Study on ongoing subsidence in Uppsala City using Sentinel-1 radar data

Abstract: Many cities in Sweden are partly located on clay and because of that, some urban city centres are undergoing significant subsidence. To measure subsidence in cities, precise leveling has been the traditional technique, but the interest for the Persistent Scatter InSAR (PSI) technique has increased in the last years, in this application. With the PSI technique, a mm-accuracy can be obtained and the analyses can be done over large areas. In this study, a validation between the PSI and the precise leveling techniques was performed for a selection of buildings located in areas that are facing great subsidence. A correlation between the subsidence rate achieved in the PSI analyses and near-surface soil type was also done, to easier identify risk zones. The city of Uppsala was chosen as study area, because it is partly built on deep layers of clay and the consulting company Bjerking AB has established a leveling network with metal pegs on many buildings. One ascending and one descending PSI analysis was performed, with Sentinel-1 data from the period mid-2015 to mid-2019, and the PSI analyses were done in SARPROZ. After the PSI analyses, comparative permanent scatters (PS) points and metal pegs were identified creating validation pairs. 15 different validation pairs were identified in four different objects, which was one or two buildings. The PSI analyses showed that Uppsala is undergoing significant subsidence in some parts, with an annual rate of about 6 mm/year in the line-of-sight (LOS) direction, which corresponds to about 7.5 mm/year in the vertical direction. The areas of greatest deformation were exclusively found on postglacial clay. The standard deviation of the time series were calculated around their linear regression lines, which was a measure of how temporal coherent the points were. The mean of this standard deviation for the PS points in the 15 validation pairs was 1.5 mm. This standard deviation increased to 2.3 mm in the time series where the direction was transformed from LOS to vertical and where the movements were in respect to the benchmarks. Between the PSI and the precise leveling techniques, in the validation, the vertical subsidence rate differed less than 1 mm/year in all validation pairs and the mean of all differences was 0.56 mm/year. Based on these results, Sentinel-1 data can measure urban subsidence in a satisfactory way, when the PSI technique is applied.