Salinity goes from normal to nan

[rduran]

My model blew up in first step ... after inspecting the initial conditions (first file to be written out by roms) I noticed all my 3D array for initial salinity was nan (or more precisely whatever that _fill_value was (I did salt(salt>1e10)=nan in matlab to plot and inspect file)
I also noticed that mask_psi had some weird values of 2 were land meats ocean. The rest looked fine. Now needless to say all the values in the actual initial file are ok. The salinity in my ini.nc is ok and the mask_psi is ok too.

I used the scripts provided by Hernan (thank you all for that) with some modifications to suit my needs. Any ideas on why such a thing would happen?

thanks,
Rodrigo

[kate]

Check your forcings and try with a much shorter timestep.

[rduran]

Kate thank you, as usual you have been very helpful.

Reducing the time step seems to be working. I had considered this possibility but I thought my Courant numbers to be small enough so I (prematurely) discarded that option. Here are my previous and new values:

364,366c384,386
< Minimum barotropic Courant Number = 3.28251109E-02
< Maximum barotropic Courant Number = 6.15726228E-01
< Maximum Coriolis Courant Number = 1.94561327E-02
---
> Minimum barotropic Courant Number = 9.11808636E-03
> Maximum barotropic Courant Number = 1.71035063E-01
> Maximum Coriolis Courant Number = 5.40448129E-03


the latter bunch are working hurray!!!

However I expect potential problems with artificial pressure gradients:

Maximum grid stiffness ratios: rx0 = 1.841874E-01 (Beckmann and Haidvogel)
rx1 = 1.864604E+01 (Haney)

Is that a reasonable (as sad as it is) expectation?

Is there a script to estimate rx0 and rx1 a priori when one is changing sigma coordinates?