Remote sensing estimation of water volume changes of typical lakes in Xinjiang, China from 1990 to 2021

Inland lakes are important surface water resources in arid Central Asia. Due to the superimposed influence of natural factors and human factors, the hydrological characteristics of arid lakes show significant temporal and spatial variations. However, data shortage in this area makes it difficult to carry out detailed and long-term quantitative monitoring of hydrological regimes for different lakes. Based on the Google Earth Engine Platform (GEE), we firstly selected the Landsat 5/7/8 remote sensing image data that completely covered the Saram Lake and Ebinur Lake during 1990-2021, and used the multi-remote sensing index decision tree method to extract the continuous long time series of lake area. Combined with lake water level extracted by CryoSat-2 and ICESat-2 alti-meter satellites, we constructed the storage capacity curve based on the relationship between lake area and water level, and estimated the water volume change information of the lakes. Finally, combined with the hydrological, climate and population factors data of the basin, the correlation analysis and random forest method were used to quantitatively compare and analyze the factors of water quantity variation between the two lakes. The results showed that both Saram Lake and Ebinur Lake had expanded during 1990-2021, though with quite different water conditions. The area of Saram Lake increased by only 1.3%, with little interannual variation. The water volume increased by 1.12 km³ at a growth rate of around 0.04 km³·a^-1. Conversely, the area of Ebinur Lake experienced a 30.1% expansion and exhibited significant annual fluctuation, averaging approximately 0.01 km³·a^-1. Annual precipitation and glacial meltwater were the main factors affecting the water content of the Saram Lake, with contribution rates of 33% and 27%, respectively. However, temperature and precipitation were the main factors affecting the water quantity change of Ebinur Lake, and their contribution rates in the process of water quantity change were both 28%. The aim of this study was to use remote sensing technology to reveal the characteristics of lakes' dynamic change and the difference of its response to their external environment in arid areas with the shortage of measured data, which would provide scientific reference for lake ecological environment and water resources protection in arid areas.