Machine Learning Approaches to Soil Erosion Risk Mapping: A Comparison between Logistic Regression and Fast Large Margin
Journal of the Civil Engineering Forum · 18.04.2026 · 264-276
DOI: 10.22146/jcef.24796 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of the Civil Engineering Forum, online erschienen am 18.04.2026. Verfasst von Muhammad Ramdhan Olii und 4 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-SA 4.0).
Abstract (Originalsprache)
Soil erosion is a critical environmental issue that accelerates land degradation, reduces agricultural productivity, and increases sedimentation in water bodies. Despite its importance, spatial prediction of erosion risk remains a challenge due to the complex interaction of topographic and vegetation-related factors. Previous studies have often overlooked the integration of topographic and remote sensing indices into advanced predictive models, thereby limiting the accuracy of erosion risk mapping. This study aims to evaluate the spatial distribution of soil erosion risk in the Tamalate Watershed, Gorontalo Province, Indonesia, by integrating topographic and remote sensing conditioning factors into Logistic Regression (LR) and Fast Large Margin (FLM) models. Eight conditioning factors—Normalized Difference Moisture Index (NDMI), Terrain Ruggedness Index (TRI), Stream Power Index (STI), Soil Adjusted Vegetation Index (SAVI), Normalized Difference Tillage Index (NDTI), Topographic Wetness Index (TWI), Sediment Power Index (SPI), and Vegetation Condition Index (VCI)—were analyzed using multicollinearity diagnostics and weighted scoring to quantify their relative importance. The results revealed that NDMI (0.253 in LR; 0.258 in FLM) and TRI (0.193 in LR; 0.244 in FLM) were the most influential factors controlling erosion risk, followed by STI (0.186 in LR; 0.166 in FLM). Spatially, both models classified most of the watershed into moderate risk (44.26% in LR; 48.31% in FLM) and high risk (26.09% in LR; 22.35% in FLM) categories, while very high-risk areas were minimal (
Zitieren
APA
Olii, M. R., Mokarram, M., Anshari, E., Olii, R. S. N., & Pakaya, R. (2026). Machine Learning Approaches to Soil Erosion Risk Mapping: A Comparison between Logistic Regression and Fast Large Margin. Journal of the Civil Engineering Forum, 264-276. https://doi.org/10.22146/jcef.24796
Vancouver
Olii MR, Mokarram M, Anshari E, Olii RSN, Pakaya R. Machine Learning Approaches to Soil Erosion Risk Mapping: A Comparison between Logistic Regression and Fast Large Margin. Journal of the Civil Engineering Forum. 2026:264-276. doi:10.22146/jcef.24796
BibTeX
@article{Olii2026,
author = {Olii, Muhammad Ramdhan and Mokarram, Marzieh and Anshari, Erwin and Olii, Rizky Selly Nazarina and Pakaya, Ririn},
title = {Machine Learning Approaches to Soil Erosion Risk Mapping: A Comparison between Logistic Regression and Fast Large Margin},
journal = {Journal of the Civil Engineering Forum},
year = {2026},
pages = {264--276},
doi = {10.22146/jcef.24796}
}
EUVIMED