I have a weird phenomenon going on in QGIS.

I want to derive a channel network and also the catchment areas using "watershed basin" from SAGA. In the end I want to have the flow direction and flow accumulation. I created a small DEM and I also got a few raster layers from open web data to have a bit of context and orientation/information in my DEM. As you can see in the picture, I also have a hydrographic network raster layer which shows me all the channels etc.

As you can see, there is a lake in the middle and I think because of the lake the algorithm calculates the watershed basin very very wrong. Flow direction seems to be right.

Do you know a workaround regarding this problem?

I also tried to work with Strahler Order but the DEM size is too small for it, so its very faulty

Watershed basin here

Filled DEM

1 Answer 1


This issue is not specific to QGIS and SAGA, but applies to all similar algorithms. The algorithm does work unlike the title suggests. Some considerations:

  • When there's a lake, you need to check if the DEM has the bathymetry of the lake, elevation of the water surface or nodata values. It's not clear from the lower picture what's the case here.
  • A fill sinks algorithm (that needs to be applied on a raw DEM) will fill the lake so the water will flow to a lowest point. This will result in artefacts. Also with the Strahler order you'll see parallel lines in the lake, because of a low constant slope. You can also compare the DEM before and after filling to see what happened.
  • If you don't want to include the lake in your analysis, you need to mask the lake.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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