热带气旋定量降水预报的动力相似方案

该文提出一个热带气旋定量降水预报(QPF) 的动力相似方案。方案应用热带气旋初始参数、历史过程的天气形势场和物理量场及数值预报产品, 构造预报区域内当前时刻至未来时刻环境要素场的多元客观相似判据。通过定义非线性的相似指数综合评估历史热带气旋样本与预报热带气旋在多元判据下的连续动态相似程度, 以此找到相似样本。应用相似样本的历史降水量记录进行相似指数的权重综合, 得到热带气旋未来6~48 h降水量的定点、定量预报值。预报试验表明该方案具有一定的预报技巧。

A Dynamic Similitude Scheme for Tropical Cyclone Quantitative Precipitation Forecast

It is difficult to make tropical cyclone quantitative precipitation forecast for a specified location.The forecasts from numerical weather models are generally on latitude/longitude grids and extra errors can be brought when these predictions are interpolated to a location not on the grids. A prediction scheme based on dynamic similarity principle is proposed for tropical cyclone quantitative precipitation forecast at several hundred stations in China.Tropical cyclone precipitation generates in a certain synoptic situation. As a matter of fact, a similar tropical cyclone and interaction with environmental fields usually bring similar rainfall process.The key point of the scheme is to constitute prediction scheme with NWP product to seek similar process. For the sake of looking for similarity from more past records, put to use some key technique.Using the objective criterions which are made up of initial parameters of tropical cyclone and of the surround field and the numerical prediction products the similar samples are found out from past tropical cyclones by computer.The scheme is to constitute reasonable criteria for selecting tropical cyclones similar to the predicted one in both the storm feature itself and the large scale atmospheric environment interacting with the storm. Factors considered include initial tropical cyclone parameters, initial and historical large scale atmospheric environment, and future trend of large scale atmospheric environment. A similarity index is set up using a successive-dynamic method, which is non-linear in nature and can quantitatively describe the degree of similarity. By this index, several historical tropical cyclones are selected as cases similar to the predicted one to different extent. A consensus prediction for the precipitation at each station in future 48-hour at 6-hour intervals is then made using the index as weight coefficients for the precipitation of historical cases. Independent tests are carried out for tropical cyclones causing precipitation different in both strength and time duration to China and promising results are obtained. In these tests, the numerical predictions from global circulation models are used to describe the future trend of large scale atmospheric environments. The tropical cyclone quantitative precipitation forecast is forecasted by assessment upon similarity and weight of effects on account of local surround. Using the dynamic similitude scheme for tropical cyclone quantitative precipitation forecast to different rainfall intensity and last time tests is made. Tests demonstrate that this technique has positive prediction skill comparing to the climatology and persistency method.