Satellites provide the capability to monitor weather systems from outer space, and satellite data has been incorporated in numerical weather predictions through various data assimilation approaches. The three-dimensional variational data assimilation (3DVAR) approach has been used to improve objective data analysis and numerical model conditions in weather/research centers throughout the world. The 3DVAR system developed by the National Center for Atmospheric Research (NCAR) is used in this study along with the Fifth Generation Penn State University (PSU)/NCAR mesoscale model (MM5).
For the Polar Regions, the Moderate Resolution Imaging Spectroradiometer (MODIS) instruments onboard the NASA Earth Observing System (EOS) satellites Terra and Aqua provide frequent, high density observations. The MODIS level 2 data are available at 5 km resolution on 20 vertical pressure levels. The MODIS-retrieved variables that are assimilated as a new observational data source into the MM5/3DVAR system include temperature, dew point temperature (humidity), geopotential height (or pressure), and total precipitable water. In performing the 3DVAR analysis, the “MM5 global background error” produced by NCAR is used as the source of the background error covariances. The ‘observation’ error statistics are obtained by statistically studying and comparing the MODIS retrievals with conventional sounding data.
An extreme rain event which occurred in the summer of 2003 is chosen for our case study. This event produced more than 8 cm of rain in a 24-hour period over parts of interior Alaska, resulting in local flooding. The moisture was advected into the affected area by an unusually strong mid-tropospheric flow from the west and northwest. Numerical experiments have been conducted for the purposes of investigating the feasibility of assimilating MODIS data via 3DVAR, as well as its effect on the simulation of the extreme rain event. Preliminary analysis indicates that MODIS data provides detailed mesoscale information and that its assimilation contributes to improved forecasts. Detailed analysis of the results will be presented at the conference.