This File: README_gom09_conf.txt Path: Ada: /home/jaisonk/WORK/Projects/gom_e-cesm/conf/gom09 Date: June/03/2020 1. Use gom09.h as the header file reference. Please note the following important differences (there are many) with the gom03.h file @ github roms_ihesp/Apps/gom03/gom3.h: - Most importanly, the options are grouped according to a process/mechanism. I strongly recomment to use this style for readability. - no "CCSMCOUPLED" --> is this need to be present in .h file anymore? Doesn't the RCESM automatically edit the .h file to add whatever options relevant for the coupling? - define CURVGRID --> Curvilinear grid - undef SPLINES --> This is the official recommendation on ROMS Forum. - define RI_HORAVG & define RI_VERAVG --> I had serveral cases of ROMS blowup few months into the R-CESM runs and this is one option helped me to fix them. The instantaneous winds from WRF/CAM can have pretty crazy patterns near islands/coast and really high values over few grid points. So, this is like a blanket solution for such situations even if not ideal. 2. Please use gom09.in file as the reference for your E-CESM experiment. - Grid size (Lm Mm, & N) are adjusted for gom09_POPemb_grd.nc - NtileI and NtileJ (processors along X and Y) are 12 and 12. Edit them as needed. - A DT of 30 seconds is good for testing or running first few days/weeks at the least. Later we can increase the DT to higher values (say 120 or 180 seconds) (if we have sufficient time for such tests). - NTIMES: This seems like controlled by time settings in other files now. But I prefer some consistency and provide a realistic number here. - NRREC: 0 for runs from initial condition and the 1 or relevant record number for restart runs. - NRST : Change as needed - Output history/avg/dia --> all set to 3-hour outputs and 1-file per day - Based on previous 9-km runs following values are chosen for TNU2 and VISC2: TNU2=30 m2/s and VISC2=50 m2/s - Change DSTART and TIME_REF as required. - Change Path to INI, BRY and GRD files as required.