Sub-pixel TROPOMI SIF enables process-based estimation of soybean carbon–water coupling across China

Accurately characterizing regional differences and relative environmental contributions to crop carbon assimilation, water consumption, and water-use efficiency (WUE) is essential for improving our understanding of crop carbon–water processes. We developed a process-based crop carbon–water framework constrained by solar-induced chlorophyll fluorescence (SIF). In this framework, crop-distribution information was used to separate the soybean SIF signal to reduce mixed-pixel effects. The retrieved soybean SIF was then used to estimate net photosynthesis, and canopy conductance ( G c ) was incorporated as a key input into a Penman–Monteith scheme to estimate transpiration ( T ), from which WUE is derived. For soybean across China during 2019–2025, the framework-derived multi-year growing-season mean estimates of gross primary production (GPP), G c , T , and WUE were 1151.8 gC m −2 season −1 , 0.00725 m s −1 , 699.7 mm season −1 , and 1.672 gC kg −1 H 2 O, respectively. Pixel-wise analysis revealed substantial spatial heterogeneity in the strength of interannual T –WUE co-variation among soybean-growing regions. Within the present modelling framework, partial-correlation and dominance analyses further showed that atmospheric dryness had a greater relative explanatory contribution to model-estimated WUE variability than soil water supply in more than 89% of soybean areas. Overall, the proposed SIF-based framework provides consistent large-scale estimates of T and WUE, highlights the comparatively greater explanatory contribution of atmospheric dryness to soybean WUE variability, and provides a regional diagnostic basis for identifying soybean areas where such atmospheric dryness may deserve closer attention in irrigation and water-productivity assessment.