Spatiotemporal variation and attribution analysis of water bodies in Bayinguoleng Mongol autonomous prefecture based on Google Earth Engine

Bayinguoleng Mongol Autonomous Prefecture has long been characterized by water scarcity. Dynamic monitoring of water bodies and analysis of their spatiotemporal changes are crucial for understanding the distribution and trends of these water resources. This knowledge is essential for assessing the impact of climate change on water bodies in Bayinguoleng Prefecture and for safeguarding the ecological stability of the northwestern region. This study leverages Landsat series imagery from 2003 to 2023 and employs the robust and efficient random forest model, incorporating the Enhanced Vegetation Index (EVI) as a spectral feature, to extract water body information. Experimental results demonstrate that: (1) the random forest model coupled with EVI is highly effective for extracting water body information in Bayinguoleng Prefecture; (2) the water area of Bayinguoleng Prefecture increased by 2575.769km² between 2003 and 2023, with the most pronounced increases occurring in the south, east, and northwest; (3) precipitation and temperature exert a significant influence on water body changes in Bayinguoleng Prefecture.