首页 | 本学科首页   官方微博 | 高级检索  
     检索      

海平面变化的小波和自回归模型集成预测试验
引用本文:袁林旺,谢志仁,钟鹤翔.海平面变化的小波和自回归模型集成预测试验[J].海洋科学,2008,32(4):31-35.
作者姓名:袁林旺  谢志仁  钟鹤翔
作者单位:南京师范大学,地理科学学院,江苏,南京,210097
基金项目:国家自然科学基金,国家重点基础研究发展规划(973计划)前期研究专项
摘    要:针对海面变化预测时间序列模型中趋势组份和周期(准周期)组份的提取和预测问题,基于吴淞站1955~2001年月平均潮位序列,采用小波分析(WA)与自回归(AR)模型相结合的方案,对小波分解的不同尺度分量序列,借助于时间序列模型进行分量预测,再对它们进行叠加建立预测模型,进行了月平均潮位预测试验.以1955~1996年数据为基础建立模型,1997~2001年数据作为验证,结果表明两种方法的结合使用显示了较好的效果,具有较高的精度.

关 键 词:海平面变化  预测  小波分析(wA)  自回归(AR)模型
收稿时间:2004/11/24 0:00:00
修稿时间:2007/11/9 0:00:00

Prediction experiment of sea-level changes based on the wavelet analysis and auto-regressive model
YUAN Lin-wang,XIE Zhi-ren,ZHONG He-xiang.Prediction experiment of sea-level changes based on the wavelet analysis and auto-regressive model[J].Marine Sciences,2008,32(4):31-35.
Authors:YUAN Lin-wang  XIE Zhi-ren  ZHONG He-xiang
Abstract:Based on the monthly average tidal records of Wusong tidal gauge station from 1955 to 2001,a prediction experiment is made using wavelet analysis(WA) and auto-regressive(AR) models in this paper.WA provides a way of analyzing the local behavior of functions,it may be used to isolate the sea-level change records to reveal the hierarchy of features and the scaling behavior.Since components are the filtered versions of the raw data,their behavior is more regular than that of the original signal and more predictable accordingly.Using AR model to predict each component in practice we can take advantage of each significant component,the summation of all the predicted components is the predict results.Analyzing the data of 1955-1996 to predict the sea-level change of 1997-2001,comparing the prediction results with the original data, we find that these two series are fairly comparable.It was shown that the integration use of these two methods provides an efficient way to predict sea-level changes.
Keywords:sea-level change  prediction  wavelet analysis(WA)  auto-regressive(AR) model
本文献已被 CNKI 维普 万方数据 等数据库收录!
点击此处可从《海洋科学》浏览原始摘要信息
点击此处可从《海洋科学》下载免费的PDF全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号