Pattern Informatics (PI) Algorithm Applied To The Predictive Analysis Of Intermediate-term Seismic Hazard In The Yunnan-Sichuan Region

Pattern Informatics (PI) algorithm, based on the statistical physical concepts of seismic activity, was applied to the analysis of the seismicity in the Yunnan-Sichuan region. The catalogue of earthquakes above ML3.0was used as the input data. The algorithm was applied to a grid of0.2°×0.2°, with the time window for learning, for anomaly identification, and for forecast being10years,5years, and5years, respectively. The forecast was expressed by the’PI hotspot clusters’. Statistical test used the Receiver Operating Characteristic (ROC) test.Retrospective case studies were conducted to the April20,2013, Lushan, Sichuan M7.0earthquake and the July22,2013, Minxian-Zhangxian, Gansu M6.6earthquake, respectively, which shows that the’target earthquakes’were located within the’PI hotspot clusters’. After the2013Lushan earthquake, the’PI hotspot clusters’ remained around the’triple junction’among the Longmenshan, the Xianshuihe, and the Anninghe fault zone, indicating future seismic hazard at the time scale of half decade.Major to great earthquakes in the Yunnan-Sichuan region since the1980s were investigated using the PI algorithm. As an extension of the algorithm, tectonic block boundaries which determined the distribution of major to great earthquakes in continental China were considered, and the PI hotspots were transformed to the’hot segments’of the tectonic block boundaries. It was observed that all these major to great earthquakes located in the’hot segments’of the tectonic block boundaries.Considering the quality of the earthquake catalogues as input data, investigation was conducted to examine the robustness of the algorithm against the uncertainty of magnitude. Through a series of simulated experiments by perturbing the catalogue with different levels of stochastic variations in magnitudes (together with the concept of completeness which determines the deviation from the Gutenberg-Richter frequency-magnitude distribution), it was obtained that the PI algorithm performs stably with the uncertainties of magnitudes, showing one of the advantages of the PI algorithm. Applying the PI algorithm to the seismicity in the Yunnan-Sichuan region (by considering recent earthquakes including the Lushan earthquake and the Minxian-Zhangxian earthquake), this thesis tried to consider the interdisciplinary regional geodynamics (in the perspective of tectonic blocks), and the intersection between the predictive analysis of seismicity and the routine works of seismological observation and interpretation (in the perspective of the stability of the algorithm against the perturbation of magnitude).