I tested my case with LSM initial conditions (from RAP) for the five supported suites in version 5.0. In the configure file, I set sfc_flux_spec = .false. and lsm_ics = .true.. I found that the surface values (fluxes and soil) from GSD_v1 were not reasonable. Although I initialized the RUC LSM with 9 layers of soil, the output soil conditions only have 4 layers, which is the number for Noah and NoahMP. So I wonder if RUC LSM is not supported in this version of CCPP-SCM or only the top 4 layers of soil were used?
By the way, my case also showed the same results as you mentioned in https://github.com/NCAR/ccpp-scm/pull/218: NoahMP (i.e., RRFS_v1alpha) has much higher Bowen ratio than Noah (i.e., GFS_v15p2 and GFS_v16) with the same initial conditions.