Some InSAR data considerations when studying Earthquakes

Sigurjon Jonsson(1)

(1) ETH Zurich, ETH Hoenggerberg (HPP), 8093 Zurich, Switzerland

Abstract

We can study sources of large earthquakes by estimating fault-model parameters from InSAR deformation data. These parameters include fault location and depth, fault geometry as well as variations in slip magnitude and direction on the fault plane. The large number of studies that have already been published on this topic may seem to indicate that fault parameter estimations have become a simple routine task. However, there are still many outstanding problems in the way InSAR data are post-processed and treated in these estimations. These problems include data error characterization, phase unwrapping, data sub-sampling, estimation of model-parameter uncertainties, etc. Here we discuss some of these problems and possible methods to improve model parameter estimations.

Unwanted signals or errors in InSAR deformation data come from various sources, such as from inaccurate orbit information, poor DEMs, unwrapping mistakes, interferometric decorrelation, and water vapor inhomogeneities in the atmosphere. The effect of some errors can be reduced by careful post-processing analysis while other errors remain in the data. Atmospheric errors are particularly difficult as they vary between interferograms and also within individual interferograms. These errors are spatially correlated with increasing noise power at larger spatial scales, but most researchers simply ignore these correlations, leading to biased model-parameter estimates, especially when the signal-to-noise ratio is low. We present results on under what circumstances the inclusion of the full data covariance matrix is important, as well as describing some practical methods on characterizing atmospheric errors.

Another important issue is InSAR data sub-sampling. Interferograms contain millions of data points that make inversion calculations impractical. Several sub-sampling methods have been tried, such as regular sub-sampling, circular, quadtree, resolution based sub-sampling, etc. Here we discuss several schemes and compare them in for simple earthquake model parameter estimations. In conclusion, it is clear that our understanding of how to treat InSAR data in source inversions has improved during the last few years, although more research is still needed to better identify the methods that lead to the most reliable results.

 

Workshop presentation

 

  Higher level                 Last modified: 07.10.03