I am attempting to model suspended sediment in a coastal area. When forcing a coupled ROMS/SWAN run with tides, winds, and waves, I have found that using UV_QDRAG has worked successfully in many model grids.
Since I expect that the waves have a significant effect on the BBL in the area, I expect that results will be significantly better if I use one of the other bottom stress models (e.g., SSW_BBL). Without this added forcing, the difference between the ROMS and ROMS/SWAN results is present, but not particularly dramatic. When I switch to one of the BBL models, though, after a simulated day or two (and seemingly much better results), the model always crashes. I can make small changes to the grid, and the blow-up ends up in a completely different part of the grid.
There are of course many parameters involved, but I note that it is always right after a SWAN timestep. Although my "curvilinear grid" is essentially uniform, but rotated, I have increased the number of iterations that SWAN does, as suggested in <http://130.161.13.149/swan/forum/topic.asp?TOPIC_ID=296>, and accordingly, the accuracy reported by SWAN does seem to increase, but it has no effect on the end result, or the time at which the model crashes.
Have others experienced similar errors, or is there something obvious that I might be doing wrong?
[Incidentally, after changing a number of options, recent runs do not seem to print out the sediment ripple length and height correctly anymore -- the variables are in the output files, but only as all zeros. Previously this was not the case, though, so I do not think this is associated with the cause of the crashes.]
Model blow-up with BBL model
Re: Model blow-up with BBL model
Alternatively, is there an easy way to use a simple bottom stress model, but modify the code to determine sediment suspension with the local wave parameters via wave-current interaction (i.e., use UV_QDRAG as a momentum boundary condition, but SSW_BBL as a suspended sediment boundary condition)?