Analysis of manual and semi-automatic shallow landslides inventories and its suitability in predictive models

Abstract

Shallow landslides are a frequent kind of mass movement in Brazilian territory. The creation of shallow landslides inventory is important to conditioning factors studies. They are important to analysis of susceptibility, vulnerability and risk, besides being useful for urban planning. Thus, the general aim of this research is to demonstrate how construction of shallow landslides inventories from distinct methods and database may influence their accuracy and suitability in predictive mapping. The following procedures were adopted: (1) manual mapping of shallow landslides based on two distinct databases, Google Earth Pro images and Rapid Eye (5 m); (2) semi-automatic mapping of shallow landslides by OBIA method based on Rapid Eye (5 m); (3) comparison of manual and semi-automatic inventories; and (4) definition of susceptibility to shallow landslides through statistical analysis using the informative value of a watershed. The main product of this project will be the definition of the suitability of shallow landslides inventories from distinct methods and database, it will be possible to infer which methodology is most efficient for mapping past landslides events. (AU)