Landslide Susceptibility Mapping Using Novel Hybrid Model Based on Different Mapping Units

Landslide is a most widespread geohazards around the world. Reasonable landslide susceptibility mapping can aid decision-makers in landslide prevention. Therefore, drawing regional landslide susceptibility map is of great significance to landslide prevention. This research mainly aims to explore a new method and carry out the landslide susceptibility mapping based on terrain mapping unit (TMU) and grid cells mapping unit (GCMU) in Wuqi County, Yan’an City, China. Firstly, the landslide inventory map was prepared based on 717 landslides that were extracted. Secondly, the index of entropy model (IOE) was applied to quantify landslide predisposing factors based on TMU and GCMU, respectively. Finally, the systematically developed forest of multiple trees (SysFor) was integrated with IOE to construct a novel hybrid model (IOE-SysFor) and the landslide susceptibility maps based on TMU and GCMU were obtained. Statistical indices and receiver operating characteristic curve (ROC) were applied to evaluate the results and compare the approaches. From the results, it indicated that the accuracy of landslide susceptibility maps generated by the IOE-SysFor based on TMU and GCMU was satisfactory, furthermore, the performance of IOE-SysFor model running on the TMU was better than GCMU. Therefore, the TMU was considered as a more suitable landslide mapping unit for similar regions and the approach developed by this study can provide a reference for future researches.