Impacts of multiple climate factors and vegetation changes on evapotranspiration in southwest China from 1982 to 2018

Abstract

Evapotranspiration (ET), as an essential element of the global water cycle, serves as a crucial connection for the transfer of water, carbon, and energy between the land biosphere and the atmosphere. This research concentrated on the dynamic variations in ET over 1982–2018 in Southwest China (SWC), investigating the driving mechanisms behind ET changes, particularly climate variability and vegetation dynamics. We selected long-term remote sensing and reanalysis datasets to evaluate ET dynamics in SWC. After validation against eddy covariance observations, the GLASS ET dataset with the highest accuracy was chosen for further analysis. The spatial patterns of ET across SWC and its various geomorphological subdivisions were analyzed. Overall, ET exhibited a north-to-south increasing trend, with more than 80 % of the areas in each geomorphological subdivision showing a significant upward trend in ET. Subsequently, ridge regression was employed to analyze the relative contributions of temperature (T), solar radiation (Rs), specific humidity (H), precipitation (P), wind speed (U), and leaf area index (LAI) to ET variability. T and H were found to be the dominant factors influencing ET changes, with their relative contributions exceeding 20 % during both study periods (1982–1999 and 2000–2018). Furthermore, the dominant drivers of ET changes exhibited phase-dependent variations across different geomorphological subdivisions. Additionally, the relative contribution of vegetation factors (LAI) to ET increased across all subregions. Ridge regression effectively addressed the multicollinearity issue among environmental factors, offering greater stability compared to traditional multiple linear regression. Finally, a double-log elasticity model was used to further quantify and validate the effects of different driving factors on ET, enhancing the credibility of the results. The elasticity coefficient for T was the highest and positive, while the elasticity coefficient for H was the highest and negative. These findings enhance the understanding of ET change mechanisms in SWC and are essential for elucidating regional vegetation-hydrology-climate interactions.