融合多轨道InSAR与岩层移动模型的矿山三维时序形变估计

It is feasible to retrieve mining-induced 3-D displacements by integrating single-track InSAR measurements with mining deformation model. However, this method has some drawbacks. For example, prior model parameters required in this method are generally difficult to be accurately determined. Moreover, its weighting function is empirical, and time samples of single-track SAR datasets are limited. These drawbacks not only significantly limit the practical applications of this method, but also dramatically reduce the accuracy and temporal resolution of its estimates. To circumvent these, multi-track InSAR measurements with much denser time samples than single-track ones will be introduced to fuse with the mining deformation model, so that we could develop a new method for mining-induced 3-D time-series displacement retrieval. This new method will capable of adaptively determining the required prior model parameters and improving the accuracy and significantly enhancing temporal resolution of 3-D time-series displacement estimates. More specifically, a single-track InSAR-based algorithm will firstly be investigated for estimate mining 3-D displacements with the adaptive estimation of prior model parameters. Then, a theoretically rigorous weighting approach of mining 3-D displacement estimates will be developed based on the probability density function of interferometric phase errors. Finally, a new method will be proposed for retrieving mining 3-D time-series displacements with significantly enhanced temporal resolution using a generalized weighting least square solver and multi-track InSAR 3-D displacement observations. The proposed methods in this program will address the difficulty of prior model parameters determination in previous method. Meanwhile, the results estimated by the proposed methods will offer finer and more reliable scientific evidence for the controlling of mining-related geohazards and the understanding of mining deformation kinematics.

利用单轨道InSAR观测值与岩层移动模型可以实现矿山三维形变估计。但其存在岩移参数难以准确获取，定权方法过于经验，时间采样不足等问题，制约了该方法的实用性，降低了估计结果的精度和时间分辨率。为此，本项目将充分利用时间采样更密集的多轨道InSAR形变观测值，并将其与岩层移动模型融合，发展一种岩移参数自适应获取、精度能够改善、时间分辨率大幅提升的矿山三维时序形变估计方法。具体地，本项目将首先研究岩移参数自适应获取的单轨道InSAR矿山三维形变估计算法，然后基于干涉相位误差概率密度函数发展理论更严密的多轨道InSAR三维形变定权方法，最后提出基于广义加权最小二乘和多轨道InSAR融合的、时间上大幅加密的矿山三维时序形变估计方法。本项目将解决岩移参数难以准确获取对InSAR矿山三维形变监测的限制，提高估计结果的精度和时间分辨率，从而为我国矿山地质灾害防控与变形机理研究提供更精细、更可靠的科学依据。

地下矿产资源开采容易导致岩体移动变形，进而诱发地表沉降、山体滑坡、建构筑物破坏等地质灾害。因此，地表在三维空间中的动态变形精细监测对于矿区地质灾害防灾减灾至关重要。InSAR具有大范围、高空间分辨率地表形变监测能力，但该技术仅能监测地表在雷达视线向的一维历史形变，其难以准确反映矿区地表在三维空间中的历史变形过程和演化趋势，极大限制了InSAR在矿山领域的应用前景。虽然近年来国内外学者在InSAR矿山三维形变监测与预测领域开展了探索性研究，取得了一定进展，但仍然存在大形变梯度矿区一维视线向形变InSAR监测困难、联合岩移模型与InSAR的矿区三维形变监测预测依赖矿区实测参数等关键瓶颈问题。.为突破上述瓶颈，项目组将矿山开采沉陷学与影像大地测量学融合，首先提出了二维椭圆高斯函数支持下的大形变梯度矿区一维形变InSAR监测方法，有效提升了InSAR在大形变梯度矿区的监测能力。在此基础上，引入广义测量平差理论，发展了先验参数自适应估计的矿区三维时序形变InSAR监测方法，克服了现有方法未能顾及先验参数动态性的局限，同时解决了先验参数依赖人工收集的难题。为了进一步提高InSAR三维形变观测值时间分辨率，项目组充分利用多轨道InSAR密集时间采样的特点，提出了基于广义最小二乘的矿区三维时序形变InSAR高时空分辨率监测算法，将观测值时间分辨率从目前的十几天提升至几天。最后，引入矿区地表单点动态移动规律，建立了联合Weibull与Kalman滤波模型的矿区三维时序形变InSAR预测方法，为矿山三维形变预测提供了全新的思路与视角。项目研究成果丰富了矿山地表三维形变InSAR监测与预测理论，进一步推动了InSAR业务化应用于矿山地质灾害防控与治理。