稳健估计b值及中强震异常特征研究

利用M估计的稳健回归方法，选取ψ函数为正态密度型函数，计算震级-频度关系中的b值，对基于随机泊松分布的模拟地震样本及天然地震样本，用稳健估计方法与最小二乘方法计算b值后进行比较得出，对于天然地震样本，稳健估计得到的b值更加符合震级-频度关系，作为应用实例，将最小二第六估计法b值与稳健估计b值更加符合震级-频度关系，作为应用实例，将最小二乘估计法b值与稳健估计b值的偏离度作为中强震的前兆判定指标，对新疆地区b值进行时间扫描，在对北天山及南天山东段Ms≥5、南天山西段Ms≥6级地震前异常特征进行讨论后，认为该指标作为中甜地震预报指标具有一定的实用性。

Research on robust estimation of b value and abnormal characteristics before moderate and strong earthquake

In this paper, using robust regression and choosing a function as normal density type function, we calculate b value in the relation between magnitude and frequency. After compared the earthquake samples from simulated Poisson distribution and natural earthquake samples, we think that b value calculated by robust estimation is accordant with reality. As an application example, difference (Db) between b value by least square estimation and by robust estimation is used as a discriminating index for precursor of moderate and strong earthquake. After discussing anomalies before earthquake with MS > 5 occurred in the North Tianshan and the east of South Tianshan zone, earthquake MS > 6 occurred in the west of South Tianshan zone, we also think that Db is certain practicable predicting index for moderate and strong earthquakes.