基于数据同化理论的InSAR技术黄河上游滑坡监测预报

LANDSLIDE MONITORING AND FORECASTING IN UPPER YELLOW RIVER BY InSAR TECHNOLOGY BASED ON DATA ASSIMILATION THEORY

黄河上游地区滑坡地质灾害分布广泛,活动频繁,危害严重.传统的滑坡识别和监测方法均存在局限性.InSAR技术由于其精度高、可获取毫米级形变等特点被广泛应用于滑坡监测中.但InSAR技术对影像相干性的要求较高,导致其数据的离散程度较大,无法获取到连续的形变数据,对实际应用中利用监测的结果进行预报的能力造成了较大的影响.而利用数据同化理论的建模方法,可以对多尺度、多来源、多类型的数据进行协同处理以消除误差.本文通过SBAS-InSAR技术获取地表的毫米级形变,经由卡尔曼滤波算法对观测结果进行了数据同化.单点实验的结果表明,卡尔曼滤波之后的模拟结果较同化之前有了明显的提高,验证了数据同化算法在提高数值模拟的精度上的可行性.通过数据同化理论来产生连续的预报数据,为InSAR技术监测和预报滑坡形变提供了一个新的思路.

The landslides in the Upper Yellow River region are frequent and widely distributed,causing serious damage.Traditional landslide recognition and monitoring methods have limitation,while the InSAR technology,due to its high precision and extraction of millimeter-scale deformation,is widely used in landslide monitoring.However,the technique has a high requirement for image coherence,resulting in large dispersion degree of data and impossibility of obtaining continuous deformation data,which greatly impacts the forecasting application of the monitoring results.With the modeling method of data assimilation theory,the multiscale,multisource and multitype data can be coprocessed to eliminate errors.In this paper,the millimeter-scale deformation of surface is obtained by SBAS-InSAR technology,and the observed data is assimilated by Kalman filtering(KF)algorithm.The single point experiment shows that the simulation results after KF are significantly improved compared with those before the assimilation,which verifies the feasibility of data assimilation algorithm in improving the precision of numerical simulation.The continuous forecast data is generated by data assimilation theory,which provides a new way for InSAR technology to monitor and forecast landslide deformation.