From prediction to ecological insight: exploring soil erodibility through integrated spatial modelling
Uloženo v:
| Název: | From prediction to ecological insight: exploring soil erodibility through integrated spatial modelling |
|---|---|
| Autoři: | Perović, Veljko, Čakmak, Dragan, Pavlović, Dragana, Matić, Marija, Kostić, Olga, Mitrović, Miroslava, Pavlović, Pavle |
| Zdroj: | Landscape Ecology |
| Informace o vydavateli: | Springer Science and Business Media LLC, 2025. |
| Rok vydání: | 2025 |
| Témata: | Soil erodibility, SHAP, SEM, LIME, Environmental variables, XGBoost |
| Popis: | Soil erosion is a key process of landscape degradation that affects ecosystem stability and spatial functionality. By demonstrating how environmental variables influence soil erodibility in heterogeneous areas, this study improves understanding at the landscape scale, which is critical for sustainable land management. The objectives were (1) to develop an integrated and explainable modelling framework for predicting soil erodibility and (2) to disentangle the relative and interactive effects of climatic, topographic and vegetation variables using interpretable machine learning and structural models. The study was conducted in a geomorphologically diverse region in Serbia. Environmental variables representing climate, topography and vegetation were used to assess their influence on soil erodibility. Predictive analysis was combined with local interpretability and structural assessment techniques to capture the spatial relevance and underlying relationships. Climatic variables were the most influential determinants of soil erodibility, with hydrological and thermal regimes showing dominant effects. Topographic predictors, particularly elevation and valley depth, showed moderately positive associations, while vegetation indices had limited explanatory power. The analysis revealed threshold responses and spatial heterogeneity, suggesting non-linear, context-dependent dynamics. Climate and vegetation were negatively associated with erodibility, while topographic complexity made a positive contribution. This study presents an integrated framework that combines predictive modelling with ecological interpretation to explain soil erodibility in complex terrain. The approach improves modelling transparency and identifies key environmental interactions that influence erosion risk. In line with soil protection policy, the framework supports spatially informed mitigation planning and provides a transferable tool for landscape-level decision making. |
| Druh dokumentu: | Article |
| Jazyk: | English |
| ISSN: | 1572-9761 |
| DOI: | 10.1007/s10980-025-02172-3 |
| Přístupová URL adresa: | https://radar.ibiss.bg.ac.rs/handle/123456789/7661 https://radar.ibiss.bg.ac.rs/bitstream/id/20815/s10980-025-02172-3.pdf |
| Rights: | CC BY NC ND |
| Přístupové číslo: | edsair.doi.dedup.....b40c208b3ceeca35af99215f1036e716 |
| Databáze: | OpenAIRE |
Full Text Finder 
Nájsť tento článok vo Web of Science | Abstrakt: | Soil erosion is a key process of landscape degradation that affects ecosystem stability and spatial functionality. By demonstrating how environmental variables influence soil erodibility in heterogeneous areas, this study improves understanding at the landscape scale, which is critical for sustainable land management. The objectives were (1) to develop an integrated and explainable modelling framework for predicting soil erodibility and (2) to disentangle the relative and interactive effects of climatic, topographic and vegetation variables using interpretable machine learning and structural models. The study was conducted in a geomorphologically diverse region in Serbia. Environmental variables representing climate, topography and vegetation were used to assess their influence on soil erodibility. Predictive analysis was combined with local interpretability and structural assessment techniques to capture the spatial relevance and underlying relationships. Climatic variables were the most influential determinants of soil erodibility, with hydrological and thermal regimes showing dominant effects. Topographic predictors, particularly elevation and valley depth, showed moderately positive associations, while vegetation indices had limited explanatory power. The analysis revealed threshold responses and spatial heterogeneity, suggesting non-linear, context-dependent dynamics. Climate and vegetation were negatively associated with erodibility, while topographic complexity made a positive contribution. This study presents an integrated framework that combines predictive modelling with ecological interpretation to explain soil erodibility in complex terrain. The approach improves modelling transparency and identifies key environmental interactions that influence erosion risk. In line with soil protection policy, the framework supports spatially informed mitigation planning and provides a transferable tool for landscape-level decision making. |
|---|---|
| ISSN: | 15729761 |
| DOI: | 10.1007/s10980-025-02172-3 |