自组织网络与广义回归网络耦合的副热带高压指数预测
PREDICTING THE SUBTROPICAL HIGH INDEX BY COUPLING SELF-ORGANIZING FEATURE MAP AND GENERALIZED REGRESSION NEURAL NETWORK
-
摘要: 利用亚洲夏季风系统中各成员变化活动与西太平洋副高存在的不同程度的时延相关性,从1995~2004年NCEP/NCAR逐日再分析资料中,提取了亚洲夏季风系统各成员变化活动的特征指标及其对应的超前三候的西太平洋副高(简称副高)面积和脊线指数。在此基础之上,建立了自组织网络与径向基函数网络串级耦合的副高指数预测模型。该模型首先用自组织网络对各指标样本按其自身相似原则进行无监督分类,随后用广义回归网络分别对分类出的各指数样本子集进行有监督的训练建模和预测。模型的预测试验结果表明:副高指数的预测结果与其实际值之间的相关系数达到0.89,明显优于单一的神经网络模型预测效果。Abstract: Based on time-lag correlation between the activity of individual members of the Asian monsoon system and the Western Pacific Subtropical High,character index of the members' variation and the corresponding subtropical high three pentads ahead were extracted from the NCEP/NCAR daily reanalysis dataset during the ten years from 1995 to 2004. The subtropical high index predicting model was built that coupled Self-Organizing Feature Map(SOFM) and Generalized Regression Neural Network(GRNN) in series. The model sorted the index samples without supervision according to the characteristic similarity by the SOFM Neural Network,and then built and trained the model by the GRNN based on the index sample subset. The prediction experiment results show that the correlation coefficient between the predicted result and the actual value reached 0.89 and that experiment results were much better than that by single Neural Network prediction model.
-
[1] 吴国雄,丑纪范,刘屹岷,等.副热带高压形成和变异的动力学问题[M].北京:科学出版社,2002. [2] 喻世华,杨维武.副热带季风环流圈的特征及其与东亚夏季环流的关系[J].应用气象学报,1991,2(3):242-247. [3] ZHU Qiangen,HE Jinhai,WANG Panxing.A study of circulation differences between East Asian and Indian summer monsoon with their interaction[J].Adv Atmos Sci,1986,3(4):466-477. [4] 陶诗言,朱福康.夏季亚洲南部100毫巴流型的变化及其与西太平洋副热带高压进退的关系[J].气象学报,1964,34(4):385-394. [5] 杨修群,黄士松.马斯克林高压的强度变化对大气环流影响的数值实验[J].气象学报,1989,47(1):125-138. [6] 张韧,余志豪,蒋全荣.南海夏季风活动与季内北太平洋副高的形态和西伸[J].热带气象学报,2003,19(2):113-121. [7] 舒锋敏,简茂球.亚洲季风区感热凝结潜热对副高带季节演变的影响[J].热带气象学报,2006,22(2):121-130. [8] 巩远发,何金海,段廷扬,等.北太平洋中纬度负海温异常对副热带高压影响的数值试验[J].热带气象学报,2006,22(4):386-392. [9] 喻世华,王绍龙.西太平洋副热带高压中期进退的环流机制[J].海洋学报,1989,11(3):372-37. [10] 徐海斌,张韧,刘科峰,等.小波分解与SOFM-BP网络结合的西太平洋副高数值预报产品优化技术[J].热带气象学报,2007,23(3):265-270. [11] 张韧.基于前传式网络逼近的太平洋副高活动诊断预测[J].大气科学,2001,25(5):650-660. [12] 张韧,蒋国荣.用神经网络方法建立太平洋副高预报模型[J].应用气象学报,2000,11(4):474-483. [13] MARTIN T Hagan,HOWARD B Demuth.神经网络设计[M].北京:机械工业出版社,2002. [14] 余丹丹,张韧,洪梅,等.亚洲夏季风系统成员与西太平洋副高的相关特征分析[J].热带气象学报,2007,23(1):78-84.
点击查看大图
计量
- 文章访问数: 847
- HTML全文浏览量: 0
- PDF下载量: 1187
- 被引次数: 0