The development of an automated correction �procedure for digital photogrammetry for the study of wide, shallow, gravel-bed rivers

Abstract

This paper develops an automated correction procedure for dealing with point errors associated with through-water photogrammetry, for application in the study of clear-water, shallow gravel-bed rivers. The procedure involves combining digital photogrammetry and image analysis techniques to: (i) correct for the effects of refraction at an air–water interface; and (ii) eliminate and reinterpolate points where the bed has not been ‘seen’. The correction procedure was applied to raw digital elevation models (DEMs) generated using digital photogrammetry from 1:3000 scale aerial photography of a small reach of the North Ashburton River, New Zealand. The accuracy of corrected and uncorrected DEMs is evaluated using an independent data set. A measure of ‘geomorphological usefulness’ as well as DEM external reliability is obtained from calculations of water depth distributions and mean bed level. Results show that digital photogrammetry, used in conjunction with image analysis techniques, can successfully be used for extracting high-resolution DEMs of gravel river beds. In exposed areas, errors are small and random, tending to cancel out over large numbers of points. Where water is shallow, and following correction, point elevation errors are statistically no different from those for exposed zones. In deeper water, despite an improvement following application of the correction procedure, elevation errors scale with water depth. The geomorphological potential of photogrammetric survey of large, gravel river beds is demonstrated by the ease and accuracy of calculations of water depth distribution (important for the assessment of a river's ecological and recreational characteristics) and mean bed level (important for the calculation of reach-scale sediment volumes). Copyright © 2000 John Wiley & Sons, Ltd.