Impacts of satellite data assimilation with different model vertical levels on QPFs downstream of the Tibetan Plateau

Abstract

Measurements from various sounding instruments onboard polar-orbiting meteorological satellites quantify contributions to the total radiation at various microwave or infrared frequencies from different levels of the atmosphere. Satellite data assimilation adjusts model profiles of temperature and water vapor by minimizing the differences between observations and model simulations to search for the maximum likelihood estimate of the atmospheric states. The accuracy and precision of satellite data assimilations depend on the model vertical resolution. Sensitivity studies are carried out to compare the data assimilation and forecast results over a domain centered on the Tibetan Plateau (TP) using three different model vertical resolutions: 43, 61, and 92 vertical levels from the surface to ~ 1 hPa. The NCEP Gridpoint Statistical Interpolation (GSI) analysis system and the Advanced Research Weather Research and Forecasting (ARW) model are used with a domain size of 600 × 500 grid boxes at a 15-km horizontal resolution. It is shown that the ARW/GSI system with the coarsest (highest) model vertical resolution outperforms the remaining two for the 24-h short-range (48-h medium-range) quantitative precipitation forecasts (QPFs) downstream of the TP. The satellite data assimilation at the highest model vertical resolution produced more significant positive impacts on the 36-h forecasts of a mid-tropospheric trough located to the northeast of the TP that lead to a localized precipitation event. Improvements in the QPFs with the 92-vertical-level configuration come mainly from the best match of rainfall distributions between observations and forecasts.