【摘要】：衛(wèi)星紅外遙感數(shù)據(jù)的震前異常是進行地震預測預警探索的一種有效輔助信息,其相關研究成為近年來的研究熱點之一。本文針對已有的震前紅外異常研究方法中存在缺乏大量震例驗證和多種紅外遙感數(shù)據(jù)綜合應用的缺陷,循序漸進地提出三種異常分析算法,并采用實震進行驗證。主要研究工作包括：1、針對CUSUM算法能檢測統(tǒng)計過程中均值的微小變化,本文提出基于CUSUM算法的射出長波輻射(OLR)數(shù)據(jù)(1°×1°)地震異常分析算法。首先,以樣本觀察值數(shù)據(jù)序列均值為期望的均值近似計算累積和；其次,采用局部平滑方法對累積和進行平滑；第三,利用局部最大值和最小值構(gòu)造特征曲線；最后采用雙閾值控制法進行異常檢測。汶川地震和阿根廷地震的實驗結(jié)果表明,此算法能有效地檢測出地震前后的異常,具體表現(xiàn)為：震前,異常度由小及大,在地震前后異常度達到最大,地震結(jié)束后,異常度逐漸減小。2、提出基于小波和短時傅里葉變換的平均功率譜地震異常分析算法。算法假設2.5°x2.5°的OLR數(shù)據(jù)構(gòu)成為加性模型,采用“db8”小波挖掘隱藏在OLR數(shù)據(jù)中與地震密切相關的成分,并利用短時傅里葉變換分析其相關的頻譜特性,提出特定頻率范圍的頻譜平均值作為地震異常分析的依據(jù)。對國內(nèi)外11個≥7.0的強震進行實驗分析,實驗結(jié)果表明：地震前后異常特征表現(xiàn)為上升-維持-下降的形態(tài)。3、針對震前紅外異常研究中缺乏多種紅外遙感數(shù)據(jù)綜合分析的問題,提出了基于隨機漫步的異常分析算法。將分辨率為2.5°x2.5°的OLR數(shù)據(jù)和NCEP/NCAR再分析資料的表面溫度數(shù)據(jù)、潛在溫度數(shù)據(jù)和壓力數(shù)據(jù)結(jié)合起來,綜合對8個2008-2013年國內(nèi)外≥8.0的強震進行實驗分析。統(tǒng)計隨機漫步結(jié)果中大于2倍方差的強度、位置與天數(shù),并進行四種數(shù)據(jù)對比。實驗結(jié)果表明,8個地震中每個地震至少都有兩種數(shù)據(jù)存在同步的大于2倍方差的異常,且異�；径汲霈F(xiàn)在震前2周內(nèi)。
[Abstract]:The pre-earthquake anomaly of satellite infrared remote sensing data is an effective auxiliary information for earthquake prediction and early warning exploration, and its related research has become one of the research hotspots in recent years. In view of the defects of the existing methods of infrared anomaly research before earthquakes, which lack a large number of seismic examples and comprehensive application of various infrared remote sensing data, three anomaly analysis algorithms are proposed step by step and verified by real earthquakes. The main research work is as follows: 1. Aiming at the small change of mean value in the statistical process detected by CUSUM algorithm, this paper proposes an algorithm for seismic anomaly analysis based on CUSUM algorithm for emitting long wave radiation (OLR) data (1 擄脳 1 擄). Firstly, the cumulative sum is approximately calculated by the mean value of the data sequence of the sample observation value; secondly, the cumulative sum is smoothed by the local smoothing method; thirdly, the characteristic curve is constructed by using the local maximum value and the minimum value. Finally, the double threshold control method is used to detect anomalies. The experimental results of the Wenchuan earthquake and the Argentine earthquake show that the algorithm can effectively detect the anomalies before and after the earthquake. The results show that the anomaly degree changes from small to large before and after the earthquake, and the anomaly degree reaches the maximum before and after the earthquake. The anomaly degree decreases gradually. 2. An average power spectrum seismic anomaly analysis algorithm based on wavelet transform and short time Fourier transform (STFT) is proposed. The algorithm assumes that 2.5 擄x 2.5 擄OLR data is an additive model, uses "db8" wavelet to mine the components closely related to earthquakes in OLR data, and analyzes its related spectral characteristics by using short time Fourier transform (STFT). The average frequency spectrum in a specific frequency range is proposed as the basis for seismic anomaly analysis. The experimental results of 11 strong earthquakes 鈮，

本文編號：2377974