Machine Learning Analysis of Lake Van and Lake Urmia Using Satellite Data

As vital components of the hydrological cycle and ecological balance, lakes play a crucial role in maintaining environmental stability. The significance of these aquatic ecosystems underscores the necessity for continuous monitoring and detailed analysis of their dynamics to achieve sustainable water resource management and balance economic-environmental considerations in watershed areas. This study utilizes remote sensing capabilities and geographic information systems to examine changes in Lake Urmia’s surface area over a 33-year period (1990–2023). Landsat satellite imagery, spectral indices (NDVI and MNDWI), and climate data including annual precipitation and temperature were employed as key variables influencing the lake’s condition. Based on comprehensive analyses, predictions of future water level trends were generated using the Random Forest machine learning algorithm. Findings reveal a substantial and continuous decline in the lake’s water level throughout the study period. This reduction coincided with fluctuations in surrounding vegetation cover, primarily driven by agricultural expansion and excessive groundwater extraction. While climatic data analysis indicates relatively stable precipitation patterns in the lake’s basin, temperature data show a gradual but significant increasing trend. These observations clearly emphasize the dominant role of human activities and inefficient water resource management in the lake’s depletion. Projections for the 2024–2032 period indicate a continuing decline in water levels under current water resource utilization patterns. These findings highlight the urgent need for integrated water resource management approaches, protection of feeder watersheds, implementation of sustainable land use planning, and establishment of continuous monitoring systems to prevent irreversible degradation of this valuable ecosystem.