Landslide susceptibity mapping using Analytical Hierarchy Process and Fuzzy- Analytical Hierarchy Process approches: A case study in Binh Dinh province, Viet Nam

Abstract

Binh Dinh Province, Vietnam, has recently experienced frequent landslide events,
highlighting the urgent need for effective hazard assessment. This study aims to evaluate landslide
susceptibility using the Analytical Hierarchy Process (AHP) and Fuzzy Analytical Hierarchy
Process (Fuzzy-AHP) models. Ten conditioning factors were considered in both models:
elevation, slope, aspect, Topographic Wetness Index (TWI), Standardized Precipitation Index
(SPI), Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index
(NDWI), distance to roads, distance to rivers, and geological characteristics. The resulting
susceptibility maps were classified into five categories: very low, low, moderate, high, and very
high. Model validation was conducted using the Receiver Operating Characteristic (ROC) curve,
with Area Under the Curve (AUC) values exceeding 0.80, Root Mean Square Error (RMSE) values
around 0.2, and accuracy scores above 0.8 for both models—indicating excellent predictive
performance. Notably, the Fuzzy-AHP model slightly outperformed the AHP model. The analysis
revealed that approximately 15% of the area falls within high and very high susceptibility zones,
30% within the moderate zone, and the remaining areas within low or very low susceptibility
zones. These findings confirm the effectiveness and reliability of both the AHP and Fuzzy-AHP
approaches for landslide susceptibility assessment. The resulting maps provide valuable guidance
for local authorities and stakeholders in implementing disaster risk reduction strategies, early
warning systems, and sustainable land-use planning.