268
12.3
l(eMotion
12.3.1
Overview
A.J. SEPHTON AND K.C. PARTINGTON
Ice motion is of interest in a number of contexts, identified by Carsey and Holt (1987),
as: latent heat advection, oceanic surface stress, passive tracing of currents, production
of open water by way of ice divergence and shear, and interaction between icebreaker
navigation and ice structure. In addition, ice motion is of importance in an operational
context, such as in giving an indication of where the ice is compressing and therefore
of danger to a vessel (perhaps forming pressure ridges), where (navigable) leads are
likely to form, and for drilling platforms to take evasive action if threatened by the movement of heavy ice.
Sea ice moves in response to forcing by surface winds and ocean currents, and to
stresses from and within the ice itself. In the central ice pack and marginal ice zone
(MIZ), the complex nature of the relative ice motion under divergence and shear is a
very active research area. Convergent stress conditions in the ice cover result in the production of pressure ridges, which contain a large fraction of the total sea ice volume
and are operational hazards, whereas divergent stress conditions produce areas of open
water (leads and polynyas). Through these areas of open water, significant exchanges
of heat and moisture occur. The varying coverage of leads and polynyas also result in
varying fluxes of brine in the sea.
In the Arctic, large-scale ice motion is commonly along atmospheric surface isobars,
and has been measured by tracking buoys and individual floes (e.g., Colony and
Thorndike 1984). The effect of wind and ocean currents causes most of the ice to rotate
around the Arctic basin. However, ice is continually swept south through the Fram
strait, and the warm West Spitzbergen current flows north, creating an ice-free area to
the south-west of Svalbard. The presence of land creates zones of shear, such as that
found within the area north of Alaska. Over scales less than 100 km however, ice motion
can be quite irregular as floes ranging in size up to tens of kilometers translate and
rotate as rigid bodies between other tightly-packed floes.
In order to resolve the spatial structure of this smaller-scale ice motion, a large number of measured trajectories of floes is needed, with a data sampling density of approximately 1 km- 2 .As a means of achieving such sampling rates over large areas, the use of
sequential satellite images is now generally considered to be the most effective method
of studying small-scale ice motion, the resolution of the imaging system being the limiting factor in the size of feature that can be identified. Estimates of ice motion made
from satellite image data have revealed a mean motion of some 5 km / day in the pack
ice, and approximately 15 km / day in the MIZ (Curlander et al. 1985). Ito and Muller
(1982) have reported observing ice velocities in satellite imagery of up to 34 km / day,
whilst velocities up to 80 km / day are not unknown.
Two main categories of automatic and semiautomatic methods have been developed
in recent years to track the motion of sea ice in remote sensing data between image
pairs, covering the two main types of ice condition, i.e., pack ice and marginal ice. In
pack ice, floe boundaries are generally indistinguishable but the overall motion is
mainly translational (over short time scales); in marginal ice, floe boundaries are gen-
12.3
l(eMotion
12.3.1
Overview
A.J. SEPHTON AND K.C. PARTINGTON
Ice motion is of interest in a number of contexts, identified by Carsey and Holt (1987),
as: latent heat advection, oceanic surface stress, passive tracing of currents, production
of open water by way of ice divergence and shear, and interaction between icebreaker
navigation and ice structure. In addition, ice motion is of importance in an operational
context, such as in giving an indication of where the ice is compressing and therefore
of danger to a vessel (perhaps forming pressure ridges), where (navigable) leads are
likely to form, and for drilling platforms to take evasive action if threatened by the movement of heavy ice.
Sea ice moves in response to forcing by surface winds and ocean currents, and to
stresses from and within the ice itself. In the central ice pack and marginal ice zone
(MIZ), the complex nature of the relative ice motion under divergence and shear is a
very active research area. Convergent stress conditions in the ice cover result in the production of pressure ridges, which contain a large fraction of the total sea ice volume
and are operational hazards, whereas divergent stress conditions produce areas of open
water (leads and polynyas). Through these areas of open water, significant exchanges
of heat and moisture occur. The varying coverage of leads and polynyas also result in
varying fluxes of brine in the sea.
In the Arctic, large-scale ice motion is commonly along atmospheric surface isobars,
and has been measured by tracking buoys and individual floes (e.g., Colony and
Thorndike 1984). The effect of wind and ocean currents causes most of the ice to rotate
around the Arctic basin. However, ice is continually swept south through the Fram
strait, and the warm West Spitzbergen current flows north, creating an ice-free area to
the south-west of Svalbard. The presence of land creates zones of shear, such as that
found within the area north of Alaska. Over scales less than 100 km however, ice motion
can be quite irregular as floes ranging in size up to tens of kilometers translate and
rotate as rigid bodies between other tightly-packed floes.
In order to resolve the spatial structure of this smaller-scale ice motion, a large number of measured trajectories of floes is needed, with a data sampling density of approximately 1 km- 2 .As a means of achieving such sampling rates over large areas, the use of
sequential satellite images is now generally considered to be the most effective method
of studying small-scale ice motion, the resolution of the imaging system being the limiting factor in the size of feature that can be identified. Estimates of ice motion made
from satellite image data have revealed a mean motion of some 5 km / day in the pack
ice, and approximately 15 km / day in the MIZ (Curlander et al. 1985). Ito and Muller
(1982) have reported observing ice velocities in satellite imagery of up to 34 km / day,
whilst velocities up to 80 km / day are not unknown.
Two main categories of automatic and semiautomatic methods have been developed
in recent years to track the motion of sea ice in remote sensing data between image
pairs, covering the two main types of ice condition, i.e., pack ice and marginal ice. In
pack ice, floe boundaries are generally indistinguishable but the overall motion is
mainly translational (over short time scales); in marginal ice, floe boundaries are gen-