A novel multi-objective model calibration method for ecohydrological applications

Abstract

The reliability and applicability of hydrological models within ecohydrological frameworks are major concerns. Two multi-objective model calibration strategies were formulated to achieve a balanced representation of ecologically relevant hydrologic indices. Two approaches were employed: 1) a performance-based method constraining the targeted hydrologic indices and 2) an unconstrained signature-based method explicitly incorporating the targeted hydrologic indices into multiple objective functions. Both strategies were successful in representing most of the selected hydrologic indices within a ±30% relative error acceptability threshold while yielding consistent runoff predictions in a watershed. The performance-based strategy was preferred since it showed a lower dispersion of near-optimal Pareto solutions when representing the selected indices based on water balance and Flow Duration Curve characteristics. Still, the overall representation of low flow magnitude and timing, rise and fall rates, and duration and frequency of extreme flows was limited in terms of interannual variability due to the hydrological model structural inadequacies.