Spatial Modeling of Desertification Occurrence Probability in the Central Regions of Iran Using Machine Learning Algorithms

Allahqolipour, Ehsan; Mirbagheri, Babak; Shakiba, Alireza

doi:10.22111/jneh.2026.54843.2159

Spatial Modeling of Desertification Occurrence Probability in the Central Regions of Iran Using Machine Learning Algorithms

Articles in Press

Document Type : Research Article

Authors

¹ MSc. in Remote Sensing and GIS, Center for Remote Sensing and GIS Research, Faculty of Earth Sciences, Shahid Beheshti University, Tehran, Iran

² Assistant Professor, Center for Remote Sensing and GIS Research, Faculty of Earth Sciences, Shahid Beheshti University, Tehran, Iran

³ Associate Professor, Center for Remote Sensing and GIS Research, Faculty of Earth Sciences, Shahid Beheshti University, Tehran, Iran

10.22111/jneh.2026.54843.2159

Abstract

In recent years, desertification has become one of the most significant global environmental challenges, highlighting the need to identify areas prone to desertification and the factors driving this process. In the present study, desertification was modeled using machine learning algorithms across the provinces of Isfahan, Markazi, Qom, Kerman, and Yazd during 2002–2010. The trained algorithms were then applied to predict desertification and validate their performance for a second period, 2011–2018. In this research, the Random Forest (RF), Support Vector Machine (SVM), Gradient Boosting Machine (GBM), and Extreme Gradient Boosting (XGBoost) algorithms were employed to predict the spatial distribution of desertification. To identify desertified areas, remote sensing indices were utilized, while precipitation, air temperature, elevation, direct soil evaporation, wind speed, air humidity, rural population density, and distances from cities and roads were considered as input variables. After optimizing the hyperparameters, the Area Under the Receiver Operating Characteristic Curve (AUC) was calculated to evaluate model performance and determine the most accurate algorithm. The results for the first period indicated that RF outperformed the other algorithms, achieving the highest accuracy with an AUC value of 0.879. In addition, the AUC values for SVM, GBM, and XGBoost were 0.872, 0.872, and 0.861, respectively. RF also demonstrated strong temporal stability by maintaining the highest predictive accuracy (AUC = 0.800) during the second period. During this period, the AUC values for GBM, XGBoost, and SVM were 0.780, 0.770, and 0.564, respectively. Moreover, RF achieved the highest accuracy in Qom (0.655), Kerman (0.818), Markazi (0.898), and Yazd (0.798). Finally, wind speed was identified as the most influential variable contributing to desertification in the study area. This study demonstrated that machine learning algorithms, particularly RF, are effective tools for predicting desertification risk in central Iran.

Keywords

Main Subjects

Environmental Hazards

References

References [in Persian]

Akbari, Morteza, Hamid Reza Karim Zadeh, Reza Modares, and Bahareh Chakoshi. (2007). “Assessment and Classification of Desertification Using RS & GIS Techniques (Case Study: The Arid Region, in the North of Isfahan).” Iranian Journal of Range and Desert Research, 14(2):124–42. [In Persian]

Boali, Abdalhossein, Hamidreza Asgri, Ali Mohammadian Behbahani, Abdolrassoul Salmanmahiny, and Babak Naimi. (2021). “Evaluation of Desertification Intensity Based on Quantitative and Qualitative Changes in Groundwater and Soil Criteria Using Madalus Model and Geostatistical Methods.” Environmental Sciences, 19(4):85–106. doi: 10.52547/envs.2021.37277. [In Persian]

Fakhrabadi, Amir, Mohamad Salighe, Mehri Akbari, and Mohammed Hossai Nasserzadeh. (2019). “Evaluation of Desertification Intensity in Kashan Plain.” Geographical Research, 34(4):527–38. doi: 10.29252/geores.34.4.527. [In Persian]

Geldizadeh, Faezeh, Roshan, Gholamreza, & Ghanghermeh, Abdolazim. (2023). Monitoring the temporal-spatial changes of desertification in Gavakhoni Basin based on climatic criteria. Climate Change Research, 4(13), 55–68. doi.org/10.30488/ccr.2023.396869.1133. [In Persian]

Halabian, Amir Hossien, and Mahmod Soltanian. (2017). “Assessment and Forecasting the Desertification Changes in the East and South of Isfahan by CA-Markov Model.” Journal of Spatial Analysis Environmental Hazards, 3(4):71–88. doi: 10.18869/acadpub.jsaeh.3.4.71. [In Persian]

Halabian, Amirhossein, and Zohreh Jamshidian. (2019). “Analysis of Desertification Changes with Emphasis on Precipitation in Qom Province.” Emergency Management, 7(2):105–19. [In Persian]

Heydari Alamdarloo, Esmail, Hassan Barabadi, Dehghan Dehghan Rahimabadi, Hassan Khosravi, and Javad Rafieh Sharifiabad. (2022). “Evaluation of Control Strategies for Desertification Using SWOT Model in Yazd-Ardakan Plain.” Integrated Watershed Management, 2(1):1–14. doi: 10.22034/iwm.2022.550355.1025. [In Persian]

Kamali Maskooni, Ehsan, Mohammad Amin Kamali, and Ali Khanamani. (2021). “Investigation and Preparation of Desertification Map Based on Iranian Model of Desertification Potential (IMDPA) with an emphasis on two criteria of soil and vegetation (Case study: Faryab-Kerman Province).” Journal of Environmental Science and Technology, 12(12):163. doi: 10.22034/jest.2021.30618.3917. [In Persian]

Silakhuri, Esmaeil, Majid Onagh, Amir Sadodin, and Esmaeil Filekesh. (2014). “Comparing Efficiency of Iranian Desert Hazard Assessment Models, Namely MICD and IMDPA (Case Study: Sabzevar Region).” Journal of Water and Soil Conservation, 21(4):1–28. [In Persian]

Zeaiean Firouzabadi, Parvis, Reza Talkhabi Hamid, and Leila Hosseinjani. (2014). “CHANGE DETECTION OF MIGHAN DESERT BASIN USING ETM+, TM, MSS IMAGES AND CLIMATE DATA FROM 1973 TO 2011.” Journal of Applied Research in Geographical Sciences, 13(31):173–89. [In Persian]

References [in English]

Chen, W., Pourghasemi, H. R., Zhang, S., & Wang, J. (2019). 21—A Comparative Study of Functional Data Analysis and Generalized Linear Model Data-Mining Techniques for Landslide Spatial Modeling. In H. R. Pourghasemi & C. Gokceoglu (Eds.), Spatial Modeling in GIS and R for Earth and Environmental Sciences (pp. 467–484). Elsevier. https://doi.org/10.1016/B978-0-12-815226-3.00021-1

Feng, K., Wang, T., Liu, S., Kang, W., Chen, X., Guo, Z., & Zhi, Y. (2022). Monitoring Desertification Using Machine-Learning Techniques with Multiple Indicators Derived from MODIS Images in Mu Us Sandy Land, China. Remote Sensing, 14(11), 2663. https://doi.org/10.3390/rs14112663

Lackóová, L., Kaletová, T., & Halászová, K. (2023). Are Drought and Wind Force Driving Factors of Wind Erosion Climatic Erosivity in a Changing Climate? A Case Study in a Landlocked Country in Central Europe. Land, 12(4), Article 4. https://doi.org/10.3390/land12040757

Saha, S., Saha, M., Mukherjee, K., Arabameri, A., Ngo, P. T. T., & Paul, G. C. (2020). Predicting the deforestation probability using binary logistic regression, random forest, ensemble rotational forest, REPTree: A case study at the Gumani River Basin, India. Science of The Total Environment, 730, 139197. https://doi.org/10.1016/j.scitotenv.2020.139197.