Case Example Continued

This is a continuation of the post below, so start down there.

-------------------------------------------------------------

At top, graphic is the WRF 18 hour rainfall forecast from 12 UTC 4 August, 2008, after the model was initialized using the NWS NAM initial fields. Convective precipitation is over-forecast to a fairly serious degree (relative to the radar rainfall estimates shown in previous post).

-------------------------------------------------------------

Mike subjectively corrected the PW initial field by removing 3 mm of PW from the lowest 100 mb of the TWC sounding (admittedly a brute force procedure). The results of a re-run of the WRF model are shown in the bottom graphic. Forcing the PW toward the GPS values apparently minimized the impact of the bad humidity data from the TWC RRS sounding, since the new forecast was much more similar to the radar rainfall estimates.

--------------------------------------------------------------

This simple example indicates primarily that bad humidity data from RRS sondes does contaminate the NAM initial conditions. The NOAA folks in Boulder, who run the RUC model, have been experimenting with using the GPS PW in initializing the model. I'll see if I can chase down a reference.