Modifiying the land sea mask

I would like to run an OpenIFS setup (T159) coupled to a low resolution ocean. The ocean model and OpenIFS do not fully agree as to which surface points should be wet points:

I used Joakims skript for generating oasis mask files to remove the OpenIFS sea points over what the ocean considers as land. The changes are written into the both masks.nc file for Oasis and the lsmoro file for OpenIFS input.

I also did the same for the EC-Earth runoff mapper file where I had merge some runoff basins and invent new discharge points.

I re-generated the Oasis remapping weight files with the new mask and tried to start the coupled model with the new lsmoro only to crash right after initialization. Turns out the lsmoro part of the changes was apparently not effective. OpenIFS still tries to get SST (and SIC) from the Ocean model at the points that are now no longer ocean as in the Oasis mask. The result is that the areas I have now designated as dry points have a surface temperature of 0 K, leading to the crash.

I double-checked the lsmoro files before and after my modifications and I can not see the problem there. I would say the modifications seem to have worked. I would have expected OpenIFS to calculate land surface temperatures for these points now.

That leads me to the question if there is any other place / file where I have to change the land sea mask for OpenIFS to start considering these points as land and not expect values from the coupler?

Cheers, Jan