Assimilation of the pseudo-water vapor derived from extrapolated radar reflectivity to improve the forecasts of convective events

In this study, an approach is proposed to improve the numerical weather prediction (NWP) of convective event forecasts by combining very short-term extrapolated reflectivity (0-1h) with a numerical model. After direct assimilation of real radar observations, the pseudo-water vapor estimated from future extrapolated reflectivity is assimilated using a cycled three-dimensional variational assimilation (3DVAR) system. To filter extrapolated reflectivity values with large errors and extract reliable and useful convection information, a proposed regional filtering procedure is performed on the extrapolated reflectivity. In addition, regional filters of different scales and the reflectivity of different extrapolation times are examined. The assimilation results for two convection events with different characteristics show that assimilation of the pseudo-water vapor estimated from extrapo-lated reflectivity can improve the reflectivity and accumulated precipitation forecasting. The improvement effect is more significant in case 1, which is a well-organized squall line event, and less effective in case 2, which is a faster developing and evolving local convection event. In both cases, the regional filtering method can limit the wet bias in the analysis field and suppress spurious forecasts of reflectivity and precipitation caused by assim-ilating the pseudo-water vapor estimated from extrapolated reflectivity without regional filters. Additionally, the assimilation of a shorter (30 min) time window may prevent the introduction of more error into the analysis field, but the improvement of the forecasting skill may be limited.