Simulation results change when I only made modification to Registry/Registry.EM_COMMON

[Songning_W]

Hello,

I met this problem here with WRF-v4.4.1. After I made some modifications to Registry/Registry.EM_COMMON (without changing any part of the other codes) and recompile, the simulation results changes. The modification includes two new variables in "NSSL WRF diagnostics" part, and add an variable to the nwp_output package line. The Registry file is attached.

The two plots uploaded here are results from a clean version of WRF-4.4.1 and a version with the modification to Registry.

Does anybody have a clue on why just modifying the Registry can change the result of the simulation?

Thanks,
Songning

 

[Ming Chen]

Songning,
I found that you added two new state variables in Registry.EM_COMMON, i..e

> state real UP_HELI_MAX_02 ij misc 1 - rh02 "UP_HELI_MAX_02" "MAX UPDRAFT HELICITY 0-2 km" "m2 s-2"
> state real UH_02 ij misc 1 - r "UH_02" "UPDRAFT HELICITY 0-2 km" "m2 s-2"

and these two variables are listed in nwp_output package.

However, you didn't modify any code to calculate/pass the two variables to the package. But memories are still allocated for the newly added variables. I am suspicious that somehow this affects the calculation of other variables and lead to the difference in your result. It is beyond my expertise to address how the memory allocation could affect the model run, and hope someone in the community may have more insight.