Modeling Nonlinear Soil-Vegetation Relationships in the Zouji Shoush Watershed: A Machine Learning Approach with Uncertainty Quantification

Heidary, Kohzad; Dinarvand, Mehri; Peyrov, Saba; Arami, Seyed Hossein

doi:10.22059/ije.2026.412192.1909

Modeling Nonlinear Soil-Vegetation Relationships in the Zouji Shoush Watershed: A Machine Learning Approach with Uncertainty Quantification

Document Type : Research Article

Authors

¹ Assistant Professor, Soil Conservation and Watershed Management Research Department, Khuzestan Agricultural and Natural Resources Research and Education Center, Agricultural Research Education and Extension Organization, AREEO, Ahvaz, Iran.

² Associate Professor, Forests and Rangelands Research Department, Khuzestan Agricultural and Natural Resources Research and Education Center, Agricultural Research Education and Extension Organization (AREEO), Ahvaz, Iran.

³ Assistant Professor, Forests and Rangelands Research Department, Khuzestan Agricultural and Natural Resources Research and Education Center, Agricultural Research Education and Extension Organization (AREEO), Ahvaz, Iran.

10.22059/ije.2026.412192.1909

Abstract

Research Topic: Modeling Vegetation Cover Dynamics in Arid Regions Using Deep Learning and Multi-Source Data
Objective: This study aimed to model vegetation cover percentage based on soil characteristics in the Zojy Shush watershed, Khuzestan Province, Iran. Two topographic units (hill and valley) were considered. Machine learning algorithms were employed alongside interpretability and uncertainty quantification approaches.
Method: Following sampling from 120 points, nine soil characteristics (pH, phosphorus, potassium, organic carbon, EC, lime, clay, silt, and sand) and vegetation cover percentage were measured. Three algorithms—Random Forest, XGBoost, and Gradient Boosting—were employed with cross-validation. Feature importance was assessed using Gini, permutation, and SHAP indices, while uncertainty was evaluated via the bootstrap method.
Results: In the topographically-separated model for the Valley region, Gradient Boosting achieved a median R² of 0.73, outperforming XGBoost (0.68) and Random Forest (0.14). In contrast, for the combined dataset, XGBoost showed superior performance (median R² = 0.50). In contrast, the integrated model (combining both hills and valleys) yielded negative R² values due to topographic heterogeneity. Clay (0.245), organic matter (0.156), and phosphorus (0.130) were identified as the most important predictors. Gradient Boosting provided the best uncertainty coverage (60%), with epistemic uncertainty being dominant across the models. Topography-specific modeling (separate models for hills and valleys with performance around R² ≈ 0.78) demonstrated a significant advantage over a unified model (which yielded negative R² values). Furthermore, phosphorus, pH, and silt were identified as primary limiting factors for vegetation cover in 75%, 50%, and 25% of the study area, respectively.
Conclusions: This study revealed that soil-vegetation relationships in arid regions are nonlinear and dependent on topographic heterogeneity. Phosphorus was identified as the primary limiting factor in calcareous soils. Despite a weak linear correlation, clay emerged as the most important predictor due to its threshold-based relationship with vegetation cover. The predominance of epistemic uncertainty suggests that enhancing the modeling framework is more effective than merely increasing measurement accuracy for improving predictions in such complex environments.

Keywords

Main Subjects

Water Resources Management in Natural Ecosystems

References

Baltensweiler, A., Walthert, L., Hanewinkel, M., Zimmermann, S., & Nussbaum, M. (2021). Machine learning based soil maps for a wide range of soil properties for the forested area of Switzerland. Geoderma Regional, 27, Article e00437. https://doi.org/10.1016/j.geodrs.2021.e00437.
Cao, J., Zhang, Z., Wang, C., Liu, J., & Zhang, L. (2019). Susceptibility assessment of landslides triggered by earthquakes in the Western Sichuan Plateau. Catena, 175, 63-76.
Dietze, M. C. (2017). Ecological forecasting. Princeton University Press. doi: 10.22098/mmws.2025.17919.1633.
Food and Agriculture Organization of the United Nations. (2021). The state of the world's land and water resources for food and agriculture: Systems at breaking point (SOLAW 2021). FAO. https://doi.org/10.4060/cb9910en
Forkuor, G., Hounkpatin, O. K., Welp, G., & Thiel, M. (2017). High resolution mapping of soil properties using remote sensing variables in south-western Burkina Faso: A comparison of machine learning and multiple linear regression.
Hawkins, D. M. (2004). The problem of overfitting. Journal of Chemical Information and Computer Sciences, 44(1), 1–12. https://doi.org/10.1021/ci0342472
Heidary, K., Najafinejad, A. , Mohammadian Behbahani, A. , Ownegh, M. and W. Dekker, L. (2025). Application of Machine Learning Algorithms in Analyzing the Effect of Fire on Soil Water Repellency in Tushan Watershed, Golestan Province. Watershed Management Research, (In Persian). doi: 10.22092/wmrj.2025.370992.1637
Heidary, K. , Najafinejad, A. , Mohammadian Behbahani, A. , Ownegh, M. and W. Dekker, L. (2026). Application of Machine Learning Algorithms in Analyzing the Effect of Fire on Soil Water Repellency in Tushan Watershed, Golestan Province. Watershed Management Research, 39(2), 20-43. doi: 10.22092/wmrj.2025.370992.1637
Hengl, T., Mendes de Jesus, J., Heuvelink, G. B. M., Ruiperez Gonzalez, M., Kilibarda, M., Blagotić, A., ... & Kempen, B. (2017). SoilGrids250m: Global gridded soil information based on machine learning. PLOS ONE, 12(2), Article e0169748. https://doi.org/10.1371/journal.pone.0169748.
Hinsinger, P. (2001). Bioavailability of soil inorganic P in the rhizosphere as affected by root-induced chemical changes: A review. Plant and Soil, 237(2), 173–195. https://doi.org/10.1023/A:1013351617532
Jenny, H. (1941). Factors of soil formation: A system of quantitative pedology. McGraw-Hill.
Kopittke, P. M., Menzies, N. W., Wang, P., McKenna, B. A., & Lombi, E. (2019). Soil and the intensification of agriculture for global food security. Environment International, 132, Article 105078. https://doi.org/10.1016/j.envint.2019.105078
Lundberg, S. M., & Lee, S. I. (2017). A unified approach to interpreting model predictions. In I. Guyon, U. V. Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, & R. Garnett (Eds.), Advances in neural information processing systems (Vol. 30, pp. 4765–4774).
McBratney, A. B., Mendonça Santos, M. L., & Minasny, B. (2003). On digital soil mapping. Geoderma, *117*(1–2), 3–52. https://doi.org/10.1016/S0016-7061(03)00223-4
Meyer, H., Reudenbach, C., Wöllauer, S., & Nauss, T. (2019). Importance of spatial predictor variable selection in machine learning applications–Moving from data reproduction to spatial prediction. Ecological Modelling, 411, Article 108815. https://doi.org/10.1016/j.ecolmodel.2019.108815
Minasny, B., & McBratney, A. B. (2016). Digital soil mapping: A brief history and some lessons. Geoderma, 264, 301–311. https://doi.org/10.1016/j.geoderma.2015.07.017
Mohammadian Behbahani, A. , Heidary, K. and Hosseinalizadeh, M. (2025). Modeling soil water repellency in loess soils of northern Iran using machine learning. Water and Soil Management and Modelling. (In Persian)
Mousavi, S. R. , Parsayi, F. , Rahmani, A. , Sedri, M. H. & kohsar Bostani, M. (2020). Spatial Prediction Some of the Surface Soil Properties Using Interpolation and Machine Learning Models. Journal of Soil Management and Sustainable Production, 10(3), 27-49. doi: 10.22069/ejsms.2021.17251.1916. (In Persian)
Padarian, J., Minasny, B., & McBratney, A. B. (2020). Machine learning and soil sciences: A review aided by machine learning tools. Soil, 6(1), 35-52.
Schratz, P., Muenchow, J., Iturritxa, E., Richter, J., & Brenning, A. (2019). Hyperparameter tuning and performance assessment of statistical and machine-learning algorithms using spatial data. Ecological Modelling, 406, 109–120. https://doi.org/10.1016/j.ecolmodel.2019.06.002.
Shahhosseini, M., Hu, G., & Archontoulis, S. V. (2020). Forecasting corn yield with machine learning ensembles. Frontiers in Plant Science, 11, 1120. https://doi.org/10.3389/fpls.2020.01120
Vaysse, K., & Lagacherie, P. (2017). Using quantile regression forest to estimate uncertainty of digital soil mapping products. Geoderma, 291, 55–64.
Wadoux, A. M. J. C., Minasny, B., & McBratney, A. B. (2020). Machine learning for digital soil mapping: Applications, challenges and suggested solutions. Earth-Science Reviews, 210, Article 103359. https://doi.org/10.1016/j.earscirev.2020.103359.
Winzeler, H. E., Mancini, M., Blackstock, J. M., Libohova, Z., Owens, P. R., Ashworth, A. J., Silva, S. H. (2024). Vegetation masking of remote sensing data aids machine learning for soil fertility prediction. Remote Sensing, 16(17), Article 3297. https://doi.org/10.3390/rs16173297
Yang, R. M., Zhang, G. L., Liu, F., Lu, Y. Y., Yang, F., Yang, F., ... & Li, D. C. (2016). Comparison of boosted regression tree and random forest models for mapping topsoil organic carbon concentration in an alpine ecosystem. Ecological indicators, 60, 870-878.
Yu, X., Liu, Y., Niu, S., Zhao, W., Fu, C., & Chen, Z. (2021). Structure, functions, and interactions of dryland ecosystems. In X. Yu & Y. Liu (Eds.), Dryland social-ecological systems in changing environments (pp. 69–107). Springer. https://doi.org/10.1007/978-981-16-3128-3_3
Zeraatpisheh, M., Ayoubi, S., Jafari, A., Tajik, S., & Finke, P. (2019). Digital mapping of soil properties using multiple machine learning models in the semi-arid region of central Iran. Geoderma, 337, 159-171. https://doi.org/10.1016/j.geoderma.2018.09.006
Zhang, L., Zhang, Z., Luo, Y., Cao, J., & Tao, F. (2021). Combining optical, fluorescence, thermal satellite, and environmental data to predict county-level maize yield in China using machine learning. Remote Sensing of Environment, 253, 112185. https://doi.org/10.3390/rs12010021.
Zornoza, R., Acosta, J. A., Bastida, F., Domínguez, S. G., Toledo, D. M., & Faz, A. (2015). Identification of sensitive indicators to assess the interrelationship between soil quality, management practices and human health. Soil, 1(1), 173–185. https://doi.org/10.5194/soil-1-173-2015.