Hi all,
In ROMS-2.2 to restart the model using the last run output we have to switch on the NRREC value to -1. But in my experience I had find that when I switch on to NRREC value to 0 then only the model is running from the last step onwards. When I give NRREC value to -1 then it is running freshly i.e., from the first time step onwards.
nageswararao wrote:Hi all,
When I give NRREC value to -1 then it is running freshly i.e., from the first time step onwards.
This actually happens if you set NRREC=1. NRREC=-1 means restart from the latest available field in the ININAME file. I didn't experience any weird behaviour on this using both ROMS-2.1 and ROMS-2.2.
The problem you might be encountering is one with ANA_INITIAL. The way the code is written, it calls ana_initial instead of reading the restart file if ANA_INITIAL is #defined. I posted to the ROMS blog about this.
Thanks Jacopo and Kate for ur reply.
But when I gave NRREC value as 1 also it is running from the starting time step onwards. At the same time I undefined ANA_INITIAL in cppdefs.h.
I am also getting another problem that when I restart the model, the results for first time step are very bad i.e., I am getting very low temperature and salinity values,.... and for second time step the values are good. In my case, I am writing the weekly values in the average file, the first week values are very low and from second week they are good. Why it is happening
Thanks in advance.
I don't know what to say without knowing a whole lot more about what you are trying to do. It sounds like things are acting quite strangely, so I suggest you try to find access to a powerful debugger such as totalview and see for yourself what is happening.
I have the same problem with the very first values of history files - they are just slightly different, but the change is perceivable and occurs in every history file. This has happened with all 6 roms-2.2 configured on different machines. What I do to obtain time series for analysis is to skip this first value and then, for the analysis, to interpolate the time series on a regular time vector. I don't know why it happens, but I thought it had to to with the time stepping in roms (barotropic and baroclinic) - i.e. barotropic & baroclinic time step versus the time instant for the output.
When I give NRREC value to -1 then it is running freshly i.e., from the first time step onwards. 