Geology Reference
In-Depth Information
in winter. Another study that used visible and infrared
imagery to manually map the distribution of leads in the
western Arctic is presented in Miles and Barry [1998].
They presented a 5 year lead climatology for Arctic ice by
utilizing Landsat multispectral scanner (MSS) images
(80 m resolution) with the operational line scanner (OLS)
images (600 m resolution). Röhrs and Kaleschke [2010]
have developed a semiautomated algorithm to detect sea
ice leads in the Arctic from a pair of passive microwave
images with a 6.25 km pixel size.
To quantitatively analyze the large‐scale relationships
between the wind field and the lead orientations, Barry
et al. [1989] used the following correlation coefficient:
12
/
2
2
N
N
Opennings in ice
1
rN
B
cos
sin
(9.8)
Consolidated
ice sheet
i
i
i=1
i=1
where δ is the difference between the lead orientation ϕ
and the geostrophic wind direction δ = ( ϕ θ ); N is the
number of examined grid cells.
A large r B indicates that ϕ is correlated with θ and vice
versa. Therefore, r B can be used as an indicator of the
variability of lead patterns, which can be attributed to
atmospheric forcing. For data obtained on 12 May 1983
(similar to the data shown in Figure  9.9) the average r B
over 62 grid points was 0.80 over an interval of 2 days,
0.91 between two successive days, and 0.90 for the same
day. From several series of weekly VIS and TIR images,
Barry et al. [1989] noted that in some cases lead orienta-
tions may change within 24 h of redirection of the main
wind field while in other cases they exhibit substantial
persistence of orientation.
Quantitative approaches, either automated or semiau-
tomated, have been developed to identify leads in remote
sensing data. One approach involves examination of
the  probability distribution of a measured radiometric
parameter that exhibits a substantial difference between
ice and OW (e.g., albedo or brightness temperature).
Ideally, in a scene that comprises sea ice with leads, the
distribution should be bimodal with separate modes rep-
resenting main ice cover and leads (whether open or
refrozen). In this case a simple threshold can be selected
to separate the two entities. This idealization may not be
fulfilled in many cases when the observations from the
two entities overlap, and therefore this approach may not
be robust. This is particularly true in the case of a rela-
tively coarse resolution sensor such as AVHRR when the
footprint includes a mix of ice and small leads. The
“mixed” pixels eliminate the bimodality of the distribu-
tion. In spite of this possible drawback Lindsay and
Rothrock [1995] used this approach to detect subpixel‐
sized leads in a series of AVHRR imagery acquired
throughout 1989 in the Arctic using observed thermal
Refrozen leads
Fresh leads
Figure 9.8 Bands of leads formed in the Beaufort Sea, imaged
by AVHRR in February 2013. The clockwise motion of the
Beaufort Sea Gyre is indicated in the top image. Note the prop-
agation of leads with time (top to bottom images). Original
open leads that appear dark in the top images became refrozen
with brighter signatures in the bottom image. Note the fresh
coastal lead near Banks Island in the bottom image (frames
from animation available from NASA's Earth Observatory,
annotated by the authors).
1  and 2 May 1985, respectively. The sea level pressure
data, an indicator of the geostrophic wind direction, were
obtained from the day before the lead observations in the
DMSP remote sensing images. Note the tendency for
leads to be arranged roughly parallel to the geostrophic
wind direction. This correlation was observed on other
dates from different years in the same study. The authors
concluded that major lead orientations could be
“modeled” given appropriate geostrophic wind data.
They also found that winter months were less likely to
show a strong relationship between lead patterns and
wind fields, as a result of the compact sea ice conditions
Search WWH ::




Custom Search