As there are few in situ observations in and around Antarctica, it is important to assess how assimilating remotely-sensed observations, such as satellite-observed radiances, can fill this observational gap. Thus, a month-long study was conducted over the Antarctic to examine forecast and analysis sensitivity to the assimilation of microwave radiance measurements. Several experiments, using both cyclic and non-cyclic initial conditions, were configured to quantify the impact of radiance data assimilation (DA) and explore different approaches of radiance bias correction (BC). DA was performed using the Weather Research and Forecasting (WRF) model's three-dimensional variational (3DVAR) algorithm, and the analyses initialized 72-hr Advanced Research WRF model forecasts.

The results demonstrate the critical importance of properly bias correcting raw radiance observations. When assimilating radiances using a “cold start” BC technique, forecasts and analyses were degraded compared to those from parallel experiments that only assimilated conventional (i.e., non-radiance) observations. However, when BC parameters were “spun-up” for several months before the assimilation period, radiance DA yielded forecast and analysis improvements compared to when only conventional observations were assimilated. The same general results regarding radiance DA were obtained for both the cyclic and non-cyclic experiments, but cycling led to poorer overall analyses and forecasts.

This presentation will discuss these results and describe a suggested method for radiance BC within limited-area domains.