Diffuse groundwater recharge estimation confronting hydrological modelling uncertainty

Diffuse recharge is vital in determining the availability of renewable groundwater. Estimating diffuse recharge, however, is a great challenge plagued with uncertainty due to limitations in direct observation and process understanding. Hydrological model is functional for diffuse recharge estimation at catchment scale. It can be used independently or in combination with flow recession analysis to estimate recharge but it is accompanied with uncertainties owing to model structures and parameterization schemes. Confronting the uncertainties in hydrological modelling, in this paper, we used three models (GR4J, SIMHYD and IHACRES) together with nine parameterization schemes to investigate the effectiveness and uncertainties in the estimates of diffuse recharge for unregulated headwater catchments. The parameterization schemes were configured by three objective functions and three calibration periods conditioned on climate. It was found that the capability of hydrological model in yielding more reasonable groundwater recharge estimates largely depended on the objective function used and the calibration period configured. Model calibrated using inverse NSE (NSEInv) was judged to be more likely to yield better recharge estimates. The uncertainties due to the choice of parameterization scheme were found to be comparable with those from model structures. The use of different models, objective functions and calibration periods allowed reflection on the level of uncertainties one can minimally expect in estimating catchment scale recharge.