Subsidence monitoring of the Seattle viaduct tunnelling project with Homogeneous Distributed Scatterer InSAR

摘要

Subsurface excavations may cause subsidence due to redistribution of stress in the overburden or changes in the groundwater state. In the case of tunnelling under urban areas, this may result in damage to existing high value infrastructure within the surface zone of influence. Synthetic aperture radar interferometry (InSAR) is a useful tool for urban subsidence monitoring because it provides measurements that are wide-scale and ongoing at regular time intervals. However, many existing InSAR methods such as persistent scatterer InSAR and differential InSAR do not provide sufficient spatial resolution to adequately measure short-scale spatial subsidence gradients which represent the greatest risk to infrastructure. MDA Systems Ltd. (MDA) has developed an InSAR method, Homogeneous Distributed Scatterer (HDS-InSAR), which generates higher spatial densities of measured coherent targets compared to many current methods. This is achieved through a novel combination of spatial phase filtering and coherence based target selection. This method exploits both persistent point and coherent distributed scatterers by using adaptive multilooking over statistically homogenous pixel neighbourhoods to improve the spatial resolution of the resulting subsidence maps. Furthermore, a matched parametric temporal model is fit to the deformation signal in order to enhance the detection of small amplitude subsidence. These techniques in combination provide a means for measuring small scale deformations within an urban environment and correlating their occurrence with known deformation drivers. Results are presented for the ongoing Seattle viaduct tunnelling project which consists of a 3.2 km long bored road tunnel under the city. Two satellite datasets (RADARSAT-2 Spotlight mode stacks from opposing look directions) are analysed to derive 2D (vertical + east-west) deformation maps over the area of interest spanning a time period from 2012-2015 and these results are correlated with known groundwater extraction events.