How to reconstruct the wrfrst & wrfbdy to make it possible to Restart with modified terrain height.

[jianglizhi]

Hello, I would like to conduct sensitivity test about terrain height on an em_real case. The number of domain are one.
The simulations preiod is from July 27, 1800UTC to July 29, 0600UTC. I want to evaluate the effact terrain height in the period just after July 28 0900UTC (Restart Time Point), which I was concerned.
I would like to restart using the wrfrst file and wrfbdy file at the Restart Time Point.
Because the conditions could change after modifying the terrain, I do not want to conduct a cold run from the earliest point, July 27, 1800 UTC.
And I don't want to start the run just several hours before the Restart Time Point using IC/LBC from the reanalysis dataset.
Since modifying the terrain height (HGT) would also influence the geopotential height and pressure in the model vertical coordinates, I think I should reconstruct the atmospheric state in the wrfbdy and wrfrst files so that they match the new vertical coordinates.
My question is: is there any solution to reconstruct the wrfrst file on a grid with changed terrain?
I want to use the state from the generated wrfrst and apply some interpolation solution (like real.exe).
As far as I'm concerned, real.exe interpolates the reanalysis fields into user-defined model levels.
I want to make an interpolation from the user-predefined fields into a newly user-defined grid.
Can anyone give some advice? Any help would be appreciated.

 

[jianglizhi]

COMMENTS:

After some research, I found that it is difficult to reconstruct the wrfrst file with new vertical layers because WRF is based on an eta vertical coordinate system that depends on surface pressure. Changing the terrain height will definitely affect the pressure at the surface level.

Another problem that needs to be solved is that the state variables used in some of the model's internal processes (e.g., microphysics) should transition seamlessly. This approach is high-risk, so I do not think reconstructing the wrfrst file is a good idea.

Perhaps a cold start at the restart time point using the atmospheric state recorded in the wrfrst file would work:
extract the variables from the wrfrst file, interpolate the necessary variables onto the updated terrain height, and then write the updated variables into a revised wrfinput file before restarting the model.

As for the wrfbdy file, since it only affects the lateral boundary, it may not need to be modified.
For the wrfinput file, it may be feasible to interpolate the 3D variables (e.g., U, V, W, PH, THM, T, P, Q*). Using the pressure at the top and the corresponding geopotential height at the top, we can calculate the surface pressure at the updated terrain height. Given the updated surface pressure and eta values, we can compute the 3D pressure variables and then use the hydrostatic relation to update the corresponding geopotential height. Finally, we can interpolate the 3D variables onto the updated geopotential height.
For some surface variables (e.g., Q2, T2, TH2, U10, V10), interpolation into the surface level might also be possible. I am unsure whether it is better to directly use the variables from the old wrfinput file.