Effect of Landslide Factor Combinations on the Prediction Accuracy of Landslide Susceptibility Maps in the Blue Nile Gorge of Central Ethiopia

Abstract

Database construction for landslide factors (slope, aspect, profile curvature, plan curvature, lithology, land use, distance from lineament & distance from river) and landslide inventory map is an important step in landslide susceptibility modelling. Using the frequency ratio model, the weights for each factor classes were calculated and assigned in GIS so as to add these factors and produce landslide susceptibility index maps based on mathematical combination theory. However, before combining them, their independence among each other should be ascertained. For this, the correlation matrix of logistic regression was applied and this showed that most of the correlations between factors were either absent or very insignificant suggesting that all landslide factors are independent. From a set of eight landslide factors, a total of 247 landslide susceptibility map combinations can be generated. However, for simplification, only 28 landslide susceptibility maps were chosen. Then the best landslide susceptibility map was selected based on high prediction accuracy. But, when there is similarity in the prediction accuracies of different combinations, the landslide susceptibility index difference values can be used as another selection criterion. Hence, the susceptibility map from a combination of all landslide factors except distance from river was found to be the best one. Among the 28 representative combinations, landslide susceptibility maps with the same prediction accuracy of 87.7% have been found in spite of their dissimilarity in their difference values. The combination, with a limited number of landslide factors and the highest prediction accuracy of 87.7%, was found from a combination of slope, lithology, land use and distance from lineament. In order to validate the prediction model, landslides were overlaid over the landslide susceptibility map and the number of landslides that fall into each susceptibility class was calculated. From this analysis 0.39%, 1.84%, 9.1%, 32.04% and 56.63% of the landslides fall in the very low, low, medium, high and very high landslide susceptibility classes respectively. Since 88.67% of the landslides fall in the high and very high susceptibility classes, the landslide susceptibility map can be considered reliable to predict future landslides.

The Performance of Land Use Change Causative Factor on Landslide Susceptibility Map in Upper Ujung-Loe Watersheds South Sulawesi, Indonesia

“…Moreover, success rate and predictive rate value for all methods were near to the interval of 0.02 indicating that all the methods were more reliable to a predictive landslide in the future. The proximity of success rate and predictive rate values show how the method helps in landslide prediction in the future (Meten, Prakashbhandary, & Yatabe, 2015 Figure 5. AUC of ROC of landslide susceptibility of with and without LUC causative factor using FR, and LR method; a) success rate and b) predictive rate…”

Performance of frequency ratio and logistic regression model in creating GIS based landslides susceptibility map at Lompobattang Mountain, Indonesia

“…The closeness of success rate and predictive rate values show how the logistic regression helps in landslide prediction in the future (Meten et al 2015a). The AUC curve determined by using validation dataset should be approximately equal to the AUC curve determined by using the training dataset, but it is generally lower than the success curve, because the landslide data on validating area are not used for modelling (Ngadisih et al 2013).…”

Landslide susceptibility map using certainty factor for hazard mitigation in mountainous areas of Ujung-loe watershed in South Sulawesi

“…Moreover, success rate and predictive rate value for all method were a closeness with interval 0.01 that indicates all the method more reliable to a predictive landslide in the future. The closeness of success rate and predictive rate values show how the method helps or good in landslide prediction in the future (Meten et al, 2015). Furthermore, validation with the percentage of landslide fell into LSM class high and very high, CF model with value 85.28% was a good result to predict landslide occurrence.…”