A Multi-Objective Regional Sensitivity Analysis of the Variable Infiltration Capacity Model in the North of the Iberian Peninsula

Abstract

<p><span>The use of evaporation data as part of the hydrological modelling exercise is becoming a standard practice in many hydrologic studies that aims to strengthen the predictive capability of hydrologic models. The most common application of evaporation data can be found during a post-calibration phase in which the streamflow-only calibration is evaluated against an appropriate metric for evaporation (i.e. mean values ratio for the study period, efficiency metric, correlation coefficient, bias ratio, …). However, the parameter sensitivities for the evaporation performance are rarely taken into account during the pre-calibration stage, and the parameters modified during calibration are selected merely based on the streamflow performance.</span></p><p><span>This study explores the parameter sensitivities of the Variable Infiltration Capacity (VIC) model for the streamflow and evaporation performances simultaneously. A total of 19 parameters were modified during a multi-objective regional sensitivity analysis exercise focused on the Nash-Sutcliffe Efficiency for the daily streamflow and the monthly evaporation simulations. The VIC model was implemented for a total of 31 headwater catchments belonging to the Duero River Basin, a basin located in the North of the Iberian Peninsula. The model was forced with a gridded dataset of daily precipitation and temperature extracted from SPREAD/STEAD at 0.05º resolution. The streamflow observations were provided by the Spanish Centre for Public Work Experimentation and Study (CEDEX), and the evaporation data were gathered from the outputs of the Global Land Evaporation Amsterdam Model (GLEAM) version 3.5a. </span></p><p><span>The results of this work show that only 9 parameters are important to the two metrics analysed. Among these parameters, 5 are soil-related, 2 belong to the routing scheme and the other 2 to the vegetation parameterization. These results highlight the need to include additional vegetation parameters in order to achieve a good performance for both the streamflow and evaporation simulations.</span></p><p><span><strong>Keywords:</strong></span><span> VIC model, regional sensitivity analysis, Duero River Basin, streamflow, evaporation</span></p><p><span>ACKNOWLEDGEMENTS: </span><span>All the simulations were </span><span>conducted</span><span> in the ALHAMBRA cluster (http://alhambra.ugr.es/) of the University of Granada. This work was funded by FEDER/Junta de Andalucía-Consejería de Economía y Conocimiento, project B-RNM-336-UGR18, by the Spanish Ministry of Economy and Competitiveness project CGL2017-89836-390R with additional support from FEDER Funds, and by FEDER/Junta de Andalucía-Consejería de Transformación Económica, Industria, Conocimiento y Universidades (project P20_00035). The first author was supported by the Ministry of Education, Culture and Sport of Spain (FPU grant FPU17/02098).</span></p>