The“Ensemble Dynamic Factors”approach to predict rainstorm

In this paper, we describe the definition and physical meaning of several important physical variables, such as the generalized potential temperature, the moist thermodynamic advection parameter, the thermodynamic helicity, the vertical flux of the thermodynamic divergence,the moist potential vorticity, the solenoidal vorticity and the thermodynamic solenoidal vorticity, the second-order potential vorticity,the convective vorticity vector, the wave-activity density and so on. Case studies show that these parameters have close correlation to the evolution of precipitating systems and can detect the occurrence and development of rainfall. This is mainly due to the following two reasons. First, these dynamic parameters can describe the common dynamic and thermodynamic features of precipitating systems. Second, since most of these parameters contain the generalized potential temperature which is related to the condensation latent heating and relative humidity,they implicitly reflect the structure of atmospheric moisture. Based on these parameters, an“ensemble dynamic factors”approach to predictheavy rainfall is developed. In this approach, the precipitation forecasting equation as a function of a single dynamic factor is built first using the GFS reanalysis data. Then, according to the correlation coefficients between the analyzed precipitation from different parameters and the observed precipitation, weighting functions, which measure the contribution of the precipitation obtained from a single parameter to the totalprecipitation, are developed. Based on these weighting functions, a weighted average of the precipitations from all the dynamic parameters is conducted, which gives the final precipitation forecast. This approach combines the advantages of multiple dynamic parameters, and can reflect the common characteristics of the rainfall processes. Statistical verification with a long time series shows that the precipitation forecastscore of the ensemble dynamic factors is higher than that of the GFS model, although both of them overestimate the precipitation intensity. The “ensemble dynamic factors”approach to predict precipitation is able to generate the product of precipitation forecast, and thus can provideassistance to forecasters.