Search

Regeneration after landslides is how an ecosystem recovers itself following a landslide event. Natural systems can recover and regenerate over time. The main objective of the current work is to identify and characterise the regeneration pattern after a great landslide event in 2013 in the Guerrero state (Mexico) using remote sensing, geographic information system (GIS), and machine learning techniques. Remarkably, the authors consider normalised difference vegetation index (NDVI) as a proxy of part of this "regeneration" in the present work. A first methodology attempt presented here was to monitor and characterise the "regeneration" after severe landslide events. First, the authors calculated and identified the (1) losses, (2) speed and (3) recovery time through a continuous change detection and classification algorithm (CCDC) in Google Earth Engine (GEE) from a NDVI-Landsat time series (from 1984 to 2021). Second, these three factors were introduced as variables in a not supervised machine-learning model to get 5 clusters that characterise the different regeneration patterns followed from 2013 to 2021 in the landslide zones. Finally, we studied 16 variables like elevation, slope, or soil pH. The authors included those variables in a supervised machine learning classification to find the most important main drivers related to the NDVI regeneration of the landslide areas. The results showed that the cluster called Group 0 had a low loss of NDVI value, speed and medium recovery time. Group 1 showed a medium–high loss, high speed and low recovery time. Group 2 reached a high loss, a low speed, and an increased recovery time. Group 3 with a medium loss, a medium speed, and a low recovery time. Finally, Group 4 with a high loss of NDVI, a medium speed, and a medium recovery time. Distance to the edge of the landslide, precipitation, land cover type and lithology are highlighted as the main drivers in landslide regeneration. [ABSTRACT FROM AUTHOR]

Landslides are recognized as high-impact natural hazards in different regions around the world; therefore, they are extensively researched by experts. Landslide inventories are essential to identify areas that are likely to be affected in the future, thereby enabling interventions to prevent loss of life. Today, through combined approaches, such as remote sensing and machine learning techniques, it is possible to apply algorithms that use data derived from satellite images to produce landslide inventories. This work presents the performance of five machine learning methods—k-nearest neighbor (KNN), stochastic gradient descendent (SGD), support vector machine radial basis function (SVM RBF Kernel), support vector machine (SVM linear kernel), and AdaBoost—in landslide detection in a zone of the state of Guerrero in southern Mexico, using continuous change maps and primary landslide factors, such as slope angle, terrain orientation (aspect), and lithology, as inputs. The models were trained with 2/3 of ground truth samples of 671 slidden/non-slidden polygons. The obtained inventory maps were evaluated with the remaining 1/3 of ground truth samples by generating a confusion matrix and applying the Kappa concordance coefficient, accuracy, precision, recall, and F1 score as evaluation metrics, as well as omission and commission errors. According to the results, the AdaBoost classifier reached greater spatial and statistical coherence than the other implemented methods. The best input layer combination for detection was the continuous change maps obtained by the linear regression and image differencing detection methods, together with the slope angle, aspect, and lithology conditioning factors. [ABSTRACT FROM AUTHOR]