HowTo: Run at different horizontal resolutions
De LMDZPedia
Révision de 23 janvier 2021 à 19:10 par Lfairhead (discussion | contributions)
What one needs to know to be able to run at a different model resolution
This page contains some information and tips to help you setup a successful run at a different (horizontal) resolution
- You will need input and forcing files at the sought resolution
- You will need to adapt the (dynamics) time step
As explained in the LMDZ Tutorial courses "dynamics: ..." (e.g. here) there is a simple rule of thumb to know to what extent you need to change the "day_step" (number of dynamical steps per day):
day_step(max(iim,jjm)=N) ~ day_step(max(iim,jjm)=M) * N/M
Check what the reference values (in the gcm.def.* files in the LMDZ/Deflist directory) are for a given N are to evaluate what is needed for M
Example: currently, in gcm.def_144x142 i.e. for iim=144,jjm=142 , day_step=672 so to run at twice that resolution you should halve the time steps and thus use day_step=1944 (and adapt iphysiq accordingly, i.e. double it here) - You might need to change the dissipation constants
This is more tricky and less linear than for the time steps; in practice some trial and error will be needed but overall you will need to increase dissipation (i.e. lower le teta* constants) - At present, physics need to be run at a timestep of 15 minutes so you need to adjust iphysiq according to the formula
iphysiq = 900 * day_step / 86400
As guidelines, and starting values, you could use or extrapolate from those given in the table below:
Resolution | day_step | tetagdiv | tetagrot | tetatemp | iphysiq |
---|---|---|---|---|---|
96 x 95 (CMIP5) | 480 ( = 180 s) | 5400. | 5400. | 5400. | 5 ( = 15 min) |
144 x 142 (CMIP6) | 672 ( ~ 128 s) | 3600. | 21600. | 10800. | 7 ( = 15 min) |
256 x 256 (QUEST) | 1344 (~ 64 s) | 1200. | 7800. | 3800. | 14 ( = 15 min) |
If you have values for other resolutions that work well for you, don't hesitate to contact us so that we can include them in the table