Geology Reference
In-Depth Information
calculated (average actual couplet thickness
6.8 cm, 5.5 cm brick, 1.3 cm mortar). When the
photograph was taken with the focal plane per-
pendicular to the wall, the couplets were found
to have an average thickness of 31 pixels, result-
ing in a resolution of approximately 2.2 mm of
wall length per pixel. Importantly, because the
focal plane is parallel to the wall surface, there
is no trend in the thickness data (Fig. 7b). Fourier
power spectrum calculated using standard tech-
niques (Davis, 1986) from grey-scale data derived
from the parallel focal plane photograph dis-
plays a prominent single peak at 31 pixels. Thus,
when the 'outcrop' is vertical and the focal plane
is placed in a parallel orientation, the positional
data contained within the grey-scale series is reli-
able. Unfortunately, this is not the case when
there is an angle between the focal plane and
the surface.
When the camera's focal plane is turned 20° to
the wall surface, as illustrated in Fig. 6, the fore-
shortening that is produced results in a strong
linear trend of decreasing apparent brick-mortar
couplet thickness up-section, as measured in pix-
els (Fig. 7e). Interestingly, while the cyclic pattern
of the brick-mortar couplets is still evident in the
grey-scale series, the trending in the data produces
a signifi cant degradation in the spectral power of
the couplet (Fig. 7f). The single strong peak found
in the parallel data is broken into a number of
smaller peaks of much lower spectral power posi-
tioned from 33 to 26 pixels, the range of apparent
couplet thicknesses found in the series (Fig. 7e).
If the angle between the focal plane and the
surface is increased to 30°, an angle of step-back
from the vertical fairly common to natural out-
crops such as the Picco di Carnizza, the validity
of interpretations drawn from grey-scale analysis
becomes even more questionable. With increasing
angle, the slope of the thickness trend increases,
resulting in a further degradation of the quality of
the spectral power of the couplet (Fig. 7i). Now,
the known peak at 31 pixels is further reduced
in power, with additional peaks down to a thick-
ness of 20 pixels present. Obviously, in this
simple example it would be possible to detrend
the data so as to return a result similar to the
parallel example (Fig. 7a-c). In a natural setting,
however, establishing the correct stretching factor
for detrending would be extremely diffi cult.
Unquestionably, the analysis of grey-scale ren-
derings of fi eld photographs can be considered
valid only when great care is taken in the selection
of the geometry of the photographic angle relative
(a)
3
2
1
(b)
3
2
1
Fig. 6. Cartoon illustrating the modelling of outcrop
step-back by changing the angle between a vertical brick
face and the focal plane of the camera.
weathering and erosion. As a result, an outcrop
surface nearly always steps back at an angle from
the vertical. This recession of the face of the rocks
away from an observer results in a foreshortening
of rocks lower in the section relative to those
in stratigraphically higher positions. In order to
evaluate the effect of outcrop angle on grey-scale
analysis of fi eld photographs, a simple experiment
was conducted. Modelling of different outcrop
surface geometries was accomplished by photo-
graphs taken of a brick wall with the focal plane
of the camera tilted at various angles relative to
the vertical brick surface (Fig. 6).
With the focal plane held parallel to the brick
surface, the degree of vertical distortion of the
image is minimized. As the angle is changed,
and the camera tilted up, the upper portion of the
image becomes compressed relative to the lower
portion thus modelling the effect produced by
the geometry of back-stepping outcrop surface
photographed with a vertical focal plane (Fig. 6).
By changing the angle between the brick surface
and the camera's focal plane from 0° to 20° to 30°, it
is possible to evaluate the role of outcrop surface
geometry on the grey-scale technique. Photographs
of the brick surface were scanned, cropped to a
half-brick width, and then sampled following the
process described by Cozzi et al . (2005) using the
image analysis shareware Scion Image.
In all three trials, the grey-scale series clearly
shows the presence of brick-mortar couplets
(Fig. 7a, d and g). Using the position of local
minima in the grey-scale values of the data series
allows for the thickness of each couplet to be
Search WWH ::




Custom Search