Cloud-Resolving 4D-Var Assimilation of Doppler Wind Lidar Data on a Meso-Gamma-Scale Convective System

Abstract

Abstract The authors evaluated the effects of assimilating three-dimensional Doppler wind lidar (DWL) data on the forecast of the heavy rainfall event of 5 July 2010 in Japan, produced by an isolated mesoscale convective system (MCS) at a meso-gamma scale in a system consisting of only warm rain clouds. Several impact experiments using the nonhydrostatic four-dimensional variational data assimilation system (NHM-4DVAR) and the Japan Meteorological Agency nonhydrostatic model with a 2-km horizontal grid spacing were conducted in which 1) no observations were assimilated (NODA), 2) radar reflectivity and radial velocity determined by Doppler radar and precipitable water vapor determined by GPS satellite observations were assimilated (CTL), and 3) radial velocity determined by DWL were added to the CTL experiment (LDR) and five data denial and two observational error sensitivity experiments. Although both NODA and CTL simulated an MCS, only LDR captured the intensity, location, and horizontal scale of the observed MCS. Assimilating DWL data improved the wind direction and speed of low-level airflows, thus improving the accuracy of the simulated water vapor flux. The examination of the impacts of specific assimilations and assigned observation errors showed that assimilation of all data types is important for forecasting intense MCSs. The investigation of the MCS structure showed that large amounts of water vapor were supplied to the rainfall event by southerly flow. A midlevel inversion layer led to the production of exclusively liquid water particles in the MCS, and in combination with the humid airflow into the MCS, this inversion layer may be another important factor in its development.