Surface temperature assimilation improving geostationary meteorological satellite surface-sensitive brightness temperature simulations over land

This study explores a possibility of improving Advanced Himawari Imager (AHI) surface-sensitive brightness temperature (TB) simulations over land by assimilating land surface temperature (LST) observations from the National Basic Meteorological Observing Stations of China. The Gridpoint Statistical Interpolation 3D-Var regional data assimilation (DA) system is modified to add LST as a new control variable and its background error variances, horizontal correlations and cross-correlations. The background covariances of LST with other control variables are calculated separately for daytime and nighttime samples in summer and winter seasons. A control experiment (ExpCTL) and three LST DA experiments with (ExpLST) and without (ExpLST_NBC) bias correction or with an average of LST within 2(degrees) x 2(degrees) grid boxes (ExpLST_SO) are conducted. Considering the fact that surface station observations are point measurements while the satellite TBs measure the total radiation effect of earth's surface within fields-of-view, a bias correction is found necessary for LST DA during daytimes (ExpLST). The biases are quantified by the differences from the Moderate-resolution Imaging Spectroradiometer LST retrievals to compensate for the representative differences. The analyzed fields are then used as input to the Community Radiative Transfer Model to simulate TBs of AHI surface-sensitive channels overland. A long-period statistics shows that ExpLST significantly reduces the observations minus simulations (O-B) biases and standard deviations of surface-sensitive TBs in terms of reducing the diurnal variations and season dependences of TB biases over different surface types, which also outperforms ExpLST_NBC and ExpLST_SO at daytime. This study suggests a potential benefit of combining the use of LST observations for assimilating surface-sensitive infrared TBs.