Trade-Offs and Synergies in Multi-Variable Distributed Hydrological Model Calibration: A Transboundary Blue Nile Basin Study

Abstract

Calibration of distributed hydrological models often suffers from parameter equifinality and a lack of internal state realism, particularly in data-scarce transboundary basins. This study explores the trade-offs and synergies among eight calibration schemes applied to the mesoscale Hydrologic Model (mHM) in the Blue Nile Basin (BNB). The calibration schemes considered include a no-calibration baseline (i.e., default model parameters), single-objective formulations, and triple-objective formulations by incorporating three key hydrological state variables, including streamflow (Q), remotely sensed (RS) soil moisture (SM), and evapotranspiration (ET) with novel multi-objective schemes. The performance of the eight schemes is assessed using a suite of statistical metrics, alongside hydrological signature, extreme event, and annual trends. Results show that the default (Scheme 1) provides a reasonable baseline (median NSE = 0.76). Q-only calibration (Scheme 2) yields high efficiency (median NSE = 0.88) but compromises internal consistency. In contrast, SM-only and ETonly schemes exhibit saturated state errors: high correlations with satellite variables mask large overestimations of peak flows and volumetric biases. Dual-objective schemes (Q+ET and Q+SM) partially improve internal states but retain specific biases. The triple-objective scheme (Scheme 8; Q+SM+ET) delivers the most reliable performance, with median NSE = 0.90, KGE = 0.88, and strong agreement with independent ET (PCC = 0.86), SM (PCC = 0.89), and TWS (PCC = 0.73), as well as captures key hydrological signatures, extremes, and decadal trends while avoiding biases of single-variable schemes. This multivariable framework leverages synergies among datasets to provide physical plausibility and reliability of hydrological modeling across the transboundary BNB.