一种局地非线性气候动力统计模型及其预报试验

根据反演建模理论，在引入一次样条函数的基础上，设计了一种局地非线性气候动力统计模型及其一整套反演方案。其实质是用逐段线性化的研究局地气候演化的非线性特征。作为初步试验，利用云南省18个测站1956年1月－1990年12月逐月温度距平和逐月雨量距平率的时间序列，反演得到一组近似描述云南省局地气候系统的非线性动力统计方程，应用反演获得的此非线性动力统计方程进行了云南省18个测站1991年1月－1994年12月逐月温度距平和逐月雨量距平率的预报试验。试验结果表明，温度距平和雨量距平率的拟合准确率分别约为78％和64％；温度距平和雨量距平率的外推预报准确率分别为75％和63％。表明此模型具有一定的拟合和预报能力，同时具有良好的稳定性。

A Local Nonlinear Dynamical and Statistical Climate Model and Its Forecasting Test

After introducing the linear spline function, a local nonlinear dynamical and statistical climate model and its inversion method was designed according to the inversion theory. Studying the nonlinear evolution features of the local climate system with stepwise linear method is the essential of the model. As the primary examination, the local dynamical and statistical climate model was retrieved with the monthly temperature anomaly and rainfall anomaly ratio of 18 observatories of Yunnan provincefrom January 1956 to December 1990. The forecasting ability of this model was tested with the same time series but from January 1991 to December 1994. The fitting accuracy of temperature anomaly and rainfall anomaly ratio is 78% and 64% respectively. The forecasting accuracy of temperature anomaly and rainfall anomaly ratio is 75% and 63% respectively. These examination results indicated that this model has efficient fitting and forecasting ability, and has better stability.