River peak flows that exceed the conveyance of the main channel can substantially alter channel bed morphology over event timescales. Consecutive peak flows over short periods may also result in incomplete river bed recovery between events, affecting future flood hazards; for example, intra-flood sediment deposition can reduce conveyance capacity in the main channel for subsequent peak flows. However, these changes in flood hazard can go undetected because measurements of pre- and post-peak flow bathymetry are often unavailable. Here, we explore whether spatial variation in active flow width during peak flows can be leveraged to predict intra-flood bed elevation changes and post-flood recovery. We analyze variation in river bed elevation for different flow conditions across three 10 km sections of the Waal River, the Netherlands, using bi-weekly bathymetry data obtained over the last 20 years. We compare bed elevation variation during high vs low flows to spatial changes in active flow width during peak flows estimated from Digital Terrain Models of the floodplain. We find that streamwise variation in bed elevation between high and low flows is often explained by changes in active flow-width during peak flows: peak flows tend to erode where floodplain geometry causes flow constriction and tend to deposit where floodplain geometry causes flow expansion. During low flows, inverse bed elevation changes occur at the same locations as the river responds to the post-flood morphology. The degree of incision or deposition during peak flows is often dependent on the spatial gradient in flow constriction or expansion, however, the strength of this relationship is not consistent along-channel. The difference between high and low-flow bed elevation changes are amplified where the main channel nears the floodplain and can force peak flows inundating the floodplain to cross the main channel. These results suggest that spatial variation in floodplain geometry, specifically of features that constrict flow or force peak flows to cross the main channel, can be leveraged to understand and predict hot spots of river bed elevation change during peak flows.@en