地学传感网矿区地面灾害监测数据计算和灾变建模可信度理论方法研究

GIS-based sensor network and real-time GIS are becoming one of the research hotspots in the field of Geomatics. A deep investigation into the trust theory and method of mine geohazard monitoring data computing and catastrophe modelling of geosensor network would play a significant role in forming a technical service system for monitoring and early warning of geohazards in mine areas. Such a system in turn holds great values for both scientific research and practical application once it could provide, in time, an accurate and reliable forecast of the geohazard and give guidance on minimizing the losses. Based on the database which was fed with the monitoring data in the past over 10 years by the geosensor network we established, this proposal will focus on the trust theory and method of data statistical features, filtering and denoising, and the pertinent calculation of factors causing geohazards，and catastrophe modelling, in order to break through the key scientific points geosensor network computing. The theory so developed would significantly improve the applicability, reliability and timing of models for monitoring data trend prediction, spatial variation and intelligent geohazard forecasting. For achieving this, the collaboration of couple of disciplines is necessary, such as surveying data processing, spatial statistical analysis and disaster dynamics. This service will be testified and then modified if needed in a pilot study area for gaining a reliable performance. The outcome of the proposed study would not only input new and deep understandings in academic research and improve its practical application, but also provide a solid base of theory, methodology and technique for informationization surveying and mapping as well as geographical condition monitoring of the nation.

传感网GIS、实时GIS是国际国内地球空间信息科学领域的研究热点。研究地学传感网地面灾害监测数据计算和灾变建模可信度理论方法，建立矿区地面灾害监测可信服务，对矿山及时准确科学预报、预防和减小地面灾害的损失，具有重要的科学意义和应用价值。基于研究实验区的地学传感网系统和十多年的地面地下监测数据，应用统计分析、误差处理、总体最小二乘法、灾害动力学等理论和方法，研究地学传感网数据统计特征、滤波去噪、地学孕灾环境因子关联计算和灾变建模的可信度理论方法，进一步突破地学传感网计算关键科学问题，解决地学传感网应用服务中的数据趋势时空预计、灾害事件辨识和灾变预测模型实用性、可靠性和时效性差的问题。应用地理信息服务技术，在实验区对计算和服务的效能进行实证和评价。项目研究可提升地学传感网基础理论研究和整体应用服务水平，也为信息化测绘和地理国情监测提供理论、方法和技术支撑。

应用统计学习和数字信号处理方法研究了地学传感器时序数据统计特征、异常值识别和去噪计算。对多源DEM数据的精度从指标、相对误差和绝对误差等各个方面进行了评价,并基于评价结果对DEM 数据的误差进行了校正。从基础因子、发育因子和诱发因子三个方面建立了矿区地面灾害孕灾地学环境特征因子数据集。开展了地学传感网数据与孕灾地学环境数据的关联计算，为可靠性预测计算建立基础。.对滑坡区域场景从滑坡局部区域、滑坡体区域和滑坡子区域三个层次理解。利用高分辨率无人机影像分别从纹理特征和显著性特征两个方面实现影像上滑坡特征的提取与分析。组合霍夫变换、阴影检测和Harris角点检测方法，实现塌陷区典型塌陷目标的检测。采用灰度共生矩阵对地裂缝影像的纹理特征进行分析。.利用地学传感网(GSN)监测系统，建立了星地协同InSAR监测模式，实现大面积宏观监测中异常的早期识别和发现，准确定位。运用SAR偏移量追踪技术计算了研究区域的大梯度形变。分析了不同参数对偏移追踪计算结果的影响，研究了基于纹理特征的偏移量追踪法，SAR偏移量追踪技术可对露天矿生产作业区的形变进行有效监测。.对露天矿地面沉降灾害变形、蠕动、滑动和稳定阶段的分类事件和事件特征进行分析，建立成灾模式空间特征辨识模型。应用移动最小二乘法、总体最小二乘法和卡尔曼滤波方法建立可信分析预测模型。建立了滑坡敏感性分区模型，分析矿区地下开采和地面塌陷与滑坡的关系。选取15个条件因子作为评价指标，构建了敏感性评价模型。.构建矿山地面沉陷灾害链，明确建模要素，分析灾害链的控制流关系，对矿山地面灾害链进行灾变过程分析。基于开源的移动GIS平台体系架构和地理信息服务技术，开发了移动客户端任务化地面灾害信息采集软件。.项目成果对于提升监测数据趋势时空预计、灾害事件辨识和灾变智能预测模型的实用性、可靠性和时效性，具有重要的科学意义和应用价值。