Page 1 of 1

WRF climate runs - reducing boundary data requirements

PostPosted: Mon Feb 22, 2021 10:17 am
by rmcgrath

For climate runs I would expect that after the first boundary file has been used only the upper-air boundary data that cover the relaxation zone would be required, not data over the full domain (surface data are different). If so, this would save significantly on handling data volumes from, say, a global model, for long runs.

Unfortunately, I get small but not insignificant, differences when comparing forecast runs launched from a common initial dataset driven by "full" area and masked area boundary data (first boundary dataset is a full set in both cases). To avoid NaNs due to missing data in the masked region I simply filled the missing data points with nearest neighbour data from the relaxation zone. The surface boundary fields were identical (i.e. "full") for both runs.

Have I missed something?

Any comments appreciated.