Adequately forecasting moist processes resulting from mesoscale and synoptic weather system dynamics is an active problem in the realm of operational meteorology. Numerical weather prediction has been a beneficial tool for studying and forecasting such processes. A number of parameterizations have been developed to facilitate the solution while suppressing numerical instabilities and controlling budgets of conserved quantities. However, the ideal model must be initialized with an analysis that adequately resolves variations in the moisture concentration and cloud cover on the same scale as the simulation grid spacing to attain the most accurate forecast. Due to a very sparse upper-air observation network across the United States, the only way to accomplish this is with satellite products.
A methodology has been developed for an experiment with several parallel regional Weather Research and Forecasting (WRF) model simulations initialized with satellite-based retrievals. The intent is to clarify the impact of observations, in the form of retrievals, from the Geostationary Operational Environmental Satellite (GOES) Sounder on 12, 24, and 36-hour WRF model forecasts of precipitable water, low-level relative humidity, precipitation, and sky cover. Two experimental analyses are built from a CIMSS Regional Assimilation System (CRAS) pre-forecast spin-up. The CRAS assimilates precipitable water and cloud products derived from the GOES Sounder. An experimentation period between late September and early October 2011 found that the majority of impact in the experimental simulations compared to the control is recognized in the total precipitable water field over the first 12 hours. In some cases, this resulted in an improved precipitation forecast. Cloud cover results were inconclusive, though a new technique developed for use in the CRAS outperformed the current WRF cloud fraction approach.
|Updated 18 December 2011|
|Space Science and Engineering Center|
University of Wisconsin-Madison