Geology Reference
In-Depth Information
from the turbulent conditions associated with the main-
stream current of the Arctic Ocean [
Sinha,
1995]. Ice
grew without any development of snow cover and (sur-
prisingly) remained that way for the entire winter sea-
son. Since there was very little snow coverage, less brine
was incorporated and retained near the top surface of
the ice cover. The growth rate was such that the FHF ice
demonstrated very little vertical variation in its salinity
(3‰-4‰). Ice was columnar‐grained S3 type with very
similar ice characteristics, except for the snow cover,
noted in Mould Bay for the 1981-1982 growth season
[
Sinha,
1983c] as briefly described earlier and details of
5 year Mould Bay project in Chapter 5.
The morphological properties of grains and subgrains
can be quantified from the thin‐section images using any
commercially available image analysis software. Both hor-
izontal and vertical sections are required for an adequate
description of the grain structure in a polycrystalline mass.
For directionally solidified columnar‐grained S2 or S3
type of ice, horizontal thin sections are preferable for the
analysis, as they provide a satisfactory description of the
cross‐sectional structure of the grains.
Images of thin sections taken through cross‐polarized
and parallel‐polarized light provide complimentary infor-
mation, and often images taken with scattered light are
necessary for sea ice analysis. Although the grain bounda-
ries in sea ice are most easily identified using cross‐polar-
ized light, the subgrain boundaries are distinguished more
clearly using parallel‐polarized light. Both types of images
in full color are, therefore, necessary because the absence
of details in one polarized image is usually provided by
its counterpart. Moreover, images augmented by scattered
lights are also found to be appropriate for the illumination
of inclusions and thereby identifying the subgrain bound-
aries with extremely small lattice mismatch. Colored pho-
tographs are preferred for accurate identification of grain
boundaries in thin‐section photographs. One of the most
common sources of error if gray tone images are used is
the transformation of different interference colors into the
same gray value [
Eicken and Lang,
1991], thereby making
difficult the accurate identification of grain and subgrain
boundaries. Color, brightness, contrast, and sharpness in
the digitized images of thin sections can be adjusted as
required. The lack of detail potentially can interfere with
the ability of the image analysis to differentiate between
the grain boundaries and the grains themselves.
Once the digital photographic images are available,
grain and subgrain boundaries in the images can be black-
ened with the freehand tool using any available image
processing software. The utmost care is required to ensure
that inter‐ and intragranular boundaries are outlined
completely, i.e., not one pixel is absent from the darkened
margins. Punctured boundaries cause adjacent grains to
be identified as one object by the image analysis software.
The efficiency of digitized image analyzes is demonstrated
by the enhanced ability to differentiate between grains
and subgrains. Since adjacent subgrains (or cross sections
of platelets) are separated by small‐angle boundaries with
extremely small mismatch of the crystal lattice, the color
gradient between subgrains is not usually well defined.
Therefore, the mosaic pattern formed by intersecting rows
of brine inclusions is used to outline the substructure.
Close inspection of the orientation and shape of the
inclusions provide the information necessary to deter-
mine the general direction of the subgrain boundary. To
some extent, the accuracy of this method is compromised
by the intrasubgrain brine inclusions because often rows
of brine pockets do not necessarily mean any lattice mis-
match on the either side. An alternative method for the
manual delineation of grain and subgrain boundaries in
the image is to use an automated edge detection technique
available in any image processing software.
In addition to the area (which is the summation of all
pixels contained within a crystal), major and minor axes,
a few other parameters can be used to characterize the
dimensions and the shape of the crystal. These include:
(1) perimeter, which is the summation of all pixels form-
ing the boundary of the crystal; (2) feret diameter, which
is the diameter of a fictitious circular shape that has the
same area as the crystal being measured. It is equal to
(4 × area/
π
)
0.5
; (3) shape factor, a measure of how nearly
circular crystal is. It is equal to (4 × area/perimeter
2
). (4)
Compactness factor, an alternative method for assessing
the degree of circularity of a crystal. It is given by the
ratio (perimeter
2
/area).
A computer‐assisted approach to calculate the geomet-
rical parameters of grains and subgrains in digital images
of thin sections of sea ice is presented in
Johnston and
Sinha
[1995]. The approach involves four steps. The first
entails digitizing the colored photographs of thin sections
into 256 gray tone images. In the second step the digitized
images of the same thin section (from cross‐ and paral-
lel‐polarized images) are input to a desktop publishing
software to determine the location of grain boundaries
(Figure 4.44a). The absence of details in one polarized
image is usually provided by its counterpart. In a case
where boundaries could not be discerned from either
image, they should be blackened with the freehand tool
of the software. The mosaic pattern formed by intersect-
ing rows of brine inclusions can be used to outline the
substructure. In the third step the digitized image with
defined grain and subgrain boundaries is converted to a
binary image with grains presented in white and their
boundaries in dark tone. This is shown in Figure 4.44b.
This step is preferred because color is not essential to
the actual analysis. In the fourth step an image analysis
