One of the most significant problems facing forecasters is the preparation of Quantative Precipitation Forecasts (QPF). It is well known that for numerical weather prediction forecasts, meso- and larger scale errors in the initial condition amplify rapidly. This generally leads to the degradation of QPF forecast guidance with increasing forecast lead time.

In the Winter Storm Reconnaissance (WSR) program, operationally implemented at the NWS in 2001, adaptive observations are taken over upstream areas sensitive to the development of critical winter weather events selected in real time by operational forecasters. The Ensemble Transform Kalman Filter (ETKF) technique is used to estimate the effect of a specific set of additional observations on reducing forecast error within a preselected geographical region of interest by reducing errors in the initial conditions. During the WSR 2002 program (22 Jan - 20 March), GPS dropsondes were adaptively released by the NOAA G-IV and the USAF C-130 planes along one of 53 preselected flight tracks, promising the largest forecast error reduction. All dropsonde data were assimilated operationally in the ETA and Global Forecast System (GFS; formerly MRF and AVN systems) at NCEP.

The impact of the adaptive observations is evaluated by running a parallel analysis/forecast cycle of the GFS system. QPF forecasts made from analyses with and without the dropsonde data are verified. A comparison of verification results for the parallel cycles reveals the impact of the adaptive observations on QPF. Beyond a case by case analysis of individual storms, results averaged over all cases will also be presented to explore the general effect of the adaptive observations on the moisture characteristics of winter storms.