应用统计物理方法分析川滇地区地震危险性

SEISMIC HAZARD ANALYSIS OF THE SICHUAN-YUNNAN REGION,WESTERN CHINA

川滇菱形块体及其边界断裂带(21°～33°N,96°～108°E)是中国大陆地震活动最强烈的地区之一,该地区发生的一系列大地震造成了巨大的人员伤亡和财产损失。文中分别应用基于地震空间相关性和地壳形变场的预测方法分析了该地区未来的地震危险性。由Kagan等(1994)提出的基于地震空间相关性的方法,假定未来发生地震的概率与历史发生地震的频度成正比,根据历史地震目录建立统计学模型估计未来发生地震的概率。回溯性检验表明,这种方法对于评估地震复发周期较短的断裂带的地震危险性有较高的有效性,但对于地震复发周期较长的断裂带,如龙门山断裂带,很难给出一个理想的预期。由Shen等(2007)提出的基于地壳形变场的方法,假定长期地震危险性与地壳构造应变率成正比,根据由GPS观测获得的应变率场建立统计学模型评估未来的地震危险性。回溯性检验表明,川滇地区过去30a间发生的地震与区域应变率的大小没有明显的对应关系,但过去500a间发生的地震与应变率场有很高的相关性,表明由10a时间尺度的大地测量资料得到的地壳应变场可以很好地反映数百a时间尺度的地震危险性。

As the southeastern margin of the Tibetan plateau,the Sichuan-Yunnan region is carved by ample tectonic active faults. Many strong earthquakes have taken place in the past, and some of which resul- ted in massive damages and great losses of human lives. In this study we attempt to estimate the seis- mic hazard potential using two different approaches. Our first approach follows Kagan and Jackson （1994） and uses earthquake catalog data to estimate the mid- to long-term spatial probability of strong earthquakes, based on the assumption that earthquake likelihood is proportional to the intensity and quantity of past events. A retrospective test shows that the model offers reasonable ＇ forecasts＇ statisti- cally over the last 37 years, based on the catalog data of previous 470 years for model constraints. Our second approach is to use secular geodetic strain rate deduced from GPS velocity data to constrain earthquake probability model, assuming that likelihood of strong earthquakes is spatially proportional to the geodetic strain rate. A retrospective test of the model with earthquake occurrence of the past 30 years shows that the model ＇forecasted＇ poorly, but did reasonably well when comparing with the cata- log data of the past 500 years, suggesting that the geodetic strain rate obtained at the decade scale may still be a good indicator of long-term earthquake activity in the region, but only at a time scale of hundreds of years.