Development of a near real time regional satellite data assimilation system for high impact weather forecasts

Satellite data play more and more important role in numerical weather prediction (NWP), especially for high impact weather (HIW) forecasts. Last year's hurricane Sandy (2012) forecast has shown a very good example of HIW forecasts. It was demonstrated that with and without satellite data in NWP have a big difference in predicting hurricane Sandy's landing. Although satellite data assimilation has made a good progress, challenges still exist, particularly in the application of satellite data in regional NWP model due to its high spatial/temporal resolution and limited spatial coverage. How to best represent the information from satellite observation and how to get value added information from these satellite data into regional NWP model, including both radiance and retrieved products, still need investigation. In order to enhance the application of satellite data in regional NWP for HIW forecasts, we have developed a flexible near real time regional satellite data assimilation/forecast system for tropical storm forecasts (SDAT), the SDAT is a research test-bed to study the impacts of different satellite data on tropical storm forecasts. The system mainly consists of the community Gridpoint Statistical Interpolation (GSI) assimilation system and advanced Weather Research Forecast (WRF) model. In addition to assimilate regular conventional and satellite radiance data obtained from NCEP (National Centers for Environmental Prediction), the system is also able to assimilate derived products such as hyperspectral IR (infrared) soundings, total precipitable water (TPW), layer precipitable water (LPW). To set up the system parameters, a series of experiments have been carried out to test the impacts of different initialization schemes, different background error matrix, different NCEP global model date sets, and different WRF model horizontal resolutions. Based on a couple of historical tropical storm case data, tests have been done for different satellite data impacts study. We plan to run this assimilation/forecast system in near real time in this coming hurricane season. The results from this year's cases will be presented in the meeting and the lessons learned from this process will help us better utilize the current and future satellite data.