The impact of assimilating FY-3D Microwave Humidity Sounder II radiance data on the analysis and forecast of two advection fog cases

Abstract

Accurate prediction of dense fog is critical for daily life and economic activities. However, its high sensitivity to numerical initial conditions underscores the pivotal role of data assimilation in improving prediction quality. The impacts of data assimilation on a warm advection fog over the South China Sea and a cold advection fog in the North of China in 2021 were investigated using the radiance data from the FY-3D Microwave Humidity Sounder II (MWHS2) and the conventional observational data from the Global Telecommunication System (GTS). The three-dimensional variational assimilation (3DVAR) data assimilation scheme was applied to introduce the MWHS2 radiance data, using the Radiative Transfer for Tovs (RTTOV) model as the observation operator. Experiments with and without data assimilation were performed to show the impact of data assimilation on the analysis and forecast of the warm and cold advection fog events. Experimental results showed that the assimilation experiments outperformed the control experiment in simulating the warm advection fog over the west coast and sea areas of the Leizhou Peninsula and its northeast sea areas. Specifically, for visibility below 1 km, the data assimilation experiment (Exp_Da) improved the Fraction Skill Score (FSS) by 11.82 % and 6.56 %, and the Equitable Threat Score (ETS) by 19.92 % and 14.38 %, compared to the control experiment, for the 3-h and 6-h forecasts. It was found that Exp_Da significantly improved the representation of meteorological variables, including temperature advection and relative humidity, which were crucial for better predicting fog coverage. In the cold advection fog case in the North of China, the assimilation experiment resulted in an improved simulation of cold advection, particularly at 1000 hPa. In summary, assimilating both satellite and conventional observational data improves the simulation and prediction of warm and cold advection fog events. However, accurately simulating thin fog remains challenging. Visibility in the range of 1 km to 2 km has not been well simulated, necessitating further advancements in model development and fog diagnostic methodologies.