long SWAT simulations for 2005 to 2020 with a daily timestep using historic weather data to evaluate present-day conditions. We also ran SWAT simulations using future weather predictions from climate models during baseline (2000-2019) and future periods (2040-2059 and 2080-2099) to evaluate effects of climate change. Historic (Jan 1980-Mar 2021) daily precipitation and temperature data developed and maintained by the USDA Agricultural Research Service was the starting point for developing weather inputs to SWAT. We retrieved gap-filled, continuous daily data at 33 real ground-based NOAA GHCN weather stations in the study area from this dataset. From this station network, we created virtual station data at the centroid of every subbasin in the model domain by spatial interpolation using inverse-distance-weighted averages of the timeseries data at the original 33 stations. For daily evapotranspiration and solar radiation values, we ran the Weather Generator that accompanies SWAT. Elevation data came from two sources, the more recent Upper Texas Coast 2-meter Topographic Lidar Digital Elevantion Model (DEM) LIDAR, where available in our study area (mainly Harris County), and the USGS National Map 1-arc second DEM for the rest of the domain. We merged these two datasets and resampled the elevation to a common 30 m x 30 m grid. Soil data came from the USDA Natural Resources Conservation Service Soil Survey Geographic Database (SSURGO). Customization of SSURGO soils included merging map units with like soil parameters and renaming them by their dominant component name. The result was a new simplified soil spatial boundary dataset of major soil components. Land cover data came from the USGS National Land Cover Database (2019). We used SWAT default values for runoff curve number and impervious surface fraction corresponding to each land cover class, except for high intensity, developed and industrial areas. We used Texas state land use codes to identify industrial land parcels that were sites for potential petrochemical releases. We parameterized these areas with higher runoff curve numbers characteristic of high impervious surface area and low vegetation. We used SWAT default assumptions for the management and timing of crop and plant growth with the exception of perennial forests and grasses. Plant growth in SWAT affects the model’s water balance via soil water uptake and evapotranspiration, which only happen during a plant’s growth stage and cease once the plant has reached maturity. For perennial forests and grasses, we increased the plant maturity time to ensure continuous water cycling and maintain evapotranspiration. Lakes and reservoirs in SWAT are important storage and flood control features affecting the timing and volume of water and sediment flowing through the stream network. We modeled four major reservoirs in the study area: Addicks and Barker Reservoirs, Lake Houston and Lake Conroe.
Provenance and Historical References:
Historic (Jan 1980-Mar 2021) daily precipitation and temperature data developed and maintained by the USDA Agricultural Research Service: White, M. J., Gambone, M., Haney, E., Arnold, J., & Gao, J. (2017). Development of a station based climate database for SWAT and APEX assessments in the US. Water (Switzerland), 9(6), 1–9. https://doi.org/10.3390/w9060437 We retrieved gap-filled, continuous daily data at 33 real ground-based NOAA GHCN weather stations in the study area from this dataset: https://www.ncei.noaa.gov/products/land-based-station/global-historical-climatology-network-daily For daily evapotranspiration and solar radiation values, we ran the Weather Generator that accompanies SWAT. Elevation data came from two sources. Upper Texas Coast 2-meter Topographic Lidar Digital Elevantion Model (DEM) LIDAR: https://doi.org/10.7266/2MYPTJ7Y USGS National Map 1-arc second DEM: https://www.geoplatform.gov/metadata/457a0cd3-83b2-4f58-9fd6-f87582a15951 Soil data came from the USDA Natural Resources Conservation Service Soil Survey Geographic Database (SSURGO): Soil Survey Staff, Natural Resources Conservation Service, United States Department of Agriculture. Web Soil Survey. Available online at https://websoilsurvey.nrcs.usda.gov/. Land cover data came from the USGS National Land Cover Database (2019): Dewitz, J., and U.S. Geological Survey, 2021, National Land Cover Database (NLCD) 2019 Products (ver. 3.0, February 2024): U.S. Geological Survey data release, https://doi.org/10.5066/P9KZCM54.