Using SBAS-InSAR with Sentinel-1 Data to Measure Ground Subsidence and its Possible Effects on Catchment Modelling

Abstract

A major geohazard facing urban areas in the 21st century is subsidence with studies showing several cities affected across the globe. Historically, subsidence has been accurately measured to millimeter precision using ground-based methods like precise spirit levelling, GNSS observations, among others. However, these methods provide discrete data, are expensive, labor-intensive and time-consuming to replicate continuously across large areas. These limitations are overcome using Synthetic Aperture Radar (SAR) data which in several studies on Earth has been proven to provide similar accuracy continuously over large areas. The techniques that have been utilized include classic InSAR or multi-temporal InSAR methods such as PS-InSAR or SBAS-InSAR that provide a long-term continuous monitoring option for ground surface changes. Unfortunately, none of these have been applied to monitor subsidence in any city in Uganda. Typically, ground surface is represented by Digital Elevation Models (DEMs) which form the basis for catchment modelling. In an area undergoing subsidence, the ground settles either uniformly or at various rates. This change in the ground renders previously developed DEMs inaccurate representations of the ground and in effect inaccurate catchment models. This study used SBAS-InSAR to measure subsidence on Mbarara city over a period of 35 months using 147 SAR images. It then incorporated the subsidence data into SRTM DEM to create a modified SRTM DEM. Whereas the subsidence varies across the city, the average velocity was found to be 1.2cm/year ±0.08cm for the city. All the sampled locations within the city showed subsidence ranging from 0.4cm/year to 1.8cm/year with higher subsidence values observed in the low-lying areas compared to the higher elevation locations in the city. The results compare within millimeters with data observed at the IGS GNSS station in Mbarara. Two catchment models were developed for the city based on SRTM and modified SRTM DEMs. The comparison of the models shows significant differences in stream networks in the low-lying areas. These compare very closely with streams on the ground. The two models also varied in catchment area by 226sqkm. The study demonstrated that SBAS-InSAR can be used to monitor ground subsidence accurately and that subsidence should be factored into for effective catchment modeling