HR: 1340h
AN: H53C-1264    [Abstracts]
TI: Modelling Fluvial Sediment Budgets Under Uncertainty
AU: * Brasington, J
EM: jb10016@cam.ac.uk
AF: University of Cambridge, Dept. of Geography, Cambridge, CB2 3EN United Kingdom
AU: Wheaton, J M
EM: joe.wheaton@soton.ac.uk
AF: University of Southampton, School of Geography, Southampton, SO17 1BJ United Kingdom
AU: Williams, R D
EM: richdwilliams@hotmail.com
AF: University of Cambridge, Dept. of Geography, Cambridge, CB2 3EN United Kingdom
AB: The last decade has witnessed a resurgence of interest in morphological estimation of fluvial sediment budgets, facilitated by major advances in survey technology, including airborne lidar and photogrammetry and ground-based GPS. The reliability of such morphologically inferred budgets is controlled by: a) uncertainty in the flux boundary conditions; b) survey frequency; and c) DEM quality. A body of research has sought to identify the sensitivity of derived parameters to these controls and establish a methodological framework for data quality control and assurance. To date, most interest has focused on evaluating the uncertainty in budget estimates due to DEM errors. These arise as a largely unknown function of survey point quality, sampling strategy and interpolation methods. A commonly adopted procedure for managing these uncertainties involves specifying a minimum level of detection threshold (LOD) to distinguish actual surface changes from inherent noise. Determining the LOD requires both a theory of change detection and a metric of DEM quality. Typically this is achieved by applying the classical statistical theory of errors and a measure of DEM precision derived from check data or point precision estimates. Research presented here aims to demonstrate that simple thresholding of DEMs of difference may, however, significantly underestimate the information that can be optimally retrieved through DEM differencing. Analyses are based on a series of annual surveys of the River Feshie in the Scottish Highlands using high-quality rtkGPS and lower precision airborne photogrammetric data. A new methodology for change detection is presented which incorporates: (i) a stepwise analysis of errors arising during DEM construction; (ii) the development of a spatial filter to group areas of scour and fill; and (iii) alternative methods for analysing change data which relax the assumptions of the LOD approach. These latter strategies explicitly incorporate uncertainties in DEM data and permit sediment budget calculations to be presented in a stochastic framework. The results suggest that enhanced data retrievals are possible and have the potential to recast the geomorphic interpretation of process rates.
DE: 1815 Erosion and sedimentation
DE: 1821 Floods
DE: 1824 Geomorphology (1625)
DE: 1894 Instruments and techniques
DE: 1640 Remote sensing
SC: Hydrology [H]
MN: 2004 AGU Fall Meeting