Dynamical processes occurring on the hierarchical branching structure of a river network tend to heterogeneously distribute fluxes on the network, often concentrating them into “clusters,” i.e., places of excess flux accumulation. Here, we put forward the hypothesis that places in the network predisposed (due to process dynamics and network topology) to accumulate excess sediment over a considerable river reach and over a considerable period of time reflect locations where a local imbalance in sediment flux may occur thereby highlighting a susceptibility to potential fluvial geomorphic change. We develop a dynamic connectivity framework which uses the river network structure and a simplified Lagrangian transport model to trace fluxes through the network and integrate emergent “clusters” through a cluster persistence index (CPI). The framework was applied to sand transport in the Greater Blue Earth River Network in the Minnesota River Basin. Three hotspots of fluvial geomorphic change were defined as locations where high rates of channel migration were observed and places of high CPI coincided with two of these hotspots of possibly sediment‐driven change. The third hotspot was not identified by high CPI, but instead is believed to be a hotspot of streamflow‐driven change based on additional information and the fact that high bed shear stress coincided with this hotspot. The proposed network‐based dynamic connectivity framework has the potential to place dynamical processes occurring at small scales into a network context to understand how reach‐scale changes cascade into network‐scale effects, useful for informing the large‐scale consequences of local management actions.

Long‐term prediction of environmental response to natural and anthropogenic disturbances in a basin becomes highly uncertain using physically based distributed models, particularly when transport time scales range from tens to thousands of years, such as for sediment. Yet, such predictions are needed as changes in one part of a basin now might adversely affect other parts of the basin in years to come. In this paper, we propose a simplified network‐based predictive framework of sedimentological response in a basin, which incorporates network topology, channel characteristics, and transport‐process dynamics to perform a nonlinear process‐based scaling of the river‐network width function to a time‐response function. We develop the process‐scaling formulation for transport of mud, sand, and gravel, using simplifying assumptions including neglecting long‐term storage, and apply the methodology to the Minnesota River Basin. We identify a robust bimodal distribution of the sedimentological response for sand of the basin which we attribute to specific source areas, and identify a resonant frequency of sediment supply where the disturbance of one area followed by the disturbance of another area after a certain period of time, may result in amplification of the effects of sediment inputs which would be otherwise difficult to predict. We perform a sensitivity analysis to test the robustness of the proposed formulation to model parameter uncertainty and use observations of suspended sediment at several stations in the basin to diagnose the model. The proposed framework has identified an important vulnerability of the Minnesota River Basin to spatial and temporal structuring of sediment delivery.

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.