Earthquakes are one of the major natural disasters threatening human society, often resulting in significant casualties and substantial economic losses. Understanding the seismogenic processes and genesis mechanisms of strong earthquakes is of great significance for earthquake prevention and disaster mitigation. Accurate characterization of coseismic displacement fields plays a crucial role for revealing earthquake rupture processes and deformation patterns. Image geodesy has become an effective approach for retrieving coseismic displacement fields, with commonly used techniques including interferometric synthetic aperture radar （InSAR） and pixel offset tracking based on image matching. Among these, the InSAR technique features large spatial coverage, high precision, and all-weather, day-and-night observation capability, and is commonly used to measure slow surface deformation along the range direction. In contrast, pixel offset tracking is more suitable for estimating large-scale and high-gradient surface deformation and can obtain two-dimensional deformation fields at the sub-pixel level. According to the image source, pixel offset tracking techniques mainly include SAR-based pixel offset tracking （POT） and optical image-based optical image correlation （OIC）.

In recent years, Sentinel-1 and Sentinel-2 imagery have become commonly used remote sensing data sources for monitoring co-seismic surface deformation due to their relatively high spatio-temporal resolution. The POT method based on Sentinel-1 imagery and the OIC method based on Sentinel-2 imagery are widely used to retrieve co-seismic deformation fields and have been successfully applied in numerous earthquake studies worldwide. Although the two methods are based on similar principles and both derive deformation through image matching, differences in the imaging mechanisms of optical and SAR imagery lead to variations in the processing procedures and the resulting deformation fields. For example, POT provides range and azimuth displacements, whereas OIC yields east-west and north-south offsets. At present, detailed comparative analyses of POT and OIC for mapping co-seismic deformation fields associated with strong earthquakes remain limited. A comprehensive comparison of these two methods can provide valuable insights for accurately mapping co-seismic deformation fields, enhancing our understanding of seismogenic processes and earthquake genesis mechanisms, and supporting earthquake prevention and disaster mitigation.

To analyze and compare the characteristics of OIC and POT, several strong earthquakes with strike-slip characteristics were selected as case studies, including the January 8, 2022 Menyuan M_W6.7 earthquake and the July 2019 Ridgecrest M_W6.4 and M_W7.0 earthquakes. Based on Sentinel-1 and Sentinel-2 imagery, their coseismic deformation fields were derived using POT and OIC respectively. To evaluate the accuracy of the two methods, InSAR-derived coseismic deformation fields were calculated for the earthquake events, and corresponding GNSS deformation values were also obtained. For the subsequent accuracy assessment, the InSAR line-of-sight deformation was further converted into east-west and vertical deformation components, and three-dimensional deformation was calculated by combining the range and azimuth displacements derived from POT. Using the InSAR and GNSS measurements as references, the relative accuracies of POT and OIC were quantitatively evaluated, thereby assessing the feasibility of these two methods for strong earthquake monitoring. Furthermore, the differences between POT and OIC and the underlying causes of these discrepancies were analyzed, and recommendations for their application in strong earthquake studies were provided.

The results indicate that both POT and OIC can reliably retrieve coseismic deformation fields, and the derived deformation results are generally consistent with those obtained from InSAR and GNSS, demonstrating their feasibility for studying coseismic deformation associated with strong earthquakes. However, certain differences exist between the two approaches due to the distinct characteristics of the imagery used. Sentinel-1 SAR imagery is less affected by weather conditions; however, its side-looking radar imaging geometry may introduce geometric distortions. In contrast, Sentinel-2 optical imagery has a relatively short revisit cycle but can be significantly affected by cloud and fog, which may limit data availability. Moreover, even after error correction, deformation results derived from optical imagery may still contain attitude angle-related errors that are difficult to completely eliminate. Under unfavorable weather conditions, optical imagery is more susceptible to atmospheric effects, which may lead to increased decorrelation outliers in OIC-derived results. Consequently, under poor climatic conditions, POT generally produces fewer outliers than OIC and is therefore recommended. Furthermore, POT can retrieve both range and azimuth deformation fields, which can be further combined to derive three-dimensional deformation fields, whereas OIC mainly provides horizontal deformation. Therefore, when vertical deformation information is of particular interest, the POT method is more suitable.