Groundwater interacts with surface water features nearly in all types of landscapes. Understanding these interactions has practical consequences on the quantity and quality of water in either system, because the depletion or contamination of one of the systems will eventually affect the other one. Many studies have shown that the use of heat as natural tracer in conjunction with water level measurements is an effective method for estimating water flow (fluxes) between groundwater and surface water. A number of studies have explored the effects of spatial and temporal variability of groundwater–surface water flux exchanges using temperature and water level measurements; however, the effect of temporal resolution of water level and temperature data on estimating flux remains unexplored. Therefore, this study investigated the effect of temporal resolution of input data on temporal variation of groundwater–surface water flux exchanges. To this end, we calibrated a variably saturated two-dimensional groundwater flow and heat transport model (VS2DH) at hourly and daily time scales using temperatures measured at multiple depths below the riverbed of the Zenne River, located at a well-known Belgian brownfield site. Results of the study showed that the computed water flux through the streambed ranged between −32 mm/day and +25 mm/day using the hourly model and from −10 mm/day to −37 mm/day using the daily model. The hourly model resulted in detecting reversal of flow direction inducing short-term surface water flow into the streambed. However, such events were not captured if daily temperature and water level measurements were used as input. These findings have important implications for understanding contaminant mass flux and their attenuation in the mixing zone of groundwater and surface water. Copyright © 2012 John Wiley & Sons, Ltd.