基于关键区海温的华南香蕉寒害长期预报模型探讨

根据华南地区10个代表站点1961-2005年的气候资料、海温资料、历史灾情记录对华南香蕉寒害长期预报模型进行了研究。结果表明,构造的华南香蕉寒害指数与前人研究和历史灾情记录吻合,可以代表华南香蕉寒害的强弱。通过相关分析,找出了影响华南冬季寒害的前期夏季海温关键区为（5°S~5°N,170°~120°W）、（50°~60°N,180°~140°W）、（20°~30°N,140°~110°W）、（30°~40°N,140°~150°E）、（40°~50°N,150°~170°E）,关键区海温对华南冬季寒害影响具有明确的物理意义。利用逐步回归分析,以关键区海温为自变量建立了华南香蕉寒害长期预报模型。其中1961-1990年的数据用于建模,1991-2004年数据用于模型检验。模型拟合准确率和预报准确率均高于88%,这表明此模型具有较高的精度。

Study on Seasonal Forecast of Banana Chilling Injury in South China Based on Sea Surface Temperature of Key Areas

Based on the climate data, SST, historical records of chilling injury from 10 selected meteorological observation stations in South China in the period of 1961-2005, the long-term forecasting model of banana chilling injury in South China is studied. Results show that the banana chilling injury index (ICI) defined in this article coincides well with historical records of chilling injury and could effectively indicate the degree of chilling injury. According to the result of correlation analyses, the key areas of pre-summer sea surface temperature that affect chilling injury in South China are identified: (5°S～5°N, 170° ～ 120°W) , (50°～60°N, 180°～140°W)、(20°～30°N,140°～110°W)、(30°～40°N,140°E～150°E)、(40°～50°N,150°E～170°E).The key areas of pre-summer sea surface temperature are of significant physical implication to chilling injury in South China. Taking sea surface temperature in key areas as an independent variable, stepwise regression analysis is used to establish long-term forecasting model for the banana chilling injury in South China. The data of 1961～1900 is used for establishing the model and data of 1991～2004 are used for testing the model. Accuracy of model fitting and prediction is higher than 88% , which indicates that the model has a high accuracy.