Pixel-Tracking on High-Resolution Imagery Spanning the Darfield, New Zealand Earthquake

Cornell University, Ithaca NY

Investigators

Abstract

This project is aimed at assessing damage to infrastructure and the potential for aftershocks immediately after destructive earthquakes such as the two that hit the region around Christchurch, New Zealand, during the past year. In this case, the close temporal association of the two events resulted in a case where we have more pre-seismic imagery from a wide range of satellite platforms before the second earthquake than is usual. We propose to use a combination of visual inspection and mapping of the images and automated cross-correlation (pixel-tracking) to characterize the coseismic and postseismic damage and deformation associated with each earthquake. We will also model the characteristics of each earthquake in order to assess the impact of stress changes on neighboring faults. Our primary goals are to 1) Quantify the evolving strain associated with the Darfield and Christchurch events, 2) Identify and map any observations of triggered slip on nearby faults, 3) Map liquefaction, landslides and detect incipient slumps along the Banks Peninsula and around Sumner and Lyttelton, and 4) Generate a slip distribution for both earthquakes that accounts for unknown aspects of the fault geometry and revise previous estimates of Coulomb stress change induced on neighboring faults. The work in this proposal will help advance the field of image-based satellite geodesy, which has many demonstrated applications of interest to NSF, including earthquake science, glaciology, groundwater management, landslide forecasting and mitigation, etc. This particular sequence of two earthquakes occurring in close (but not immediate) temporal and spatial proximity is of great interest to researchers studying the modes of interaction between earthquakes on nearby faults. The use of satellite imagery to study earthquakes at the present time is limited in several ways ? coseismic and postseismic deformation studies rely on a small number of platforms and sensor types (e.g. SPOT imagery, SAR, LANDSAT), while other, often more high-resolution observation types are typically ingested into GIS software packages and inspected visually for signs of damage (e.g., building collapse, flooding, liquefaction). Our study may open up these data types for use in constraining the magnitude of deformation as well, which would speed up the potential response time for relief workers, and aid in the intelligent deployment of instruments that can monitor adjacent sections of fault that may be stressed by the initial earthquake.