Strategies to enhance the reliability of flow quantile prediction in the gauged and ungauged basins

Abstract

AbstractThe precise reproduction of different flow regimes in both gauged and ungauged watersheds is crucial for managing environmental flow and water quality requirements. However, the ability of hydrological models to reproduce flow quantiles (FQs) is often influenced by the process of calibrating the most dominant parameters through traditional parameter estimation methods. This research proposes a systematic parameter estimation approach to improve the credibility of the hydrologic model in reproducing FQs in gauged and ungauged watersheds through the following steps: (a) implementation of parameter sensitivity analysis to identify the dominant parameters, (b) initial estimation of the sensitive parameter values, (c) an iterative search for the optimal value of the dominant parameter in reproducing FQs and (d) regionalization of parameters to estimate FQs in the ungauged watersheds. The analysis shows the highest sensitivity of the runoff curve number (CN2) in simulating the hydrologic process in all test watersheds. Moreover, the best value of CN2 was found to be different for each flow quantile. Therefore, CN2 was updated for reproduction of FQs, which resulted in an overall average improvement of the regionalized model accuracy (across all test watersheds and flow quantiles) by 37 and 46% during the calibration and validation periods, respectively. The spatiotemporal dynamics of the water balance components that often control the behaviours of FQs in both the gauged and ungauged watersheds were also quantified. The results show a wide range of spatiotemporal variations for the majority of the water balance components.