Extended Abstract (372K)8.5
Impact of assimilating rainfall derived from radar and satellites on rainstorm forecasts over the southwestern United States
J. Xu, University of Arizona, Tucson, AZ; and X. Gao, S. Sorooshian, and Q. Xiao
The Four-Dimensional Variational (4DVAR) data assimilation technique is a powerful tool to improve model initial conditions for weather forecasts. In this study, the MM5-4DVAR system is used to investigate the impact of rainfall assimilation on the forecasts of convective rainfall over the mountain areas in southwestern United States. The assimilation rainfall is derived from radar and/or satellite information. This study evaluates the 4DVAR rainfall assimilation skill through a convective rainstorm event occurred over southern Arizona during 5-6 August 2002. Six experiments consist of 2 options of assimilation window (i.e., 3-hour and 6-hour) and 3 options of rainfall data sources (i.e., radar-derived, satellite-derived, and radar-satellite rainfall) were conducted. The results show that using rainfall assimilation can produce more realistic moisture divergence and temperature fields in initial conditions, thereby, forecast rainstorm closer to observations in both quantity and pattern. The crucial minimization process in the 4DVAR is found sensitive to the assimilation rainfall data source and the length of assimilation window. The effective duration of forecast depends on the length of the assimilation window. For the study case, 3-hour assimilation window in the 4DVAR system works well for the 6-hour rainfall forecast, but 12-hour rainfall forecast requires the use of 6-hour assimilation window.
Session 8, Assimilation of Observations (Ocean, Atmosphere, and Land-surface) into Models (Room 618)
Thursday, 15 January 2004, 1:30 PM-3:00 PM, Room 618
Previous paper Next paper
Recorded presentation