Morphological matching of seismic observation date with different time length using dynamic time warping method
-
摘要: 本文将动态时间规整方法引入到地震观测资料的形态匹配分析中，以解决因时间尺度不一致的两列观测数据无法定量比对的问题。基于动态时间规整技术方法原理，通过测试数据验证了动态时间规整方法的可行性，并利用云南西部地区的断层实际观测数据，分析了1996年丽江MS7.0地震前的跨断层观测数据异常形态与当前数据异常形态的相似性问题。结果表明：① 动态时间规整方法可用于地震资料时间长度不一致时的相似性匹配；② 时间不一致的两列观测数据可用累积规整路径距离来定量表征，累积距离越短，曲线形态越一致；③ 动态时间规整方法可用于给定模板的前兆数据相似度的计算机自动提取，可提高当前仅依靠人工判别的工作效率；④ 从模式识别的角度考虑，当前下关跨断层水准观测数据变化形态与1996年丽江MS7.0地震和2008年汶川MS8.0地震前的水准数据变化形态较为一致。
-
关键词:
- 动态时间规整（DTW）方法 /
- 数据异常形态 /
- 模式识别 /
- 1996年丽江MS7.0地震
Abstract: In this paper, the dynamic time warping （DTW） method is introduced so as to solve the problem that the two observation data cannot be compared due to their different time length. On the basis of the principle of dynamic time warping technology, this paper verifies the feasi-bility of identifying the conformity of seismic observation data by testing sample data. At the same time, using the actual fault observation data in western Yunnan, the similarity between the anomalous morphology before the 1996 Lijiang MS7.0 earthquake and the current anomalous morphology is studied. The results show that: ① The DTW algorithm can be used for similarity matching of precursor data in seismological field; ② Time-inconsistent precursor observation data can be expressed by cumulative warping distance, the shorter the cumulative distance, the more consistent the curve shape; ③ The DTW algorithm can be used for automatic extraction of precursor data similarity of template-making, which can improve the current work efficiency in the case of only relying on manual tracking precursor curve changes; ④ Considering from the point of view of pattern recognition, the data of Xiaguan cross-fault leveling are consistent with those before Lijiang MS7.0 earthquake in 1996 and Wenchuan MS8.0 earthquake in 2008. -
-
![]()
图 5 动态时间规整方法的模板形态特征自动识别结果
①—⑤分别对应于图4中最小极值点，不同颜色表示识别出的不同的观测时段
Figure 5. Automatic recognition of template morphological features by using DTW algorithm
①−⑤ correspond to the minimum extreme points in Fig. 4，different color represents the different recognized period
![]()
图 6 1970年丽江MS7.0地震和下关跨断层水准观测场地的空间分布
Figure 6. Location of 1970 Lijiang MS7.0 earthquake and Xiaguan cross-fault leveling observation site
![]()
图 7 下关跨断层水准观测时间序列
Figure 7. Leveling time series observed at Xiaguan cross-fault leveling observation site
![]()
图 8 下关跨断层水准动态时间规整模式识别累积路径距离随时间的变化
Figure 8. Variation of cumulative path distance recognized by DTW pattern with time in Xiaguan cross-fault leveling observation site
![]()
图 9 下关跨断层水准观测的时间序列DTW模板形态特征自动识别结果
图中①—④对应于图8中极小值点，不同颜色表示识别出的不同的观测时段
Figure 9. Automatic recognition result of DTW template morphological features of the time series recorded by Xiaguan cross-fault leveling observation site
①−④ correspond to the minimum extreme points in Fig. 8，different color represents different recognized period
![]()
图 10 下关跨断层水准不同时段模板特征识别结果及其规整路径
①—④对应于图8中极小值点
Figure 10. Recognition results and warping path in different time intervals for Xiaguan cross-fanlt leveling observation site
①−④ correspond to the minimum extreme points in Fig. 8
![]()
图 11 下关跨断层水准模板时间序列和动态时间规整后的不同时段时间序列形态对比
①—④分别表示与图8中极小值点相似度较高的4个时段
Figure 11. Morphological comparision between Xiaguan cross-fault leveling template and DTW results in different periods
①-④ represent four high similar recognized periods correspond to the minimum extreme points in Fig. 8
-
陈棋福. 2002. 中国震例（1995—1996）[M]. 北京: 地震出版社: 1−200. Chen Q F. 2002. Earthquake Cases in China （1995—1996）[M], Beijing: Seismological Press: 1−200 （in Chinese）.
蒋光亮,陈应禄,江起达,李忠富,胡伦,杨仁华. 1998. 场兆与源兆的追踪识别[J]. 地壳形变与地震,18(3):85–91. Jiang G L,Chen Y L,Jiang Q D,Li Z F,Hu L,Yang R H. 1998. Tracing recognition of field and source precursor[J]. Crustal Deformation and Earthquake,18(3):85–91 (in Chinese).
焦青,周俊萍. 1998. 跨断层垂直位移速率的动态演化特征与强震的关系[J]. 地震,18(3):265–273. Jiao Q,Zhou J P. 1998. The temporal evolution characteristics of cross-fault vertical displacement rate and their relation to strong earthquakes[J]. Earthquake,18(3):265–273 (in Chinese).
刘琦,张晶. 2011. S变换在汶川地震前后应变变化分析中的应用[J]. 大地测量与地球动力学,31(4):6–9. Liu Q,Zhang J. 2011. Application of S transform in analysis of strain changes before and after Wenchuan earthquake[J]. Journal of Geodesy and Geodynamics,31(4):6–9 (in Chinese).
刘强,余庆坤. 2004. 2003年大姚6.2和6.1级地震的断层形变异常特征[J]. 地震研究,27(4):301–307. doi: 10.3969/j.issn.1000-0666.2004.04.003 Liu Q,Yu Q K. 2004. Anomalous characteristics of fault deformation for Dayao MS6.2 and MS6.1 earthquake[J]. Journal of Seismological Research,27(4):301–307 (in Chinese).
罗兰格,程万正. 1995. 各类地震异常短临前兆标志体系及综合预报方法研究[J]. 地震,15(增刊):118–124. Luo L G,Cheng W Z. 1995. Short-term and impending precursory symbol system of various earthquake anomalies and study on comprehensive prediction method[J]. Earthquake,15(S):118–124 (in Chinese).
马振兴. 1984. 卡尔曼滤波在地震数据处理中的应用[J]. 石油地球物理勘探,19(2):140–147. Ma Z X. 1984. Kalman filtering theory applied in seismic data processing[J]. Oil Geophysical Prospecting,19(2):140–147 (in Chinese).
邱泽华,唐磊,张宝红,宋茉. 2012. 用小波-超限率分析提取宁陕台汶川地震体应变异常[J]. 地球物理学报,55(2):538–546. Qiu Z H,Tang L,Zhang B H,Song M. 2012. Extracting anomaly of the Wenchuan earthquake from the dilatometer recording at NSH by means of wavelet-overrun rate analysis[J]. Chinese Journal of Geophysics,55(2):538–546 (in Chinese).
王海涛,朱令人,杨马陵,蔡仲琼,白超英. 1991. 地震前兆观测数据的信息熵分析:以乌鲁木齐10号井地下水氡为例[J]. 地震,11(5):13–18. Wang H T,Zhu L R,Yang M L,Cai Z Q,Bai C Y. 1991. Analysis of information entropy of seismically precursory observation data: Taking the groundwater radon in Urumuqi No.10 well for an example[J]. Earthquake,11(5):13–18 (in Chinese).
王琼,朱令人. 2002. 时间序列的投影寻踪自回归在新疆地震预报中的应用[J]. 内陆地震,16(2):118–125. doi: 10.3969/j.issn.1001-8956.2002.02.004 Wang Q,Zhu L R. 2002. Application of projection pursuit autoregression of time sequence to earthquake prediction in Xinjiang[J]. Inland Earthquake,16(2):118–125 (in Chinese).
王炜,马钦忠,林命週,吴耿锋,吴绍春. 2005. 主成分分析及地震活动参数的约简[J]. 地震学报,27(5):524–531. doi: 10.3321/j.issn:0253-3782.2005.05.007 Wang W,Ma Q Z,Lin M Z,Wu G F,Wu S C. 2005. Primary component analysis method and reduction of seismicity parame-ters[J]. Acta Seismologica Sinica,27(5):524–531 (in Chinese).
吴大铭,韩大宇. 1983. 用多道维纳滤波方法处理唐山地震前后的大灰厂三种形变资料[J]. 地震学报,5(1):31–38. Wu D M,Han D Y. 1983. Processing three kinds of deformation data of the Dahuichang station before and after Tangshan earthquake by the multi-channel Wiener filtering method[J]. Acta Seismologica Sinica,5(1):31–38 (in Chinese).
郗钦文,杨林章,黎凯武. 1994. 地球对引潮力的响应及综合分析[J]. 中国地震,10(增刊1):83–89. Xi Q W,Yang L Z,Li K W. 1994. Earth’s response to tidal generating force and its synthetic analysis[J]. Earthquake Research in China,10(S1):83–89 (in Chinese).
张晶,黎凯武,武艳强,李春光. 2011. 断层活动协调比在地震预测中的应用[J]. 地震,31(3):19–26. doi: 10.3969/j.issn.1000-3274.2011.03.003 Zhang J,Li K W,Wu Y Q,Li C G. 2011. Application of fault motion coordination ratio in earthquake prediction[J]. Earthquake,31(3):19–26 (in Chinese).
张雁滨,蒋骏. 1997. 潮汐线应变观测的地表平面应变状态[J]. 地壳形变与地震,17(1):70–75. Zhang Y B,Jiang J. 1997. Plane strain of tidal line strain observation on the earth’s surface[J]. Crustal Deformation and Earthquake,17(1):70–75 (in Chinese).
张昭栋,王立忠,高玉斌,张铸钢. 1993. 用褶积滤波处理井水位对固体潮响应的滞后[J]. 地震,13(4):23–29. Zhang Z D,Wang L Z,Gao Y B,Zhang Z G. 1993. Response lag of well water level to solid tide processed by convolution filtering[J]. Earthquake,13(4):23–29 (in Chinese).
周硕愚,高文海,董慧凤,宋永厚. 1984. 前兆观测系统的动态灰箱分析法[J]. 地壳形变与地震,4(4):376–385. Zhou S Y,Gao W H,Dong H F,Song Y H. 1984. The analytical method of dynamic grey box using for the observational system of seismic precursor[J]. Crustal Deformation and Earthquake,4(4):376–385 (in Chinese).
Itakura F. 1975. Minimum prediction residual principle applied to speech recognition[J]. IEEE Trans on Acoust,Speech,Signal Process,23(1):67–72.
下载:
计量
- 文章访问数: 1677
- HTML全文浏览量: 625
- PDF下载量: 62
