On 18th December 2023,an Ms 6.2 earthquake centered in Jishishan County,Gansu Province triggered a prototypical earthquake-induced liquefaction flowslide in Jintian Village and Caotan Village,Zhongchuan Township,Minhe County,Haidong City,Qinghai Province.This hazard caused numerous buildings to be engulfed and interred by several meters of deposits.Owing to the abruptness of its occurrence and extremely high fluidity,it was mistakenly identified as a"sand surge".Through on-site investigation and remote sensing imagery analysis,we confirm this flowslide to be an earthquake-triggered liquefied flowslide event,and explore its causal mechanisms.The results show that:(1)This flowslide was induced by seismic vibratory loading causing liquefaction of the saturated silt layer(loess layer)underlying the plateau,forming a landslide that transformed into a mudflow moving along the valley,rather than an in-situ"sand surge"in the conventional sense.(2)This flowslide includes two failure modes:diffuse failure and lateral spreading.(3)Earthquake-induced soil liquefaction frequently occurs in saturated granular materials(silts,fine sands,etc.)exhibiting prominent strain softening characteristics.The occurrence of such geological hazard is abrupt.Post-instability,the landslide mass flows significant distances like a fluid,readily resulting in catastrophic consequences.This warrants heightened attention.
Objectives:With the development of Shenzhen city,China,land renovation is more frequent.At the same time,affected by the subtropical monsoon climate,the area under the jurisdiction has abundant rainfall and dense vegetation coverage,making it difficult to identify the hidden dangers of geological hazards widely distributed on artificial slopes and natural slopes.Therefore,it is necessary to develop a set of hazard evaluation system of geological disaster that can solve the unique terrain and climate conditions in Shenzhen,so as to achieve the purpose of preventing disasters in advance and reducing casualties.Methods:(1)On the basis of high-precision digital elevation model of Shenzhen city obtained by airborne light detection and ranging(LiDAR),about 3 500 slope disaster prone points in Shenzhen are obtained through data collection,remote sensing interpretation and field verification.The sample library expanded 330％after proofreading.(2)Taking 3 major factors(8 factors)of terrain,geological structure and human engineering activities into comprehensive consideration,and based on the rainfall-induced disaster mechanism,a rainfall collection factor is proposed,and the weight of evidence method is used to complete the geological disaster hazard evaluation model under rainfall-induced conditions.(3)The threshold determination method of"key point control"under the actual background of single disaster is proposed,and the classification of the risk assessment model is completed.Results:The area under curve value of receiver operating characteristic curve model reaches 0.903,indicating that the model has a good effect on disaster forecasting.LiDAR technology can improve the identification accuracy of geological hazards in cities under dense vegetation coverage.Conclusions:Based on airborne LiDAR data,through a series of means such as expansion of disaster database,analysis of disaster distribution law,establishment of disaster evaluation factors,and classification of risk levels,it can form a refined evaluation system for the hazard evaluation of the slope in densely vegetated areas under the influence of the subtropical monsoon climate.
Objectives:In mid-to late-April 2024,an extreme heavy rainfall event occurred in Shaoguan City,Guangdong Province,inducing a large number of landslides in Jiangwan Town,Shaoguan.People lost connection with the outside world for nearly 36 hours,which aroused widespread social concern.Rapid-ly and accurately identifying the basic characteristics of landslides,development and distribution patterns and formation conditions is crucial for disaster emergency decision-making and risk elimination and dispos-al.Methods:Using the post-disaster optical remote sensing images and combining with deep learning mod-el,the rainfall-induced landslides in Jiangwan Town,Shaoguan,were quickly and automatically identified.Results:After manually calibration,a total of 1 192 landslides were deciphered,with a total area of about 3.14 km2.The scale of the landslides was dominated by small and medium-sized landslides,which were mainly distributed as an aggregated belt along the river in the northeast-southwest direction,with a signifi-cant characteristic of concentrated occurrence.Spatial statistical analysis showed that the landslides were mainly distributed on concave slopes with slopes of 10°-30° in the range of 200-300 m elevation.Further quantitative analysis of the geomorphic controlling factors of landslides using the random forest model and SHAP theory reveals that different topographic and geomorphic factors have different degrees of nonlinear effects on landslide formation,and that multiple factors such as elevation,slope,and catchment conditions are coupled to jointly control the formation of landslides.Conclusions:This paper highlights the great ad-vantage of deep learning-based intelligent identification and analysis technology in the emergency investiga-tion and formation conditions analysis of landslide disasters,which can provide important technical support for the rapid assessment of disaster losses and risk identification.
在高频次观测、任务突发等动态业务场景下,任务数据往往具有动态增量性,需要影像理解模型能够对新的任务及影像数据快速作出响应.然而,由于现实世界的动态开放性,新增任务数据往往偏离历史观测数据,需要人工干预以实现模型快速调整和更新.
随全球气候变化,高位崩塌灾害频发,造成了巨大人员伤亡与财产损失.然而,当前崩塌模拟软件尚存明显的不足,如地形精度较低、未考虑岩体结构特征及无法实现块体运动过程中碰撞碎裂等现象,且利用中央处理器(central pro-cessing unit,CPU)的运算效率较低也限制了模拟崩塌的规模.为此,急需一种能更真实反映崩塌运动过程的快速模拟方法.利用无人机摄影技术,结合现场调查,提取危岩岩体结构特征,建立精细化的崩塌地质-力学模型;利用集成PhysX物理引擎与中央处理器-图形处理器(central processing unit-graphics processing unit,CPU-GPU)并行计算能力的Unity3D平台,研发了大规模崩塌运动过程模拟软件,再现崩塌坠落-撞击-碎裂-堆积的全过程;系统可输出崩塌三维运动轨迹、速度、能量及弹跳高度,为崩塌防治设计提供依据.以贵州纳雍谢家岩崩塌为原型案例,开展崩塌运动过程三维模拟与验证,模拟结果显示落石堆积范围与现场调查坡底落石堆积范围较吻合,单体块石运动特征符合现实规律.
高精度重力观测是地震学研究的一个重要手段,近年来在中国地震预测预报工作中发挥了重要作用.首先简要介绍了重力观测仪器的发展现状与分类,然后从重力观测网规模、重力数据处理流程、重力观测结果解释等方面对中日两国的研究现状进行了对比,并在此基础上系统分析了中国地震重力观测研究面对的挑战,力争为后续工作提供一些借鉴与启示.
交通领域的传统线性规划方法仅在静态网络中求解有限规模的资源调度问题.笔者面向城市巡游出租车长周期运营过程优化目标,使用融合了监督学习神经网络机制和奖励的深度强化学习技术替代线性规划,将动态交通网络中表征乘客和驾驶员出行行为下的时空变化特征、状态属性特征和交互关系特征等领域知识转换映射为包含状态、行为、转移概率和奖励函数等元组的马尔可夫过程,基于序贯决策思想在强化学习框架内的多智能体合作型随机博弈场景下,求解多对象司乘匹配和车辆行为选择策略优化任务的组合动态优化问题,实现顾及交通参与主体调度平台和驾驶员双向的营运收益最大化,有益于在现实层面破解"人找车难、车找人难"的出租车运力资源时空间不均衡现状.
频谱分析是以波数为自变量对位场异常频谱进行分析研究,从而解决实际工作中的异常转换、滤波、正反演等问题.位场的频谱特征包含了地下地层的空间分布、岩性及构造特征,有助于地球物理资料的解释、分析,为进一步了解研究区域概况提供信息.位场异常谱中的振幅谱与地下地质体的埋深、宽度与物性都有直接的关系;相位谱则反映了地下地质体的水平位置与倾角信息.计算了多种类型模型的频谱曲线,分析频谱曲线特征,并详细说明了模型参数变化对频谱曲线的影响;阐述了振幅谱反演、相位谱反演方法,提出了频率域联合反演方法,精确地反演模型的几何参数.将此方法应用于武清凹陷地层分析,反演得到了两个观测剖面的目标层深度和宽度,与实测地震资料对比,表现出较好的反演结果.
肩负"两统一"职责的国家自然资源部自2018年组建后开展了自然资源统筹管理工作,国土空间的立体化管理是其核心内容.国内外二十余年的三维地籍研究不断支撑着土地立体化管理模式的推进,实现城市空间创新管理模式应用,三维地籍需要面向全流程进行研究,同时支撑规划、建设、运维等多个时空阶段的土地利用活动和管理,实现统一的地理空间治理.
北斗三号全球卫星导航系统(global BeiDou-3 navigation satellite system,BDS-3)正联合北斗二号(区域)卫星导航系统(BeiDou-2 navigation satellite system,BDS-2)在3个或3个以上频率信号上为用户提供高精度的定位、导航和授时服务.为深度融合BDS-2和BDS-3三频数据,建立了 BDS-2/BDS-3三频无电离层组合和三频非组合精密单点定位(precise point positioning,PPP)模型,推导了 接收机频间偏差(inter-frequency bias,IFB)和 BDS-2 与 BDS-3 时间延迟偏差(timedelay bias,TDB)的数学表达式.对PPP定位性能及模型偏差特性进行分析,结果表明:BDS-3能显著增强PPP定位性能,BDS-2/BDS-3融合静态PPP在东、北、天3个方向上的收敛时间和定位精度分别为15.8、7.3、22.3 min和1.2、1.0、1.8 cm,动态PPP收敛时间和定位精度分别为27.3、10.3、33.8min和2.2、1.7、4.4 cm;三频无电离层组合与非组合PPP的收敛性能和定位精度基本相当,且三频观测值对定位性能提升不明显;IFB和TDB均具有较好的天内稳定性,BDS-2和BDS-3信号IFB天内标准差(standard deviation,STD)可分别达到7.9 cm和3.6 cm,而TDB天内STD可达到8.8 cm.
Objectives: The Beijing section of Beijing Tianjin Intercity Railway is 50km long. The railway line crosses the subsidence area from dongbalizhuang to Dajiaoting in Beijing. The uneven subsidence along the line poses a certain threat to the safe operation of high-speed railway. The South-to-North Water Diversion Project has brought new water sources to Beijing， the exploitation of groundwater in the plain area has decreased， and the development trend of land subsidence has slowed down. Methods: Taking the Beijing section of Beijing Tianjin Intercity Railway as the research area， this paper selects Radarsat-2（2011-2015） and Sentinel-1（2016-2019）satellite images， and uses permanent scatterer InSAR （PS-InSAR） technology to obtain the time-series subsidence information along and near the railway. The differences of subsidence evolution along the railway before and after the South-to-North Water Diversion in the study area is analyzed. On this basis， combined with the geological data and groundwater dynamic monitoring data along the railway， the response relationship between compressible layer thickness difference， groundwater level change and land subsidence is analyzed. Results: before the South-to-North Water Diversion （2011-2014）， the average length of the area with annual subsidence greater than 30mm along the railway is 13.25km， accounting for about 26.50% of the total length of Beijing section. After the South-to-North Water Diversion （2015-2019）， it has been reduced to about 8.87km， accounting for about 17.74% of the total length of Beijing section; Similarly， the mean value of the annual subsidence extreme value along the railway decreases from 105mm to 78mm， and the extreme value position moves eastward obviously from DK17 to near DK20. Conclusions: The subsidence along the railway is consistent with the development trend of regional subsidence funnel， which is mainly affected by the decline of groundwater level caused by long-term over exploitation of groundwater and the thickness of compressible layer. After the South-to-North Water Diversion Project， the subsidence along the railway has slowed down significantly. There is no obvious interaction between the normal operation of Beijing Tianjin Intercity Railway and the subsidence along the railway.
Length of day （△LOD） predictions play an important role in tracking and navigation of deep-space detector， precise determination of artificial satellite orbit and climate forecasting. In full consideration of the time-varying characteristics of △LOD， this work presents the application of a hybrid technique for predicting △LOD. The △LOD predictions are generated by means of the combination of （1） singular spectrum analysis （SSA） extrapolation for the linear trend， annual and semiannual oscillations in △LOD based on an iterative interpolation strategy， and （2） autoregressive integrated moving average （ARIMA） stochastic prediction of SSA remaining residuals， referred to as SSA+ARIMA. In order to evaluate the effectiveness of this approach， the △LOD predictions up to 365 days into the future are calculated year-by-year for the 2-year period from Jan 1，2000 to Dec 31， 2001 using the data covering the previous 10 years from International Earth Rotation and Reference Systems Service （IERS） C04 series.. The prediction results are analyzed and compared with those obtained by machine learning methods such as back propagation neural network （BPNN）， general regression neural network （GRNN） and Gaussian Process （GP）. It is shown that the accuracy of the predictions are better that by machine learning methods in terms of the mean absolute error （MAE） of predictions， especially for medium and long-term predictions. Compared with the predictions obtained by the BPNN， GRNN and GP， the MAE of the proposed SSA+ARIMA predictions up to 30 days and 365 days in future is reduced by 39% and 61%， respectively.
近年来,中国积极向中亚地区提供援助,以帮助改善当地的区域发展不平衡,但目前缺乏针对该地区援助效果的客观评估.针对该问题,根据2012-2019年的夜间灯光遥感数据构建中亚地区的夜光均衡指数,宏观分析该地区的社会经济发展情况,结合援助信息构建该地区一级行政区划的面板数据,并建立计量经济模型,从经济活动均衡化角度客观评估中国对中亚地区的援助效果.研究结果表明,近年来,中国向中亚地区提供的援助对促进该地区经济活动的均衡化有显著的促进作用;能源类和工业类的援助项 目促进区域经济活动均衡化的效果更突出;该地区的中等偏下收入国家在接受能源类的援助后,经济均衡化发展更明显.该研究证明了中国对中亚地区的援助切实改善了当地的社会经济发展水平,为实现人类命运共同体贡献了力量.
通过地质调查提前了解地质灾害发生的历史和现状,对最终实现潜在灾害的识别和预警具有重要意义.目前,传统人工地面调查手段难以发现并查明茂密植被覆盖或地形高陡等复杂山区的重大地质灾害及隐患,而航空遥感作为一种多功能综合性探测技术,因其独特视场角、不受地面条件限制等优势可高效地获取地质灾害发育分布特征和时空演化规律.首先,概述了地质灾害领域常用的航空遥感平台类型及发展趋势,分析了不同荷载传感器信息处理技术优势及主要解决的地质灾害问题.其次,综述了航空遥感技术在地质灾害基础地形测绘、早期识别、调查评价、中长期监测、应急处置5个应用阶段的重点研究成果,并论述了不同阶段的各类技术方法要求及优劣性.最后,总结航空遥感技术在地质灾害领域应用研究的不足之处,并阐明了未来发展趋势和建议.
现有的分布式空间数据管理系统侧重于对离散点集、点序列等类型数据的索引与查询,对线、面类对象的支持不足.针对该问题,基于开源HBase数据库,提出了一种面向矢量数据的自适应分布式管理方案(vector-oriented adaptive management scheme based on HBase,VA-HBase).该方案先对点、线、面等矢量对象采用两级索引结构,主索引 自适应剖分对象,寻找其合适存储层级,二级索引在全局粗粒度网格上记录覆盖对象的最小存储层级,然后设计最简字节流方法简化编码长度,优化存储空间,最后基于此设计了高效的范围查询算法.实验结果表明,所提出的VA-HBase方案能有效压缩各类矢量对象的存储空间,查询时能维持稳定的过滤性能,查询效率高于GeoMesa等对比方案约2～10倍,当数据集增大时,VA-HBase显示出良好的扩展性.
海洋二号B星(Haiyang-2B,HY2B)是中国海洋动力环境监测的首颗组网卫星,主要用于探测海面高度、海面风场、重力场等多种海洋动力环境参数,高精度卫星轨道是完成上述任务的前提与关键.选取HY2B卫星2021年1月的星载GPS观测数据,从数据完整性与多路径效应两个方面分析了观测质量,采用简化动力学定轨方法进行了精密定轨研究.结果显示,相比海洋二号A星(Haiyang-2A,HY2A),搭载同一型号接收机的HY2B卫星伪距多路径误差有所下降;基于国产星载双频GPS接收机可以实现HY2B卫星径向2 cm、三维优于3 cm的定轨精度;同时,验证了相位中心变化(phase center variation,PCV)模型对精密定轨的改进作用.目前的星载GPS数据与定轨方法可以满足高精度海洋测高任务的需求.
近年来,深度学习技术推动了高分辨率遥感影像智能理解的发展.然而,基于深度卷积神经网络的高分辨率遥感影像应用往往只能获得模型的最终预测结果,而无法解释决策过程中影响模型泛化性的原因.此外,深度学习的对抗样本问题也使得这些应用存在安全隐患.因此,笔者将模型黑盒问题和对抗样本问题归结为遥感影像理解的可信问题.受人类视觉系统启发,遥感影像智能模型的黑盒问题需要考察不变性表征与泛化性的关系.它包含不变性表征的解释性和交互性.同样,对抗样本问题本质是遥感影像不变性表征的稳定性和鲁棒性.
林地是国家重要的自然资源和经济资源,掌握林地分布状况对林地资源调查管理具有重要意义.针对传统林地提取方法精度较低且边界不规则的问题,设计了一种联合多尺度注意力机制与边缘约束的林地提取方法.首先,构建一种端到端的多尺度注意力神经网络模型,充分提取影像中林地的上下文特征,对不同尺度下的林地进行语义描述,实现高精度的林地像素级表达;其次,构建边缘约束规则,对提取结果进行边界优化,提高林地提取结果的可读性.为证明方法的有效性,以中国四川省绵阳市三台县作为实验区,建立数据集并进行林地提取实验,结果显示,所提方法提取的精确度为81.9％,召回率为75.6％,F1分数为78.1％,交并比为64.2％.结果证明,所提方法在遥感影像林地提取应用上效果良好.
随着人工智能的发展,利用高分影像进行滑坡等地质灾害识别逐渐成为研究热点.滑坡目视解译需依赖专家经验,传统滑坡自动识别方法又易将滑坡和裸地、道路等地物混淆.针对以上问题,提出了基于模拟困难样本的掩模区域卷积神经网络(mask region-based convolutional neural network,Mask R-CNN)滑坡提取方法.在现有样本的基础上,利用滑坡的形状、颜色、纹理等特征模拟更为复杂的滑坡背景进行困难样本挖掘增强,并将得到的困难样本输入Mask R-CNN网络进行滑坡精细检测分割.在实际研究区域中,由于滑坡数量有限,因此在频率域进行小样本学习,在减少数据需求的同时,保证分割识别的准确度.中国贵州省毕节市的实验结果表明,基于模拟困难样本的Mask R-CNN方法检测精度为94.0％,像素分割平均准确率为90.3％,可实现低虚警率下的高性能检测分割;采用频率域学习,在一半数据输入量的情况下,模型检测精度仍可得到提升.利用中国甘肃省天水地区的滑坡区域进行实际验证,进一步证明了所提方法的有效性.
点云密度是激光雷达(light detection and ranging,LiDAR)技术的重要参数,对森林遥感反演指数的提取有重要影响.以1 600 m× 1 450 m大小的无人机(unmanned aerial vehicle,UAV)LiDAR数据为实验数据,采用分级随机抽稀法对实验数据进行抽稀,获取不同密度的点云数据集,利用不同密度数据集提取郁闭度、间隙率、叶面积指数、点云高度和密度分位数等森林遥感反演指数,并与原始数据提取的森林遥感反演指数进行差值比较.(1)当点云密度较小时,提取的郁闭度略微偏低,而间隙率略微增加,点云密度对郁闭度、间隙率的影响极小.(2)当点云密度较大时,对叶面积指数的影响不大,但当点云密度较小时,对叶面积指数的影响较大,个别区域可能出现叶面积指数突变.(3)当点云密度较大时,点云密度对高度、密度分位数的影响不明显,但当点云密度降至3.6点/m2时,可能会出现个别区域密度、高度密度分位数突变的情况.点云密度对森林遥感反演指数有重要影响,合适的点云密度有利于更准确地描述森林结构形态,过小的点云密度影响森林遥感反演指数的提取.
