Deltaic avulsions over the past half-century captured by satellite imagery

River avulsions pose a hazard to human life and property on densely populated river deltas and are responsible for distributing freshwater and nutrients to wetland ecosystems. Predicting when and where avulsions will take place is crucial for hazard mitigation and sustainable management of river deltas, but testing predictive models is challenging due to limited field data. Existing models predict that avulsions are set up through gradual sedimentation of the riverbed and are triggered when overbank floods forge a new, steeper path to the shoreline. Previous work on a handful of well-studied low gradient deltas suggests that avulsions preferentially occur at a node located approximately one backwater length upstream of the shoreline, whereas the apex of steeper fan deltas is tied to a change in topographic confinement. To test deltaic avulsion models at a broader scale, we present a compilation of more than 80 avulsions on 41 river deltas across the globe that have occurred over the past 34 years and were captured by NASA Landsat. We constrain channel pathways from remote sensing data and local topography, flood discharges, and sea level from published data where available. Our compilation supports the hypothesis that avulsion location correlates with the backwater length on lowland river deltas, and with canyon-fan transitions for steeper fan deltas. New channels forge a path that is shorter than both the old path and the median of all possible paths to the shoreline, except in cases of abandoned-channel reoccupation. Avulsions tend to initiate during flood events with greater than a two-year recurrence period, when overbank flooding is expected, and the parent channel can remain partially active for up to eight years before being completely abandoned. These results will help to test predictive models of avulsion location and timing on fans and deltas.