Regarding landslide inventory mapping (LIM) as a task similar to change detection, current methods for LIM using bi-temporal optical remote sensing images are generally derived from change detection methods. In practice, not all changed regions belong to landslides, e.g., new roads, canals, and vegetation. Therefore, an ideal strategy is supposed to present two steps: discriminating changed and unchanged regions, and detecting landslides apart from other changed regions. Owing to the complexity and uncertainty of landslides, it is difficult to simultaneously separate landslides with unchanged and other changed regions by a single model. Addressing this problem, in this letter, we apply a generative adversarial network (GAN) in a Siamese neural network, and then propose a GAN-based Siamese framework (GSF) for LIM. The GSF comprises two cascaded modules, namely, domain adaptation and landslide detection. The former module aims to make a cross-domain mapping between prelandslide and postlandslide images with adversarial learning, then translate paired images into the same domain to suppress the domain discrepancies of bi-temporal remote sensing images. Meanwhile, the latter module aims to perform pixel-level landslide detection with a Siamese model. By training this cascaded framework, our method learns to produce landslide inventory maps without any preprocessing or postprocessing. Extensive experiments and comparison with other state-of-the-art methods verify the efficiency and superiority of our method.

On 4 and 6 July 2019, an M_w 6.4 foreshock and an M_w 7.1 main shock successively struck the city of Ridgecrest in eastern California. These two events are the most significant earthquake sequences to strike in this part of California for the past 20 yr. We used both continuous global positioning system (GPS) measurements and interferometric synthetic aperture radar (InSAR) images taken by the Sentinel-1 and ALOS-2 satellites in four different viewing geometries to fully map the coseismic surface displacements associated with these two earthquakes. Using these GPS and InSAR measurements both separately and jointly, we inverted data to find the coseismic slip distributions and fault dips caused by the two earthquakes. The GPS-constrained slip models indicate that the M_w 7.1 main shock was predominately controlled by right-lateral motions on a series of northwest-trending faults, while the M_w 6.4 foreshock involved both right-lateral slipping on a northwest-trending fault and left-lateral slipping on a northeast-trending fault. The two earthquakes both generate significant surface slip, with the maximum inferred slip of 5.54 m at the surface. We estimate the cumulative geodetic moment of the two earthquakes to have been 4.93 × 10¹⁹ Nm, equivalent to M_w 7.1. Furthermore, our calculations of the changes in static Coulomb stress suggest that the M_w 7.1 main shock was promoted significantly by the M_w 6.4 foreshock. This latest Ridgecrest earthquake sequence ruptured only the northern part of the seismic gap between the 1992 M_w 7.3 Landers earthquake and the 1872 M 7.4–7.9 Owens Valley earthquake. The earthquake risk in this area, therefore, remains very high, considering the significant accumulation of strain in the Eastern California Shear Zone, especially in the southern part of the seismic gap.