Surface roughness values (z0) saved during run and restart for ABLTerrainSolver

[nitiesharya]

Hi.
I am using ABLTerrainSolver with shear stress condition at the surface for flow over topography. The flow field seems perfectly fine. When I checked the stress provided by the model, the values are quite reasonable. However, the surface roughness values that are written in the Rwall file are different from that I provide at the beginning (some zero and even negative). I thought the simulation is run with constant surface roughness which is true as I output the value of the surface roughness, it shows the value I provide at the start of the simulation in the setup file.

The main issue is when I restart the simulation I have to change the surface roughness value (to the initial value) in the Rwall file in the latest time folder. The ustarMean values provided by the solver changes (visible in the log file) and the simulation blows up after some time. Since a constant surface roughness is used in the simulation, the ustarMean value should ideally be equal to the value at the latest time step.

However, if I run the simulation without restarting it, the results provided by the simulation are perfectly fine.

I want to know the source of this discrepancy and, if possible, a solution. I am using a time-varying inlet fed from a precursor simulation
There is no heating or cooling, so the ABL has a neutral stratification.