Satellites have been observing Arctic sea ice cover
since the late 1970s, and over the last few decades
the Arctic has undergone drastic changes.
These observations document a continued decline of sea ice,
with a record minimum summer extent reached in 2007,
and again this year in 2012.
Surface air temperatures in the Arctic have increased much more
than the increase observed in the global mean,
a phenomenon called 'polar amplification,' and a
phenomenon that's projected by climate models.
Yet, climate model projections still show a very large spread
in terms of simulated Arctic ice cover over the next 100 years.
At MIT, our goal is to improve the representation of sea ice,
and the underlying ocean, in model simulations,
and to develop methods to combine these model simulations
with available observations of Arctic sea ice cover,
to then produce a best possible estimate of the state of the Arctic,
its evolution today, and its potential evolution in the future.
This animation is an observation-constrained model simulation,
looking at effective ice thickness in the Arctic, and
the simulation goes between 1992 and 2009.
So, we're looking at the North Pole here, and an ocean that
is enclosed on the one side by Siberia; on the other side by Canada and Greenland.
This part of the world gets very cold in the winter, and
sea ice forms as the ocean surface freezes and ice forms on top of it.
As you follow the animation, a couple of striking things stand out:
we see an increase and decrease in Arctic ice cover from season to season.
In late winter, most of the Arctic is covered with sea ice,
and with the arrival of spring, the ice cover shrinks,
until it reaches a minimum extent, which is usually sometime in September.
At that point, larger parts of the Eastern Arctic,
for example the Siberian shelf, are ice free.
At the same time, thick ice piles up against the coast
of Canada and Greenland, something that you see here in red.
All ice that survives the melt season is called perennial, or multi-year ice,
and this ice can survive for more than 5 years in the Arctic.
However, observations show that over the last few decades,
Arctic sea ice has been getting younger, and thinner.
In addition to melting, another important process in the loss
of Arctic sea ice is its drift.
Ice is moved around along with the wind and ocean currents.
As the ice flows, it converges, it will deform and pile up into ridges,
that can be easily seen as sails above the sea surface,
and extensive keels below the sea surface.
There is evidence that sea ice in the Arctic has become overall weaker,
which means it more easily deforms, more easily opens up gaps
within the ice pack, called leads, and it more easily drifts.
In 1893, Fritof Nansen undertook a drift aboard the Fram,
following evidence of an Arctic Ocean current known as the Transpolar Drift.
Nansen froze his ship into the ice, and waited for the drift
to carry her towards the North Pole. The drift took about 3 years.
In 2006, the European sailboat 'Tara' repeated this drift
across the Arctic within the ice pack.
It look just over one year, this time, to complete this drift.
The reason for the shorter drift is likely a combination of
changed ocean currents in response to changed atmospheric circulation,
and a weaker more deformable ice pack that drifts more swiftly.
As we go forward, we need to improve our understanding of
processes in the Arctic, processes governing sea ice fluctuations
and sea ice change.
We need to improve their representation in models that we use for,
for example, projecting how sea ice in the Arctic will develop in the future.
We also need to improve our observational capabilities
so as to validate our models, and to obtain a really good
estimate of what the Arctic is doing today, in order to then
forecast what the Arctic is doing in the future.
These forecasts will undoubtably be used, for example,
for economic planning, for resource exploration in the Arctic,
and given the very delicate environment of the Arctic,
the resource exploration needs to be done with a very good understanding
of the potential risks involved.
All of this requires very good understanding,
good models, and very good observations.