I ran two ROMS model simulations. In both cases, I have two run segments - initial run and then one which begins after a perfect restart (I use the PERFECT_RESTART CPP option). In the first run, I specify a point source via the UV_PSOURCE CPP option and in the second, I use Q_PSOURCE CPP option instead. The first run completes both run segments successfully. The second run however completes the first segment successfully but blows-up immediately upon doing the (perfect) restart (runs for 2 time steps). I did some debugging and found out that it blew-up (in diag.F) because rho exceeded 200.0 units which is the limit set for it in the ROMS.
Has anybody else experienced this? Could this be a bug - or something that is missing for a perfect restart in the restart NetCDF file?