QUANTIFYING AVULSION ACTIVITY ON RIVER DELTAS FROM 1984-2012

Avulsion is the process whereby channelized flow partially or completely reroutes to a new or preexisting channel on the adjacent floodplain. This process constructs delta landforms and stratigraphy, yet not much is known about avulsion activity because avulsions occur infrequently and are hard to observe. Here we use the Landsat satellite data to quantify avulsion activity on deltas throughout the world from 1984 to 2012. From a database of 1,342 deltas, we identified 70 avulsions using timelapse videos created in Google Earth Engine. For each avulsion we collected morphometric data to quantify behavior as well as explore hydrodynamic controls on their location. On each avulsion, we measured (1) channel centerline lengths of the parent channel (P_L) and avulsion channel length (A_L) from the shoreline to the avulsion location; (2) shoreline length (S_L) between the river mouth of the parent and avulsion channel; (3) take-off angle between the parent channel and the avulsion channel; and (4) avulsion duration from initiation until all water flows into the avulsion channel. Our results show that average avulsion duration was 6.9 years with a standard deviation of 3.5 years. We observed a positively skewed unimodal distribution for all take-off angles in which the mean of the distribution was ~49 degrees. We found that S_L = 2.21P_L^0.8044 (R² = 0.569) indicating larger avulsions create more widely spaced river mouths at the shoreline. The ratio of P_L to A_L was nearly 1:1 and ~55% of all avulsion channels were shorter than the corresponding P_L suggesting those channels avulsed to a steeper slope. This insight is consistent with slopes of the parent channel (S_P) and avulsion channel (S_A) 30-m resolution topography data. We tested the hypothesis that P_L scales with the backwater length. Backwater length typically scales as the characteristic channel depth (h) divided by S_p. To estimate h we used hydraulic geometry power-law relationships in conjunction with normal-flow equations. Results indicate that backwater length explains some of the variation in P_L we found in Landsat imagery, but does not explain all the avulsions we observed.