Situated between the petroliferous Cenozoic Zhujiang(Pearl)River Mouth Basin and the mud volcano-rich Mesozoic Dongsha Basin in the middle sector of the northern South China Sea,the Weitan Banks area has been previous...Situated between the petroliferous Cenozoic Zhujiang(Pearl)River Mouth Basin and the mud volcano-rich Mesozoic Dongsha Basin in the middle sector of the northern South China Sea,the Weitan Banks area has been previously mapped as a basement high that is composed of Mesozoic magmatic rocks.In this study,we present several favorable indicators for petroleum geology that were detected from geophysical profiling and benthic sampling in the area.A conspicuous hill was discovered,named“Zhongwei Hill”,~80 m high above the~340 m deep seafloor and~1 km broad,in a depression with more than 7 km thick sedimentary strata.The Zhongwei Hill was seismically imaged with a mushroom-shaped structure and containing a cake-like crown,fluid flow pipes,and an~10 km broad anticline at depth.Thus,the hill represents a source-plumbing-eruption system.Shallow gas zones linked to deep fracture were found at or near the hill.Stratigraphic correlation indicates that the deep strata comprise the Jurassic and Paleogene strata,the major hosts of hydrocarbon source rocks.In addition to the hill,there are number of mounds from which three bottom water samples were collected and the samples are rich in dissolved methane with concentrations high up to~900 nmol/L,much higher than the background level(0.5–2 nmol/L).The benthic samples are rich in coarse sediment clastics,authigenic carbonate nodules,and deep-water habitats likely feeding on methanotrophic community.Given these observations and the context,we propose that the Zhongwei Hill represents a mud volcano,likely thermally driven,that seeps methane from Jurassic and Paleogene source layers,thus poses a favorable clue for significant hydrocarbon generation capacity in transitional zone of the Zhujiang River Mouth Basin and the Dongsha Basin.展开更多
Automatically correcting students’code errors using deep learning is an effective way to reduce the burden of teachers and to enhance the effects of students’learning.However,code errors vary greatly,and the adaptab...Automatically correcting students’code errors using deep learning is an effective way to reduce the burden of teachers and to enhance the effects of students’learning.However,code errors vary greatly,and the adaptability of fixing techniques may vary for different types of code errors.How to choose the appropriate methods to fix different types of errors is still an unsolved problem.To this end,this paper first classifies code errors by Java novice programmers based on Delphi analysis,and compares the effectiveness of different deep learning models(CuBERT,GraphCodeBERT and GGNN)fixing different types of errors.The results indicated that the 3 models differed significantly in their classification accuracy on different error codes,while the error correction model based on the Bert structure showed better code correction potential for beginners’codes.展开更多
基金Supported by the Special Supporting Program for Cultivating High level Talents in Guangdong Province(No.2019 BT02H594)the National Natural Science Foundation of China(NSFC)(Nos.U1901217,91855101,42306239)+2 种基金the Guangdong Basic and Applied Basic Research Foundation(Nos.2022A1515011836,2021 A 1515110851)The sub bottom profiling,multi-beam sounding data and benthal samples were collected onboard R/Vs Jiageng and Dongfanghong-3 implementing the open research cruises(Nos.NORC 2017-05,NORC 2017-06,NORC 2018-06,NORC 2019-05,NORC 2019-06,NORC 2020-05,NORC 2020-06)NSFC Shiptime Sharing Project(Nos.41649905,41649906,41749906,41949905,41949906,41949905,41949906)。
文摘Situated between the petroliferous Cenozoic Zhujiang(Pearl)River Mouth Basin and the mud volcano-rich Mesozoic Dongsha Basin in the middle sector of the northern South China Sea,the Weitan Banks area has been previously mapped as a basement high that is composed of Mesozoic magmatic rocks.In this study,we present several favorable indicators for petroleum geology that were detected from geophysical profiling and benthic sampling in the area.A conspicuous hill was discovered,named“Zhongwei Hill”,~80 m high above the~340 m deep seafloor and~1 km broad,in a depression with more than 7 km thick sedimentary strata.The Zhongwei Hill was seismically imaged with a mushroom-shaped structure and containing a cake-like crown,fluid flow pipes,and an~10 km broad anticline at depth.Thus,the hill represents a source-plumbing-eruption system.Shallow gas zones linked to deep fracture were found at or near the hill.Stratigraphic correlation indicates that the deep strata comprise the Jurassic and Paleogene strata,the major hosts of hydrocarbon source rocks.In addition to the hill,there are number of mounds from which three bottom water samples were collected and the samples are rich in dissolved methane with concentrations high up to~900 nmol/L,much higher than the background level(0.5–2 nmol/L).The benthic samples are rich in coarse sediment clastics,authigenic carbonate nodules,and deep-water habitats likely feeding on methanotrophic community.Given these observations and the context,we propose that the Zhongwei Hill represents a mud volcano,likely thermally driven,that seeps methane from Jurassic and Paleogene source layers,thus poses a favorable clue for significant hydrocarbon generation capacity in transitional zone of the Zhujiang River Mouth Basin and the Dongsha Basin.
基金supported in part by the Education Department of Sichuan Province(Grant No.[2022]114).
文摘Automatically correcting students’code errors using deep learning is an effective way to reduce the burden of teachers and to enhance the effects of students’learning.However,code errors vary greatly,and the adaptability of fixing techniques may vary for different types of code errors.How to choose the appropriate methods to fix different types of errors is still an unsolved problem.To this end,this paper first classifies code errors by Java novice programmers based on Delphi analysis,and compares the effectiveness of different deep learning models(CuBERT,GraphCodeBERT and GGNN)fixing different types of errors.The results indicated that the 3 models differed significantly in their classification accuracy on different error codes,while the error correction model based on the Bert structure showed better code correction potential for beginners’codes.
