r.lake with raster as seed input

[rlwcomposto]

Hello,

I'm having some trouble using a raster layer with multiple seed points in r.lake. Is there anything I'm missing with using a raster as one of the inputs? Anything I should try?

Problem: I am trying to run r.lake with a HAND layer and stream height at ~12 gauges. I've converted the point height data into a raster layer and use it as the seed or "Raster layer with with starting point(s)" parameter and HAND as the elevation parameter.

Tried:

• Tried a few of the gauge points as water_level and coordinates parameters, using one gauge point at a time has valid results.
• Checked that all stream height values are above the HAND values. The height values are all above zero and overlap with HAND values of zero. Some of the height values are over 1 and some are under 1 (e.g., 0.12).
• Made all the pixels that do not have stream height values "no data" (maybe there should be a different fill value in the raster?)

[wenzeslaus]

Hi, the seed raster really specifies only the positions (coordinates), but there is still only one water level provided as water_level.

My understanding is that r.lake fills a lake with a bathtub model, so it would not have any way to decide which water level to use when more than one is possible. What do you expect to happen?

What about computing each gauge separately, and then combining the result in a way that upstream of a gauge only the corresponding gauge result is used, while downstream the highest one wins? Just an idea. Some scripting would be needed to do it for 12 gauges.

[marisn]

This is a limitation of r.lake – you have to fill one bathtub at a time. PRs with improvements are welcomed.

[rlwcomposto]

These answers have cleared up my misunderstanding. Thank you both!