Factors affecting topographic thresholds in gully erosion occurrence and its management using predictive machine learning models

Soil degradation induced by gully erosion represents a worldwide problem in the many arid and semi-arid countries, such as Iran. This study assessed: (1) importance of variables that control using Boruta algorithm, (2) relationship among causative gullied locations evidential belief function model (EBF), (3) development algorithms boosted regression tree (BRT) support vector machine (SVM). Based on results slope, land use, lithology, plan curvature, elevation were most important factors controlling erosion. The EBF showed predominance rangeland loess-marl deposition. concave positions, with slope 5°–20° vicinity drainage lines, illustrates preferential topographic zone and, therefore, terrain threshold for gullying. correlation rangelands weak soils positions demonstrates interactions soil characteristics, topography, use stimulate low gullies development. These relationships are consistent concept given soil, climate within landscape encourage area critical surface necessary incision. Furthermore, BRF-SVM had highest efficiency lowest root mean square error, followed BRT predicting development, compared LN-SVM algorithm. application two learning methods head cut susceptibility northern Iran maps generated these could provide an appropriate strategy geo-conservation restoration efforts gullying-prone areas.