Application Study On The Multi-method Combination Model For Earthquake Prediction

Earthquake prediction has always been one of the main issues of seismology, the preparation of a large earthquake usually includes several stages: tectonic loading, quasi-static nucleation, dynamic rupture, stress redistribution and fault strength restoration. A Single prediction model cannot completely reflect the evolution of the various stages of a large earthquake. In this paper, we combine four earthquake prediction methods with clear physical meaning for seismic hazard evaluation. The four methods are: the Pattern Informatics(PI), the Load/Unload Response Ratio(LURR), the State Vector(SV), and the Accelerating Moment Release(AMR). The following work has been completed in this paper:1. On the basis of previous studies, we make a retrospective forecasting study on the 2014 and 2015 M>5.0 earthquake tendency in Chinese mainland using the latest software of the combination model prediction. Comparison with the earthquake actually happened, most of the M > 5.0 and all the M>6.0 earthquakes during the two years occurred in the seismic hazard regions predicted.2. We also completed some basic research of the combination prediction model. We replace the circular region usually adopted in LURR practice with an area within which the tectonic stress change would mostly increase the Coulomb failure stress. This may improve the prediction effectiveness of LURR.3. We try to focus on specific areas of China mainland by changing the predicted spatial and temporal scales. In this way, we get the best prediction parameters in these areas. Taking the M>5.5 earthquakes occurred in Sichuan-Yunnan region during 2013- 2014 as an example, nine of ten earthquakes(Including six M>6.0 earthquakes), except Aug31,2013 M5.9 Zhongdian earthquake, occurred in the seismic hazard regions predicted by the combination model. Based on the above, we have improved the software for the prediction considering the different seismic activity in different regions.4. Furtherly, we try to apply the precursor observational data to the LURR analysis. By now, we have completed the LURR analysis with groundwater level data.Through the prediction practice in this paper, we found that compared with using single prediction model, proper combination of the four methods can make a temporal and spatial approach to the potential seismic hazard areas and give clearer assessment of future earthquake tendency.