Shallow landslide mapping using freely accessible images: a case study in the Ribeira Valley, Brazil

Shallow landslide mapping is an important procedure for landslide assessment and is the first step to susceptibility, vulnerability, and risk analysis. Knowing the exact location of occurrence of this kind of natural hazard makes it possible to trace spatio-temporal patters and evaluate topographic influences. Landslides are very frequent along the Brazilian south and southeastern coast, where mass movements are triggered by heavy rainfall almost every year in the summer season (Dec-Mar), causing harm to society, such as the destruction of buildings, other infrastructure, and economic and human losses. Landslide recognition and mapping are poorly developed in Brazil, since no mapping guidelines exist, as well as due to low investments in mass movement prevention and mitigation actions. Thus, this research aimed to evaluate the use of freely accessible Google Earth Pro images for shallow landslides recognition and mapping. The study area is located in Itaóca and Apiaí counties, São Paulo state, in Ribeira Valley region, Brazil. Itaóca and Apiaí were affected by mass movements in January 2014, resulting in several economic and infrastructure damages, and 25 fatalities. The most recent post-event images available in Google Earth Pro were used, dated as of 08/10/2014. The visual criteria for landslide scars recognition and mapping were the absence of vegetation, shape and size, drainage network distance, slope position, planar rupture surface, and altimetric variation. As a reference for manual mapping contour and hydrography curves of 1:10.000 scale from the Geographic and Cartographic Institute of the State of São Paulo (for areas belonging to the municipality of Itaóca) and contour and hydrography curves of 1: 50.000 scale from the Brazilian Institute of Geography and Statistics (for the sectors belonging to the municipality of Apiaí) were used. The results showed that Google Earth Pro images are suitable for landslide recognition and mapping in a tropical environment. A total of 1,850 shallow landslides scars from the 2014 event with different sizes were mapped, where the smallest has 14 m² and the largest 9,539 m². They occurred under different morphological and lithological conditions, where most landslides are concentrated at slopes between 20 and 30°, south and southeast orientation, elevations of 600 to 800 m, concave curvatures, and in Quartz-Monzonite and Biotite Monzogranite rocks. The advantage of Google Earth images is that they are very high resolution data and free to access and use for everybody. However, the periods available on the software are limited. The event occurred in January of 2014 but it was only possible to access study area images of October of 2014, nine months after the event. In this way, it is important to verify if the mapping process is influenced by environmental changes, for example, vegetation recovery, that may cause interference for the visual interpretation. The inventory can be used as a basis for further analysis, such as for creating susceptibility and hotspot maps. Such products help to better understand shallow landslide dynamics in the study area, allowing comparison with other environments, and can support spatial planning and decision making of government authorities.