高相干点稀疏的采煤沉陷区时序极化InSAR地表形变反演研究-以淮南矿区为例

Underground coal mines cause severe land subsidence, leading to many ecologic and environmental problems. With the purpose of comprehensive subsidence management and sustainable ecosystem development, subsidence measurement and monitoring are of great importance. Time series analysis of ground subsidence based on high coherent pixels is an extension to the conventional InSAR technique. It offers a practical way to reduce the temporal and geometrical decorrelation, and atmospheric artifacts comparing to conventional D-InSAR technique, therefore can be successfully applied in long-time surface subsidence monitoring with multi-temporal SAR images. However, most of the coal mining areas are located in non-urban areas with densely vegetation, where the coherence generally is low, which prevents the formation of a high quality interferogram and results in sparse high coherent pixels, therefore limits the performance of time-series subsidence analysis. In the area of Huainan, Anhui province, active coal mines are located causing serious surface subsidence. This area consists of both urban area and non-urban areas. The proposed research would take this area as an example to perform an improved high coherent pixels based on time-series DInSAR method with polarization information. It is aimed to (1) exploit the polarimetric information to perform coherence optimization, to increase the quality of interferogram and increase the number of pixel candidates; (2) analyze phase difference and select pixels that show even or odd bounce scattering properties to increase the number of pixels, especially in non-urban and densely areas; (3) perform time-series subsidence inversion analysis based on SBAS method utilizing the results of polarimetric interferometric optimization and increased pixel numbers. The subsidence inversion results would be validated and further optimized with geodetic data and numeric model.

煤矿开采后采空区的地面沉陷对生态环境造成重大影响。因此，在采煤沉陷区开展高精度的沉降监测，为地面沉陷的预警预报提供决策依据具有重要意义。在采煤沉陷区建筑物少、植被较为茂盛的地方，由于干涉图相干性差，无法找到足够密度的高相干点进行干涉相位的精确估算，因此难以实现准确的地表形变监测。本项目以淮南煤矿采煤沉陷区为典型研究区域，从多极化角度解决复杂地表覆盖区域相干性差、高相干点稀疏的问题，通过（1）极化干涉最优化研究提高干涉质量和相位估计精度；（2）极化相位差研究实现即使是植被覆盖区域也有足够多的高相干点参与地表形变反演；（3）改进SBAS方法，实现高精度的时序地表缓慢形变反演，掌握研究区域的地面沉降速率和时空分布规律，形成适用性强的地面沉降监测方法，为中国东部平原地区城镇化程度低、高植被覆盖的煤矿地区地面沉降监测服务，为地面沉陷的预警预报提供决策依据。

本项目将针对非城镇地区、植被覆盖地区的地表沉降监测需求，以淮南煤矿区作为典型研究区域，从多极化角度解决复杂地表覆盖区域相干性差、高相干点稀疏的问题，通过极化干涉最优化和极化相位差研究，提高干涉对的干涉质量和相位估计精度，提供高质量的差分干涉图集，增加候选相干点的数量，实现了即使是小数据集、植被覆盖区域也有足够多的高相干点参与地表形变反演的目的。研究结果表明，经过相干优化后的干涉图的质量有显著提高，对于提高干涉差分的质量，特别是植被覆盖区域的相干性，从而提高相位估计精度具有重要的意义。通过极化相位差挑选出来的奇偶反射点具有较高的可靠性，可以和通过相干图挑选出来的高相干点互相补充，共同作为高相干候选点，从而使得非城镇地区、植被覆盖地区也能获得足够多的高相干点参与时序InSAR的地表反演工作。在此基础上，改进并优优化了SBAS InSAR方法，将极化干涉最优化和极化相位差的思想融入到算法流程中，实现了高精度的时序地表缓慢形变反演，获得了研究区域地面平均沉降速率的空间分布以及地面沉降的时空变化特征。形变结果与水准测量的结果进行了比较，得到了较好的一致性。通过本项目的研究形成了适用性更强的地面沉降反演方法，为中国东部平原地区高植被覆盖、城镇化程度较低的煤矿区域地面沉降监测服务，为地面沉陷的地质灾害预警预报提供决策依据。