Scale-dependent drivers of water use efficiency across China: integrating stable isotopes, remote sensing, and machine learning

Feng Jiang, Xiaoyi Shi, Fuxi Shi, Zhenyi Jia, Xin Song, Tao Pu, Yanlong Kong, Shijin Wang, Lizong Wu, Jia Jia, Zhenzhen Zhang, Jie Wang, Wenqing Han,
Scale-dependent drivers of water use efficiency across China: integrating stable isotopes, remote sensing, and machine learning,
CATENA,
Volume 260,
2025,
109403,
ISSN 0341-8162,
https://doi.org/10.1016/j.catena.2025.109403.
(https://www.sciencedirect.com/science/article/pii/S0341816225007052)
Abstract: Water use efficiency (WUE) serves as a crucial metric for terrestrial carbon–water coupling, yet systematic gaps persist in understanding the spatial patterns and drivers of leaf-level intrinsic WUE (iWUE) versus ecosystem-scale WUE (WUEEco). Combining machine learning with 1,446 leaf δ13Cp records, we investigated the spatial heterogeneity and main drivers of iWUE and WUEEco across different life forms and climate zones in China. Results showed that inverse spatial patterns, where iWUE peaked in arid northwestern grasslands (60.46 μmol mol−1). In contrast, WUEEco exhibited maxima in humid southeastern forests (1.82 g C/kg H2O). Hierarchical partitioning and structural equation modeling revealed that elevation indirectly influenced iWUE (17.72 %) and WUEEco (25.64 %) through its modification of climatic conditions. Vegetation factors (e.g., leaf area index) and climatic factors (e.g., relative humidity) emerged as key drivers of iWUE (24.06 %) and WUEEco (15.31 %), primarily through their regulation of photosynthesis–transpiration coupling processes. Among four machine learning models, Random Forest has the best performance in iWUE prediction (R2 = 0.73, NRMSE = 0.122, MBE =  − 0.078), providing a high-resolution national iWUE dataset. This study highlights the importance of scale in understanding carbon–water interactions and provides a valuable reference for water resource management under climate change.
Keywords: iWUE; WUEEco; Spatial pattern; Machine learning; Driving factor; China