accordingly. The average error without regard to sign was 6.5 mm and the
maximum error was 20 mm, or 10% of the length. The assumptions made were
that the camera axes were parallel and vertical and that the lens centre to
image distance was exactly known. These conditions can almost be achieved
with a lot of time-consuming use of spirit levels but such is not the norm
for underwater work.
The same set of stereopairs were re-measured (standard postcard prints
were used with a simple ruler) and the observations processed in a computer
program which made none of the above assumptions. The image to lens centre
distance was solved for simultaneously with the camera orientations as its
value does seem to vary, photograph to photograph, by up to 200 microns.
This might depend on how firmly the camera lens is pushed into the Nikonos
camera body (wear in the sealing rubber 0 - ring) or be due to movement of
the pressure plate, or both. The results gave an average error without
regard to sign of 1.5 mm with a maximum error of 7 mm.
Over 3000 stereopairs have been photographed by Dr. T. Done of AIMS and
whilst comparatively few have been analysed, the advantages of photography as
the means of data capture are again emphasised. To accumulate such a large
data base by conventional methods would have involved vast amounts of ship
time. The experiments on growth patterns can be done "hands off" so that
specimens studied are not destroyed. The photography can be done by assistant
staff and, using the accuracy attainable from photogrammetry, time series can
be studied for finer variations over shorter time spans.
In a second study, the 15 mm wide angle Nikkor lens was used with a
single camera to take 360 overlapping photographs of a coral reef surface 10
metres by 40 metres in extent. Thirty-three strips, each ten metres long,
were photographed down the reef face from a depth of 2.7 metres to 26.2 metres.
Ground control was provided by steel pegs placed at the ends and in the middle
of each strip. Slope distances were measured from peg to peg and later these
distances were reduced to horizontal as depths were obtained for each peg. A
total of 243 distances were simultaneously adjusted in the program VARYCORD
by courtesy of the Division of National Mapping, Canberra, and co-ordinate
values for the 99 pegs adopted. The author is currently (late 1981) attempt-
ing to perform a block adjustment of these photographs so that the spatial
distribution of the coral communities can be mapped.
170