LANDSLIDE SUSCEPTIBILITY ZONATION OF TAKENGON AND SURROUNDING AREAS USING A COMBINATION WEIGHT OF EVIDENCE (WOE) AND LOGISTIC REGRESSION (LR) METHOD

• Main Author: PAMELA
• Format: Theses
• Language: Indonesia
• Subjects: Geologi, hidrologi & meteorologi
• Online Access: https://digilib.itb.ac.id/gdl/view/29899
• Institution: Institut Teknologi Bandung

<p align="justify">Takengon is an area prone to earthquake and landslide disaster. Earthquake induced large numbers of landslide has occurred in this area, contributing to significant casualties. On July 2, 2013, Central Aceh earthquake has induced large numbers of landslides in Takengon area, which resulted in casualities of 39 people. Most of the landslides occur along the Bireuen-Takengon road in the vicinity of Gayo highlands area. This location was chosen to assess the landslide susceptibility of Takengon, using a statistical method, referred to as the weight of evidence (WoE), logistic regression (LR) and combined LR-WoE. This combined method proposes, taking advantage of their individual merits and overcoming their limitation. The 251 landslides were randomly divided into two groups of modelling/training data sets (70%) and validation/test data sets (30%). Twelve thematic maps of evidence are slope degree, slope aspect, lithology, land cover, elevation, rainfall, lineament, peak ground acceleration, curvature, flow direction, distance to river and roads were used as landslide causative factors. The results showed that the LR-WoE method had a better accuracy than the WoE and LR methods, with values of 0,890 success rate and 0.853 predictive, higher than that of the WoE method (0,880 success and 0,830 prediction) and LR method (0,843 success and 0,79 prediction). Based on Seed Cell Area Index (SCAI) analysis, the zonation maps generated from LR-WoE method analysis for very high have more precise results than the other two methods. Based on the analysis of spatial domains the LR-WoE method have a greater pixel accuracy and no pixel errors than the WoE and LR methods. It is therefore concluded that the combined method of WoE and LR can provide a promising level of accuracy for landslide susceptibility mapping.<p align="justify">