Ecohydrological processes are a key element to consider in functional river restorations. In the framework of a LIFE+ European restoration program, we have investigated the potential for airborne thermal-infrared remote sensing to map surface water-groundwater exchanges and to identify their driving factors. We focused our attention on anastomosing channels on an artificial island of the Upper Rhine River (Rohrschollen), where a new channel was excavated from the floodplain to reconnect an older channel in its upstream part. These hydraulic engineering works led to an increased inflow from the Rhine Canal. Here, we propose an original data treatment chain to (a) georeference the thermal-infrared images in geographic information system based on visible images, (b) detect and correct data errors, and (c) identify and locate thermal anomalies attributed to groundwater inputs and hyporheic upwellings. Our results, which have been compared to morpho-sedimentary data, show that groundwater upwelling in the new channel is controlled by riffle-pool sequences and bars. This channel is characterized by large bedload transport and morphodynamic activity, forming riffles and bars. In the old channel, where riffle-pool sequences no longer exist, due to impacts of engineering works and insufficient morphodynamic effects of the restoration, thermal anomalies appeared to be less pronounced. Groundwater inputs seem to be controlled by former gravel bars outcropping on the banks, as well as by local thinning of the low-permeability clogging layer on the channel bed.