Statistical Seismological Problems Associated With CSEP-CN Testing Region In The Context Of CSES

As an encouraging work facilitating the China Seismic Experiment Site(CSES)and the step 2.0 of Collaboratory for the Study of Earthquake Predictability(CSEP),several statistical seismicity problems need to be solved and explored.Our work focuses on the CSES region 21~32~o N,97.5~105.5~oE and the earthquake catalog recorded since 1970/01/01 to 2019/04/01.This thesis introduces the statistical seismological features of seismicity and the distribution of regional fault system.Then several M_C analysis methods are given to evaluate the time-dependent completeness of earthquake catalogue,containing catalog-based and network-based methods.The different analysis of the catalog-based method shows that since 1970 the overall completeness magnitude located between M_L2.5 and M_L 3.0.The PMC method suggests that the network has a high detect ability for the target events of magnitude M_L2.5 in the mostly whole region.After the detail analysis on the quality of the catalog and existed problem such as data missing follow a strong events,we applied different earthquake forecast models to give experiments for the study region,such as the Pattern Informatics(PI)algorithm,Reasenberg-Jones model,Epidemic-type Aftershock Sequence(ETAS)model and Earthquake Nowcasting method.The result obtained from the PI algorithm suggests that the PI model parameters set has an optimal distribution on the dimension of grid size,forecast time length and the start point in the retrospective forecast experiment.Then we choose the best parameters set to give the‘hotspot'distribution and find that the alarmed region which has high earthquake potentials likely located on the Longmenshan region and the South part of the CSES region.At the same time,the time length of the background window/anomaly identification window/forecast window should be changed when PI algorithm is used to explore the forecast experiment in the region which has a low seismicity level.The results of the Reasenberg-Jones model reveal that this model fits good to the 2008 M_S8.0 Wenchuan and 2013 M_S7.0 Lushan earthquake sequence.Then the forecasting on the occurrence rate and the probability of the aftershocks are obtained based on the forecast routine of Reasenberg-Jones model.The statistical test using the N-test method suggested that the performance of the model has its instability in the temporal distribution.We use the ETAS model to fit the earthquake sequence of the CSES since 1970 to 2019 and output the occurrence rate of the background events and the clustered events,which turns out that the conditional intensity at the end of the study period mostly located on the central and south part of study region.The statistical-based delustering method separates the background events and the clustered events using the probability threshold as a background events,which suggests that the aftershocks triggered by 2008 Wenchuan earthquake promote the increasing of both background activity and clustering events.We use the Nowcasting method which is widely used in economics and meteorology to evaluate the state of the study system and find that the Earthquake Potential Score(EPS)value which can indicate the current level of hazard reached 0.68 for the target earthquake of M_?5.0.Other evaluations for different target events is also listed and be expected to applied into the Annual Consultation in China for the decision of the alarmed and noticed region.All of this work is expected to participate into the scientific research on the statistical problem analysis and give a reference to the application of different probability forecast models in CSES region.