Improvement of fault model estimations by using multiple datasets and full data covariances: An application to the Kleifarvatn earthquake, Iceland.
Henriette Sudhaus(1) and Sigurjon Jonsson(1)
(1) ETH Zurich, Schafmattstr.30, 8093 Zurich, Switzerland
We take the moderate-sized Kleifarvatn earthquake, which occurred on 17 June 2000 on the Reykjanes Peninsula, Iceland, as an example to demonstrate the strength of combining independent geodetic datasets to suppress model parameter trade-offs and to show the benefit of accounting for correlated errors in InSAR data in earthquake source modelling.
Despite the Kleifarvatn earthquake fault has already been modelled twice using geodetic data (descending InSAR and descending InSAR plus GPS), we investigated the surface deformation in the epicentral region again to improve previous source models. We suspected that existing models were suffering from model parameter trade-offs and saw a high potential for an improvement by additionally using ascending InSAR data together with the GPS and descending InSAR observations.
The basic assumptions about the fault model are the same in all three studies, i.e., that a uniform slip took place on a simple, planar and rectangular fault embedded in an elastic half-space. The data processing steps and the algorithms used for the non-linear model parameter optimization are basically the same as well. The only important difference, apart from the amount of input data, is that we analyzed the characteristics of the spatially correlated error within the InSAR images and weighted the data accordingly in the optimization. From the results of the error analysis we also produced multiple modified data-sets by adding synthetic realizations of data noise to the original data. The results of many optimization runs using these modified input data provide a distribution of source models which outline the confidence intervals of the optimum source model parameters. The estimated confidence intervals show that most, but not all, model parameters of the two previous models fall within the 95% confidence.
This is the first time such confidence intervals are estimated for the Kleifarvatn earthquake. However, due to the overlap of the used data and methods we can approximately reproduce earlier optimizations and estimate confidence intervals for the two older source models as well.
From these results we show how the inclusion of GPS or the different InSAR data decreases the source parameter trade-offs and uncertainties.
Furthermore, we compare the effect of the different approaches in data weighting on the resulting source models, i.e., by using weights based on the noise variance only or by using weights that consider the correlation of data errors.
Keywords: ESA European
Space Agency - Agence spatiale europeenne,
observation de la terre, earth observation,
satellite remote sensing,
teledetection, geophysique, altimetrie, radar,
chimique atmospherique, geophysics, altimetry, radar,