For the past two winter seasons, the Forecast Systems Laboratory has run special high-resolution versions of the Rapid Update Cycle (RUC) model in support of the PACJET Experiment and distributed model guidance products to the NWS Western Region for local office display within AWIPS. These PACJET forecasts have been completed using a new version of the RUC model (featuring enhancements to the moist physics, land-surface model, and analysis procedure) that has been shown to produce better precipitation forecasts and is currently being implemented at NCEP. The forecasts have benefited from the assimilation of experimental rapid-scan satellite-derived winds (derived from 7.5 min visible and infrared satellite imagery as part of the GOES rapid-scan Winds Experiment - GWINDEX) in addition to the conventional observations. Real-time RUC forecasts have been complemented by retrospective data impact tests that quantify the forecast improvement from the rapid-scan winds.

In this paper, we will present a summary of the real-time forecast system and the products provided to the NWS forecasters, highlighting the evolution of our activities from 2001 to 2002. Changes include increasing the grid resolution from 20-km to 10-km, using LDM to better automate the AWIPS file transfer procedure, and adding a forecaster evaluation procedure. Selected case-study comparisons illustrating the differences between the 10-km PACJET RUC and 40-km operational RUC will be shown.

The remainder of the paper will focus on initial results from the retrospective test designed to quantify the forecast impact from the special rapid-scan satellite winds. In this test, two parallel sets of 1-h cycled RUC forecasts were produced for a meteorologically active period. The first experiment utilized all conventional observations (rawinsonde, 404 and 915 MHz profiler, RASS, VAD, ACARS, and surface reports), while the second experiment, supplemented the conventional observations with the rapid-scan satellite winds. Initial results from the retrospective test indicate that inclusion of the rapid-scan satellite wind observations improved short-range (0-12 h) wind predictions by up to 10% for this case.