Excessive cooling leads to blow up

[sonaljit.m]

Hello,

I am running a simulation with HYCOM fields reconfigured for my ROMS grid. Momentum fluxes are taken from MERRA, and net heat flux is taken as an hourly time series with a diurnally varying shortwave radiation, obtained from a mooring station. Plot of the net heat flux is given below.

I am also providing a heat flux sensitivity dQdSST (around -40) to the model.

The simulation runs for 1 month, and then blows up. It is blowing up because the temperature is reducing to very low values, and crossing 0 at certain regions. Below, is a time lapse video of the surface temperature during the last 4 days before blowing up.
https://www.youtube.com/watch?v=DJSUROvtbRs
The cooling is happening on the northern side of the masked regions, and is quickly spreading out over 2 to 3 days. Below, I am showing a vertical transect of the temperature along longitude -64.2 at days 13, 18, 20.

Day 13

Day 18

Day 20

It seems this cooling is generating at the bottom (even though my bottom net-heat flux is zero), and then spreading along the entire water column. Is anyone familiar with such a problem (or a similar one)? I couldn't figure out why this cooling is happening.

Thanks,
Sonaljit.

[kate]

It looks to me like your net heat flux has a lot of cooling. There's probably no need to suspect bottom cooling with 400 W/m2 of surface cooling. Are you sure you aren't missing something like incoming longwave radiation?

[sonaljit.m]

The variables I am providing in the forcing file, are the following:

Temperature fluxes -> swrad(srf_time), lwrad(lrf_time), latent(lhf_time), sensible(shf_time), shflux(shf_time) (all units in W/m^2)
For Precipitation -> swflux(swf_time,eta_rho,xi_rho) (cm/day)
For bottom temperature and salinity -> bhflux(bhf_time), bwflux(swf_time,eta_rho,xi_rho) (they are all zero)
Surface wind stress -> sustr(sms_time,eta_u,xi_u), svstr(sms_time,eta_v,xi_v) (N/m^2)
Bottom momentum stress -> bustr(sms_time,eta_,xi_), bvstr(sms_time,eta_v,xi_v) (N/m^2, all zero)
sea surface temperature & salinity -> SST(sst_time,eta_rho,xi_rho), SSS(sss_time,eta_rho,xi_rho)
Net heat flux Sensitivity to SST dQdSST(sst_time,eta_rho,xi_rho) (around -42 everywhere)

Here's a plot of swrad, lwrad, latent and sensible:

(Latent and sensible fluxes are very less in magnitude compared to the rest.)
Their summation shflux, is below:

dQdSST is provided as a spatial 2D array over different times of sst_time. A contour plot of dQdSST at the first time, is below:

I'm calculating dQdSST as per Barnier et al, 1994 (https://www.sciencedirect.com/science/a ... 6394000349)

Average wind stress in my forcing file, is around 0.06 N/m^2.

[kate]

I still think your longwave flux is off. That's got to be upward longwave, not net longwave. Longwave has a downward component too.