Recent years,we have witnessed the increasing research interest in developing machine learning,especially deep learning which provides approaches for enhancing the performance of microearthquake detection.While consid...Recent years,we have witnessed the increasing research interest in developing machine learning,especially deep learning which provides approaches for enhancing the performance of microearthquake detection.While considerable research efforts have been made in this direction,most of the state-of-the-art solutions are based on Convolutional Neural Network(CNN)structure,due to its remarkable capability of modeling local and static features.Indeed,the globally dynamic characteristics contained within time series data(i.e.,seismic waves),which cannot be fully captured by CNN-based models,have been largely ignored in previous studies.In this paper,we propose a novel deep learning approach,TransQuake,for seismic P-wave detection.The approach is based on the most advanced sequential model,namely Transformer.To be specific,TransQuake can exploit the STA/LTA algorithm for adapting the three-component structure of seismic waves as input,and take advantage of the multi-head attention mechanism for conducting explainable model learning.Extensive evaluations of the aftershocks following the 2008 Wenchuan MW 7.9 earthquake clearly demonstrates that TransQuake is able to achieve the best detection performance which excels the results obtained using other baselines.Meanwhile,experimental results also validate the interpretability of the results obtained by TransQuake,such as the attention distribution of seismic waves in different positions,and the analysis of the optimal relationship between coda wave and P-wave for noise identification.展开更多
The special seismic tectonic environment and frequent seismicity in the southeastern margin of the Qinghai-Tibet Plateau show that this area is an ideal location to study the present tectonic movement and background o...The special seismic tectonic environment and frequent seismicity in the southeastern margin of the Qinghai-Tibet Plateau show that this area is an ideal location to study the present tectonic movement and background of strong earthquakes in China's Mainland and to predict future strong earthquake risk zones. Studies of the structural environment and physical characteristics of the deep structure in this area are helpful to explore deep dynamic effects and deformation field characteristics, to strengthen our understanding of the roles of anisotropy and tectonic deformation and to study the deep tectonic background of the seismic origin of the block＇s interior. In this paper, the three-dimensional （3D） P-wave velocity structure of the crust and upper mantle under the southeastern margin of the Qinghai-Tibet Plateau is obtained via observational data from 224 permanent seismic stations in the regional digital seismic network of Yunnan and Sichuan Provinces and from 356 mobile China seismic arrays in the southern section of the north-south seismic belt using a joint inversion method of the regional earthquake and teleseismic data. The results indicate that the spatial distribution of the P-wave velocity anomalies in the shallow upper crust is closely related to the surface geological structure, terrain and lithology. Baoxing and Kangding, with their basic volcanic rocks and volcanic clastic rocks, present obvious high-velocity anomalies. The Chengdu Basin shows low-velocity anomalies associated with the Quaternary sediments. The Xichang Mesozoic Basin and the Butuo Basin are characterised by low- velocity anomalies related to very thick sedimentary layers. The upper and middle crust beneath the Chuan-Dian and Songpan-Ganzi Blocks has apparent lateral heterogeneities, including low-velocity zones of different sizes. There is a large range of low-velocity layers in the Songpan-Ganzi Block and the sub-block northwest of Sichuan Province, showing that the middle and lower crust is relatively weak. The Sichuan Basin展开更多
文摘Recent years,we have witnessed the increasing research interest in developing machine learning,especially deep learning which provides approaches for enhancing the performance of microearthquake detection.While considerable research efforts have been made in this direction,most of the state-of-the-art solutions are based on Convolutional Neural Network(CNN)structure,due to its remarkable capability of modeling local and static features.Indeed,the globally dynamic characteristics contained within time series data(i.e.,seismic waves),which cannot be fully captured by CNN-based models,have been largely ignored in previous studies.In this paper,we propose a novel deep learning approach,TransQuake,for seismic P-wave detection.The approach is based on the most advanced sequential model,namely Transformer.To be specific,TransQuake can exploit the STA/LTA algorithm for adapting the three-component structure of seismic waves as input,and take advantage of the multi-head attention mechanism for conducting explainable model learning.Extensive evaluations of the aftershocks following the 2008 Wenchuan MW 7.9 earthquake clearly demonstrates that TransQuake is able to achieve the best detection performance which excels the results obtained using other baselines.Meanwhile,experimental results also validate the interpretability of the results obtained by TransQuake,such as the attention distribution of seismic waves in different positions,and the analysis of the optimal relationship between coda wave and P-wave for noise identification.
基金supported by China earthquake scientific array exploration Southern section of North South seismic belt(201008001)Northern section of North South seismic belt(20130811)+1 种基金National Natural Science Foundation of China(41474057)Science for Earthquake Resllience of China Earthquake Administration(XH15040Y)
文摘The special seismic tectonic environment and frequent seismicity in the southeastern margin of the Qinghai-Tibet Plateau show that this area is an ideal location to study the present tectonic movement and background of strong earthquakes in China's Mainland and to predict future strong earthquake risk zones. Studies of the structural environment and physical characteristics of the deep structure in this area are helpful to explore deep dynamic effects and deformation field characteristics, to strengthen our understanding of the roles of anisotropy and tectonic deformation and to study the deep tectonic background of the seismic origin of the block＇s interior. In this paper, the three-dimensional （3D） P-wave velocity structure of the crust and upper mantle under the southeastern margin of the Qinghai-Tibet Plateau is obtained via observational data from 224 permanent seismic stations in the regional digital seismic network of Yunnan and Sichuan Provinces and from 356 mobile China seismic arrays in the southern section of the north-south seismic belt using a joint inversion method of the regional earthquake and teleseismic data. The results indicate that the spatial distribution of the P-wave velocity anomalies in the shallow upper crust is closely related to the surface geological structure, terrain and lithology. Baoxing and Kangding, with their basic volcanic rocks and volcanic clastic rocks, present obvious high-velocity anomalies. The Chengdu Basin shows low-velocity anomalies associated with the Quaternary sediments. The Xichang Mesozoic Basin and the Butuo Basin are characterised by low- velocity anomalies related to very thick sedimentary layers. The upper and middle crust beneath the Chuan-Dian and Songpan-Ganzi Blocks has apparent lateral heterogeneities, including low-velocity zones of different sizes. There is a large range of low-velocity layers in the Songpan-Ganzi Block and the sub-block northwest of Sichuan Province, showing that the middle and lower crust is relatively weak. The Sichuan Basin
