水库诱发地震预测方法探讨

Discussion on the Prediction Methods for Reservoir induced Seismicity

摘要:

在系统整理和分析已有的预测方法的基础上，根据预测方法的特点将其归为定性预测、定量预测和基于监测的预测法三大类。详细介绍了震级上限法、经验类比法、高震级b值与低震级b值之比预测法、加卸载响应比法、概率预测法、模糊综合评判法和综合影响参数法，并对其优点、存在的问题和发展方向进行了讨论。认为由于目前对水库诱发地震机理认识的不足，水库诱发地震问题的复杂多样性及震例的有限性，当前的预测方法均未能较好地对水库诱发地震的实际情况进行合理的评估，应加强从水库诱震机理的角度对预测方法进行研究分析，在实际应用时应考虑水库的实际情况，结合多种预测方法，并基于水库监测数据的综合分析进行预测。

Abstract:

Earthquakes can be triggered by water storage and drainage，near reservoirs，and may be referred to as Reservoir Induced Seismicity（RIS）.One area of research is to predict the possibility，location，probable maximum of induced earthquake，and earthquake development trend according to the seismic activities after impoundment.Usually the intensity of reservoir-induced earthquakes is small，but reservoir-induced earthquakes may play a role in triggering shallow-focus earthquakes that have a greater effect on the dam.RIS may also induce large earthquakes，such as the Xinfengjiang reservoir earthquake.Therefore，scientific forecasting and corresponding precautionary measures are the premise and basis of earthquake disaster preparedness and mitigation.The existing prediction methods were classified as three categories in this paper according to their characteristics，including qualitative forecasting，quantitative prediction，and monitoring and forecasting.The qualitative forecasting method is mainly based on the understanding of the mechanism of reservoir-induced earthquakes，and various predictions can result from different analytical processes.The quantitative prediction is used to predict the possibility and probable maximum of induced earthquakes，and combines empirical knowledge of reservoir earthquakes with mathematical theory.The monitoring and forecasting method is mainly based on monitoring data，and the object of study is an existing reservoir.Each of the prediction may use a variety of methods including earthquake up-limitation magnitude，empirical analogy method，high magnitude b-value and low magnitude b-value ratio prediction method，load-unload response ratio theory，probability prediction，fuzzy comprehensive evaluation，and comprehensive impact parameter.In this study，advantages，existing problems，and directions of development were investigated for these methods.The results show that the current forecasting methods are inefficient for evaluating RIS due to a lack of understanding of the mechanism，the complex multiplicity of problems and the data limitations of earthquake cases. Research on mechanisms for reservoir induced earthquakes should be strengthened to improve ability to predict RIS.In practical application，the likelihood and possible magnitude of RIS should be determined by a combination of several methods and considering site specific conditions.Predictive methods should be based on the seismic，specific hydrological and geological environment data of the local and reservoir area.Predictive methods that are based on monitoring data are likely to be the main direction of research on the topic of RIS.