Changes in climate and land cover/land use (LCLU) have a significant impact on the water–carbon cycle in the Jinsha River Basin. This study quantifies the impacts of climate change and LCLU on runoff and gross primary productivity (GPP) in the Jinsha River Basin. Using a comprehensive dataset spanning 1982–2018, including meteorological, hydrological, land cover, soil properties, and leaf area index (LAI) data, we employed the process-based WaSSI-CN (Water Supply Stress Index–Carbon Nexus) model to simulate coupled water–carbon processes. The model operates at a spatial resolution of 0.1° × 0.1° and simulates key components of the water–carbon balance, including runoff, evapotranspiration (ET), and GPP. It integrates hydrological processes (potential ET via the Penman–Monteith equation, soil moisture accounting, and runoff generation via the SAC-SMA scheme) with carbon cycle dynamics regulated by water availability and vegetation characteristics.

Robust model validation demonstrates high simulation accuracy. For monthly runoff across seven key hydrological stations, Nash–Sutcliffe Efficiency (E_NS) values range from 0.72 to 0.88, and coefficients of determination (R²) range from 0.70 to 0.90. Simulated GPP (WaSSI_GPP) was evaluated against three independent satellite-based products (MOD17A2, VPM, and PML_V2) for 2003–2015. WaSSI_GPP (905 g/(m²·a)) shows strong agreement with PML_GPP (1,056 g/(m²·a)), MOD_GPP (607 g/(m²·a)), and VPM_GPP (699 g/(m²·a)), capturing consistent spatial gradients (higher downstream values >1,200 g/(m²·a) and lower upstream values of 200–900 g/(m²·a)) as well as seasonal dynamics (peak July—August). Discrepancies primarily arise from differences in retrieval algorithms and uncertainties inherent in the reference datasets.

Analysis of basin-wide trends reveals significant environmental changes over the past 37 years. Annual precipitation increased significantly (Z = 2.08, p<0.05) at a rate of 14.22 mm/10a, reaching a mean of 658.6 mm/a. Mean annual temperature rose markedly (Z = 6.17, p<0.01) by 0.58 °C/10a, totaling approximately 1.8 °C of warming. Vegetation restoration efforts, particularly the “Grain for Green” program, were highly effective: forest area expanded by 7%, while grassland decreased by 6%. This change was reflected in a significant basin-wide increase in LAI (0.034 m²/m²/10a) and a detected Pettitt breakpoint in 1997. Spatially, runoff averaged 267.6 mm/a and exhibited a non-significant decreasing trend (Z =－0.55). A pronounced spatial heterogeneity emerged: significant runoff declines dominated the southeastern and middle–lower reaches (e.g., －17.85% post-2001 in the mid-reaches), in contrast to increases in the northwestern upper reaches (Z = 2.60 above Zhimenda). Conversely, GPP showed a highly significant basin-wide increase (Z = 4.96, p<0.01), averaging 907.9 g/(m²·a), with 66.88% of the basin exhibiting significant upward trends, strongest in the upper and middle reaches.

Employing scenario simulations with defined baseline (1983–1997) and change period (2004–2018), we isolated the respective contributions of climate change and LCLU. Annual natural runoff decreased by 6.44 billion m³ (−5.2%) relative to the baseline. Climate change was the dominant driver, contributing 86.0% (−5.54 billion m³) to this reduction, primarily through increased evapotranspiration (ET) induced by warming, particularly pronounced in the mid-reach dry–hot valleys. Although increased precipitation partially offset water losses, it was insufficient to compensate for enhanced ET. LCLU (mainly forest expansion) contributed 14.0% (−0.90 billion m³) to the runoff decline, largely through increased canopy interception and transpiration. In contrast, annual GPP increased substantially by 72.6 g/(m²·a) (+8.4%). Climate change remained the primary driver (72.4% contribution, +52.6 g/(m²·a)), mainly through precipitation-induced alleviation of water stress and warming-related extension of the growing season. LCLU contributed 27.6% (+20.0 g/(m²·a)), driven by increased forest cover and higher LAI.

Climate change and LCLU-induced vegetation greening synergistically enhanced the terrestrial carbon sink (increasing GPP) while jointly exerting negative pressure on water yield (reducing runoff), with climate change remaining the dominant driver in both processes. This quantification (climate change: 86.0% contribution to runoff decrease and 72.4% to GPP increase; LCLU: 14.0% and 27.6%, respectively) provides important scientific evidence for managing trade-offs among hydropower generation, water security, and ecosystem carbon sequestration under ongoing environmental change in the Jinsha River Basin and similar large river systems.