Polar Winds from MODIS: Algorithm and Processing Improvements

The generation of winds over the polar regions from the Moderate Resolution Imaging Spectroradiometer (MODIS) data began four years ago by adapting the winds algorithm used for geostationary satellites. It was recognized that deriving winds from polar orbiting satellites presented new challenges due to both the satellite orbit and the unique polar environment. These issues were identified but were not to be addressed until the usefulness of the polar winds was evaluated.

A positive evaluation by the ECMWF and the NASA Global Modeling and Assimilation Office (GMAO) of a case study dataset lead to the near real-time production of winds at CIMSS. The wind information is now retrieved by several numerical weather prediction (NWP) centers worldwide and is used in both operational and experimental forecast systems. We have also begun transitioning the software for use in NESDIS operations. These successes have prompted efforts aimed at improving the polar winds product in several areas: the timeliness of the data, correction for parallax, improvements in the height assignment, and the impact of the background guess field.

The timeliness of the winds is critical for assimilation by NWP models. The current delay of up to 5 hours is too long for some assimilation systems. We are reducing that time by using data from MODIS Direct Broadcast receiving stations in the high latitudes and by deriving winds from alternating passes of Terra and Aqua. The improved timeliness will be reported.

Recent improvements to the retrieval algorithm have also been made. A correction for the effect of parallax has been introduced. This is expected to improve the quality of the winds from single-satellite derived winds, but more importantly, it will allow for the derivation of winds from alternating passes of Terra and Aqua. This reduces the time required for a triplet of images from 200 to 100 minutes, but also has an affect on coverage. The software was also updated to use the higher bit-depth of MODIS, which we expect to have a positive effect on the cloud height determination. Previously, the data bit depth was reduced from 12 to 8 bits.

As part of the process of transitioning the code to NESDIS operations, differences in the winds were noted when using different background guess grids. These differences and the resolution will be presented.