An examination of Battlescale Forecast Model Initializations over complex terrain

The Battlescale Forecast Model (BFM) has been initialized in a series of tests using a combination of Navy Operational Global Atmospheric Prediction System (NOGAPS) grids and RAOBs. The BFM is a PC-based mesoscale weather model currently being validated in a joint effort of the Environmental Verification and Analysis Center (a research center of the University of Oklahoma) and the Battlefield Environment Division of the Army Research Laboratory.

This paper examines initial errors in surface temperature and pressure and attempts to explain the source of these errors. The series in question stretched from 11 to 28 February 1999 and consisted of 12Z initializations only. Ingest data included 12Z soundings and forecast grids from the previous 00Z NOGAPS run. The BFM domain consisted of an 81x81grid centered at 36N and 111W with a resolution of 12 kilometers.

BFM 0-hour grids were compared to surface observations within a grid-cell surrounding each model data point. The verification data set was further restricted to observations taken at elevations within 10 meters of the model surface.

The presence of substantial nocturnal inversions and the mountainous nature of the model domain resulted in large variations in observed surface temperatures and pressures. This variability, in conjunction with the coarse data sets used for initialization, resulted in considerable differences between observed and 0-hour model surface temperatures and pressures.