Monitoring non-linear ground motion above underground gas storage using GNSS and PSInSAR based on Sentinel-1 data

Abstract

Several methods allow accurate measurement of terrain surface motions. Global navigation satellite systems (GNSSes) and interferometry with synthetic aperture radar (InSAR) stand out in terms of measurement accuracy among them. In principle, both methods make it possible to evaluate a three-dimensional vector of the motion of points on the terrain surface. In this work, we dealt with the evaluation of motions in the up-down (U-D) and east-west direction (E-W) over underground gas storage (UGS) from InSAR. One crucial step in breaking down PSInSAR line of sight (LOS) measurements to U-D and E-W components is getting time series derived from individual tracks to the same time frame. This is usually performed by interpolation, but we used an innovative approach: we analyzed individual time series using the Lomb-Scargle periodogram (LSP), which is suitable for periodic noisy and irregularly sampled data; we selected the most significant period, created LSP models, and used them instead of the original time series. Then, it was possible to derive time series values for any arbitrary time step. To validate the results, we installed one GNSS receiver in the Tvrdonice UGS test area to perform independent measurements. The results show a good agreement in the evaluation of motions by both methods. The correlation coefficient between horizontal components from both PSInSAR and GNSS was 0.95 in the case of the E-W component, with an RMSE of 1.75 mm; for U-D they were 0.78 and 2.35 mm, respectively. In addition to comparing the motions in the U-D and E-W directions, we also created a comparison by converting GNSS measurements to a line of sight of the Sentinel-1 satellite to evaluate the conformity of InSAR and GNSS measurements. Based on descending track, the correlation coefficient between LOS from both methods is, on average, 0.97, with an RMSE of 2.70 mm.