西太平洋副热带高压季节内异常活动的非线性动力-统计预测方法研究

West Pacific subtropical high (WPSH) is an important system which has a great impact on summer climate change in East Asia, especially in our country. The forecast of WPSH is a difficult problem of severe weather forecasting and meteorological scientific research, which has important scientific significance and application value. Aiming to tackle the difficult prediction of morphological variation and the abnormal activities such as westward extension, northward jump or continuous maintain of WPSH, the low rate of medium and long term forecasting accuracy of WPSH, WPSH and its surrounding weather systems are diagnostic analyzed and key influencing factors and prediction factors are objectively extracted. Rough set methods combined with intelligent predictive algorithms such as hybrid hierarchy genetic radial basis network are used to solve incomplete information of small sample cases and forecasting difficult problem of abnormal activities such as westward extension and north hop of WPSH within season. Memory function of dynamical system are introduced to expand,the dynamic forecasting model idea of Youth Fund and 7-15 days medium and long term trend prediction of WPSH are investigated. The "two-step" power-quarter forecast system of WPSH is established to carry out seasonal and interannual variability prediction of WPSH anomalies. In the project, the forecast ideas and approaches of WPSH in Youth Foundation are further expanded,and the forecasts accuracy of the abnormal intraseasonal activities and medium and long term trend prediction is greatly improved. The research provides theoretical and technical support for season activities and variation of WPSH..

西太平洋副热带高压是影响夏季东亚地区天气气候和我国天气变化的重要系统。副高预测一直是气象科学研究和灾害性天气预报的难点问题，具有重要的科学意义和应用价值。本项目针对副高西伸、北跳或持续维持等异常活动和形态变异预测困难，中长期预报准确率偏低等问题，对副高及其周围的天气系统进行诊断分析，客观提取副高关键影响要素和预报因子；将粗糙集等方法与混合递阶遗传径向基网络等智能预测算法相结合解决副高西伸北跳等季节内异常活动“小样本案例”信息不完备和预报困难等问题；引入动力系统记忆函数，完善和拓展青年基金中的动力预报模型反演思想，开展副高7-15天的中长期趋势预测研究；建立副高的“两步法”动力季度预测系统，开展对于副高的季节和年际变率异常的预测。通过研究，进一步完善和拓展青年基金中的副高预测思想和方法途径，提高副高的季节内异常活动和中长期趋势预报的准确率，为副高季节内活动和变异提供理论和技术支持。

本项目是国家自然科学基金 2014 年资助的面上项目，研究起止日期为2014 年1 月至2017年12 月，资助经费 71 万元。本项目旨在针对副高西伸、北跳或持续维持等异常活动和形态变异预测困难，中长期预报准确率偏低等问题，开展副高异常活动的非线性动力-统计预测方法研究。研究内容包括五个方面：客观提取副高关键影响要素和预报因子；开展副高季节内异常活动的预报研究、开展副高7-15天的中长期趋势预测研究、开展副高年际变率的预测研究以及副高的非线性动力机制研究。课题组围绕副高非线性动力-统计预测方法开展了系统深入的研究，取得了积极的研究进展和富有创新特色的研究成果，提出了遗传算法、ANFIS等智能预测算法相结合解决副高西伸北跳等季节内异常活动的预报思路以及引入动力系统记忆函数，完善和拓展传统的动力预报模型反演思想，开展副高7-15天的中长期趋势预测的科学方法和技术途径。在核心刊物发表录用研究论文21篇，包括15篇SCI，1篇EI。申请国家发明专利一项。研究成果提取和凝炼出有创新特色的副高预测思想和途径，对副高异常活动预报和中长期预报具有科学意义和应用价值。此外，先后有1名博士和2 名硕士研究生参加了本课题研究，通过科研实践，研究生的基础理论和科研能力得到了全面锻炼和提高，达到了科学研究与人才培养双赢的目的。