The construction of a river barrage can increase groundwater levels upstream of the barrage during the rainy season. Analytical and statistical approaches were applied to evaluate the relationship between groundwater and river water at the Changnyeong–Haman river barrage in Korea using time series data of water level and electrical conductivity from June 2011 to September 2014. An artificial neural network based time series model was designed to filter out the effect of rainfall from the groundwater level data in the study area. Aquifer diffusivity and river resistance were estimated from the analytical solution of a one-dimensional unit step response function by using the filtered groundwater level data. River resistance increased in response to groundwater level fluctuations. Cross-correlation analyses between the groundwater and the river water showed that the lag time increased during the observation period for both the water level and the electrical conductivity while the cross-correlation function declined for the same period. The results indicated that a constant river stage maintained at the river barrage can weaken the hydrologic stress and reduce the exchange of material between the river and the adjacent aquifer because of the deposition of fine sediment on the river bottom and walls. Copyright © 2016 John Wiley & Sons, Ltd.