Review of recent advances in seismic location methods
Precise seismic location is essential for many scientific and technical problems such as the recognition of the Earth's internal structure and seismogenic processes,refined fault structure detection,resource and energy exploration and development,and earthquake early warning.Given the rapid development of seismic location methods,a timely review of the latest advances is important and necessary.The existing review papers mainly cover the progress of conventional seismic location methods,but there are few systematic summaries of seismic location methods involving machine learning.To facilitate readers to understand the principles of seismic location methods and the latest cutting-edge advances,this paper first introduces the newly developed seismic location methods in recent years,such as the source scanning class method,the double difference class location method,and GrowClust,etc.We then focus on the latest seismic location methods involving machine learning,including the fully machine learning-based seismic location methods and the machine learning-assisted seismic location processes.According to the adopted neural networks,including convolutional neural networks,graph neural networks,and recurrent neural networks,the machine learning-based location methods can be further categorized into different groups.For machine learning-assisted location process,we introduce three popular workflows,EasyQuake,QuakeFlow,and LOC-FLOW.By elaborating the practical applications of LSTM-FCN model and LOC-FLOW method,the location results of representative methods are compared.Finally,this paper analyzes and outlooks the problems and prospects of seismic location methods involving machine learning,pointing out that the lightweighting of machine learning models is an important research direction and the joint of multiple seismic location methods is an important goal for the development of seismic location.
NETL
NSTL
万方数据
