Hello WRF community,
I am new to WRF and this forum, but I believe (after some searching) that this particular question was not asked. I wish to simulate KH waves (scale of several kilometers) over some fairly complex terrain; I wish to run the model at 333 m resolution which necessitates 4 nested domains and to keep the number of points reasonable some of the domains unfortunately need to intersect complex topography.
The REAL case is for a downslope windstorm in the June of 2005, and the model gives fairly good mean wind, but after 2 days some weird instabilities occur that cover large portions of inner domains (333 m and 1 km), and seem to originate in the outer domains. The instabilities in the innermost domain have wavelengths of several km (haven't checked the outermost domains), and occurred from 12 to 18 UTC (again, in the innermost domain). Figures are attached of U10 and V10 fields at 15 UTC in several domains, as well as the terrain. I also attached namelist.wps and namelist.input files.
Domain 1 (9 km) - terrain (left) and V10m (right)


Domain 2 (3 km) - terrain (left) and U10m (right)


Domain 3 (1 km) - terrain (left) and V10m (right)


Domain 4 (0.333 km) - terrain (left) and V10m (right)


What I tried:
Pechudin
I am new to WRF and this forum, but I believe (after some searching) that this particular question was not asked. I wish to simulate KH waves (scale of several kilometers) over some fairly complex terrain; I wish to run the model at 333 m resolution which necessitates 4 nested domains and to keep the number of points reasonable some of the domains unfortunately need to intersect complex topography.
The REAL case is for a downslope windstorm in the June of 2005, and the model gives fairly good mean wind, but after 2 days some weird instabilities occur that cover large portions of inner domains (333 m and 1 km), and seem to originate in the outer domains. The instabilities in the innermost domain have wavelengths of several km (haven't checked the outermost domains), and occurred from 12 to 18 UTC (again, in the innermost domain). Figures are attached of U10 and V10 fields at 15 UTC in several domains, as well as the terrain. I also attached namelist.wps and namelist.input files.
Domain 1 (9 km) - terrain (left) and V10m (right)


Domain 2 (3 km) - terrain (left) and U10m (right)


Domain 3 (1 km) - terrain (left) and V10m (right)


Domain 4 (0.333 km) - terrain (left) and V10m (right)


What I tried:
- feedback = 0 and 1; found a paper that claims feedback = 0 can generate spurious waves at highest wavenumbers due to change in numerical phase speed (makes sense), however this seems to not make a difference
- originally I wanted to use the ASTER 1' dataset in the finest domain; changing to default 30s topography in all domains did not remove the waves
- changed dt/dx from recommended minimum 6 to 3
- w_damping = 1 and epssm = 0.2 because I was getting vertical CFL violations in the steepest terrain
Pechudin