Currently running regcm4/swat/model2swat.py on the nrel system and it is processing about 2 years of data per day. This is too slow for my enjoyment. Clearly, recomputing the raster_stats every timestep is not efficient. Perhaps we have enough memory to load up the raster mask for each huc12 and then just use it to sample?
Currently running
regcm4/swat/model2swat.py
on thenrel
system and it is processing about 2 years of data per day. This is too slow for my enjoyment. Clearly, recomputing theraster_stats
every timestep is not efficient. Perhaps we have enough memory to load up the raster mask for each huc12 and then just use it to sample?