Impact of Hyperspectral Infrared Sounding Observation and Principal-Component-Score Assimilation on the Accuracy of High-Impact Weather Prediction

Abstract

Observations from a hyperspectral infrared (IR) sounding interferometer such as the Infrared Atmospheric Sounding Interferometer (IASI) and the Cross-Track Infrared Sounder (CrIS) are crucial to numerical weather prediction (NWP). By measuring radiance at the top of the atmosphere using thousands of channels, these observations convey accurate atmospheric information to the initial condition through data assimilation (DA) schemes. The massive data volume has pushed the community to develop novel approaches to reduce the number of assimilated channels while retaining as much information content as possible. Thus, channel-selection schemes have become widely accepted in every NWP center. Two significant limitations of channel-selection schemes are (1) the deficiency in retaining the observational information content and (2) the higher cross-channel correlation in the observational error (R) matrix. This paper introduces a hyperspectral IR observation DA scheme in the principal component (PC) space. Four-month performance comparison case studies using the Weather Research and Forecasting model (WRF) as a forecast module between PC-score assimilation and the selected-channel assimilation experiment show that the PC-score assimilation scheme can reduce the initial condition’s root-mean-squared error for temperature and water vapor compared to the channel-selection scheme and thus improve the forecasting of precipitation and high-impact weather. Case studies using the Unified Forecast System Short-Range Weather (UFS-SRW) application as forecast module also indicate that the positive impact can be retained among different NWP models.