Analysis of the Impact of Mapping Units on Landslide Susceptibility: A Comparative Study of Grid Units and Slope Units

The mapping unit is a classification of land surfaces based on specific criteria, serving as the fundamental unit for spatial data extraction in landslide susceptibility analysis. In the landslide susceptibility analyses, a grid unit is frequently employed due to its ease of generation as uniform grid cells of a designated size. However, the utilization of grid units does have certain limitations. Specifically, these units often fail to accurately represent the actual site topography. Consequently, they result in an incomplete consideration of valley and drainage lines, which are critical factors in landslide occurrence. In contrast, slope units, delineated based on hydrological criteria (e.g., ridges, valleys), offer a more topographically accurate representation. This is due to the fact that they integrate spatial data and topographic factors more effectively into the analysis than grid units.

This study aims to compare the impact and performance of grid units and slope units in landslide susceptibility analysis. To this end, the study utilizes various analytical techniques to evaluate the influence of conditioning factors across these mapping units. The study area, designated as Jecheon-si, Chungcheongbuk-do, Republic of Korea, was selected to assess the impact of mapping units due to its experience with several landslides in August 2020. The analysis incorporated a range of conditioning factors, including elevation, slope aspect, slope angle, standard curvature, planar curvature, profile curvature, Specific Catchment Area (SCA), Topographic Wetness Index (TWI), Stream Power Index (SPI), forest type, forest density, forest stand height, timber diameter, timber age, soil texture, soil depth, slope shape, topography, land use, and lithology. In order to assess the significance and contribution of these factors, visualization techniques were employed, including SHAP (Shapley Additive Explanations) plots, summary plots, and dependence plots. These methods facilitated a comparative analysis of factor importance and influence on landslide susceptibility using the two mapping units. Additionally, correlation analysis among the conditioning factors and trend identification within each unit were conducted to enhance the accuracy and interpretability of the results. The landslide susceptibility analysis was implemented using a Multi-Layer Perceptron model, and the performance of the model was evaluated using the Area Under the Curve (AUC). Finally, the results of the study were analyzed to compare and evaluate the relative advantages and limitations of the slope unit and the grid unit in landslide susceptibility assessment.

 

Acknowledgments

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (RS-2023-00222563) and by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (RS-2024-00463587).