How well can displacement be resolved close to earthquake surface ruptures using optical image correlation?

The study of natural hazards like earthquakes requires accurate measurement of ground displacement. When paired with high-resolution satellite imagery, Optical Image Correlation (OIC) has proven to be highly effective in mapping near-field ground displacements for large earthquakes, offering detailed and precise data. This is crucial for understanding fault mechanics and the generation of strong ground motions during shallow earthquakes.

OIC has several advantages over field or traditional geodetic methods. First, it is robust against image noise, allowing meaningful data extraction from various types of imagery, even when separated over long time periods. Second, unlike InSAR, OIC does not suffer from decorrelation close to fault ruptures, thus providing rich data in the near-field region and offering insight into shallow fault characteristics. Third, OIC has subpixel resolution, enabling the detection of small (cm-scale) displacements. Fourth, OIC provides dense displacement measurements that would be difficult to replicate with field methods. Finally, OIC can help to identify subtle ground features and long-wavelength displacement signals, including those associated with off-fault deformation. OIC has been widely used to characterize near-field displacements during several recent surface-rupturing earthquakes. Displacements measured by OIC typically surpass field measurements due to the latter's inability to capture smaller, distributed deformations away from the primary fault rupture. OIC data can thus help us to more accurately infer the width of the fault zone, encompassing both on-fault and off-fault deformation. 

Studies on the 2019 Ridgecrest earthquake used various optical datasets and correlation methods to explore near-field displacement and the extent of off-fault deformation. However, the choice of correlation approach used can impact the magnitude and nature of the observed deformation, which, in turn, may impact subsequent analysis of the strain field. 

This study aims to analyze multiple correlation algorithms (MicMac, COSI-Corr, Ames Stereo Pipeline and AmpCor) and optical datasets (Pleiades, WorldView, Spot and ADS80), spanning a range of resolutions, incidence angles, and temporal variations. We explore how correlation techniques influence displacement values and whether they can artificially smooth discrete fault offsets, creating apparent (artificial) off-fault deformation. Using synthetic tests and the 2019 Ridgecrest earthquakes as a case study, we explore the variability in off-fault deformation and fault zone width, depending on the processing approach adopted. Ultimately, we highlight the limitations of OIC in quantifying off-fault deformation, thus providing constraints on the extent to which such data can be used to address aspects of fault mechanics.