Development of Regionally Focused Algorithm for AIRS Temperature and Humidity Retrievals Using a Moving-Window Technique

Abstract

Regionally focused algorithm for Atmospheric Infrared Sounder (AIRS) temperature and humidity retrievals was developed. We first employed regression model with a moving window technique. This is done by relating the AIRS measurements to temperature and humidity profiles with consideration of regionally and seasonally changing local climatology. Regression coefficients were obtained from four-year (2006-2009) of ECMWF interim data over East Asia and simulated AIRS radiances. Result showing a notable improvement of mean biases, compared to the regression retrieval which does not consider local features, suggests that the moving-window technique can produce better regression retrievals by including the local climatology in the regression model. For further improvement of the regression retrieval, one dimensional variational (1DVAR) physical model was also included in our algorithm. Error covariance matrix for the moving-window regression was obtained by using pre-developed regression retrieval and its error covariance. To assess the performance of 1DVAR using the mowing-window regression as a priori, error statistics of the physical retrievals from clear-sky AIRS measurements during four months of observation (March, June, September, and December of 2010) were compared

the results obtained using new a priori information were compared with those using a priori information from a global set of training data which are classified into six classes of infrared (IR) window channel brightness temperature. This comparison demonstrated that the physical retrieval from the moving-window regression shows better result in terms of the root mean square error (RMSE) improvement. For temperature, RMSE improvements of 0.1 – 0.2 K and 0.25 – 0.5 K were achieved over the 150 – 300 hPa and 900 – 1000 hPa layers, respectively. For water vapor given as relative humidity, the RMSE was reduced by 1.5 – 3.5% above the 300 hPa level and by 0.5 – 1% within the 700 – 950 hPa layer. As most of improvements due to use of the moving-window technique were shown in situations in which the relationship between measured radiances and atmospheric state is not clear, we investigated a possible use of surface data for further improving AIRS temperature and humidity retrievals over the boundary layer. Surface data were statistically and physically used for our AIRS retrieval algorithm. Results showing reduced RMSEs at both the surface level and the boundary layer, suggest that the use of surface data can help better resolve vertical structure of temperature and moisture near the surface layer by alleviating the influences of incomplete channel weighting function near the surface on the retrieval. In conclusion, developing regionally focused algorithm, the inclusion of climate features in the AIRS retrieval algorithm can result in better temperature and humidity retrievals. Further improvement was also demonstrated by adding surface station data to the channel radiances as pseudo channels. Since the hyperspectral sounder is available on the geostationary platform, the development of regionally focused algorithm could enhance its applicability to enhance our ability to monitor and forecast severe weather.