WRF crashes at KF cumulus call

Greetings,

I have run a quasi-global WRF (3.5) simulation. I wish to use the parent domain output (forced by CFSR up to 50 hPa) to force an inner domain using ndown. I successfully generate the new wrfinput and wrfbdy file. When I submit wrf.exe however, I get the following error at initialization as the model tries to integrate for the first time:

d01 2008-10-01_00:00:00 in kf_eta_cps
-------------- FATAL CALLED ---------------
FATAL CALLED FROM FILE: <stdin> LINE: 2122
QG, QG(NK).LT.0
-------------------------------------------

My nested domain boundaries are sufficiently far away from the parent domain boundaries.

To be clear, p_top_requested is 5000 (50 hPa), but there is data in the met_em files up to 1 hPa. Based on what I've read, it seems as though my p_top_requested was set too high in the parent domain experiment.

I don't want to go back and rerun a three-year-long quasi-global simulation. Any tips on how I can circumvent this error?

Thanks in advance,
-Stefan

Update --

I ran the parent domain simulation with p_top_requested = 5000. I then ran ndown with the same, generating the wrfbdy and wrfinput files. For wrf.exe in the nest, I changed p_top_requested to 100, and the simulation ran fine.

Something seems fishy here....

-Stefan

FIXED!

The problem was the result of one of my classic bone-headed maneuvers in which I defined p_top_requested TWICE. Wrf.exe ran fine with a p_top_requested = 5000. Sorry!

-Stefan