Prediction of Landuse Changes Applying Knowledge-Based and Markov Chain Methods

Document Type : Research Article

Authors

• Tayebeh Irani ¹
• Hirad Abghari ²
• ali akbar rasooli ³

¹ PhD student in watershed management, Faculty of Natural Resources, Urmia University, Iran

² Faculty of Natural Resources, Urmia University, IRAN.

³ Department of Environmental Sciences, Macquarie University, Sydney, Australia

10.22067/geoeh.2023.82510.1369

Abstract

Therefore, this study aimed to investigate the current status and predict the future state of land use in the Gadarchai watershed, located in West Azerbaijan Province, which is one of the important sub-watersheds of the Lake Urmia basin. For this purpose, Sentinel-2 satellite images for 2016, 2020, and 2022 were obtained from the European Union's Copernicus website. Then, preprocessing methods were applied in various software environments, and the relevant images were sent to the eCognition software environment. In this environment, using various basic and object-oriented methods (especially segmentation and production of different coefficient layers), the nearest neighbor classification method was implemented, and land use maps were produced. Finally, the Markov chain model was used to predict changes in land use in future years. To verify the accuracy of the Markov chain model, the predicted land use change map for 2022 was compared with the 2022 classification map. The research results showed that with the application of basic methods, especially nearest neighbor classification, it is possible to produce land use maps with high accuracy (90% kappa coefficient). Also, by applying the Markov model, land use change maps with an acceptable accuracy level (around 80%) are possible. The final results indicate that by the year 2028, agricultural land use (13.89%), dry farming (14.1%), residential areas (0.33%), and salt pans of Lake Urmia (26%) will increase. It should be noted that the soil use class will decrease by 10.26%, and pastures will decrease by 5.35%. Overall, the final models demonstrate the high accuracy of basic and object-oriented methods and the suitable performance of the Markov model in the process of studying land use changes. The findings of this research can be useful for future environmental planning, especially in recommending sustainable land use and proper land management.

Keywords

• The Godarchai Watershed
• Sentinel2 images
• Knowledge-Based
• Markov Chain Model
• Prediction of Landuse Changes

Main Subjects

• مدیریت مخاطرات

---