地震预测模型优化方法研究

在分析现有基于观测指标“异常-正常”的“二态”地震前兆模式不足的基础上, 提出了异常可表现为多种状态的“多态”前兆模式, 给出“多态”前兆模式下预测效能判定指标, 进一步提出了基于预测效能最优的单项预测模型参数的选择方法。 在广义(时间、 空间、 时-空联合的预测指标)“多态”前兆模式下, 将所建立的模型应用于我国华北和南北带地区, 分别确定了两地区活断层预测地震(M_S≥6)空间分布的最优模型(分别为断层周围20 km、 30 km)及其预测效能(分别为0.42、 0.34)。

Study on efficiency evaluation and optimization of earthquake forecast method

The paper analyzed the deficiency of “binary state” (either abnormal or normal) earthquake precursory model and introduced the “multi-state” precursory model which refers to the fact that multi states of the abnormal of an earthquake forecast index exists. The efficiency evaluation method is put forward in the paper according to the “multi-state” precursory model. Based on the new concept, an optimization method for parameters of earthquake forecast model under the condition of maximization of forecast efficiency is developed. According to the general “multi-state” precursory model in which the forecast index can be in spatial, temporal, or spatial-temporal united domain, the model optimization method is applied to the North China and North-South Seismic Zone in China. The results demonstrate that optimization models to forecast location of potential earthquakes with M_S≥6 in relation to the latest (Q₃~Q₄) active faults are the region around the faults, whose widths are 20 km and 30 km respectively in the North China and South-North Seismic Zone in China. The corresponding forecast efficiency values are 0.42 and 0.34.