基于InSAR的缓变型区域滑坡早期动态识别与监测技术研究

Early detection and monitoring of landslides is a major and popular issue in the research field of geo-hazards, and this issue has been developed rapidly with the advancement of InSAR technology recently. Due to the influences of complex regional topography and vegetation cover, improving the data accuracy of InSAR in monitoring the development of progressive landslides has become a challenge and urgent problem to be solved. In this research, Wudongde hydropower reservoir area, where landslides occur extensively, is investigated to further promote the application and monitoring accuracy of InSAR on this issue. The main tasks of this research include: (1) considering the influences of complex regional topography and satellite attitude parameters, a computational model can be established to quantitatively analyze the correlation between local incident angle and interference characteristics, and then the satellite attitude parameters can be optimized accordingly, which can properly receive SAR data; (2) to improve the insufficient monitoring accuracy of the current InSAR time series analysis method in applying to large, mountainous or vegetation-covered area, a computational system of PS-DS InSAR by combining Persistent Scatterers (PS) and Distributed Scatterers (DS) can be constructed; (3) properly identify and obtain the position and development of landslides in this area, and conduct simulation of a representative landslide deformation based on the characteristics of triggering and sliding mechanism as well as the deformation-time correlation. Based on the anticipated results from this research, early detection and monitoring of progressive regional landslides with greater monitoring accuracy can be achieved in mountainous areas of complex regional topography and vegetation cover.

滑坡监测预警是灾害研究领域的难点。近年来InSAR技术的发展，使缓变型区域滑坡早期动态识别监测成为可能，但限于滑坡频发区域复杂地形、山区及植被覆盖等特征，识别监测精度成为亟待攻克的难题。本研究以乌东德水电库区为例，开展以下研究工作：1）考虑区域复杂地形特征与卫星姿态参数响应关系对数据干涉质量的影响，建立地面局部入射角几何计算模型，定量分析局部入射角与数据干涉质量的变化关系，反演卫星数据的最佳姿态参数，确定合理的SAR数据获取方案；2）基于现有时序InSAR对大范围、植被覆盖山区算法精度的不足，构建永久散射体PS与分布式散射体DS联合解算的PS-DS InSAR算法体系；3）结合滑动机理、成因类型、形变时空变化特征，进行区域滑坡形变滤波分析及单体滑坡滑动主向形变场模拟。基于以上研究解决InSAR技术在复杂地形山区及植被条件下识别监测滑坡精度不足的难题，实现区域滑坡早期精准识别与高精动态监测。

滑坡灾害早期识别与监测预警是地质灾害研究领域的热点和难点。近年来InSAR技术的发展，使缓变型区域滑坡早期动态识别监测成为可能，但限于滑坡频发山区复杂地形及茂密植被覆盖，识别监测精度成为亟待攻克的难题。本研究以西南典型高山峡谷滑坡易发区乌东德水站电库区为例，开展了以下研究工作：1） 针对滑坡灾害易发的高山峡谷地形陡峭区域，InSAR 技术受地形与卫星侧视成像影响难以全面有效覆盖监测的技术难题，提出和建立基于地形与卫星姿态参数响应关系的综合因子地面局部入射角GLIA计算模型，定量分析反演不同GLIA下数据干涉质量特征，得出高山峡谷地形复杂区域InSAR技术最佳干涉质量的GLIA范围区间为30°-60°，且其0°-90°的范围干涉结果相对可靠，但叠掩（小于0°）和阴影（大于90°）区由于几何扭曲和失相干干涉结果是不可用的，并精确计算出无法有效监测的叠掩区和阴影区，实现了对有效监测区域的精准定位；2）针对山区植被茂密覆盖干涉点目标提取不足的技术难题，提出和建立基于 PS 与 DS联合结算的 PS-DS 算法体系，即在优选PS点基础上，通过K-S检测对时序图像空间进行同质像素簇SHP探测选取分布式散射体DS，而后将PS和DS点作为相同性质的干涉点目标进行联合结算提取地表微小形变值。结果表明相同地形与数据覆盖情况下研究区干涉点目标的提取密度提高三分之一以上，实现InSAR技术山区植被覆盖区域地表微小形变的精准提取；3）结合区域滑坡成因类型、滑动机理、形变时空演化特征，实现了对乌东德水电库区区域滑坡的精准识别和定位；并对库区坝址附近单体金坪子滑坡进行了详细研究，建立地表三维形变向视线向形变转换的计算模型，实现PS-DS InSAR单点监测结果与地面高精度单点监测结果的精准验证。已研发形成一套确实可行的基于InSAR技术高山峡谷复杂地形与茂密植被覆盖区域滑坡早期识别监测算法体系，解决了InSAR 技术受高山峡谷复杂区域地形及地表茂密植被覆盖调查监测精度难以满足实践需求的技术难题，实现了区域滑坡早期动态识别与精准监测，为区域灾害隐患早期识别与监测预警提供高效可行的技术方法和有效途径，解决滑坡灾害无法早期监测预警的技术难题。