nolake sensitivity experiment: Is modifying only LU_INDEX in geo_em.d0*.nc sufficient?

[845968376]

Hi all,


I am conducting a nolake sensitivity experiment in WRF and would like to confirm the correct and physically consistent procedure.


So far, I have only modified the land-use type by changing the LU_INDEX over lake grid cells in geo_em.d0*.nc (i.e., converting lake to land). The model runs successfully, but I noticed an issue in the simulation results:

Even after removing lakes in geo_em, the high surface skin temperature (TSK) over former lake areas persists throughout the simulation period. This warm patch is clearly inherited from the ERA5 initial conditions, and it contradicts the purpose of my nolake experiment.

This leads me to the following questions:

1. Is modifying only LU_INDEX in geo_em.d0*.nc sufficient for a nolake experiment?
2. Since TSK is interpolated from the driving data (ERA5) during real.exe and does not automatically adjust to land-use changes,
   should TSK over former lake grid cells be explicitly modified in wrfinput_d0*？ How？just use T2m？（but T2m also tend to too high）
3. Besides TSK, are there other surface or soil variables that should also be adjusted to remove residual lake thermal or hydrological memory?
4. Would extending the spin-up time help eliminate the warm TSK signal inherited from ERA5?
   
   • My current spin-up time is about 10 hours.
   • Since this study focuses on a heavy rainfall event, what would be a reasonable spin-up duration (e.g., 24 h, 48 h, longer)?

Any guidance or best practices would be greatly appreciated.

Thank you very much.