Paul ARIAS, Universidad Técnica del Norte, Ecuador
Miguel Angel SAZ , Universidad de Zaragoza, Spain
Severino ESCOLANO, Universidad de Zaragoza, Spain
Land use and land cover (LULC) is one of the main dynamic factors in landslide triggering. The research aimed to determine the effects of land-use change on landslide susceptibility for the past and future scenarios. The study area selected was a basin in the Andean Region of Ecuador, the Mira basin. For this purpose, land cover maps were developed for the years 1996, 2018. Also, landslide inventory was developed for the 2018 year. Two future scenarios were created for the years 2030 and 2040. Multi-Layer Perceptron Neural Network built-in TerrSet software was implemented to forecast LULC. The landslide susceptibility analysis was performed using the Frequency ratio analysis, researching changes to susceptibility over time. The effects of land-use change in landslide susceptibility were determined using geographically weighted regression (GWR). The results show an environmental dynamic because land-use change influences landslide susceptibility in time. There is an increase of susceptibility in the past (1996-2018) and in the future for 2030 and 2040 scenarios. This is due to the decrease in forest and pasture. In all scenarios analyzed, an adjusted R-squared value greater than 0.52 was obtained. The results suggest an effect of change of forest and pasture over susceptibility. This demonstrates that there is no actual landscape management that might contribute to the reduction in landslide susceptibility to the future.
Mots clés : Land-use|future-scenarios| landslide|susceptibility|Frequency ratio
A105057AP