Assimilating FY3D‐MWRI 23.8 GHz observations in the CMA‐GFS 4DVAR system based on a pseudo All‐Sky data assimilation method

Abstract

AbstractIn the current operational four‐dimensional variational (4DVar) data assimilation (DA) system of the Global Forecast System developed by the China Meteorology Administration (CMA‐GFS), all microwave observation data in cloud and precipitation regions are discarded during pre‐processing. This study implemented a Pseudo All‐Sky DA (referred to as PAS‐DA hereafter) subsystem in the operational cycle version of the CMA‐GFS (CMA‐GFS v3.2). The term “pseudo” in this study indicates that the Jacobians of brightness temperature with respect to hydrometeors from the adjoint radiative transfer model were temporarily neglected. Specifically, a liquid hydrometeor sensitive channel (23.8 GHz V pol.) of the MicroWave Radiation Imager on the platform of FengYun‐3D (FY3D‐MWRI) was selected to assess the impact of the all‐sky assimilation approach using the PAS‐DA subsystem. In the observation error model, we proposed a new cumulative distribution function (CDF) bias correction method for the cloud proxy in consideration of large discrepancies between the probability density functions (PDFs) of the observed‐cloud‐proxy and simulated‐cloud‐proxy (known as “cloud bias”). Results of single‐observation experiments justified that the present PAS‐DA subsystem could extend analysis increments to cloud regions, meanwhile correcting the errors of humidity analysis according to the mislocation of cloud distributions. In addition, the forecast experiments that were run for 1 month of the 6 hr PAS‐DA cycle in July and August 2021 demonstrate obvious superiority of the PAS‐DA over the current operational clear‐sky DA cycle: (a) root‐mean‐square errors (RMSEs) of humidity analysis were reduced by about 10% in the tropics, (b) significant improvements in humidity forecasts could be sustained for 96 hr and (c) many other forecast scores in the tropics and Southern Hemisphere also benefit from the PAS‐DA approach. Therefore, the PAS‐DA could be used for all‐sky assimilation studies and particularly for understanding how the all‐sky assimilation approach works, although the PAS‐DA serves as a transition scheme from the clear‐sky to “true” all‐sky DA.