(RESOLVED) Using monthly input for WRF

[jbasconcillo]

I will try to run monthly WRF using monthly ERA5 input data hence the interval_seconds in the namelist should be the total number of seconds in a month. My issue is the varying number of seconds per month. Has anyone done monthly simulations before and how did you resolve this issue?

 

[kwerner]

Hi,
Unfortunately there is no option for doing this, using the same namelist every month. Even if you were able to modify the WPS code to allow for this, the WRF model also requires lateral boundary conditions (LBC) updated at a constant interval, so code modifications for LBC updates at varying intervals wouldn't be compatible with how WRF and the WPS use datasets.

Additionally, you may want to consider the disadvantages to using monthly data for LBCs. This will likely lead to large errors around the boundaries of the WRF domain after much less than a month, since, for example, the boundary conditions would not provide any sort of diurnal cycle. If interested, take a look at this slide presentation (Dynamical Downscaling Overview and Best Practices) from our lab's Regional Climate Tutorial to see some additional suggestions for setting up a long simulation.

 

[jbasconcillo]

Thank you. Yes, I have come to realize these mistakes when I was trying to design the experiment. I want to run 10-year simulations and I thought that the output file sizes would be huge. I thought that using monthly data will be better to minimize output file size.

Now, I will instead use daily output from 6 hourly input data.

I would like to know your thoughts if that still stands to reason.

 

[kwerner]

Yes, 6-hourly data should suffice.

 

[kwerner]

Hi,
I received additional references that you may or may not find helpful.

• Using bias-corrected data for climate simulations

• Details of a climate study done for the Gulf of Mexico