Region-time-length algorithm and its application to the study of intermediate-short term earthquake precursor in North China

摘要

The Region-Time-Length Algorithm (RTL algorithm) is introduced and improved in the paper. Compared with the original definition, the influence of rupture length on RTL function is emphasized and the weights of epicentral distance function, time function, and rupture length function are ensured to be equal. The retrospective examinations of RTL algorithm in North China have indicated that the anomalies obtained by the improved RTL algorithm show the short or intermediate-short term precursory features in most cases. There are two types of RTL anomalous patterns before the main shock. For the Ⅰ-type, the variation pattern of the VRTL, numerical values of the VRTL(x,y, z, t) function, is complete and most of them have shown the changing pattern of "rising from 0 → turning →dropping" or "dropping from 0 → turning → rising". For the Ⅱ-type, the variation pattern of VRTL is not complete,which increases or decreases quickly from 0 and there is no evident turning, the main shock generally occurs in the short period around the peak VRTL. The rising of VRTL indicates an increase of seismic activity relative to the background level, which means the enhancement of seismic activity, while the dropping of VRTL indicates the decrease of seismic activity relative to the background level, which represents the seismic quiescence to a certain extent.According to statistical examination results of RTL algorithm in North China, the methods to distinguish the intermediate and short-term anomalies and to estimate the occurrence time of the coming main shock are given in the paper. For both Ⅰ and Ⅱ-type RTL anomalies, the R-value, i.e., the forecasting score, is about 0.6 and 0.3 for the 3months forecasting period and about 0.7 and 0.4 for the 6 months forecasting period. The preliminary discussion is also made for the influences of characteristic time-span to, characteristic distance r0, and threshold magnitude Mo on computation of VRTL, as well as some other significant problems in application.