Persistent hillslope erosion delays fluvial recovery from catastrophic flooding

As large floods increase in frequency, understanding of post-flood recovery dynamics can improve predictions and inform human (non-)intervention in perturbed watersheds. Existing work has mostly focused on post-flood changes to the channel bed, often at cross-section or reach scales. Our 8-year study following 100−1000-year flooding caused by the 2011 Tropical Storm Irene in New England, USA, is an opportunity to test how disrupted hillslopes influence watershed-scale recovery. We investigate whether persistent erosion from hillslopes can extend impacts of major disturbance and whether post-event floods repeatedly re-activate hillslopes, delaying hillslope and channel co-recovery. Pairing sediment flux and hillslope erosion measurements, we find that river suspended sediment fluxes remain elevated over pre-Irene levels by 2−10 times, with much excess sediment originating from Irene-activated landslides on channel-adjacent hillslopes. At least 8 years post-Irene, landslide scars have mostly not recovered and are eroding at 0.14 ± 0.06 m/yr, >1000 times the average erosion rates for these watersheds. Although landslides occupy <0.1% of the watershed area in the study region, landslide erosion accounts for 20%−120% of post-Irene sediment transport. We propose a conceptual model for channel-flux recovery that requires hillslope restabilization to slow above-average sediment transport. This has not yet occurred; instead, persistent channel-hillslope connectivity maintains efficient routing of hillslope-derived sediment to channels, effectively extending the effects of Irene by a decade or more. Elevated sediment flux may persist indefinitely if floods continue to renew hillslope-channel connectivity, an effect we document and a likely outcome given observed increases in regional flood frequency.