Diurnal variation of tropical convection and kinematic and thermodynamic conditions was investigated for different large-scale environments of the convectively active and inactive periods by using satellite observatio...Diurnal variation of tropical convection and kinematic and thermodynamic conditions was investigated for different large-scale environments of the convectively active and inactive periods by using satellite observations and surface measurements during the Intensive Observation Period （IOP） of the Tropical Ocean Global Atmosphere/Coupled Ocean-Atmosphere Response Experiment （TOGA/COARE）. During the convectively active period, the features of nocturnal convection appear in vertical profiles of convergence, vertical velocity, heat source, and moisture sink. The specific humidity increases remarkably in the middle troposphere at dawn. On the other hand, the altitude of maximum convergence and that of the upward motion is lower during the convectively inactive period. The specific humidity peaks in the lower troposphere in the daytime and decreases in the middle troposphere. Spectral analyses of the time series of the infrared （IR） brightness temperature （TBB） and amounts of rainfall suggest multiscale temporal variation with a prominent diurnal cycle over land and oceanic regions such as the Intensive Flux Array （IFA） and the South Pacific Convergence Zone （SPCZ）. Over land, the daily maximum of deep convection associated with cloud top temperature less than 208 K appears at midnight due to the daytime radiative heating and the sea-land breeze. Over the ocean, convection usually tends to occur at dawn for the convectively active period while in the afternoon during the inactive period. Comparing the diurnal variation of convection with large-scale variables, the authors inferred that moisture in the middle troposphere contributes mostly to the development of nocturnal convection over the ocean during the convectively active period.展开更多
Assessment of past-climate simulations of regional climate models(RCMs)is important for understanding the reliability of RCMs when used to project future regional climate.Here,we assess the performance and discuss pos...Assessment of past-climate simulations of regional climate models(RCMs)is important for understanding the reliability of RCMs when used to project future regional climate.Here,we assess the performance and discuss possible causes of biases in a WRF-based RCM with a grid spacing of 50 km,named WRFG,from the North American Regional Climate Change Assessment Program(NARCCAP)in simulating wet season precipitation over the Central United States for a period when observational data are available.The RCM reproduces key features of the precipitation distribution characteristics during late spring to early summer,although it tends to underestimate the magnitude of precipitation.This dry bias is partially due to the model’s lack of skill in simulating nocturnal precipitation related to the lack of eastward propagating convective systems in the simulation.Inaccuracy in reproducing large-scale circulation and environmental conditions is another contributing factor.The too weak simulated pressure gradient between the Rocky Mountains and the Gulf of Mexico results in weaker southerly winds in between,leading to a reduction of warm moist air transport from the Gulf to the Central Great Plains.The simulated low-level horizontal convergence fields are less favorable for upward motion than in the NARR and hence,for the development of moist convection as well.Therefore,a careful examination of an RCM’s deficiencies and the identification of the source of errors are important when using the RCM to project precipitation changes in future climate scenarios.展开更多
Image-based 3D modeling is an effective method for reconstructing large-scale scenes,especially city-level scenarios.In the image-based modeling pipeline,obtaining a watertight mesh model from a noisy multi-view stere...Image-based 3D modeling is an effective method for reconstructing large-scale scenes,especially city-level scenarios.In the image-based modeling pipeline,obtaining a watertight mesh model from a noisy multi-view stereo point cloud is a key step toward ensuring model quality.However,some state-of-the-art methods rely on the global Delaunay-based optimization formed by all the points and cameras;thus,they encounter scaling problems when dealing with large scenes.To circumvent these limitations,this study proposes a scalable pointcloud meshing approach to aid the reconstruction of city-scale scenes with minimal time consumption and memory usage.Firstly,the entire scene is divided along the x and y axes into several overlapping chunks so that each chunk can satisfy the memory limit.Then,the Delaunay-based optimization is performed to extract meshes for each chunk in parallel.Finally,the local meshes are merged together by resolving local inconsistencies in the overlapping areas between the chunks.We test the proposed method on three city-scale scenes with hundreds of millions of points and thousands of images,and demonstrate its scalability,accuracy,and completeness,compared with the state-of-the-art methods.展开更多
Rainfall amount in mid-summer(July and August)is much greater over eastern than western Sichuan,which are characterized by basin and plateau,respectively.It is shown that the interannual variations of extreme rainfall...Rainfall amount in mid-summer(July and August)is much greater over eastern than western Sichuan,which are characterized by basin and plateau,respectively.It is shown that the interannual variations of extreme rainfall over these two regions are roughly independent,and they correspond to distinct anomalies of both large-scale circulation and sea surface temperature(SST).The enhanced extreme rainfall over western Sichuan is associated with a southward shift of the Asian westerly jet,while the enhanced extreme rainfall over eastern Sichuan is associated with an anticyclonic anomaly in the upper troposphere over China.At low levels,on the other hand,the enhanced extreme rainfall over western Sichuan is related to two components of wind anomalies,namely southwesterly over southwestern Sichuan and northeasterly over northeastern Sichuan,which favor more rainfall under the effects of the topography.Relatively speaking,the enhanced extreme rainfall over eastern Sichuan corresponds to the low-level southerly anomalies to the east of Sichuan,which curve into northeasterly anomalies over the basin when they encounter the mountains to the north of the basin.Therefore,it can be concluded that the topography in and around Sichuan plays a crucial role in inducing extreme rainfall both over western and eastern Sichuan.Finally,the enhanced extreme rainfall in western and eastern Sichuan is related to warmer SSTs in the Maritime Continent and cooler SSTs in the equatorial central Pacific,respectively.展开更多
Transposable elements(TEs)are key drivers of genomic variation and species evolution.Although advances in high-throughput sequencing have enabled population-scale identification of TE insertions,accurate detection acr...Transposable elements(TEs)are key drivers of genomic variation and species evolution.Although advances in high-throughput sequencing have enabled population-scale identification of TE insertions,accurate detection across large and complex genomes remains challenging.Existing tools often struggle to efficiently process large genomes,recover low-copy elements,or accurately reconstruct full-length TEs,limiting comprehensive TE analyses.Here,we present panHiTE,a population-scale TE detection framework that introduces several methodological innovations.First,panHiTE employs a dynamically updated global TE library to avoid redundant detection of previously identified elements,improving computational efficiency and enabling application to extremely large genomes,such as the 15-Gb wheat genome.Second,to recover low-copy TEs that are frequently missed in individual genomes,panHiTE realigns candidate elements across population-scale genomes,enabling accurate reconstruction of full-length TEs across accessions.Third,because long terminal repeat retrotransposons constitute a major fraction of plant genomes,panHiTE integrates a deep-learning-based detection algorithm developed in this study,achieving higher sensitivity and precision than the state-of-the-art tool panEDTA in population-scale analyses.In addition,a fault-tolerant redundancy-removal algorithm efficiently groups divergent family members,generating TE libraries with more than 50%fewer sequences while doubling the number of Perfect TEs across 26 maize genomes.These advances enable panHiTE to deliver high-resolution TE annotations and accurately resolve TE–gene positional relationships,thereby facilitating the systematic identification of TE-induced differential expression loci(TIDELs).In 32 Arabidopsis accessions,panHiTE identifies 85 TIDELs associated with diverse biological functions and metabolic pathways.Overall,panHiTE provides a robust and scalable solution for population-scale TE discovery and functional characterization in complex plant genomes.展开更多
Topic modeling is a mainstream and effective technology to deal with text data, with wide applications in text analysis, natural language, personalized recommendation, computer vision, etc. Among all the known topic m...Topic modeling is a mainstream and effective technology to deal with text data, with wide applications in text analysis, natural language, personalized recommendation, computer vision, etc. Among all the known topic models, supervised Latent Dirichlet Allocation （sLDA） is acknowledged as a popular and competitive supervised topic model. How- ever, the gradual increase of the scale of datasets makes sLDA more and more inefficient and time-consuming, and limits its applications in a very narrow range. To solve it, a parallel online sLDA, named PO-sLDA （Parallel and Online sLDA）, is proposed in this study. It uses the stochastic variational inference as the learning method to make the training procedure more rapid and efficient, and a parallel computing mechanism implemented via the MapReduce framework is proposed to promote the capacity of cloud computing and big data processing. The online training capacity supported by PO-sLDA expands the application scope of this approach, making it instrumental for real-life applications with high real-time demand. The validation using two datasets with different sizes shows that the proposed approach has the comparative accuracy as the sLDA and can efficiently accelerate the training procedure. Moreover, its good convergence and online training capacity make it lucrative for the large-scale text data analyzing and processing.展开更多
The thermal forcings of annual and interannual periodic variations are introduced into the barotropic vorticity equation,by using low order spectral model of the equation,more than 40 numerical experiments whose integ...The thermal forcings of annual and interannual periodic variations are introduced into the barotropic vorticity equation,by using low order spectral model of the equation,more than 40 numerical experiments whose integration time is larger than 100 model years are performed in order to investigate variations of large-scale flow patterns arising from both external interannual thermal forcing and internal dynamical processes.In certain parametric range,when the fre- quency of the forcing term with interannual period equals to the frequency which is created by the internal dynamical processes alone,the amplitude of interannual variations of flow patterns increases obviously,and the period becomes double.In other parametric range,the amplitude of interannual variations of flow patterns shows abrupt changes and other nonlinear behavior,along with gradual changes of interannual forcing parameters.展开更多
针对现有特征匹配方法在大尺度变化场景下匹配数量少、误匹配率高的问题,基于局部特征匹配框架ELoFTR(efficient detector-free local feature matching with transformers)提出一种新颖的特征匹配算法SA-LoFTR(scale adaptive-LoFTR)...针对现有特征匹配方法在大尺度变化场景下匹配数量少、误匹配率高的问题,基于局部特征匹配框架ELoFTR(efficient detector-free local feature matching with transformers)提出一种新颖的特征匹配算法SA-LoFTR(scale adaptive-LoFTR)。该算法首先利用一种多尺度下采样融合策略以增强注意力机制的尺度敏感性,借助多尺度机制和小波下采样同时提取局部和全局信息,提升模型对尺度变化的适应能力;其次,引入尺度自适应区域对齐模块,通过初始匹配点分布估计尺度比值,自适应裁剪并放大较小的共视区域,增加匹配点对的数量;最后,设计了一种双阈值过滤机制,通过置信度阈值和局部支持性验证联合判定有效匹配点,剔除低置信度且缺乏局部一致性的误匹配。在公共数据集MegaDepth和HPatches以及自制数据集ScaleMega上进行的大量实验验证了SALoFTR的优越性,其匹配精度和匹配数量相较于E-LoFTR均显著提升。展开更多
Large-scale virtual scene exploration is still a challenging task. The novice users caneasily get distracted and disorientated, which results in being lost in space. Assistedcamera control technology is the most effec...Large-scale virtual scene exploration is still a challenging task. The novice users caneasily get distracted and disorientated, which results in being lost in space. Assistedcamera control technology is the most effective solution for virtual environment exploration problems which requires viewpoint computation and path planning. In this paper,a novel approach for large-scale virtual scene based on viewpoint scoring is proposed.First, the scene was adaptively divided into several meaningful and easily analyzedsubregions according to the optimal view distance criterion. Second, a novel viewpointscoring method based on visual perception and information entropy fusion was developed for optimal viewpoint determination and greedy N-Best viewpoint selection algorithm was utilized for visual perceptibility calculation. Then evolutionary programmingapproach for the Traveling Salesman problem was applied for intra-subregion and intersubregion exploring path optimization. Finally, the Cubic Hermite Curve was introduced to smoothen the inflection point on the exploration path. The experimental resultsdemonstrate that the proposed method can effectively generate an automatic smooth,informative, aesthetic and non-intersecting path, with the characteristics of good exploring comfort, strong immersion and high scene information perception.展开更多
文摘Diurnal variation of tropical convection and kinematic and thermodynamic conditions was investigated for different large-scale environments of the convectively active and inactive periods by using satellite observations and surface measurements during the Intensive Observation Period （IOP） of the Tropical Ocean Global Atmosphere/Coupled Ocean-Atmosphere Response Experiment （TOGA/COARE）. During the convectively active period, the features of nocturnal convection appear in vertical profiles of convergence, vertical velocity, heat source, and moisture sink. The specific humidity increases remarkably in the middle troposphere at dawn. On the other hand, the altitude of maximum convergence and that of the upward motion is lower during the convectively inactive period. The specific humidity peaks in the lower troposphere in the daytime and decreases in the middle troposphere. Spectral analyses of the time series of the infrared （IR） brightness temperature （TBB） and amounts of rainfall suggest multiscale temporal variation with a prominent diurnal cycle over land and oceanic regions such as the Intensive Flux Array （IFA） and the South Pacific Convergence Zone （SPCZ）. Over land, the daily maximum of deep convection associated with cloud top temperature less than 208 K appears at midnight due to the daytime radiative heating and the sea-land breeze. Over the ocean, convection usually tends to occur at dawn for the convectively active period while in the afternoon during the inactive period. Comparing the diurnal variation of convection with large-scale variables, the authors inferred that moisture in the middle troposphere contributes mostly to the development of nocturnal convection over the ocean during the convectively active period.
文摘Assessment of past-climate simulations of regional climate models(RCMs)is important for understanding the reliability of RCMs when used to project future regional climate.Here,we assess the performance and discuss possible causes of biases in a WRF-based RCM with a grid spacing of 50 km,named WRFG,from the North American Regional Climate Change Assessment Program(NARCCAP)in simulating wet season precipitation over the Central United States for a period when observational data are available.The RCM reproduces key features of the precipitation distribution characteristics during late spring to early summer,although it tends to underestimate the magnitude of precipitation.This dry bias is partially due to the model’s lack of skill in simulating nocturnal precipitation related to the lack of eastward propagating convective systems in the simulation.Inaccuracy in reproducing large-scale circulation and environmental conditions is another contributing factor.The too weak simulated pressure gradient between the Rocky Mountains and the Gulf of Mexico results in weaker southerly winds in between,leading to a reduction of warm moist air transport from the Gulf to the Central Great Plains.The simulated low-level horizontal convergence fields are less favorable for upward motion than in the NARR and hence,for the development of moist convection as well.Therefore,a careful examination of an RCM’s deficiencies and the identification of the source of errors are important when using the RCM to project precipitation changes in future climate scenarios.
基金This work was supported by the Natural Science Foundation of China(Nos.61632003,61873265)。
文摘Image-based 3D modeling is an effective method for reconstructing large-scale scenes,especially city-level scenarios.In the image-based modeling pipeline,obtaining a watertight mesh model from a noisy multi-view stereo point cloud is a key step toward ensuring model quality.However,some state-of-the-art methods rely on the global Delaunay-based optimization formed by all the points and cameras;thus,they encounter scaling problems when dealing with large scenes.To circumvent these limitations,this study proposes a scalable pointcloud meshing approach to aid the reconstruction of city-scale scenes with minimal time consumption and memory usage.Firstly,the entire scene is divided along the x and y axes into several overlapping chunks so that each chunk can satisfy the memory limit.Then,the Delaunay-based optimization is performed to extract meshes for each chunk in parallel.Finally,the local meshes are merged together by resolving local inconsistencies in the overlapping areas between the chunks.We test the proposed method on three city-scale scenes with hundreds of millions of points and thousands of images,and demonstrate its scalability,accuracy,and completeness,compared with the state-of-the-art methods.
基金supported by the Strategic Priority Research Program of Chinese Academy of Sciences(Grant No.XDA23090102)the Second Tibetan Plateau Scientific Expedition and Research(STEP)program(Grant No.2019QZKK0102).
文摘Rainfall amount in mid-summer(July and August)is much greater over eastern than western Sichuan,which are characterized by basin and plateau,respectively.It is shown that the interannual variations of extreme rainfall over these two regions are roughly independent,and they correspond to distinct anomalies of both large-scale circulation and sea surface temperature(SST).The enhanced extreme rainfall over western Sichuan is associated with a southward shift of the Asian westerly jet,while the enhanced extreme rainfall over eastern Sichuan is associated with an anticyclonic anomaly in the upper troposphere over China.At low levels,on the other hand,the enhanced extreme rainfall over western Sichuan is related to two components of wind anomalies,namely southwesterly over southwestern Sichuan and northeasterly over northeastern Sichuan,which favor more rainfall under the effects of the topography.Relatively speaking,the enhanced extreme rainfall over eastern Sichuan corresponds to the low-level southerly anomalies to the east of Sichuan,which curve into northeasterly anomalies over the basin when they encounter the mountains to the north of the basin.Therefore,it can be concluded that the topography in and around Sichuan plays a crucial role in inducing extreme rainfall both over western and eastern Sichuan.Finally,the enhanced extreme rainfall in western and eastern Sichuan is related to warmer SSTs in the Maritime Continent and cooler SSTs in the equatorial central Pacific,respectively.
基金supported in part by the National Natural Science Foundation of China(grant nos.62350004 and 62332020)the Project of Xiangjiang Laboratory(grant no.23XJ01011).
文摘Transposable elements(TEs)are key drivers of genomic variation and species evolution.Although advances in high-throughput sequencing have enabled population-scale identification of TE insertions,accurate detection across large and complex genomes remains challenging.Existing tools often struggle to efficiently process large genomes,recover low-copy elements,or accurately reconstruct full-length TEs,limiting comprehensive TE analyses.Here,we present panHiTE,a population-scale TE detection framework that introduces several methodological innovations.First,panHiTE employs a dynamically updated global TE library to avoid redundant detection of previously identified elements,improving computational efficiency and enabling application to extremely large genomes,such as the 15-Gb wheat genome.Second,to recover low-copy TEs that are frequently missed in individual genomes,panHiTE realigns candidate elements across population-scale genomes,enabling accurate reconstruction of full-length TEs across accessions.Third,because long terminal repeat retrotransposons constitute a major fraction of plant genomes,panHiTE integrates a deep-learning-based detection algorithm developed in this study,achieving higher sensitivity and precision than the state-of-the-art tool panEDTA in population-scale analyses.In addition,a fault-tolerant redundancy-removal algorithm efficiently groups divergent family members,generating TE libraries with more than 50%fewer sequences while doubling the number of Perfect TEs across 26 maize genomes.These advances enable panHiTE to deliver high-resolution TE annotations and accurately resolve TE–gene positional relationships,thereby facilitating the systematic identification of TE-induced differential expression loci(TIDELs).In 32 Arabidopsis accessions,panHiTE identifies 85 TIDELs associated with diverse biological functions and metabolic pathways.Overall,panHiTE provides a robust and scalable solution for population-scale TE discovery and functional characterization in complex plant genomes.
基金This work was supported in part by the National Natural Science Foundation of China under Grant Nos. 61572226 and 61876069, and the Key Scientific and Technological Research and Development Project of Jilin Province of China under Grant Nos. 20180201067GX and 20180201044GX.
文摘Topic modeling is a mainstream and effective technology to deal with text data, with wide applications in text analysis, natural language, personalized recommendation, computer vision, etc. Among all the known topic models, supervised Latent Dirichlet Allocation （sLDA） is acknowledged as a popular and competitive supervised topic model. How- ever, the gradual increase of the scale of datasets makes sLDA more and more inefficient and time-consuming, and limits its applications in a very narrow range. To solve it, a parallel online sLDA, named PO-sLDA （Parallel and Online sLDA）, is proposed in this study. It uses the stochastic variational inference as the learning method to make the training procedure more rapid and efficient, and a parallel computing mechanism implemented via the MapReduce framework is proposed to promote the capacity of cloud computing and big data processing. The online training capacity supported by PO-sLDA expands the application scope of this approach, making it instrumental for real-life applications with high real-time demand. The validation using two datasets with different sizes shows that the proposed approach has the comparative accuracy as the sLDA and can efficiently accelerate the training procedure. Moreover, its good convergence and online training capacity make it lucrative for the large-scale text data analyzing and processing.
基金This work was supported by the National Natural Science Foundation of China
文摘The thermal forcings of annual and interannual periodic variations are introduced into the barotropic vorticity equation,by using low order spectral model of the equation,more than 40 numerical experiments whose integration time is larger than 100 model years are performed in order to investigate variations of large-scale flow patterns arising from both external interannual thermal forcing and internal dynamical processes.In certain parametric range,when the fre- quency of the forcing term with interannual period equals to the frequency which is created by the internal dynamical processes alone,the amplitude of interannual variations of flow patterns increases obviously,and the period becomes double.In other parametric range,the amplitude of interannual variations of flow patterns shows abrupt changes and other nonlinear behavior,along with gradual changes of interannual forcing parameters.
文摘针对现有特征匹配方法在大尺度变化场景下匹配数量少、误匹配率高的问题,基于局部特征匹配框架ELoFTR(efficient detector-free local feature matching with transformers)提出一种新颖的特征匹配算法SA-LoFTR(scale adaptive-LoFTR)。该算法首先利用一种多尺度下采样融合策略以增强注意力机制的尺度敏感性,借助多尺度机制和小波下采样同时提取局部和全局信息,提升模型对尺度变化的适应能力;其次,引入尺度自适应区域对齐模块,通过初始匹配点分布估计尺度比值,自适应裁剪并放大较小的共视区域,增加匹配点对的数量;最后,设计了一种双阈值过滤机制,通过置信度阈值和局部支持性验证联合判定有效匹配点,剔除低置信度且缺乏局部一致性的误匹配。在公共数据集MegaDepth和HPatches以及自制数据集ScaleMega上进行的大量实验验证了SALoFTR的优越性,其匹配精度和匹配数量相较于E-LoFTR均显著提升。
文摘Large-scale virtual scene exploration is still a challenging task. The novice users caneasily get distracted and disorientated, which results in being lost in space. Assistedcamera control technology is the most effective solution for virtual environment exploration problems which requires viewpoint computation and path planning. In this paper,a novel approach for large-scale virtual scene based on viewpoint scoring is proposed.First, the scene was adaptively divided into several meaningful and easily analyzedsubregions according to the optimal view distance criterion. Second, a novel viewpointscoring method based on visual perception and information entropy fusion was developed for optimal viewpoint determination and greedy N-Best viewpoint selection algorithm was utilized for visual perceptibility calculation. Then evolutionary programmingapproach for the Traveling Salesman problem was applied for intra-subregion and intersubregion exploring path optimization. Finally, the Cubic Hermite Curve was introduced to smoothen the inflection point on the exploration path. The experimental resultsdemonstrate that the proposed method can effectively generate an automatic smooth,informative, aesthetic and non-intersecting path, with the characteristics of good exploring comfort, strong immersion and high scene information perception.
