Current BASINS 2.0 subwatershed
delineation toolCurrent BASINS 2.0 subwatershed
delineation tool
BASINS is designed to support analysis of environmental systems within a watershed-based study area. The program facilitates watershed analysis by bringing data for surface land and water features, water quality monitoring, and pollutant sources together in one computer environment. Using ArcView as an integrating framework, BASINS allows the user to examine water quality data and related features in their spatial context, and select watershed data for input into water quality modeling programs.
BASINS hardware and software requirements can be seen on the EPA's BASINS web site. It is important to distinguish between BASINS 1.0 and BASINS 2.0. BASINS 1.0 is available from the EPA regional office on CD-ROM, and may be ordered from the EPA BASINS web site. In the experience of a few of us who have ordered it, delivery from Region 6 is fairly prompt (within a week or two). BASINS 1.0 is being superseded by BASINS 2.0, which is currently in the beta-test stage, and may be downloaded from the EPA BASINS web site. BASINS 1.0 will only run with ARCVIEW 2.1. BASINS 2.0 will only run with ARCVIEW 3.0a.
The best way to familiarize yourself with BASINS is to work through the exercises provided by the "BASINS Training Workshop" tutorial. This workshop can be downloaded from the BASINS library web site. Currently, in the Civil Engineering Learning Resources Center (LRC), BASINS 2.0 is installed on three PCs. For lack of a better way of locating them, they are the three workstations, starting one in from the aisle, on the back row, in the back room of the LRC, 3.400. These exercises work through basic utilization of all the BASINS analytical tools, using the state of Georgia as a study area. The necessary data is already present on the PCs in the LRC. These PCs all have the BASINS tutorial data installed on the hard drive, and available for use. Ann Quenzer, a former graduate research assistant, has also prepared some introductory BASINS exercises set in the Austin area. These exercises are located at http://www.ce.utexas.edu/stu/quenzeam.
Outside of the tutorial, data for use in BASINS 2.0 may be taken from the BASINS 1.0 CD-ROM for the appropriate EPA region or downloaded from the web site. There are a few specific files that have been updated for BASINS 2.0 and must be downloaded as well. Unfortunately, these files are only available for some locations in the country.
A BASINS project is initiated with the data extraction tool. This tool selects your study area by state, county, or HUC from the regional information provided on the BASINS CD. It saves the coverage information for the extent of the selected study area as a distinct folder in the BASINS data folder. The BASINS project builder tool is then used to build an ArcView project file with all the necessary BASINS scripts and referencing the data file for your study area for the available themes. Currently, there seem to be some problems with the project building process in the WindowsNT environment, but EPA has provided a workaround in the form of a specific ArcView project file, build.apr, which may be downloaded from the web site.
Tools for performing analysis in BASINS can be grouped into two modules :
b. ASSESS is designed to operate on a particular watershed. ASSESS operates on the same data as TARGET, but at a finer resolution, allowing you to examine the spatial distribution of pollutant problems, and the effects on specific stream reaches.
b. QUAL2E, which allows fate and transport modeling of both point and nonpoint source loadings to selected stream reaches. The program is accessed through a Windows interface. Stream reaches and criteria pollutants are selected within BASINS, which then formats the input for the model. BASINS also allows for nonpoint source loadings generated in an NPSM output file to be input into the QUAL2E model. Output from QUAL2E can be viewed graphically within the QUAL2E Windows interface or it can be viewed spatially with the BASINS Visualization script. This script allows model output for the selected reaches to be examined from the BASINS project view.
c. and TOXIROUTE, based on the EPA Pollutant Route (PROUTE) model. It uses a first-order decay model to simulate pollutant transport in selected reaches under either mean flow or low (7Q10) flow conditions. BASINS prepares input for point source loading from available Permit Compliance System (PCS) data and can also incorporate nonpoint source loadings generated from an NPSM output file. The model output can be viewed in a tabular format, listing concentrations for each selected reach, or it can be viewed spatially within the BASINS project view, using the Visualization script.
The process of watershed delineation described in this project is based on a series of procedures using both ArcInfo and ArcView. For this project, I used A UNIX system with ArcInfo which was separate from the PC system with ArcView. There are some steps below which are only necessary for transfering information between the two environments. In "Future Work," below, I address the issue of automating the following procedures and putting them into the BASINS environment.
Note : If you have downloaded a compressed DEM, you must uncompress it before continuing.
2. Convert the USGS DEM into an Arc/Info grid coverage.
Arc : demlattice
<"USGSfile"> <"gridDEM"> usgs
3. Merge multiple grids as necessary.
Arc : grid
Grid : <"DEM">
= merge ( "gridDEM1", "gridDEM2", etc...)
4. Projecting the DEM into the BASINS Albers projection. Currently, all BASINS 1.0 data is presented in a standard United States Albers projection. Coverages imported into BASINS must first be rendered into this projection. When created, USGS DEMs are transformed into either a WGS72 or WGS84 datum. The BASINS data uses the NAD27 datum. The difference in the initial transformations of the WGS and NAD datums prohibits simply projecting from one coordinate system directly to the other. Reprojection involves a three-step process which can only be accomplished in Arc/Info. This procedure can be found in the Arc/Info online Help section, under "steps for converting WGS-NAD".
For this project, I have prepared the projection files necessary to project from WGS72 to NAD27.
Prjstep1 Prjstep2
input
input
projection geographic
projection albers
units ds
units meters
datum WGS72 seven
datum NAD83
parameters
parameters
output
29 30 0.000
projection albers
45 30 0.000
units meters
-96 0 0.000
datum nar_c
23 0 0.000
parameters
0.0000
29 30 0.000
0.0000
45 30 0.000
output
-96 0 0.000
projection albers
23 0 0.000
units meters
0.0000
datum NAD27
0.0000
parameters
end
29 30 0.000
45 30 0.000
-96 0 0.000
23 0 0.000
0.0000
0.0000
end
Important! : At this point, you must complete another step before proceeding. From your selected reaches, you must continue to select downstream reaches until your stream network coverage will leave the boundary of the DEM to be used in watershed delineation. This is necessary to ensure proper processing by the hydrologic modeling tools. Use the projected DEM as a guide if necessary. Automation of this procedure will be addressed in future work.
2. Convert the selected reaches into a new shapefile, using Theme/Convert to Shapefile. Transfer the shapefile to your ArcInfo workspace.
3. Convert the shapefile to a line coverage.
5. This raster representation of the Rf1 stream coverage will not be useful for watershed delineation. The normal flow direction and accumulation methods used in determining streams from a DEM use the eight-direction pour point model as their basis. This model defines flow from cell to cell of an elevation grid in the direction of steepest descent. With a "filled" DEM (a DEM that has been conditioned for hydrologic modeling), there can only be one direction of flow from each cell. In this manner a linear flow direction grid is established, that, when combined with a flow accumulation grid, defines a stream network. When converting a stream vector coverage to a grid, however, this unique relationship of flow direction is not preserved. Where a vector overlaps neighboring cells, the stream grid will identify all of these cells as part of the stream path, creating the potential for cells with multiple flow directions. This problem may be solved by using the Grid command Thin. Thin reduces a linear raster feature to one cell width. Applying the Thin command to the Rf1grid will produce a stream network identical to one defined from the flow direction of a filled DEM.
6. If necessary, export the stream network grid for watershed delineation in ArcView.
2. "Burn" the Rf1 stream network onto the DEM. The process
of burning DEMs has been developed to adapt a vector stream coverage to
a DEM. Once the vector stream network is converted into a grid stream
network of unit value (all stream cells = 1, all other cells = NODATA),
it is multiplied by the DEM. This produces a grid where each stream
cell has the elevation value of the terrain it overlies. The entire DEM
is then raised by a large constant elevation, and the stream elevation
grid is reinserted. The final burned DEM then consists of the stream
cells with their corresponding original elevations, while the surrounding
terrain retains its relief, but at a much higher elevation than the stream
network. When the burned DEM is filled, further development of the
flow direction and flow accumulation will be based on the burned in stream
coverage.
(b) In Analysis/Map Calculator, multiply the unitstream grid by the DEM, {unitstream}*{DEM}. Using Theme/Properties rename the resulting map calculation, "demstrm."
(c) In Analysis/Map Calculator, add a large constant, for example 5000m, to the DEM, {DEM}+5000. Using Theme/Properties rename the resulting map calculation, "demplus."
(d) Burning the DEM requires a script. The script used here
is short and simple, but requires that the input themes and view name be
set as entered in the script. For this script, the active view must
be named "View1" and contain the themes "demstrm" and "demplus".
In the Project window, double-click on the Script icon and
type in the following script :
TheView = av.Getproject.FindDoc(ViewName)
aTheme = TheView.FindTheme(aGridName)
aGrid = aTheme.GetGrid
bTheme = TheView.FindTheme(bGridName)
bGrid = bTheme.GetGrid
listGrid = {bGrid}
outGrid = aGrid.Merge(listGrid)
outGrid.SaveDataSet("<your working directory path>/burndem".AsFileName)
outTheme = GTheme.Make(outGrid)
TheView.AddTheme(outTheme)
Unfortunately, this watershed coverage is not compatible with the NPSM input preparation. Use of these watersheds in NPSM generates an error message, to the effect that some information is missing. By comparison, the following table shows the attributes of a subwatershed prepared with the BASINS delineation tool. You can see that it includes several fields that are missing from an imported polygon coverage.
To solve this problem, add the missing fields to the attribute table of the imported subwatersheds. Open the attribute table of the imported subwatershed coverage. Select Table/Start Editing. Use Edit/Add Field to add the following fields :
Using this method, I have been able to run NPSM. Note that there are no values in the added fields. At this time I do not know what information is actually missing. I have asked the EPA technical support people for an answer, but, as yet, have not received an answer.
Here is an example of the NPSM output for the runoff from agricultural land in one subwatershed of the study area during 1993 :
Here is an example of the NPSM output for the fecal coliform load from agricultural land in one subwatershed of the study area during 1993 :
Taking a stream network along the Flint River, east of Atlanta, Georgia, I applied both methods of delineating subwatersheds to the RF1 stream reaches. The differences in the subwatersheds can be seen :
It appears obvious that total area and land use change for each subwatershed depending on how it is delineated. To look more closely at these changes in one particular subwatershed, click on the charts below :
The effect of these differences in area and land use on the subsequent runoff and load calculations can be seen in the example NPSM output in Running NPSM, above. These plots show the runoff and fecal coliform loads produced by the two methods of watershed delineation over the same period of weather. In this particular case,the peak runoff differs from 200 to 500 cfs, and the total fecal coliform load differs correspondingly. These are significant differences that will effect watershed planning and decisionmaking. While there are inherent uncertainties in estimating rainfall-runoff relationships and EMCs from land use types, the uncertainty will only be compounded by poorly defined subwatersheds. The more accurately the subwatersheds are defined, the better the runoff and loading estimates will be.
Using the procedures in this project it is will also be possible to create a watershed delineation tool for BASINS that will allow for the introduction of different digital elevation data. The work that I have done to introduce digitally delineated subwatersheds based on RF1 stream coverages into BASINS 2.0 consists of a series of procedures using both ArcInfo and ArcView. It remains necessary to use ArcInfo to project the DEM as long as this data is not provided in the BASINS projection. A digital elevation coverage, however, is promised to be included with the final release of BASINS 2.0. Using Avenue scripting then, the potential exists to automate these procedures, and reduce the user burden to simply selecting the reaches to have subwatersheds delineated. A conceptual series of scripts to accomplish this is shown here :
These scripts currently exist, with the exception of those highlighted :
2. Convert reaches to grid (and THIN) : These procedures can be accomplished by the Spatial Analyst extension to ArcView.
3. Process watersheds for import : Given the problems encountered in this project in making a subwatershed coverage compatible with the NPSM input requirements, it is probable that some manipulation of the subwatershed attribute table will be necessary.
"Spatial Hydrology of the Urubamba River System in Peru Using Geographic Information Systems (GIS)." F. Olivera. January 1996.