Recently, Zou and Van Woert (2002) developed an algorithm to retrieve the three-dimensional atmospheric wind fields using satellite soundings over the polar ocean area. In this algorithm, thermal winds computed from satellite temperature retrievals are sequentially added to a satellite-derived ocean surface wind field to obtain a first-guess wind profile, and then the first-guess wind field is forced to conserve mass in a variational procedure. Studies with limited, but different, validation datasets (Zou and Van Woert 2002; Francis et al. 2004) suggest that the satellite-derived upper-level atmospheric winds from the TOVS temperature retrievals yield good accuracy over both the Arctic Ocean and Southern Ocean and thus provide a valuable source for climate change studies in the polar region.

This study extends the atmospheric wind retrieval algorithm into the polar land area. A planetary boundary layer model has been developed and added to the previous algorithm. The temperature retrievals from AMSU (Advanced Microwave Sounding Unit) and the 10-meter surface wind field from GDAS (Global Data Assimilation System) are used as inputs to the wind retrieval algorithm. The retrieved atmospheric wind profiles are then validated against radiosonde observations over 11 different stations around the Antarctic coast. Reasonable accuracy has been achieved over most of the radiosonde stations in a one-year validation period. Comparisons with the GDAS wind profiles at the radiosonde stations are also conducted. Reasons for the agreement and disagreement between the satellite-derived winds and radiosonde observations will be discussed.