4. LARGE SCALE MAPPING OF RASKELF A suitable area for the survey, the village of Raskelf, was
chosen close to the airfield in North Yorkshire. The
4.1 Aerial Survey village was covered in two flight lines of 14 images with
the camera’s major axis parallel to the flight direction,
The results from the Newcastle survey, whilst With 50 % fore and aft overlap, a base to height ratio es
encouraging in terms of suitability for identifying areas of approaching 0.25 was possible. The flying height of
change, do not suggest that the technique lends itself ^ 500 m provided an image scale of 1: 17,000 (ground
readily to mapping at large scales. If, however, larger Pixel size of 0.16 m). From the block of ‘photography’,
scale imagery could be obtained using the camera then three successive images were chosen for further work.
the technique may be useful in terms of mapping small Further survey details can be found in Mills et a/. (1996).
areas from scratch. Unfortunately, the 4 second image
download time of the DCS200 will not allow sufficient — 4.2 Ground Control Survey E
overlap to be acquired at larger scales in an aircraft with 3
stalling speeds as high as that of the Cessna. As a Ground control for the survey was carried out using a e
result, it was decided to use a microlight aircraft to obtain Leica TC 400 total station. This is a 10 second =
imagery at a larger scale with sufficient overlap. instrument with EDM measuring to +/-5 mm +5 ppm
Although the use of such aircraft is not allowed for RMS error. A six station traverse was observed around
commercial purposes at present in the United Kingdom, the survey site and natural ground features, identifiable
the method provided the opportunity to achieve the kind on the photography, were co-ordinated on a local co-
of ground resolution, all be it over a much smaller area, ordinate system by radiation. A total of 60 points were
that will be available with cameras of the future. co-ordinated in a single morning. The traverse computed
with a misclosure of 22 mm in plan and 13 mm in height.
The least squares adjustment of the data showed the
worst GCP to have an error ellipse of major axis 16 mm
(0.1 pixel) and a minor axis of 14 mm (0.1 pixel) with an
RMS error for height of 26 mm (0.2 pixel). These figures
are comfortably within the ground pixel size of 0.16 m,
allowing a true assessment of the camera's potential for
mapping to be made. Sixteen well distributed GCPs
were chosen for controlling the three images so as to 9
leave some points for use in the assessment of absolute
accuracy. The orientation results are shown in Table 3.
Image 41 Image 42 | Image 43
X (m) 832.90 947.12 1060.49 e
LM ui Y (m) 940.87 954.28 966.48
Figure 1: The Thruster microlight aircraft. Z (m) 596.43 596.67 596.98
o (degrees) 1.678 2.694 3.749 4.4
The Thruster microlight that was used in the project can 6 (degrees) -2.730 -3.116 3.971
be seen in Figure 1. This is a 60 hp aircraft with a (degrees) 0.896 2.472 1251 The
stalling speed of 18 m/s, enabling large scale image 9 h 3 99 23 101 51 stere
acquisition with the DCS200. Details of the Thruster can |-Datum ht. (m) | 97.22 I
be found in Graham (1988). Because of the cramped | B/H ratio 0.23 | 0.22 Mea
nature of the cockpit, the camera was slung behind the No. of GCPs 8 9 8 surv
cockpit and fired by the co-pilot using a cable release. RMSE (pixels) 0.63 0.61 0.62 grea
The mount (Figure 2) was constructed of lightweight RMSE (m) 0.10 0.10 0.10 expe
aluminium with foam padding and was suspended Using ^ Table 3: Orientation parameters and GCP residuals of Neve
tapes So. a5 to.Do-froo. Of Vibration, the imagery used in the Raskelf survey. direc
usin
4.3 Photogrammetric Processing Le
Processing was carried out in a similar manner to the Th
Newcastle survey. The DEM was this time created with a 9
post spacing of 10 pixels (1.6 m), the justification for this effec
being that it was a much more continuous surface than accu
the city centre site in Newcastle and processing time heigl
could be saved. The two DEMs produced from the two ratio
overlaps were mosaiced together and the central image para
of the three overlaid onto this to create the pixel
orthophotomap. The detail was then plotted directly por
using the planimetric map option in the DMS. The vector Com
data was subsequently edited in AutoCAD and then S
overlaid onto the orthophotomap to produce Fours 3. The
s d ; om
Figure 2: The DCS200 camera mount. IS S à Borton of the 250 x 150 m area plotted T accu
566
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996
