The rates and spatial patterns of annual riverbank erosion revealed through terrestrial laser-scanner surveys of the South River, Virginia

Between A.D. 2006 and 2008, we completed annual surveys of two mercury-contaminated eroding banks, one forested and the other grass covered, along the gravel-bed, bedrock South River in Virginia. Gridded digital terrain models with a resolution of 0.05 m were created from bank topography data collected using a terrestrial laser scanner. Model comparisons indicate that the forested bank retreated nearly 1 m around two leaning trees, while elsewhere the extent of bank retreat was negligible. On the grassy bank, retreat was controlled by the creation of small overhanging clumps of turf at the top of the bank, their occasional failure, and the ultimate removal of failed debris from the bank toe. Partial autocorrelation analysis of vertically integrated bank retreat demonstrates that bank profile erosion is virtually uncorrelated at horizontal distances greater than about 1 m on both banks, a length scale of approximately half the bank height. This extensive streamwise variability suggests that widely spaced profile data cannot adequately represent bank erosion at these sites. Additional analysis of our comprehensive spatial data also indicates that traditional bank profile surveys with any spacing greater than 1 m would result in measurement errors exceeding 10%, an important conclusion for assessing annual rates of mercury loading into the South River from bank erosion. Our results suggest that three-dimensional gridded bare-earth models of bank topography may be required to accurately measure annual bank retreat in similar river systems. Copyright (C) 2010 John Wiley & Sons, Ltd.