Hello everyone,
I’m writing to ask for some guidance. Until a few months ago, I was using version 1.3 without any issues for 3-day forecasts. Recently, I upgraded to version 2.0.1 and tested a simulation that had previously worked, with the only difference being the executable from the new version (same physics and NetCDF files). To my surprise, the simulation failed.
While verifying that I was indeed using the same physics, I noticed that the new version includes the following parameterizations: BD_EXCHANGE, ENDDOLOOP2D, DOLOOP2D_R, and DOLOOP2D (also present in version 2.0). My parent grid is quite large at 600x600 with 42 vertical levels, and I have AGRIF_2WAY (2 child domains), TIDES, PSOURCE, and MPI enabled. My original simulation and many others used NP_XI=6 and NP_ETA=6 without any issues.
Considering BD_EXCHANGE, I thought the blow-up error might be related to the number of subdomains in different directions. However, I also tried NP_XI=4 and NP_ETA=8, which failed. When I set NP_XI=5 and NP_ETA=7, the simulation ran without any problems.
So, my question is: Are there any specific recommendations or considerations regarding the BD_EXCHANGE parameterization in this new version?
Thank you in advance for your insights!