人工神经网络的短期气候定性预测方法研究

采用EOF方法将众多具有一定物理意义的环境场预报因子和序列周期因子进行降维去噪处理，并结合人工神经网络前馈网络模型，进行了影响广西热带气旋年频数和夏季(6～8月)降水量短期气候预测的定性预报建模方法研究。结果表明，该方法对于影响广西热带气旋的年频数及广西夏季降水量定性趋势具有很好的预报能力，可作为一种新的短期气候预测业务预报配套方法。

A Short-Term Climate Qualitative Forecast Model by ANN

The dimension reducing processing was conducted by means of the EOF method on a number of the environmental field and series periodic factors with definite physical meanings.In collaboration with the artificial neural network (ANN) scheme,a qualitative forecast model for the annual frequency of tropical cyclones and summer (June to August) rainfall affecting Guangxi was worked out.The results indicate that this model has a better qualitative capability of predicting the annual frequecy of tropical cyclones and sumer rainfall over Guangxi and can be taken as an auxiliary operational short term climate prediction technique.