对矿区开展长期形变监测和后续形变预计对于预防矿山开采所引发的潜在安全问题和矿山生态环境保护具有十分重要的意义。利用合成孔径雷达干涉测量技术(Interferometric Synthetic Aperture Radar, InSAR)与概率积分法(Probability Integral Method, PIM)结合是较为有效的矿区形变预计方法。然而，目前矿区时序InSAR形变建模环节中大多数采用纯经验数学模型，未顾及开采沉陷机制，严重影响获取的形变观测值的精度；不准确的InSAR形变观测值将误差传递至后续形变预计，进而导致误差累积。本文将坐标-时间函数(Coordinate Time, CT)引入时序InSAR建模环节，构建了坐标-时间函数预计模型(CT-PIM)以取代传统纯经验数学模型；再基于InSAR相位观测值求解未知参数，将构建的CT-PIM直接用于矿区形变预计，避免了误差的传递，以改善形变预计精度。分别开展了模拟实验和真实实验对提出预计方法进行验证。模拟实验显示：模型预计获取的时序形变与模拟真值的均方根误差(RMSE)为±4.6 mm。真实实验利用覆盖淮安市一盐矿区的 35 景 Sentinel-1A SAR卫星影像开展，获取了研究区域2019年3月30日至2019年7月28日的预计结果，结果显示测区最大预计沉降量为152 mm。本文所提方法相比传统多速率模型，建模精度提升了38.2%；相比传统静态概率积分预计方法获取的形变结果，精度提升了39.1%。本文方法为预防盐矿区采矿引起的地质灾害问题提供了有力的工具，为盐矿区生产安全和生态环境保护提供参考。

(Objective) Long-term monitoring and the subsequential prediction of deformation for salt mining area is essential to the safety prevention and environmental protection of mining areas. The combination of Interferometric Synthetic Aperture Radar (InSAR) technique with Probability Integral Method (PIM) has proven to be powerful in predicting the deformation of mining areas. However, the single Multi-Temporal InSAR (MT-InSAR) has a limitation in that it can only obtain the deformation sequences during SAR acquisition dates, and the subsequent future displacement beyond the span of the SAR observations cannot be acquired. Besides, traditional mathematical empirical models are mostly used in time series modelling of mining areas, ignoring the underground mining mechanisms, which seriously affect the accuracy of the observations. Inaccurate InSAR deformation monitoring results will transmit errors to forward predicted subsidence, which may induce more significant errors. (Method) In this study, the Coordinate-Time (CT) Function is introduced into the time series InSAR deformation modelling, and a Coordinate Time Function prediction model (CT-PIM) is constructed to replace the traditional mathematical empirical models, which can better describe the dynamic evolution disciplines of the underground mining subsidence in the InSAR deformation modelling procedure. The unknown CT-PIM parameters can be estimated directly via the InSAR time series phase observations, and the constructed CT-PIM is directly used in the deformation prediction of the mining area, which can avoid the error propagation from the InSAR-generated deformations and improve the deformation prediction accuracy. (Result) The new approach has been tested by both simulation and real data experiments. The simulation results show the Root Mean Square Error (RMSE) between the time series deformation prediction of the model and the simulated true value is estimated to be ±4.6 mm, which implies that the proposed method is of promising accuracy. The real experiment was carried out using a total of 35 Sentinel-1A SAR images covering the salt mining area in Huaian city, and the deformation prediction of the study area from March 30, 2019 to July 28, 2019 were obtained. The results show that the maximum settlement of deformation prediction in the study area is 152 mm. The modelling accuracy with an improvement of 38.2% compared to traditional SBAS-InSAR, and the deformation prediction accuracy with an improvement of 39.1% compared to traditional static PIM prediction method. (Conclusion) The CT-PIM was used as a substitute for traditional MT-InSAR pure empirical models and was applied for predicting the dynamic deformation over salt mining area, which provides a more robust tool for the forecasting of mining-induced hazards. The above results show that the CT-PIM can describe the temporal dynamical characteristics of the mining-induced subsidence more realistically, which can avoid the secondary error propagation, as well as a reference for safety management and ensuring the environment protection.