RI地震预测模型的分析及其验证
|
摘要
地壳运动是一个极其复杂的无秩序现象,通过对地壳运动的研究很难对地震作出预测。为了准确地预测未来地震次数,利用统计学原理是一个很好的选择。RI(Relative-Intensity)方法是从统计学角度构造地震预测模型,对历史发生地震数据进行学习,预测未来将要发生的给定震级范围内的地震次数。RI方法基于这样一个假设:相同区域未来将要发生地震的相对强度和过去发生的地震相近。它将地震模型分为若干个等大的网格,并以网格为基本单位进行统计、计算,最后得到每个网格的地震预测值,对目标区域内所有网格的预测值进行累加就可以得到目标区域的预测值。RI方法在中国华北地区回顾性预测中表现出较好的性能和准确性。
The movement of crust of the earth is an extremely complex and chaotic phenomenon.It is very hard to make a prediction by surveying the movement of crust.In order to make an accurate prediction,statistic as a tool is a good choice.RI(Relative-Intensity) algorithm builds an earthquake forecast model which can make a prediction about the number of earthquakes that will occur in the future after learning the data of earthquakes in the past.RI algorithm is based on an assumption that earthquakes are considered likely to occur where earthquakes occurred frequently in the past.RI divides model into several same-sized boxes,calculates with the data within the boxes,and make a prediction of every boxes.The sum of predictions of boxes in the target area is the prediction of the target area.RI shows superior performance and accuracy in retrospective testing of North China.
The movement of crust of the earth is an extremely complex and chaotic phenomenon.It is very hard to make a prediction by surveying the movement of crust.In order to make an accurate prediction,statistic as a tool is a good choice.RI(Relative-Intensity) algorithm builds an earthquake forecast model which can make a prediction about the number of earthquakes that will occur in the future after learning the data of earthquakes in the past.RI algorithm is based on an assumption that earthquakes are considered likely to occur where earthquakes occurred frequently in the past.RI divides model into several same-sized boxes,calculates with the data within the boxes,and make a prediction of every boxes.The sum of predictions of boxes in the target area is the prediction of the target area.RI shows superior performance and accuracy in retrospective testing of North China.
引文
[1]Nanjo K Z.Earthquake forecast models for Italy based on the RI algorithm[J].Annals of Geophysics,2010,53(3):117-127.
[2]石玉涛,高原,赵翠萍,等.汶川地震余震序列的地震各向异性[J].地球物理学报,2009,52(2):398-407.
[3]王卫民,赵连锋,李娟,等.四川汶川8.0级地震震源过程[J].地球物理学报,2008,51(5):1404-1410.
[4]孙毅,韩坤亮.地震数据的无损压缩存储[J].计算机技术与发展,2011,21(8):177-180.
[5]张晓东,蒋海昆,黎明晓.地震预测与预警探讨[J].中国地震,2008,25(1):67-76.
[6]Szewiola V S.Method of forecasting seismic energy induced by longwall exploitation based on changes in ground subsidence[J].Mining Science and Technology(China),2011,21(3):375-379.
[7]Wu Anxu,Lin Xiangdong,Jiang Changsheng,et al.Seismic comprehensive forecast based on modified project pursuit regression[J].Earthquake Science,2009,22(5):563-574.
[8]李祚泳,彭荔红.BP网络学习能力与泛化能力满足的不确定关系式[J].中国科学,2003,33(10):887-895.
[9]智会强,牛坤,田亮,等.BP网络和RBF网络在函数逼近领域内的比较研究[J].科技通报,2005,21(2):193-197.
[10]聂红林,袁孝,胡伍生,等.基于BP神经网络技术的区域短期地震预测模型研究[J].现代测绘,2012,35(2):3-6.
[11]夏雅琴,陈维升,李均之,等.地震预测的新进展[J].北京工业大学学报,2006,32(1):11-14.
[2]石玉涛,高原,赵翠萍,等.汶川地震余震序列的地震各向异性[J].地球物理学报,2009,52(2):398-407.
[3]王卫民,赵连锋,李娟,等.四川汶川8.0级地震震源过程[J].地球物理学报,2008,51(5):1404-1410.
[4]孙毅,韩坤亮.地震数据的无损压缩存储[J].计算机技术与发展,2011,21(8):177-180.
[5]张晓东,蒋海昆,黎明晓.地震预测与预警探讨[J].中国地震,2008,25(1):67-76.
[6]Szewiola V S.Method of forecasting seismic energy induced by longwall exploitation based on changes in ground subsidence[J].Mining Science and Technology(China),2011,21(3):375-379.
[7]Wu Anxu,Lin Xiangdong,Jiang Changsheng,et al.Seismic comprehensive forecast based on modified project pursuit regression[J].Earthquake Science,2009,22(5):563-574.
[8]李祚泳,彭荔红.BP网络学习能力与泛化能力满足的不确定关系式[J].中国科学,2003,33(10):887-895.
[9]智会强,牛坤,田亮,等.BP网络和RBF网络在函数逼近领域内的比较研究[J].科技通报,2005,21(2):193-197.
[10]聂红林,袁孝,胡伍生,等.基于BP神经网络技术的区域短期地震预测模型研究[J].现代测绘,2012,35(2):3-6.
[11]夏雅琴,陈维升,李均之,等.地震预测的新进展[J].北京工业大学学报,2006,32(1):11-14.
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心