Chandler, Jim 
  
individual particles being difficult to discern. For the Tilti ng Flume, the median grain size was much larger (24mm) 
with each pixel occupying an area of just 0.6mm so that individual particles were deatly defined. The consequence of 
this variation is that for the Tilting flume, a large propartion of pixels lie in regions between particles and height 
estimates in these aeæ are unreli able because of the obstruction of light rays by adjacent particles. For the FCF study, 
median gainsize ad pixel sizeis comparable and areas between particles are simply not represented in the imagery. It 
is important that careful attention must be given to the structure of the surfacebeing studied and the relative size of 
pixel and particle size Comparing the two studies has shown that acaurades are not diredly dependent upon photo- 
scde. Reducing the camera devation can increase the dead ground problem, which dea eases surface acuracy. 
One superficial conclusion from these studies would be that the surface aquired for the FCF is more acurate and 
therefore of greder value than the Tilting Flume DEM. In some respeds this gatement would be valid but this is a 
simplification. It would be more predse to state that the FCF surfaceis more acerate, provided that the desired surface 
is required to represent just the bedforms, (i.e. acarate à the bedform-scde). If the fluvial study required information 
regarding the composition and inter-relationships between the grains (i.e. information at the grain-scale) then the FCF 
surface must be mnsidered inadequate. The Tilting flume DEM may be slightly inaccurate for those points between 
particles but it certainly contains information at the grainscde. The important conclusion is that the DEM must be 
generated at a resolution that is appropriate for the phenomena that are being investigated. It is interesting to note that 
this point is dated clealy by the Nyquist theorem (Graham, 1998 and although it's usage is well established amongst 
photogrammetrists for image aquisition, it israrely used for considering generated data, such as a DEM. 
In the cae of the Surwapta projed, the image pixel size was similar to the median gainsize of the particles and the 
resolution of the extraded DEM was 0.20m. As the projed desired spatial data à the bedform scde, both the pixel size 
and DEM resolution was appropriate for this cde of enquiry. 
5 CONCLUSION 
It has been shown that a measurement system based upon automated digital photogrammetry and a digital camera can 
be usefully used to measure the morphology of fluvially worked surfaces, both in a flume and in a red river. As 
expeded, acarades were found to be dependent upon normal photogrammetric controls such as de, camera 
cdibration and oHiquity of view, but only partly. It has been shown also that the relative relationship between the size 
of the particles which constitute the bed and pixel dimensions on the objed are aiticd. If the pixel sizeis larger than 
individual particles then the surfacemay be accrate à the bedform scde but will be inacarate a the grainscale. The 
concept of different scales of representation is of grea importance to the quality of any landform representation, it 
being essential for DEMsto be generated at a scde that is appropriate for the purposes intended. 
ACKNOWLEDGEMENTS 
The authors gratefully acknowledge the financial suppart provided by EPSRC (UK) grants GR/L58118 GR/L69213for 
the flume studies carried aut at Hydraulics Reseach Wallingford. The collaborative research carried out in Canada was 
suppatted by The Royal Society (UK), NSERC (Canada) and an internal research gant from the University of Western 
Ontario. 
REFERENCES 
Ahmad, A. and Chandler, J.H., 1999 Photogrammetric capabiliti es of the Kodak DC40, DCS420 and DCS460 Digital 
Cameras, Phtogramretric Record, 16(94):601-615. 
Ashmore, P.E. and Church, M. 1998 Sediment transport and river morphology: a paradigm for study. In Gravd-bed 
rivers in the Environment. P.C. Klingeman, R.L. Beschta, P.D. Komar and J.B. Bradley. (Editors), Water Resources 
Publications LLC, Highlands Ranch, Colorado, Chapter 7, 115-148. 
Brunsden, D., and Chandler, J.H., 1996 The continuing evolution of the Black Ven mudslide, 194695. Theories put to 
the test, in Brooks, S. and Anderson. M., (Eds), Advances in hillslope processes, Wiley, 869-898 
Butler, J.B., Lane, S.N. and Chandler, J.H., 1998. Assessment of DEM quality charaderising surfaceroughness using 
close range digital photogrammetry. Phatogramnetric Record, 16, 271-91. 
Butler, J.B., Lane, S.N. and Chandler, J.H., in press Application of two-dimensional fradal analysisto the 
charaderisation of gravel-bed river surfacestructure. Paper accepted for a spedal edition of Mathematical Geology. 
Chandler, J.H., 1999 Effedive application of automated digital photogrammetry for geomorphologicd research, Earth 
Surface Processes and Landforms, 24, 51-63, 
Chandler, J.H., Lane, S.N., and Shiono, K., 1998. Acquisition of digital elevation models at high spatial and temporal 
resolutions using automated digital photogrammetry. http://www-staff .|boro.acuk/-cvihc/HRWall Rep.html 
  
256 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000.