Extended Abstract
Introduction
Land use and land cover change is a multifaceted process that is influenced by human activities and natural processes that disrupt the functioning of ecosystems. The rate and intensity of land use and land cover change has increased significantly in recent decades at different levels, from local to global scale. Especially in developing countries, due to unsustainable use of resources and population pressure, the intensity of changes is greater. An important measure to prevent the unwanted and undesirable consequences of the above changes is the systematic evaluation of land use and land cover changes. In this way, the current research has been carried out with the aim of identifying and monitoring land use and land cover changes, and it has been tried to reveal the changes between the years 1999, 2010 and 2023 and predict them until 2043. The results of this prediction help create the necessary groundwork for planning and implementing sound policies regarding the optimal utilization of agricultural and forest lands.
Materials & Methods
 The current research includes a part of the western region of Gilan province, which includes the cities of Bandar Anzali, Soumesara, Foman, Masal and Razvanshahr. In this research, after downloading the Landsat images from the Google Earth Engine platform, random forest algorithm was used for classification and the Quantity and Allocation Disagreement Index were used to evaluate the accuracy of the classified images, and after ensuring the accuracy of the classification, the transformed areas between the classes with the use of GIS was calculated. To improve the accuracy of the prediction model, using the multi-criteria evaluation method, land suitability maps were created based on the physical characteristics of the land and socio-economic factors for each class and integrated with the Cellular Automata-Markov Chain model. The simulated map for 2023 was prepared and compared and validated with the ground reality map of 2023 using Allocation Disagreement, Quantity Disagreement and Figure of merit index. Finally, after confirming the validation results, the pattern of land use classes and land cover for 2043 was predicted by the Cellular Automata-Markov Chain model in the IDRISI TerrSet software platform.
Results and discussion
 The present study examines the changes in land use and land cover in the western region of Gilan province from 1999 to 2023 and estimates the changes until 2043. The basic data in this research are Landsat 5, 7 and 8 images. Image classification and change detection were done using random forest algorithm in Google Earth Engine platform and geographical information system. In order to predict land use changes until 2043, the Cellular Automata-Markov Chain model was used. Then, the accuracy of the classified maps was calculated using Quantity and Allocation Disagreement index. The QADI values for 1999, 2010, and 2023 were 0.009, 0.01, and 0.02, respectively, indicating high accuracy of classification. Also, Kappa coefficient (KC) and overall accuracy (OA) of more than 96% confirm the very good efficiency of RF algorithm in decision making and classification. The multi-criteria evaluation module using ground data was integrated with Cellular Automata-Markov Chain model to increase the accuracy of the prediction model. Validation of the model was done using three criteria: Allocation Disagreement, Quantity Disagreement and Figure of merit. The total value for these three factors was 4.65, 2.02 and 48.60 percent, respectively. The findings indicate a decrease in the area of forest lands, agricultural lands and wetlands, as well as an increase in the built-up areas and range lands in the base period of the study (1999-2023) and the expected decades. This analysis shows the continuation of the trend of reducing the area of natural resources and increasing human activities, spatially urban development in the coming periods. This information can help decision makers in natural resource management to take preventive measures and continuous management to preserve natural environments.
Conclusion
 Predicted maps play an important role in estimating land use changes and natural resources, such as forests, water bodies, biodiversity, soil, minerals and other elements. These maps serve as practical tools for informed decision-making in environmental planning and resource management and enable policy makers to prevent adverse environmental consequences. In this research, three-time Landsat satellite images and random forest classification algorithm were used in Google Earth Engine platform to detect historical changes in land use and land cover from 1999 to 2023. The accuracy of the classified images was verified using the QADI index. In order to improve the efficiency of the of cellular automata-Markov chain model, in predicting future spatial changes, the multi-criteria evaluation method was integrated with this model. Then, by confirming the validity of the simulator model, it was possible to predict land use and land cover changes for 2043. The results showed that, during the study period, built-up areas and range lands have grown by 99.75% and, 6.57 respectively, On the other hand, forest cover, wetlands and agricultural lands lost 3.33, 24.38 and 1.96%, of their area, respectively. The results of the model predicted a significant decrease in the extent of forests, wetlands, and agricultural lands, while increasing the extent of built-up areas. The decrease observed in forest, agriculture and wetland shows the alarming trend of destruction of natural resources and environment.