Accounting for the horizontal observation error correlation of precipitation observation
It is known that observations measured with the same instrument such as atmospheric motion vectors and satellite radiances have correlated errors. Satellite radiances have error correlations not only between channels but also in the horizontal. Previous studies successfully assimilated the satellite radiances with explicit consideration of inter-channel error correlations. For better use of dense observations, it would be essential to account for the horizontal observation error correlations, but no study has investigated the impact so far. In this study, we extend the NICAM-LETKF system to include the horizontal observation error correlation for assimilating dense precipitation data known as the global satellite mapping of precipitation (GSMaP) data. The results show slight improvements of the humidity analysis, especially in the southern hemisphere and the tropics, by assimilating dense GSMaP data with explicit consideration of horizontal error correlations.
Dr. Koji Terasaki RIKEN, Center for Computational Sciences