r.wrat

NAME

wrat - Water Resource Assessment Tool

USAGE

This module must be used interactively.

INPUT MAP CODES

elevation: meters, as well as cell resolution

rainfall maps: 100ths of an inch

K factor: K factor times 100 K of .35 = 35

Introduction

The Water Resource Assessment Tool is a collection of programs run within the GRASS GIS. These programs are an aid in understanding the nature of runoff in a study area based on information typically available for a GIS. The programs analyze the terrain to define drainage direction and areas, simulate runoff and peak runoff and model nonpoint source pollution and map contaminant source areas and contaminant pathways. This users guide outlines the typical analytical sequence, describes input data requirements and possible output. Some suggestion for interpreting output are included. It is assumed the user is familiar with the basics of GRASS.

Input

Map layers required: digital elevation model, land cover, soil maps of hydrologic soil group, soil texture, and erodiblity factor (K factor). Optional maps describe best managment practices for sediment, nitrogen phosphorous and chemical oxygen demand (COD).

Typical Analysis

Three interrelated areas of analysis are available: terrain analysis, hydrology and pollution (see figure X1). Because these areas are not distinct, analysis should proceed in a logical order since some portions build on previous results. Terrain analysis and runoff generation must proceed routing peak discharge, which in turn must proceed contaminant routing. The following steps provide a natural sequence of analysis through all of the tools. However, a great deal of valuable information may be gained executing only a portion of the available programs. This is a tool kit, not a prescription for studying every area.

Terrain Analysis	Hydrology		Pollution
_________________	_________________	_________________
|               |       |		|	|		|
| Elevation     |	| Land Cover	|	| Land Cover	|
| Model         |	|		|	|		|
--------+--------	| Hydrologic	|	| Soils		|
        |		| Soil Group	|	| K factor	|
________V________	--------+--------	| Texture	|
|		|	        |		--------+--------	
| Idealized	|		|			|
| Elevation	|		|			|
| Model		|		|			|
|		|		|			|
| Drainage	|		|			|
| Direction	|		|			|
|		|		|			|
| Drainage	|	________V________		|
| Accumulation	|	|		|		|
|		|	| Runoff Map	|		|
| Slope		|	-+------+--------		|
--------+--------	 |	|			|
	|	     +----	|			|
	--------+----(--------->|<-----------------------	
		|    |	________V________	_________________
		|<---+	|		|	|		|
		|	| Contaminant   |	| Best 		|
		|	| Source Areas  |	| Management	|
		|	--------+--------	| Practices	|
	________V________	|		--------+--------
	|		|	|			|
	| Peak 		|	------------+------------
	| Discharge	|		    |
	-----------------	    ________V________
				    |		    |
				    | Contaminant   |
				    | Routing	    |
				    -----------------

			figure X1.

Defining the study area

GRASS looks at the world through a rectangular "window". The window should be set to include the entire area of interest with several cells to spare in all directions. Because of difficulty interpreting terrain at the edge of the window, the analysis moves in from the window edge. The extra area is particularly important if the top of a watershed of interest is not marked by a steep decline into the adjacent watershed.

Routing routines assume that the entire watershed is within the current window. This is not always true or even practical. For instance, an analysis along the side of a long river will provide valuable information about local contributions to the larger resource. However, calculations in the long river itself can not take into account upstream effects outside the current window and are thus invalid. Yet those local contributions may be of considerable interest. Great care must be taken interpreting results of watersheds only partially contained within the operating window.

Cell resolution, the size of each cell, should be set as large as is appropriate for your purposes. In no case should the resolution be smaller than the spatial resolution of your best input layer. Doubling the cell length and width will cut computation times by 3/4s! The east-west resolution need not match the north-south resolution. However, large deviations from square cells will distort some of the neighbor concepts used by some algorithms to "feel their way" across the terrain.

Once the analysis is begun, the same window, including cell resolution, must be maintained throughout the project. The interface program tries to insure this. Changing the boundaries invalidates the drainage accumulation maps. Changing the cell resolution can have disastrous effects on the drainage direction map, such as creating dead end and circular drainages. The idealized elevation model will lose its "ideal" character if resampled. If the window must be changed, start the analysis from the beginning.

Some of the output maps may be resampled to a smaller window or different cell size after the analysis is finished. Contaminant source area maps are produced in units of mass per unit area and may be resampled. This may be useful for presentation of results or incorporation within another GIS analysis.

Temporal scale

Hydrologic simulations are based on single storms, presumably one day events. Actual or design storms may be employed.

Terrain analysis

The actual analysis may begin using the digital elevation model to produce an idealized elevation model. Real data has, for a number of reasons, sections which are difficult to drain. The idealized elevation model has no cells which can not drain to the edge of the window. This assumes complete surface drainage is possible. This condition is never completely true and in some regions, such as those with karst topography, so far from the case as to invalidate the peak discharge and contaminant routing results.

The idealized elevation model is used to determine the direction in which each cell will drain. This is done by searching the edge of the window for the lowest point and draining the watershed that leads there. Then the next lowest outlet is found and its watershed drained, and so on until the entire area is drained.

The drainage direction map is used to produce a drainage accumulation map. The values in a drainage accumulation map are the total number of cells including that cell, which drain through that cell. If multiplied by the area of a cell, the drainage accumulation map will yield the area of the watershed above and including each cell.

Hydrology

Two programs contribute to the hydrology section: runoff and peak discharge. The runoff section is based on the Soil Conservation Services Curve number method (Soil Conservation Service, USDA, 1971). Curve numbers, from 0 to 100, describe the potential for runoff. Curve numbers are inferred from the land cover map and the map of hydrologic soil groups. A depth of precipitation for a design storm or a rainfall map is required to produce a runoff map.

To this point the terrain analysis and the hydrology could progress independently. From this point on the distinctions between terrain analysis, hydrology and pollution, as separate lines of analysis begin to fade. Actually pollution analysis will always be dependent to some extent on hydrology but that will be dealt with in a later section. Peak discharge combines the results of the terrain analysis and runoff sections.

Peak discharge, the volume of water passing out of a cell per unit time, is calculated with an empirical relationship developed by Smith and Williams (1980) and employed in CREAMS and AGNPS. Watershed characteristics are interpreted from several all of the results from the terrain analysis. The resulting map of peak discharge, significant by itself, is also used in the routing of contaminants.

Pollution

Contaminant sources are mapped as the contaminant originating per unit area. Sediment, nitrogen, phosphorous and COD may be modeled. The algorithms used to predict contaminant mobilization depend on the landcover of the cell involved. The universal soil loss equation and adaptations from AGNPS and CREAMS are employed for rural, agricultural and open lands. An algorithm from SWMM is adopted for urban areas. Land cover, soil texture and erodibility factors, rainfall, runoff and peak discharge maps are the required inputs for this section.

The final part of the analysis routes contaminants. It requires the results of the preceding sections. Sediment, nitrogen and phosphorous nutrients and COD may be modeled. The routing is done as a mass balance accounting for imports of contaminants into a cell, contaminants originating within the cell, losses to infiltration and deposition and exports to the next cell. Figure X2 is a schematic representation.

          
                       rainfall with
                       contaminants
                             |
                       ______________
imports from          /             /       Surface exports
up gradient      -   /             /  -    down gradient
                    /_____________/
 
                           |
 
                   losses to infiltration
                   and deposition
 
                                  figure X2.

Some practical considerations:

GRASS only stores integers, assuming values less then one to be equal to zero. Therefore small units of measure have been employed. Rainfall and runoff are described in hundredths of inches, while quantities of sediment are in kilograms with nutrients in grams. Flows are measured in cubic feet per second. The mixing of english and metric units employs those units in common usage in the U.S. Maps may be converted to any appropriate set of units using "r.mapcalc" to multiply the map by the appropriate conversion factor.

Preparing map layers

GRASS associates numeric values with cells as their attributes. Sometimes these numbers have real meanings, as in an elevation model in which the cell's value is its elevation. Some are more arbitrary codes representing qualitative categories, such as landcover. The following tables provide guides to units and codes used in the water resource assessment tools. The GRASS tool "r.reclass" allows speedy recoding of map layers to these codes.

Input layers and data

Elevation

Values in the digital elevation model represent the altitude above mean sea level expressed in meters.

Rainfall

Rainfall maps should be rainfall for a single day, measured in hundredths of an inch. Thus a one and a quarter inch rainfall would be represented as 125. The GRASS command "s.surf.idw" can be used to interpolate a rainfall map between observation stations, however, s.surf.idw does not understand the effects of terrain on rainfall and will make significant mistakes in hilly country and for thunderstorms passing through a widely scattered set of observation points. A design storm may be preferable for assessment purposes.

Soil, texture

Values are a code for the dominant texture.

     1 = clay soils
     2 = silt soils
     3 = sandy soils
     4 = peat
     5 = water

Soil, hydrologic soil group

Hydrologic soil group is primarily dependent on the soil texture. However depth to bed rock and depth to water table may strongly influence a soil's classification.

     1 = hydrologic soil group A
     2 = hydrologic soil group B
     3 = hydrologic soil group C
     4 = hydrologic soil group D
     5 = water

Soil, erodibility K factor

K factors are a decimal value greater then zero and typically less then .5. Formally they represent the soil loss rate per unit of per erosion index unit for a specified test plot. (Agriculture Handbook Number 537 "Predicting Rainfall Erosion Losses", USDA 1978) Information about local soils should be available from your local Soil Conservation Service office. Because GRASS uses only integer values, multiply K factors by 100. Thus a K factor of .37 would be coded as 37. Use a code of 100 for water.

Landcover

The landcover map must be coded so the water resource assessment tools can recognize the landcovers indicated. The following table illustrates the 15 landcover category codes expected. When attempting to make the best fit of available data to this encoding scheme, remember to think of the areas both in terms of their ability to slow runoff and their contributions to water quality. Some creative lies may be useful for special purposes. For instance, a development built to the performance standard that peak discharge is not to exceed a field in good condition could be coded as an old field for hydrologic analysis and built up for pollution analysis.

   1 = corn
   2 = rye
   3 = oats
   4 = soybeans
   5 = hay
   6 = grass
   7 = old field (grass)
   8 = old field (shrub)
   9 = pasture
   10 = forest
   11 = wetlands
   12 = fens
   13 = water
   14 = built up
   15 = barren

Output Map layers

Many maps can be produced with the water resource assessment tools. The following sections outline how these maps are prepared, the units of measure and codes employed along with some notes as to how they should and should not be used. More complete information is available in the model documentation. The source code is also provided and contains a great deal of internal documentation. Output maps are presented in the approximate order in which they would be produced.

The idealized elevation file is a digital elevation file. Cell values represent elevation in meters. This elevation model will differ from the input DEM provided by the user, in that depressions in the data are filled in so that water landing anywhere on the idealized surface can flow to the edge. Although useful for analyzing terrain, and employed for slope measurements, the idealized elevation model is probably further removed from reality then the original data upon which it is based. The idealized elevation data may not be resampled to a different cell resolution and retain its desired "idealized" characteristics.

Drainage Direction

Cells in the drainage direction map are coded to indicate the neighboring cell into which they drain. The eight nearest neighbors are represented by the integers 1 through 8 as depicted in figure X3.

Drainage direction is similar to aspect. However they are not exactly the same thing. In an aspect map, the cells at the bottom of a V shaped valley may face each other without draining into each other. This map is produced by searching the border of the window for the lowest cell and draining all cells in that drainage. The algorithm moves up through the drainage one unit of elevation at a time until all adjacent cells of that elevation are drained, trying to assign the most direct drainage direction. Then the next lowest undrained border cell is found and that watershed is drained until the entire map is drained. The routine sometimes has difficulty at the top of watersheds if the next valley does not clearly drop away.

Drainage Accumulation

Values in the drainage accumulation map represent the number of cells, including the cell with the value, that drain through the cell. Thus the minimum drainage accumulation is one, while the maximum is the number of cells in the largest watershed in the study area. The drainage accumulation map is used to find the watershed area for each cell in the study area and to guide a search up the watershed to find the length of the longest stream. The drainage accumulation map can be used to define the stream networks by reclassifying the cells below a minimum accumulated drainage to zero. The remainder represents the streams. Drainage accumulation is drainage area represented in cells. Multiplying a drainage accumulation by the area of a cell yields the drainage area above that cell.

Runoff

The values in the runoff map represent the depth of water expected to run off that cell, from the input rainfall, based on the soil and landcover maps specified. Values could range between zero and the amount of rain that fell on that cell. Runoff, like rainfall, is measured in hundredths of inches. A value of 75 represents .75 inchs of runoff. Runoff values can be multiplied by the area of each cell to give a volume of runoff. However, care should be taken when converting units because cell sizes in GRASS are defined in meters. Upgradient values in the runoff map are summed during the peak discharge calculations and used in the contaminant source area and routing calculations.

Peak Discharge

Peak discharge is the maximum rate at which water passes through a cell measured in cubic feet per second. These estimates are based on an empirical relationship which describes the entire upstream portion of the watershed. If the entire upstream portion of the watershed is not within the current window, the model assumes the upstream edge is the top of the watershed. This is wrong and estimates of peak discharge in streams which enter the window are meaningless as are routings of contaminants based on peak discharge. The calculations are retained so that local contributions from side streams and slopes may be modeled. These local effects may be of substantial importance to those involved in the local area regardless of distant conditions.

Contaminant Source Areas

Sediment, nitrogen and phosphorus nutrients and COD can be simulated. The algorithms used to predict contaminant mobilization are dependent on the contaminant and the landcover class. In non urban areas sediment production is based on the universal soil loss equation. Nitrogen and phosphorous are modeled both in solution and associated with sediment in rural areas and only as a function of sediment in urban areas. COD is assumed to be soluble and is calculated by loading factors.

Sediment loadings are expressed as kilograms per hectare. Nutrients and COD loadings are expressed as grams per hectare.

These loadings are a function of the rainfall, the soil, slope and landcover. They do not represent the natural scatter found in natural events. Results may be used as a relative indicator of contaminant generation. As such these maps may highlight those areas in a watershed that would benefit most from conservation efforts. Subtracting two source area maps generated for different scenarios would help locate the areas of greatest change.

Routing of Contaminants

Contaminants are routed down stream assuming total conservation unless a best management practice map for the particular contaminant is used. If a best management practice map is used contaminants flux across a BMP is reduced by the percentage indicated.

The output maps indicate the totak mass of contaminant passing through a cell. These maps can be used as indicators of the contaminants delivered to any point down stream. Similar analysis for varying land cover regimes could be subtracted with "r.mapcalc" to find areas of greatest positive and negative change resulting from the changes in scenarios. Again these numbers represent relative indicators.

Interface Program

The Water resopurces assessment tool is run through an interface program invoked from the grass environment at any GRASS prompt by typing wrat. Work is organized in projects which should represent the investigation of a geographic area. The user interface is a series of menus which guide the user through the program. A project file is created which keeps track of the options used and input and output map names. The project file also creates a database window based on the active window at the time a project is started. The active window is returned to this active window for subsequent work sessions on the project.

First Menu

At this menu the user can:

1 start a new project
2 start a project based on an existing project
3 work on an existing project
4 remove project files
5 exit

The user is prompted for needed information such as the name of the project to work on with the option of listing project files if needeed. Once a project is selected the secon menu is offered. The User may do:

1 Terrain Analysis
2 Runoff Analysis
3 Contaminant Analysis
4 return to main menu

Choices 1, 2 or 3 lead to menus for each of those analysis.

In the Terrain Analysis the options are:

1 Create an idealized elevation model
2 Create a drainage direction map
3 Create a drainage accumulation map
4 Create a slope map
5 Return to the previous menu

In each case except option 5 the user is prompted for needed information. where appropriate the likely choices are supplied if not insisted upon. The drainage direction map must be based on the idealized elevation map. The drainage accumulation map needs a drainage direction map which should be the output map from option 2. In this way the user is guided along.

In the runoff analysis section there are only three options:

1 Create a runoff map
2 Simulate peak discharge
3 Return to the previous menu

In this section the options are not as straight forward and the user is coached along. The user must provide the names of several input maps and can obtain a listing by entering list instead of a map name where requested. The user must also choose between a design storm and rainfall map. The most difficult information requested of the user is the anticeedent moisture condition. This is a number between 1 and 3 inclusivly which describes the amount of moisture already in the soil which effects the amount of rain which can be absorbed during the current storm. 1 represents extreemly dry conditions and 3 extreemley wet conditions. The concept comes from the Soil Conservation Service curve number method of predicting runoff. The basic guidance from the SCS is offered with the request and a value of 2 neither wet or dry is the default. Peak discharge is quite sensative to thee anticeedent moisture condition so some care should be taken here.

The contaminant section has only three options as well:

1 Model Contaminant source areas
2 Rout contaminants through the watershed
3 Return to the previous menu

When modeling contaminant source areas the user may model just sediment or sediment and any combination of nitrogen, phosphorpus and COD. Because nutrients are associated with sediment sediment predictions are required for nutrient modeling. Most input for this section is straight forward except the number of days since the last significant rain. The amount of dirt available to wash off of urban areas accumulates between rainstorms. Rain storms wash urban areas clean. a default of 7 days is suggested because this puts urban and rural areas on roughly even footings for comparison. To complicate the issue street sweeping also removes contaminants from road ways and so if the study area has a regular streeet cla\eaning program that should be reflected by entering fewer dry days. Several minor rain storms, especially short hard rains will accomplish the same cleaning as a long large rain. SWMM assumes that a half inch of rain removes half of the available dirt and the next half inch removes half of what is left an so on. A worst case for urban areas is a rain ater an extended dry period. However the goal is to compare watersheds or source areas. Agriculturlal areas contribute the most poultion with an anticeedent moisture condition of 3 since that is when the most runoff and erosion takes place.

Summary

The tools presented here run within the GRASS GIS. A flow chart of input and analysis (figure X1) can be used as a guide for the order of analysis. The system of units and means for encoding qualitative information are supplied for both input and output map layers. Careful preparation of input maps is strongly encouraged since this effort will likely save time and trouble later in the analysis.

The ultimate utility of analyses performed with these water resource assessment tools is dependent on the interpretation of the user. Always keep in mind that these results are based on empirical relationships which represent expectations over a long term average. For this reason numerical results should be used only for comparison of areas and scenarios. Employing the entire set of tools from the tool kit may be inappropriate, and or redundant information.

You can't hurt the model by running it! So feel free to experiment!

AUTHOR Brian R. Brodeur Cook College Remote Sensing Center
<bbrodeur_dep_boston@state.ma.us>