Watershed tool returns only immediate catchment, not entire upstream catchment extent

4099
7
Jump to solution
02-22-2013 08:48 AM
DarleneWilcox
New Contributor II
I have a 1m DEM derived from LiDAR.

I wish to determine the watershed catchment above a specific pour point.  The Pour Point is in a shapefile.

I am working with the tools in Spatial Analyst Tools > Hydrology.

I did a FlowDirection, then I did a Sink, and yes, there were some, so I did a Fill.  I then repeated the FlowDirection. 

I also did FlowAccumulation and FlowLength, as well as a SnapPourPoint to make sure my Pour Points were appropriately located.  I selected a search of 3 cells, and the point moved down three cells, so I moved it UP three cells from my original location, and my original location was selected, so I know it is an acceptable pour point.

With the Flow Length (using the Upstream option and making high values dark) I can see a beautiful representation of the flow.  I want to capture the entire extent of the flow into that point.  The FlowAccumulation at that point is over 65000, and the upward FlowLength is nearly 600.

However, when I try to define the Watershed using the Pour Point, I get just this small catchment immediately above the Pour Point, about 100 grid cell count, NOT the entire catchment.  Using the FlowDirection I moved one cell downstream, and got a smaller adjacent catchment of only 40 cells -- so it didn't even include the original try's extent, let alone ALL the upstream extent!

How does one set thresholds to say give me the whole thing, not just the immediate little bit.

Thanks.  I'm attaching a graphic to show what I mean.
Reply
0 Kudos
1 Solution

Accepted Solutions
curtvprice
MVP Esteemed Contributor
Something to try:

Make sure your environment extent and snap raster are set to the flow direction grid for your run of the watershed tool.

View solution in original post

7 Replies
curtvprice
MVP Esteemed Contributor
Something to try:

Make sure your environment extent and snap raster are set to the flow direction grid for your run of the watershed tool.

View solution in original post

CourtneyBrailo1
New Contributor

This was the reason I had this same problem. Thanks for the response!

Reply
0 Kudos
MarkBoucher
Regular Contributor II
You are performing the flow direction on the filled DEM, right?
DarleneWilcox
New Contributor II
You are performing the flow direction on the filled DEM, right?


Yes I am, but it was a good question to ask!
Reply
0 Kudos
DarleneWilcox
New Contributor II
Something to try:

Make sure your environment extent and snap raster are set to the flow direction grid for your run of the watershed tool.



Holy cow, you nailed it!  I NEVER would have thought to try that, as I (obviously wrongly) assumed that if the only raster was the original DEM then the environment settings would default to the appropriate values.  Lesson learned -- THANKS!
Reply
0 Kudos
EricRice
Esri Regular Contributor
Darlene,

Just to clarify what happened a bit for anyone else who visits this thread.  You made use of the Snap Pour Point tool.  The raster that is derived from this tool will have the extent of the input points if you leave the extent settings default when you run the tool.  Usually the point layer's extent is smaller than the DEM extent.  When you move forward in the hydro workflow (watershed tool) and you use the smaller extent raster as input, the tool's default behavior is to use the smallest extent of any of the inputs.  In other words, just the area of overlap between the inputs.  You can avoid this by setting the extent to the DEM when you run the Snap Pour Point tool, or like Curtis suggested, when you run Watershed.

If you create an envelope of the points using the Minimum Bounding Geometry tool and overlay it with your original watershed result it should be quite clear why your catchments got 'cut' off.

Best,
Eric
DarleneWilcox
New Contributor II
Darlene,

Just to clarify what happened a bit for anyone else who visits this thread.  You made use of the Snap Pour Point tool.  The raster that is derived from this tool will have the extent of the input points if you leave the extent settings default when you run the tool.  Usually the point layer's extent is smaller than the DEM extent.  When you move forward in the hydro workflow (watershed tool) and you use the smaller extent raster as input, the tool's default behavior is to use the smallest extent of any of the inputs.  In other words, just the area of overlap between the inputs.  You can avoid this by setting the extent to the DEM when you run the Snap Pour Point tool, or like Curtis suggested, when you run Watershed.

If you create an envelope of the points using the Minimum Bounding Geometry tool and overlay it with your original watershed result it should be quite clear why your catchments got 'cut' off.

Best,
Eric


You're exactly right about the cause of the problem, though one point needs clarification.

The SnapPourPoint raster has the same extent as the original DEM.  My particular problem arose in that I used the point shapefile, not the raster, and you are spot on about the extent being redefined.

It doesn't seem intuitive that the raster extent would be redefined to a shapefile extent -- though I guess, technically, the 'smallest extent of any of the inputs' would include shapefiles as well as the rasters.  THIS is the real point to be made.  So, either use the SnapPourPoint raster or define the Extent and the Snap Raster in the Environments!

Thanks for explaining why this problem occurred!  Darlene