EVHY 544 – Fall 2005

Exercise: Watershed Delineation

Objective: to create watershed boundaries from raster digital elevation model
datasets and find land use percentages within those boundaries

The process of watershed delineation is somewhat simple and consists of a few

steps that need to be performed in the correct sequence.

1. Set up GIS work environment.

2. Start ArcGIS and create a new map document
3. Watershed Delineation
1. Creating a depressionless DEM
2. Flow direction
3. Flow accumulation
4. Watershed pour points
5. Delineating watersheds
4. Automatically delineating watersheds ("Basins") (not required)
5. Calculating flow length (not required)
6. Raster to vector conversion (stream network as line shape) (not required)
7. Watershed visualization (not required)
8. Clip land use layer and calculate land use for watershed (not required)

Setting up ArcGIS work environment

1. Create a directory on the hard drive (F:\temp\hydro). This will be used to

store files temporarily.

2. Open ArcMap and accept the defaults to create a new map.

3. Enable the Spatial Analyst Extension. (Tools > Extensions > check
mark next to Spatial Analyst)

4. Open Spatial Analyst Toolbar. (View > Toolbars > Spatial Analyst)

5. Import proper zip from http://fisher.lib.virginia.edu/services/class/evhy544/data/ and

export into your working directory. Note, do not drag and drop. Use the
export button in WinZip. This will preserve the directory structure
within the zipped data. Your data set may not work properly if the
directory structure is not correct.

6. Add the dem layer to your map by using the Add Data button ( ) from
f:\temp\hydro\yourarea
7. Open ArcToolbox by using the button. ( )

8. Set some environment properties (Tools > Options > Geoprocessing >
Environments)

1. Set the Current Workspace, Scratch Workspace, and Output

Extent as below:

2. Set the Cell Size as below. Note, your numeric value may differ.
 9. Save the map document as f:\temp\hydro\watershed.mxd. (File > Save
AS)

Watershed Delineation

• Creating a depressionless DEM

NOTE: This operation takes too long to perform during lab. Therefore,
a new depressionless layer is available in your downloaded data
(dem_fill). Please add that layer and proceed to step 3.

It is important to start with an elevation grid that has no depressions.

1. Open the ArcToolbox toolset Spatial Analyst Tools > Hydrology. This is
where the surface hydrology tools are located.

2. Open the Fill tool. The input surface is the dem grid. Output is
f:\temp\hydro\Fill_dem1 (the default name).

While the grid is filling, which will take a large number of minutes, you
might want to entertain yourself with the Onion.
 3. After a few minutes, a new layer, fill_dem1, will be added to the data
frame. This is identical to the dem raster, but any areas of internal
drainage are filled in.

Note the difference in the lowest elevation value in the legend; sink cells in
the original data set have been filled in.

4. Remove the dem layer from the data frame by (right clicking on the layer
and hitting Remove), since you will be working on the filled grid from this
point on.

It is important to have a depressionless DEM for all subsequent

hydrological analyses. Areas of internal drainage can cause problems
later in the watershed delineation process.

• Flow direction

1. Open the Flow Direction tool (ArcToolbox > Spatial Analysis Tools >
Hydrology > Flow Direction).
1. The input surface is the filled dem grid.
2. The output raster should be set to f:\temp\hydro\FlowDir_fill1 (the
default)
 2. Turn off display of the filled dem layer.

3. Add the layer f:\temp\hydro\yourarea\hydro to the data frame. This is

the streams layer. You should start to see some patterns when zoomed
in.

4. Note that the numbers refer to coded direction of flow.

Direction of flow must be known for each cell, because it is direction of

flow that determines the ultimate destination of water flowing across the
surface.

• Flow accumulation

1. Open the Flow Accumulation under Hydrology.

1. Set the input flow direction raster to the output of the last task.
2. Set the output raster to f:\temp\hydro\FlowAcc_flow1 (default
name).
2. Again, this may take some time. Surf the Internet at your leisure.

3. When complete, turn off the flow direction layer. The flow accumulation
layer has a value for each cell; that value represents the number of cells
upstream from that cell. Cells with higher values will tend to be located in
drainage channels rather than on hillsides or ridges.

4. Alter the legend for this layer. It will be easier to visualize high-flow
pathways by altering how cells are displayed.

1. Change the symbology method to (right click on layer >

Properties > Symbology) classified. Use 2 classes.

2. In the dialog box under the Classify button, alter the break value
for the first class to 5000.

3. Change the symbology (no color for the first class, red for the
second class).
Now the cells that are displayed in red have the flow of at least
5000 upstream cells flowing through them.

You should also see that the DEM-generated drainage network

looks somewhat like the vector streams, although if you look at
details you will see where the data sets do not line up.
 You will need to zoom in before you can see the details of the flow
network.

Flow accumulations are important because they allow us to locate cells

with high cumulative flow. Pour points must be located in cells of high
cumulative flow or the resultant watersheds will be very small.

• Watershed outlet ("pour") points

1. Create a new point shapefile using ArcCatalog

(f:\temp\hydro\ppoints.shp; copy the coordinate system parameters
from dem).

1. Open ArcCatalog, browse to your directory, right click in white area,

New > Shapefile.
 2. Hit the Edit button. Hit the Import button. Select dem.

2. Add the new point layer to your data frame and begin editing. (View >
Toolbars > Editor then Editor > Start Editing)

3. In order to better see where you are adding points, open a magnifier
window (Window > Magnifier).

4. You may need to zoom in and increase the zoom factor of the magnifier
(right-click the magnifier's title bar and select Properties) before adding
any points.

5. Use the Identify tool ( ) to examine the values of the Flow

Accumulation layer before adding any points. You will be able to tell
which direction is upstream (lower accumulation value) and which
direction is downstream (higher accumulation value).
6. Add points by selecting the Sketch tool ( ) from the Editor toolbar.

You must zoom in quite a way to do this; otherwise your pour point
may not be located within a high-flow pathway!
If your points are not in a high-flow pathway, move them before
proceeding.

7. Everything upstream from the point will define a single watershed.

8. If adding more than one point, open the attribute table for the ppoints
layer. (right click on layer, Open Attribute Table) Alter the ID field to
represent unique values for each different record (e.g., 1 and 2). Once you
have added the points and altered their IDs, stop editing the layer, making
sure to save the edits. (Editor > Stop Editing > Save Edits? > Yes)
Watersheds are defined by pour point IDs; if you do not alter the ID values
for the points, there will be only one value, and you will not generate
unique watershed areas.

9. Select Tools > Options > Geoprocessing > Environments > General
Settings > Output Extent to make sure the output extent will be correct
(you don't want the watershed truncated).
10. Also, check the Spatial Analyst > Options as well (Spatial Analyst
controls outputs from the Spatial Analyst Tools, e.g., Surface Analysis >
Hillshade).

11. The watersheds are not generated directly from point features, but by
raster cells. In order to convert the pour points to a grid, use the tool
Conversion tool in ArcToolbox (Conversion Tools > To Raster >
Feature to Raster).

1. Convert the ppoints features to a raster.

2. Use default values for all other parameters, but make sure the cell
size is the same as your “dem_fill” layer.
3. Alter the drawing order so you can see all the layers clearly. Zoom
in as needed to verify that the new cells overlap with the high-flow
pathway.

If the Analysis Extent and cell size do not match an existing layer,
there will be problems of registration between the pour grid and the
other grids necessary to delineate watersheds. Therefore, it is always
a good idea to set your cell size and analysis extent relative to an
existing grid layer.

Note that the new feature_pour1 layer cell covers the place where the
pour point was added. If the pour grid cells are not in high-flow
pathways, the watersheds you create will be too small. If the pour grid
cells are not directly "over" the flow path, you will not create
watersheds that represent large drainage areas.
• Delineating watersheds

1. Open the Watershed tool under Toolbox > Spatial Analysis Tools >
Hydrology. Note that one of the options for the input data is the point
feature dataset (ppoints). Selecting a point feature dataset will work, but
there is no way to verify your point features will fall in a high-flow pathway.
That is the reason for the conversion to raster in the last step. It is
recommended to have converted the points to a grid first, which verifies
that the pour point locations are in the high-flow pathway.

1. Select the flow direction grid as the input flow direction raster.

2. Select the raster version of the pour points as the input raster.

3. Accept the other defaults.

2. The new watershed grid will be added to the data frame.

 This large grid zones represent the area upstream from the selected pour
point. The pour point defines the entire watershed system.

3. Convert the watershed grid to a polygon shapefile called

watershed1.shp. (Toolbox > Conversion Tools > From Raster >
Raster to Polygon)
 You have just delineated your watershed, based on your elevation grid
data and the pour point you chose. If your watershed is too small, it is
because you located your pour points outside of a high-flow pathway, or
you did not fill your original input elevation grid.

Automatically delineating watersheds

Compare your manual method with an automated method. This method

takes a very long time to process and does not have the fine control needed.

1. Open the Basin tool.

1. Select the previously created flow direction grid.

The automatically delineated watersheds are defined by pour points

 at the edge of the grid.

2. Convert these to polygons as well in order to see where the

watershed boundaries are. Make sure to uncheck Simplify
polygons in order to match the actual grid zone boundaries rather
than generalizing lines in the output.

3. Zoom in. You can see a number of very small basins.

You have just let ArcGIS automatically generate a series of watersheds.
Automatic watershed delineation is easy, but does not give you the control to
create basins specifically for pour points of your own selection. For this reason,
the manual method is used almost exclusively.

Calculating flow length

Flow length shows the distance water will need to travel across the grid.

1. Open the Flow Length tool.

1. The input raster is the flow direction grid created earlier.
2. Use defaults for the other controls.
2. Alter the color ramp so you can see the differences between low flow
length and high flow length areas. In this color ramp, the red cells are at
the upper reaches of streams in the forest, and the blue cells are farther
downstream.

This shows the flow length to the ultimate pour point for each cell.
Suppose you want the flow length to the closest downstream high-flow
pathway, rather than to the ultimate outlet. This is possible, using flow
length with a weighting grid.

3. Create the weight grid by making a grid whose values are no data within
the 5000-cell high-flow path, and values of 1 elsewhere. This is done with
a setnull function in the Raster Calculator:
 This calculation means "make a new grid; where the flow accumulation
cells have a value greater than or equal to 5000 in value, make those
output cells null; where the flow accumulation cells are less than 5000 in
value, make the output cells have a value of 1")

4. Use the new grid as a weight grid in the flow length tool.

5. The output grid now has values that represent the flow length-distance to
the closest high-flow pathway rather than to the ultimate outlet.
Raster to vector conversion (stream network as line shape)

Raster data sets can represent drainage networks (e.g., the flow
accumulation cells that have at least 5000 upstream cells). When making
maps that present the results of watershed delineation you may want to show
the grid-based flow network instead of, or in addition to, the vector stream
network, especially if the two flow networks do not agree.

1. First, create a grid that represents only high-flow (5000 +) cells. This is
also done in the Raster Calculator, similar to what was done in the last
step.
 2. Open the Stream to Feature tool.
1. The input stream raster is the result of the last calculation.
2. The input flow direction raster is the flow direction grid that was
made before.

Watershed visualization

The last step for this lesson will be to visualize the watersheds created earlier
with other data.

1. Create a new data frame.

2. Add a copy of the grid layer fill_dem1 from the other data frame.

3. Create a hillshade grid from fill_dem1.

4. Add the ppoints point feature layer.

5. Create contours from dem and add them to map. Add a copy of the
watershed1.shp polygon feature layer.

6. Alter its legend so that is not filled and displayed with a red outline.

7. Alter the drawing order and legends so that features are discernible. Does
this look reasonable to you?
 8. Visualizing in ArcScene can also be helpful.

Using Land Use Data

A key to the land cover data set is located at

http://landcover.usgs.gov/classes.asp

1. Save project and remove all layers but your watershed raster.
2. Add land use layer. (landcover)
3. Clip landcover by watershed raster.
1. ArcToolbox > Spatial Analysis Tools > Extraction > Extract by
Mask
2. Input layer: landcover.
3. Mask data: watershed raster.
4. Since landcover and watershed raster layers are no longer
needed, they can be removed if so desired.
5. Normally, at this point you might want to convert this raster back
into vector data in order to calculate areas, assign other attributes,
etc. However, this is a time consuming process and will be skipped
for this lab. This does not mean we cannot do some future
calculations with the grid data.
6. Open the attribute table for your clipped land use layer. What you
see is the land use classification code and a count. The count is
the number of cells for that land use classification. Using this data,
we can calculate area and percentages of land use.

Add all the counts together and divide by each row to get
percentage of each land use classification.

Use the count to calculate area. Each cell is 0.000790600442 km2

(28.117618 m * 1 km/1000 m * 28.117618 m * 1 km/1000 m). So,
number of cells * 0.000790600442 km2/cell * 1 mile2/2.59 km2 =
square miles.