I'm trying to run r.sim.water in GRASS7 to simulate overland flow during a heavy rain event. My DEM has a resolution of 1m. I prepared it by adding values for buildings and included ditches and culverts using r.carve. I calculated xy-derivatives using r.slope.aspect.
My study area is about 8,5 sq km. I was not able to run r.sim.water with 1m resolution so I changed the resolution to 2m using r.resamp.rst. Including NULLS, the input raster size is now about 4.000.000 cells. r.sim.water did still not launch so I changed the number of walkers to the exact number of cells (instead of twice the number). Now I get values for water depth that are far higher than the actual depression underlying the cells, which is physically not possible. Changing the value for the Manning coefficient even increases those "water mountains".
I think, to avoid this I have to change the diffusion coefficient and/or the values for hmax, halpha and hbeta. However, I have no idea of how they actually influence my result or what could be reasonable values for my purpose. Does anyone have experiences with that or can tell me where I could find information about it?
Besides, the water depth after 10min is far higher if I set the duration to 60 or 120 min as if I chose, for example, only 30 min, which also doesn't seem logical to me. Can anyone tell me why this could happen?
I did not find any information about a maximum number of cells for the input DEM. However, I think r.sim.water cannot handle large areas or high numbers of walkers. Does anyone know if there are limitations for the calculation?