(1. 中南大学 地球科学与信息物理学院，长沙 410083；
2. 湖南科技大学 煤炭资源清洁利用与矿山环境保护湖南省重点实验室，湘潭 411201；
3. 中南大学 湖南省普通高校精密工程测量及形变灾害监测重点实验室，长沙 410083)
摘 要: 利用10景ALOS PALSAR影像获取山西云冈某矿区在2007年7月1日至2009年1月3日的地表时序沉降值，并使用Logistic模型拟合该矿区全盆地时序沉降。结果表明:通过交叉验证Logistic模型估计参数预测的时序沉降与InSAR监测值后发现，两者吻合较好，且其平均均值和均方根误差分别为-0.4和2.5 cm，表明在整个下沉盆地内，各点的动态沉降均符合“S”型增长，且Logistic模型能较好地描述该过程。统计该矿区全盆地Logistic模型形状参数a和b后，发现参数a和b分别服从Weibull分布和随机分布，且其数值变化较大，表明利用少量离散地表监测数据拟合的Logistic模型参数预测的全盆地动态沉降结果可靠性不高。最后，利用全盆地Logistic模型估计参数预测了该矿区2009年2月18日的地表沉降值，该值与InSAR监测结果吻合较好，均方根误差为2.15cm。
关键字: InSAR时序形变；开采沉陷；时空演化；Logistic模型；遗传算法；Levenberg-Marquard算法
（1. School of School of Earth Science and Geomatics Engineering,
Central South University, Changsha 410083, China;
2. Hunan Province Key Laboratory of Coal Resources Clean-utilization and
Mine Environment Protection, Hunan University of Science and Technology, Xiangtan 411201, China;
3. Key Laboratory of Precise Engineering Surveying and
Deformation Hazard Monitoring of Hunan Province, Central South University, Changsha 410083, China）
Abstract:10 ALOS PALSAR images were used to derive the mining ground time-series subsidence of one mining area in Yungang city, Shanxi Province, China, from July 1, 2007 to January 3, 2009, and the mining ground time-series subsidence was fitted with the Logistic model. The results show that the predicted time-series subsidence by the Logistic model has a good agreement with those of InSAR-measured with average mean and root mean square error of -0.4 and 2.5 cm, respectively, from the cross validation of both. The dynamic subsidence of all surface points in the whole basin agrees with S-shaped temporal evolution, and the Logistic model could describe this temporal evolution accurately. Subsequently, the shape parameters a and b of Logistic model are followed the Weibull and random distribution, respectively, and rapidly changes of parameters occur from their statistical histograms. This indicates that the predicted kinematic subsidence is unreliable if the parameters of Logistic model are yielded by the measurements of a few sparse observation points. At last, the subsidence of this mining area on February 28, 2009 was predicted, which have a good agreement with those of InSAR measured with root mean square error of 2.15 cm.
Key words: InSAR-derived time-series deformation; mining subsidence; spatio-temporal evolution; Logistic model; genetic algorithm; Levenberg-Marquard algorithm
