一种基于多层感知器的动态区域联合短时降水预报方法

A DYNAMIC REGIONAL COMBINED SHORT-TERM PRECIPITATION FORECASTING APPROACH BASED ON MULTILAYER PERCEPTRON

降水是最难预报的天气要素之一。在预报站点和周围站点之间建立多层感知器,利用主成分分析对环流背景场和局地气象要素差异等13个因子降维,作为多层感知器的输入,并采用贪心算法选择多层感知器的结构。通过联合多个多层感知器同时预报降水,提出一种动态区域联合短时降水预报方法。实验证明该方法降水预报效果较好,3小时降水预报能力优于ECMWF(European Centre for Medium-Range Weather Forecasts)和日本气象厅(JMA)数值模式。

Precipitation is one of the most difficult weather factors to predict.A multilayer perceptron was built between the prediction site and the surrounding sites,and principal component analysis was used to reduce the dimension of 13 factors,such as circulation background field and local meteorological factors.The result,as the input of multilayer perceptron,combining with the greedy algorithm,was used to select the structure of multilayer perceptron.Precipitation is predicted by combining multiple multilayer perceptron.A short-term precipitation prediction method based on dynamic region combination was proposed in this paper.The result of the experiment shows that the method of precipitation forecast is effective that the 3-hour precipitation forecast capability is better than the ECMWF and the JMA numerical model.