Effects of modifying variable "xtime"

[dashline]

I have an restart.2017-08-10_08.00.00.nc file, and I used it to get restart.2017-08-10_09.00.00.nc file by integrating forward one hour with "config_do_restart = true". At the same time, I changed the value of the variable xtime in the restart.2017-08-10_08.00.00.nc file to “2017-08-11_08:00:00”(i.e., I changed the time to one day later) and renamed the restart.2017-08-10_08.00.00.nc file to restart.2017-08-11_08.00.00.nc, and then I integrated one hour forward to get the file restart.2017-08-11_09.00.00.nc.
In my understanding, the values of the variables in both restart.2017-08-10_09.00.00.nc and restart.2017-08-11_09.00.00.nc should be the same (I compared the theta of the two and found that they are not). Can you tell me why this is? Does modifying this xtime make the input (e.g. solar radiation) change?

 

[mgduda]

As you've suggested, modifying the valid time for fields in a file (and modifying your namelist accordingly) does change various calculations in the model for real-data simulations, such as the position of the sun in the sky at any particular time as well as land-surface characteristics like the temporally interpolated climatological vegetation fraction and albedo.

For a dycore only simulation (as you might get with one of the idealized cases), the simulation start time should have no impact. This is even true for simulations that use only certain classes of parameterizations (e.g., microphysics). But for physical parameterizaton schemes involving radiation and land-surface, the date and time matter.

 

[dashline]

Thank you very much for your answer, it was explained very clearly!