Abstract

The ensemble streamflow prediction (ESP) method has been widely used to produce seasonal streamflow forecasts, especially in snow-influenced basins. Because the approach relies on the assumption of perfect initial conditions that are obtained from hydrological models, choices related to their implementation may have considerable impacts on forecast attributes. Here, we investigate the extent to which the choice of calibration objective function (OF) affects the quality of seasonal (Spring-Summer) streamflow forecasts in mountainous regions, and also explore possible connections between forecast skill and hydrological consistency - measured in terms of biases in hydrological signatures - obtained from the model parameter sets. To this end, we calibrate three conceptual rainfall-runoff models (GR4J, TUW, and Sacramento) using 12 different calibration metrics, including seasonal objective functions that emphasize errors during the snowmelt period, and produce hindcasts for five initialization times over a 33-year period (April/1987 - March-2020) in 22 mountain catchments that span diverse hydroclimatic conditions along the semiarid Andes Cordillera. The results show that seasonal objective functions generate satisfactory performance in terms of probabilistic skill, reliability, and correlation compared to classic OFs like the Nash-Sutcliffe Efficiency (NSE). Nevertheless, commonly used OFs provide more realistic simulations in terms of simulated hydrological signatures. Among the options tested, an OF that combines the Kling-Gupta Efficiency (KGE) and NSE(log(Q)) provides the best compromise between hydrologically consistent model simulations and good forecast performance. Overall, we do not find direct relationships between hydrologically consistent model parameter sets and the quality of seasonal ESP forecasts. Finally, the results show that ESP is most skillful in catchments with high baseflow index and high interannual runoff variability. 

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call