Characterization and removal of shadow bias from Optical Image Correlation; application to the 2013 Balouchistan earthquake
James Hollingsworth  1@  , Francois Ayoub  2@  , Marie-Pierre Doin, Simon Daout  3@  , Hugo Perfettini  4@  , Gilles Peltzer  5, 6@  , Sergey Samsonov  7@  
1 : Institut des Sciences de la Terre
Université Joseph Fourier - Grenoble 1, Institut français des sciences et technologies des transports, de l'aménagement et des réseaux, Institut national des sciences de l\'Univers, Institut de recherche pour le développement [IRD] : UR219, PRES Université de Grenoble, Université Savoie Mont Blanc, Centre National de la Recherche Scientifique : UMR5275, Université Grenoble Alpes, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers, Institut national des sciences de l\'Univers
BP 53 - 38041 Grenoble cedex 9 -  France
2 : Jet Propulsion Laboratory
4800 Oak Grove Drive, Pasadena, CA 91109-8099, USA -  United States
3 : Christian-Albrecht University zu Kiel, Department of Geosciences
Institute of Geosciences Otto-Hahn-Platz 1, 24118 Kiel -  Germany
4 : IRD/ISTerre  (IRD)  -  Website
Institut de Recherche pour le Développement, Institut de recherche pour le développement [IRD] : UR219
Adresse du siège - Le Sextant 44, bd de Dunkerque, CS 90009 13572 Marseille cedex 02 -  France
5 : Jet Propulsion Laboratory, California Institute of Technology  (JPL)
Pasadena, CA 91109 -  United States
6 : Earth and Space Science Department, University of California Los Angeles  (UCLA)
595 Charles Young Drive, Los Angeles, CA 90095 -  United States
7 : Natural Resources Canada

In recent years, optical image correlation (OIC) has become a powerful geodetic tool (in compliment to InSAR) for the retrieval of near-field earthquake ground deformations using optical satellite data. However, because optical satellite images are acquired throughout the year, the changing illumination conditions at the time of each acquisition may be expected to subtly influence the shadow content of each pixel, which in turn may bias the correlation process, resulting in contaminated displacement maps. We use the Landsat8 archive of satellite images acquired several years after the 2013 Balouchistan earthquake (Mw 7.8; left-lateral strike-slip with thrust) to investigate the influence of seasonal variations in sun illumination on the resulting deformation field retrieved with OIC. We find the displacement field varies seasonally as a function of the difference in pre- and post-image sun elevation. The amplitude of the seasonal artifacts correlate strongly with topography (slope angle), and are greatest in the north-south displacement component, which is subject to the largest variation in shadow movement. By correlating different combinations of images, we develop functions for each pixel in the correlation describing how the seasonal displacement varies as a function of the pre- and post-image sun elevation. We then remove seasonal artifacts from various correlations derived from Landsat8 combinations spanning the co- and post-seismic periods. Finally, we invert our illumination-corrected correlations using the NSBAS processing chain, which takes advantage of the large data redundancy to produce a more robust deformation time-series. These two approaches allow us to significantly increase the precision of OIC, such that we can retrieve small (


e
Online user: 1