Machine learning revealed monthly change characteristics of lakes on the Tibet Plateau over the past two decades

Study region

The Tibetan Plateau (TP) hosts the highest concentration of plateau lakes on Earth, which are undergoing significant climate-induced changes. However, the lack of high-temporal-resolution data has posed challenges for capturing these dynamics at the monthly scale.

Study focus

This study integrates optical imagery from the Google Earth Engine (GEE) platform and altimetry data from 11 satellite missions, combined with four machine learning models, to investigate the monthly change characteristics of 253 lakes on the TP from 2000 to 2022.

New hydrological insights for the region

Lake area, level, and volume have expanded, risen, and increased, respectively, during the study period. Lake volume increased by 5.62 % per year, with the growth mainly concentrated in the Inner Plateau and in lakes larger than 100 km². A decreasing trend in lake volume was observed in the Brahmaputra, Ganges, and Yangtze basins. Monthly lake area, level and volume changes mainly peaked in August, September and October. The lake water volume in the Indus, Ganges, and Amu Darya basins peaks earlier. The monthly peak water volume of lakes larger than 50 km² shows a delayed trend, while lakes smaller than 50 km² show an earlier peak trend. Lake volume changes were closely linked to the El Niño-Southern Oscillation (ENSO), while regional precipitation and runoff were likely the primary drivers of monthly variability.