search

clawpack / geoclaw

Version of Clawpack for geophysical waves and flows
http://www.clawpack.org/geoclaw
BSD 3-Clause "New" or "Revised" License
76 stars 87 forks source link

periodic BCs incompatible with netcdf topo files #539

Open bolliger32 opened 2 years ago

bolliger32 commented 2 years ago

I'm having problems running a global model with periodic boundary conditions in the E-W direction and "extrap" conditions in the N-S direction. I believe this has to do with how netcdf's are loaded here, b/c they are clipped to within a dx and dy buffer of the domain. This means there is no topography loaded that lies outside of the domain within the ghost cell portion.

This comes into play in setaux.f90. When setting for ghost cells, wrap_coords is called, which, as expected, wraps the E-W cells but does nothing to the N-S cells (since yperdom is false).

Then cellgridintegrate is called, which due to this line, raises an error that there is no topography overlapping that cell.

Does that sound like the right culprit (cc: @mandli @rjleveque)? I imagine the fix would be to increase the extent of the netcdf that gets loaded if you are in this situation.

mandli commented 2 years ago

Makes sense to me. Periodic BCs are not used in GeoClaw except for planetary wide simulations and I am not sure we have used anything but the ESRI ASCII formats.