Terrestrial Laser Scanning of grain roughness in a gravel-bed river

This paper demonstrates the application of Terrestrial Laser Scanning (TLS) to determine the full population of grain roughness in gravel-bed rivers. The technique has the potential to completely replace the need for complex, time-consuming manual sampling methods. Using TLS, a total of 3.8 million data points (mean spacing 0.01 m) were retrieved from a gravel bar surface at Lambley on the River South Tyne, UK. Grain roughness was extracted through determination of twice the local standard deviation (2 sigma(z),) of all the elevations in a 0.15 m radius moving window over the data cloud. 2 sigma(z) values were then designated to each node on a 5 cm regular grid, allowing fine resolution DEMs to be produced, where the elevation is equivalent to the grain roughness height. Comparisons are made between TILS-derived grain roughness and grid-by-number sampling for eight 2 m(2) patches on the bar surface. Strong relationships exist between percentiles from the population of 2 sigma(z) heights with measured a-, b-, and c-axes, with the closest matches appearing for the c-axis. Although strong relationships exist between TLS-derived grain roughness (2 sigma(z)), variations in the degree of burial, packing and imbrication, results in very different slope and intercept exponents. This highlights that conventional roughness measurement using gravel axis length should be used with caution as measured axes do not necessarily represent the actual extent to which the grain protrudes into the flow. The sampling error inherent in conventional sampling is also highlighted through undertaking Monte Carlo simulation on a population of 2000 clasts measured using the grid-by-number method and comparing this with the TLS-derived population of grain roughness heights. Underestimates of up to -23% and overestimates of up to + 50% were found to occur when considering the D-84, and -20% and overestimates of up to + 36% were found to occur when considering the D-50. (C) 2009 Elsevier B.V. All rights reserved.