Application of the Satellite-Based Spectral Relationship to the Vertical Localization for the Microwave Humidity Sounder in the Ensemble Kalman Filter

Abstract

Localization is an essential technique to mitigate the sampling error in the ensemble Kalman filter (EnKF). In order to effectively use the localization function within the EnKF, the specific location information of observations being assimilated is required to measure the distance between the model state variables and the observations. Since the satellite-observed radiance represents integrated quantities across the entire vertical profile of an atmospheric column, it is, however, a challenging issue to accurately assign the vertical location, especially for the satellite radiances sensitive to the variable atmospheric constituents (e.g., water vapor). In this study, we propose an efficient method for assigning the vertical location for observations of microwave humidity sounders (MHSs) onboard low-earth-orbiting (LEO) satellites, using discrete spectral characteristics between the 60 GHz oxygen and 183 GHz water vapor absorption bands. As the radiance differences between the channels are solely used as the predictors for estimating the vertical location in a multivariate regression framework, there is no need to conduct radiative transfer simulations that require additional computation costs. The estimated vertical locations are employed for the vertical localization function within the Data Assimilation (DA) Research Testbed (DART) implementation of ensemble filtering. The verification results show that applying vertical localization to observations from MHSs significantly improves the water vapor analysis derived by the DART system, particularly in the lower troposphere.