Blockly earthquake transformer:A deep learning platform for custom phase picking

作     者：Hao Mai Pascal Audet H.K.Claire Perry S.Mostafa Mousavi Quan Zhang 

作者机构：Department of Earth and Environmental SciencesUniversity of OttawaOttawaCanadaK1N 6N5 Canadian Hazards Information ServiceNatural Resources CanadaOttawaCanadaK1A 0E7 Department of GeophysicsStanford UniversityStanford94305USA 

出 版 物：《Artificial Intelligence in Geosciences》 (地学人工智能（英文）)

年 卷 期：2023年第4卷第1期

页      面：84-94页

核心收录：

学科分类：070801[理学-固体地球物理学] 07[理学] 0708[理学-地球物理学] 

基　　金：funded by a Discovery Grant(RGPIN-2018-03752)from the Natural Science and Engineering Research Council of Canada(PA) This is NRCan publication number 20220610 

主　　题：Earthquake detection Seismic phase identification Deep learning Seismology 

摘      要：Deep-learning(DL)algorithms are increasingly used for routine seismic data processing tasks,including seismic event detection and phase arrival *** many examples of the remarkable performance of existing(i.e.,pre-trained)deep-learning detector/picker models,there are still some cases where the direct applications of such models do not generalize *** such cases,substantial effort is required to improve the performance by either developing a new model or fine-tuning an existing *** address this challenge,we present Blockly Earthquake Transformer(BET),a deep-learning platform for efficient customization of deep-learning phase *** implements Earthquake Transformer as its baseline model,and offers transfer learning and fine-tuning *** provides an interactive dashboard to customize a model based on a particular *** the parameters are specified,BET executes the corresponding phase-picking task without direct user interaction with the base *** the transfer-learning module,BET extends the application of a deep-learning P and S phase picker to more specific phases(e.g.,Pn,Pg,Sn and Sg phases).In the fine-tuning module,the model performance is enhanced by customizing the model *** no-code platform is designed to quickly deploy reusable workflows,build customized models,visualize training processes,and produce publishable figures in a lightweight,interactive,and open-source Python toolbox.