Geostatistical modeling of susceptibility to landslides in the district of Cajamarca

Resumen

The geostatistical modelling of landslide susceptibility aimed to analyze the initiation of landslides and the areas potentially affected by their propagation. Landslide susceptibility depends on intrinsic factors, which were modelled using the historical landslide inventory from the southern sector of the Cajamarca district. The stochastic methodology applied allowed the generation of a landslide susceptibility zoning map through the combination of the following variables: slope, geology, geological faults, and road cuts, factors considered essential for this study. The results were computed using Excel and ArcGIS 10.5, yielding the weighting of each factor through the geostatistical kriging technique. The corresponding weights obtained were 0.05, 0.08, 0.24, and 0.13 for distance to road, distance to geological fault, slope, and geology, respectively. These results were then used to estimate the susceptibility values. Subsequently, variograms were developed and calculated in the 0°, 45°, 90°, 135°, and 70° directions, applying theoretical variogram models, nugget effect, spherical, exponential, and Gaussian to determine the anisotropy. The study concludes with the development of the landslide susceptibility map using ordinary kriging, establishing that slope and geology are the most influential intrinsic factors. Moreover, the study area exhibits a structural behavior oriented at 70° and best fits the spherical variogram model.