Day 4

Detailed paper information

Back to list

Paper title A model-backfeed scheme to optimize InSAR deformation time series estimation
Authors
  1. Bin Zhang Faculty of Geo-Information Science and Earth Observation (ITC), University of Twente Speaker
  2. Ling Chang Faculty of Geo-Information Science and Earth Observation (ITC), University of Twente
  3. Alfred Stein ITC, University of Twente
Form of presentation Poster
Topics
  • D1. Managing Risks
    • D1.01 Satellite EO for Geohazard Risks
Abstract text A model-backfeed scheme to optimize InSAR deformation time series estimation
Bin Zhang, Ling Chang, Alfred Stein
Faculty of Geo-Information Science and Earth Observation (ITC), University of Twente, Hengelosestraat 99, 7514AE Enschede, The Netherlands
InSAR deformation time series estimation is highly dependent on the outcome of the spatio-temporal phase unwrapping and the correctness of the pre-defined deformation time series model. When acknowledging temporal smoothness, a linear function of time can be assumed to hold for deformation time series modeling, which can facilitate phase unwrapping. This assumption is suited for the Constantly Coherent Scatterers (CCS) that have a strictly linear behavior over time. Using such a simple linear model, however, we may over- or under- estimate deformation parameters, such as the CCS deformation velocity that shows nonlinear behavior. To address this issue, we designed a new scheme that optimizes deformation time series estimation. It iteratively re-introduces the best deformation models of every CCS, as determined by Multiple Hypothesis Testing (MHT), into phase unwrapping. It includes both linear and nonlinear canonical functions. We name our new scheme a model-backfeed (MBF) scheme.
The MBF starts with post InSAR deformation time series modeling. The InSAR deformation time series is generated using a standard time series InSAR method, such as Persistent Scattering Interferometry (PSI). Once a number of potential nonlinear canonical functions was built as an extension to the linear function, we applied MHT to determine the best deformation model. The variance-covariance matrix of deformation estimators was obtained at every CCS. Next, we iteratively replaced the simple linear model with this model during phase unwrapping, and estimated the deformation parameters.
We illustrated our method with a study on surface subsidence of the Groningen gas field in the Netherlands between 1995 and 2020, using 32 ERS-1/2, 68 Envisat, 82 Radarsat-2, and 13 ALOS-2 images. The results show that the cumulative maximum surface subsidence has been up to 25 cm over the past 25 years in response to local oil/gas extraction activities [1]. They also show nonlinear behavior of some CCS. Taking two quality indicators, we showed that the values of the ensemble coherence increased by 10 – 33% and the values of the spatio-temporal consistency for MBF decreased by 2%-20% as compared with a standard InSAR time series analysis.
We conclude that the model-backfeed scheme can mitigate phase unwrapping errors. It can also obtain better phase unwrapping parameters than the standard InSAR time series method.
[1] Zhang, B., Chang, L., & Stein, A. (2021). A model-backfeed deformation estimation method for revealing 25-year surface dynamics of the Groningen gas field using multi-platform SAR imagery. (Under review).