Seismic Hazard Assessment Of The Active Faults In Southern Sichuan-yunnan Region Using Gps Observations

GPS (Global Positioning System), with its outstanding capabilities such as high accuracy, low expense, dense distribution and flexible survey revolutionarily improves the observations of crustal deformation and tectonic movement at all scales. In the past decade, with the implementation of the national major scientific project Crustal Movement Observation Network of China (CMONOC), massive GPS data are accumulated. Sichuan-Yunnan region is one of the areas with the most active seismicity in China Mainland. There are also dense GPS stations and multi-phase GPS observation data which provide sufficient data for this thesis.Through a systemic summary of the methods for seismic hazard analysis both at home and abroad, this thesis explores the method for seismic hazard quantitative evaluation of active faults based on GPS with southern Sichuan-Yunnan region as an example.First, the seismic moment rate and crustal deformation parameters are calculated according to the CMONOC GPS observation data. Average earthquake recurrence intervals of the major active faults in the region are calculated using GPS velocity field data, which is different with the geology as an independent method. The results are not only the addition to the geological results, but also to provide an independent check.Second, we establish the geodetic model for seismic hazard quantitative evaluation of active faults of Sichuan-Yunnan region, with the combination of GPS observation data and real-time probability model. The conditional probability of the major active faults in southern Sichuan-Yunnan region is calculated. Taking into account the complexity of earthquake recurrence behavior, the characteristic earthquake model and Gutenberg-Richter model are both used in the large magnitude earthquakes. Compared with the single characteristic earthquake model and the G-R exponential model, it better reflects the recurrence law of large earthquake.Finally, we express the effect of time-dependence on PSHA. The result of BPT model and Lognormal model is larger than the result of Poisson model when the ratio between the elapse time and average recurrence intervals is low. In addition, the probability tends to increase as the prediction time increase, which is in line with the theoretical foundation that the higher the probability, the strain accumulation time is longer in real-time probability model.