Chandler, Jim 
  
MEASURING RIVER-BED AND FLUME MORPHOLOGY AND PARAMETERISING BED 
ROUGHNESS WITH A KODAK DCS460 DIGITAL CAMERA 
J.Chandler!, S.N. Lane? and P. Ashmore? 
1. Loughborough University, Loughborough, LE11 3TU, UK 
J.H.Chandler 9 lboro.ac.uk 
2. University of Leeds, Leeds, LS2 9JT, U.K. 
S.Lane Q geog.leeds.ac.uk 
3. University of Western Ontario, London, Ontario, Canada N6A 5C2 
pashmore @ julian.uwo.ca 
KEY WORDS: DEM generation, automation, accuracy, surface representation 
ABSTRACT 
Hydraulic engineers and fluvial geomorphologists need to understand how moving water flows over stream beds, and 
results in sediment transport. One critical aspect that is becoming increasingly important is gaining knowledge about the 
exact shape and morphology of water worked sediments. This paper demonstrates how automated terrain model 
extraction software combined with images acquired using a Kodak DCS460 digital camera have been effective in 
generating digital elevation models (DEMs) to represent such complex bed morphology and derive estimates of bed 
roughness. 
The automated extraction of DEMs to represent sedimentary forms created in a flume requires careful photogrammetric 
design. In addition to the normal constraints imposed by scale and photo-configuration, the estimation and stability of 
camera inner orientation are critical. The approaches adopted and recommended are illustrated by recent research 
projects carried out on large flumes at Loughborough University and Hydraulics Research, Wallingford, both in the UK. 
The methodology has also been developed and applied to a real and large braided river channel system in the Canadian 
Rockies, using oblique imagery acquired with the DCS460. These three applications show the efficacy of the approach 
and demonstrate that morphological data has been collected at significantly higher spatial and temporal densities than is 
possible using other methods available. 
Automated digital photogrammetry now provides hydraulic engineers and fluvial geomorphologists with an ability to 
measure at the bedform scale and partly at the grainscale. Manipulation of these base morphological descriptors and 
data derived from them, is becoming increasingly necessary for understanding fully, fluvial flow and sediment transport 
mechanisms. 
1 INTRODUCTION 
Riverbed morphology develops from the action of water flow and sediment movement, which are, in turn, influenced by 
the bed morphology. This link between bed form and hydraulics functions at a variety of scales, from individual grains, 
through to bedforms and to the stream channel itself. Clearly these scales are linked, with collections of particles 
creating the bedform and the spatial distribution of bedforms creating channel-scale topography. These different scales 
of bed topography influence fluvial processes in a variety of ways. For example, particle erosion and deposition depend 
upon grain size (Shields, 1936) and grain packing and bedform geometry (Kirchner ef al., 1990). 
  
Despite the significance of riverbed topography at these different scales, measurement of other parameters such as flow 
velocity and turbulence has dominated fluvial research, whilst bed topography is often reduced to simple statistical or 
descriptive values. One of the main reasons for this emphasis has been the difficulty involved with measuring bed 
morphology, particularly over the range of spatial scales. In the few studies that have involved bedform measurement, 
the difficulties have forced researchers to measure profiles, typically across the river channel (e. g. Robert, 1998; Nikora 
et al, 1998). Traditional measurement procedures (i.e. level and staff) result in slow rates of data acquisition, which can 
compromise studies in two mutually exclusive ways. Cross-sections may be spaced too widely, so that whilst change 
across the channel is monitored effectively, it is difficult to quantify downstream change. If however, an appropriate 
density of cross-sections is maintained, then frequency of survey is inevitably reduced such that data maybe sampled at. 
a frequency that is lower than the time scale of change. If understanding is desired at the smaller grainscale, then these 
issues are compounded further because of the need for a far denser sampling strategy than conventional surveying 
methods can allow. Even if the study is transferred to the flume environment where many of the practical problems 
  
250 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000.