2

I am using r.watershed to create watershed basins using a 25*25m dem.

The problem I have is that basins are not always created at coast lines. I am experimenting with the parameter "minimum size of exterior watershed basin", which controls the minimum number of pixels required to make a basin. If i set this value to say 1-50 I get basins created at coastlines but they are very narrow. If I choose a large value 400+ i get reasonable sized basins but they are are not always created at coasts.

Image - the yellow shaded area is a basin, but the area immeadiately to the left is not a basin.

Is there a way to get basins that cover all areas on the DEM? or is this the correct "behaviour"?

basins with 400 pixels

  • 2
    In general, I think what you described is correct behavior. The threshold basin size (Mimimum exterior basin) is the minimum area considered to be a basin. At your resolution of 25x25, each cell=625 sq.m. So a threshold of 400 means a minimum basin of 250,000 sq.m or 25 hectares. Areas that drain off the map, or into the sea, and are smaller than that are not considered to be a basin. – Micha Aug 19 '16 at 12:35
  • Ok thanks. Yes I tried with a minimum threshold of 1 and the result is 1 pixel wide basins down each slope. My work around: The basin edges largely line up when using thresholds of say 1, 10, 50 and 100. I created basins with threshold 400, then fill gaps with basins threshold = 200 and so on. – BenP Aug 19 '16 at 13:02
2

Resolved. In addition to my previous comment.

To ensure that the r.watershed program creates basins with total coverage at coastal or around inland water bodies, you will need to have elevation values for these water bodies.

Initially my DEM was masked (nan values) for water bodies and these are therefore treated as an edge in the watershed program. I found that setting water bodies to a low elevation value (i.e. lower than your minimum actual elevation) creates basin features that you would "expect". These however now extend below the water line. I therefore masked the basin output back to nan values at the water bodies.

Hey presto you now have complete basin coverage and which are clipped to water bodies. This may not be the most accurate method, and having a DEM with bathymetry would be optimal, but this produces output which would be "expected" in most cases.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.