High‐resolution topography reveals that floodplains along meandering rivers in Indiana commonly contain intermittently flowing channel networks. We investigated how the presence of floodplain channels affects lateral surface‐water connectivity between a river and floodplain (specifically exchange flux and timescales of transport) as a function of flow stage in a low‐gradient river‐floodplain system. We constructed a two‐dimensional, surface‐water hydrodynamic model using Hydrologic Engineering Center's River Analysis System (HEC‐RAS) 2D along 32 km of floodplain (56 km along the river) of the East Fork White River near Seymour, Indiana, USA, using lidar elevation data and surveyed river bathymetry. The model was calibrated using land‐cover specific roughness to elevation‐discharge data from a U.S. Geological Survey gage and validated against high‐water marks, an aerial photo showing the spatial extent of floodplain inundation, and measured flow velocities. Using the model results, we analyzed the flow in the river, spatial patterns of inundation, flow pathways, river‐floodplain exchange, and water residence time on the floodplain. Our results highlight that bankfull flow is an oversimplified concept for explaining river‐floodplain connectivity because some stream banks are overtopped and major low‐lying floodplain channels are inundated roughly 19 days per year. As flow increased, inundation of floodplain channels at higher elevations dissected the floodplain, until the floodplain channels became fully inundated. Additionally, we found that river‐floodplain exchange was driven by bank height or channel orientation depending on flow conditions. We propose a conceptual model of river‐floodplain connectivity dynamics and developed metrics to analyze quantitatively complex river‐floodplain systems.

The process of channelization on river floodplains plays an essential role in regulating river sinuosity and creating river avulsions. Most channelization occurs within the channel belt (e.g. chute channels), but growing evidence suggests some channels originate outside of the channel‐belt in the floodplain. To understand the occurrence and prevalence of these floodplain channels we mapped 3064 km2 of floodplain in Indiana, USA using 1.5 m resolution digital elevation models (DEMs) derived from airborne light detection and ranging (LiDAR) data. We find the following range of channelization types on floodplains in Indiana: 6.8% of floodplain area has no evidence of channelization, 55.9% of floodplains show evidence (e.g. oxbow lakes) of chute‐channel activity in the channel belt, and 37.3% of floodplains contain floodplain channels that form long, coherent down‐valley pathways with bifurcations and confluences, and they are active only during overbank discharge. Whereas the first two types of floodplains are relatively well studied, only a few studies have recognized the existence of floodplain channels. To understand why floodplain channels occur, we compared the presence of channelization types with measured floodplain width, floodplain slope, river width, river meander rate, sinuosity, flooding frequency, soil composition, and land cover. Results show floodplain channels occur when the fluvial systems are characterized by large floodplain‐to‐river widths, relatively higher meandering rates, and are dominantly used for agriculture. More detailed reach‐scale mapping reveals that up to 75% of channel reaches within floodplain channels are likely paleo‐meander cutoffs. The meander cutoffs are connected by secondary channels to form floodplain channels. We suggest that secondary channels within floodplains form by differential erosion across the floodplain, linking together pre‐existing topographic lows, such as meander cutoffs. Copyright © 2016 John Wiley & Sons, Ltd.