Chandler, Jim 
  
throughto DEM credion and mosaicing and significantly, is being carried out by a competent fluvial reseacher without 
any formal photogrammetric qualification or training The system certainly demonstrates the value of a fluvial bed 
measurement system based upon digital photogrammetry and has become just another tool used by fluvial researchers. 
3.4 Sunwapta River, Canadian Rockies 
The projeds reported upon above have dl involved simulated river channel flumes, short camera-objed distances (< 
4.2m) and imagery acquired in the traditional verticd perspedive. These ae dl situations where the photogrammetry 
can be planned, controlled and success would perhaps be expeded. The final projed was far more challenging 
primarily because it involved measuring the changes occurring on ared river and required the use of oblique imagery at 
an approximate distance of 235m from the riverbed. 
The Sunwapta River is a large and adively braiding, gravel-bed, pro-gladal river in the Rocky Mountains of Alberta, 
Canada, between Jasper and Lake Louise. The river is fed by the Athabasca Glader and, due to the daily fluctuation in 
air temperatures, experiences low flows in the ealy morning, building to pe« flows in the ealy evening. These 
variations are d a maximum during the summer meltwater period, when peak flows cause significant transport of bed 
sediment, causing substantial and rapid (daily) change in the streambed topography. Quantification of such changes can 
provide estimates of sediment transport rates (Lane et al., 1995 Ashmore and Church, 1998. Automated DEM 
acquisition has been accomplished in flume studies for this purpose (Stojic et al., 1998), but not in the field using 
oblique imagery. The low flows and exposed bed forms in the morning provide a oppatunity to determine 
morphologicd change and to measure bedform by photogrammetric methods. It was accepted that traditional field 
survey methods would still be required to provide surfacedata beneah the low water surface but it was hoped that the 
photogrammetry could reducefield survey time significantly and provide DEMs at high spatial and temporal densities. 
Another unique atradion of the field site was the presence of a high cliff overlooking the read, from which oblique 
and digital overlapping photographs could be obtained. This viewpoint had been used previously to obtain sequences of 
35mm photography, to identify change by qualitative means (Goff and Ashmore, 1994). Images were obtained from 3 
camera stations 45m apart; ead was locaed 125m above and 235m from the center channel of the river. The Kodak 
DC3460 camera equipped with an 85mm lens was used, and combined coverage of two overlappi ng areas excealed an 
areaof the reat that was 125 long and 80m wide. Photo-scde varied, but in the centre of the reac was 1:2,750, with 
ead pixel providing ground coverage of approximately 0.055 x 0.025m. Fifteen photo-control targets were placed upon 
prominent channel bars. These aonsisted cf black and white painted beards with dimensions 0.3 x 0.3m, constructed 
locdly and secured using sted reinforcing bars hammered-into the: stream bed. Their coordinates were derived by 
theoddite intersection methods using measurements obtained from three control stations. Average predsion following a 
lea& squares variation of coordinates estimation was -bmm 
  
Digital photographs were obtained over 
a 13-day period in July/August 1999 
initially at 9.00am in the morning but 
also at 7pm in the evening for the last 
four days of the monitoring phase. 
Images were downloaded, examined to 
assess/modify exposure settings and 
backed up on a portable PC ead day. 
Additional bed monitoring was being | ios 
caried out on a repea daily basis by 
more conventional field surveying 
methods. This consisted of measuring 16 
crossstream profiles, ead between 85m 
and 140m in length and 10m apart. The 
sampling strategy involved measuring 
points at intervals of one meter, with 
additional points introduced at 
significant breds of slope and between | '? 
these profile lines, and so approximately 
3,000 points needed to be measured each 
day. Even with acces to one motorized 
Total Station (Leica TCA1800 and four 
survey teams using levels and staffs, this 
occupied 4-5 hours of ead field day, 
followed by 1-2hrs of data entry and No jen a 
processing each evening. In contras, thé | Figure 3, Slope shaded representation of DEM, Sunwapta- 30-07-99 
1050 
1000 
  
  
  
  
  
254 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000.