Coupling water balance and vegetation carbon dynamics in a parsimonious global hydrologic model

Abstract

Understanding ecohydrological responses to environmental change requires parsimonious models that can represent interactions between water balance and vegetation carbon uptake. Simple conceptual hydrologic models efficiently simulate runoff and evapotranspiration, but they usually lack dynamic vegetation representation. We extend the abcd monthly water-balance model to simulate coupled water–carbon dynamics by integrating the Horton Index framework with a two-parameter Gross Primary Productivity–Evapotranspiration (GPP–ET) functional relationship. The resulting eight-parameter framework is calibrated using observation-constrained runoff and GPP, while ET is withheld from the calibration objective and evaluated independently. Using GSWP3-W5E5 forcing, FLUXCOM-X BASE GPP and ET, and GRUN-ensemble runoff over 235 global basins, we calibrate the model for 2001–2014 and evaluate it independently for 2015–2019. During validation, the model reproduces GPP robustly, with a median Kling–Gupta Efficiency (KGE) of 0.758 and positive KGE in 97.9% of basins. ET performance is also broadly positive despite not being directly calibrated, with a median validation KGE of 0.673 and positive KGE in 87.7% of basins. Runoff skill is moderate and more heterogeneous, with a median validation KGE of 0.435 and positive KGE in 81.7% of basins. Seasonal-cycle and anomaly analyses show that the model strongly captures the climatological timing of GPP and ET while retaining meaningful anomaly skill after removing the mean annual cycle. The calibrated coupling parameters λ, β, and α exhibit geographically organized patterns consistent with their physically interpretable roles in evaporative partitioning and water–carbon coupling, while β and α remain effective model-scale rather than intrinsic physiological coefficients. These results show that a parsimonious monthly hydrologic model can be extended to diagnose first-order coupled water–carbon dynamics while retaining interpretable structure and computational efficiency.