View in EDS

Mapping Soil Erodibility Using Machine Learning and Remote Sensing Data Fusion in the Northern Adana Region, Türkiye.

Saved in:
Bibliographic Details
Title: Mapping Soil Erodibility Using Machine Learning and Remote Sensing Data Fusion in the Northern Adana Region, Türkiye.
Authors: Işik, Melek, Işik, Mehmet, Acar, Mert, Wahab, Taofeek Samuel, Koca, Yakup Kenan, Şahin, Cenk
Source: Agronomy; Feb2026, Vol. 16 Issue 3, p294, 21p
Subject Terms: SOIL erosion prediction, MACHINE learning, GEOGRAPHIC information systems, SOIL erosion, REMOTE sensing, SOILS, BOOSTING algorithms, SOIL conservation
Geographic Terms: TURKEY
Abstract: Soil erosion is a major threat to the sustainable productivity of arable lands, making the accurate prediction of soil erodibility essential for effective soil conservation planning. Soil erodibility is strongly controlled by intrinsic soil properties that regulate aggregate resistance and detachment processes under erosive forces. In this study, machine learning (ML) models, including the Multi-layer Perceptron Regressor (MLP), Random Forest (RF), Decision Tree (DT), and Extreme Gradient Boosting (XGBoost), were applied to predict the soil erodibility factor (K-factor). A comprehensive set of soil properties, including soil texture, clay ratio (CR), organic matter (OM), aggregate stability (AS), mean weight diameter (MWD), dispersion ratio (DR), modified clay ratio (MCR), and critical level of organic matter (CLOM), was analyzed using 110 soil samples collected from the northern part of Adana Province, Türkiye. The observed K-factor was calculated using the RUSLE equation, and ML-based predictions were spatially mapped using Geographic Information Systems (GISs). The mean K-factor values for arable, forest, and horticultural land uses were 0.065, 0.071, and 0.109 t h MJ−1 mm−1, respectively. Among the tested models, XGBoost showed the best predictive performance, with the lowest MAE (0.0051) and RMSE (0.0110) and the highest R2 (0.9458). Furthermore, the XGBoost algorithm identified the CR as the most influential variable, closely followed by clay and MCR content. These results highlight the potential of ML-based approaches to support erosion risk assessment and soil management strategies at the regional scale. [ABSTRACT FROM AUTHOR]
Copyright of Agronomy is the property of MDPI and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Complementary Index
Description
Abstract:Soil erosion is a major threat to the sustainable productivity of arable lands, making the accurate prediction of soil erodibility essential for effective soil conservation planning. Soil erodibility is strongly controlled by intrinsic soil properties that regulate aggregate resistance and detachment processes under erosive forces. In this study, machine learning (ML) models, including the Multi-layer Perceptron Regressor (MLP), Random Forest (RF), Decision Tree (DT), and Extreme Gradient Boosting (XGBoost), were applied to predict the soil erodibility factor (K-factor). A comprehensive set of soil properties, including soil texture, clay ratio (CR), organic matter (OM), aggregate stability (AS), mean weight diameter (MWD), dispersion ratio (DR), modified clay ratio (MCR), and critical level of organic matter (CLOM), was analyzed using 110 soil samples collected from the northern part of Adana Province, Türkiye. The observed K-factor was calculated using the RUSLE equation, and ML-based predictions were spatially mapped using Geographic Information Systems (GISs). The mean K-factor values for arable, forest, and horticultural land uses were 0.065, 0.071, and 0.109 t h MJ<sup>−1</sup> mm<sup>−1</sup>, respectively. Among the tested models, XGBoost showed the best predictive performance, with the lowest MAE (0.0051) and RMSE (0.0110) and the highest R<sup>2</sup> (0.9458). Furthermore, the XGBoost algorithm identified the CR as the most influential variable, closely followed by clay and MCR content. These results highlight the potential of ML-based approaches to support erosion risk assessment and soil management strategies at the regional scale. [ABSTRACT FROM AUTHOR]
ISSN:20734395
DOI:10.3390/agronomy16030294