How to predict when the Arctic sea ice will reach its lowest level

The Arctic sea ice extent is at its lowest winter level ever. Does it mean we will see a new minimum record this summer?

Looking at the observational record of sea ice, there is no doubt that the Arctic sea ice cover has been declining during the last few decades. The decline occurs for all seasons, but is strongest in September, when Arctic sea ice reaches its annual minimum. However, it is more difficult to identify and prove the reasons for the observed decline.

There is a strong physical argument that the decrease is caused by climate change: an increase of atmospheric CO₂ leads to an increase in long-wave radiation, which directly results in increased ice melt. But the climate system is not that straightforward. As warmer and colder summers occur in the UK, there are years with more and less sea ice in the Arctic. Natural variability also plays an important part.

I was one of three meteorologists from the University of Reading involved in a recent Nasa-led study on forecasting summer Arctic ice levels, published in the Earth's Future journal. We are part of the Centre for Polar Observation and Modelling (CPOM), studying processes in the polar latitudes that can affect Earth's polar atmosphere and global sea levels.

When we make a seasonal prediction for sea ice minimum in September, we are interested in the inter-annual variations. The climatological trend is removed and we predict the anomalies with respect to this trend. In our latest study, we use the first day on which melting occurs, sea ice concentration and melt pond area as predictors for September ice extent.

Such predictions a couple of months ahead can never be perfect, because we do not know the atmospheric conditions during summer, and these summer conditions naturally modify the sea ice state. For time scales of more than 10 days, the atmosphere behaves as a chaotic system. However, prediction skill is gained from important feedback mechanisms.

When snow and ice start melting, some of the melt water forms so called melt ponds on top of the sea ice. Ice that is covered by melt ponds absorbs more solar radiation than bare or snow-covered sea ice causing a further increase of ice melt. When leads form within the pack ice, the water absorbs more solar radiation heating the upper layer of the ocean. This results in an increase of bottom and lateral ice melt further increasing the open water fraction.

Understanding and accurately implementing these feedback mechanisms into climate models is crucial to improving long-term sea ice projections, and to be able to separate the contribution of internal variability and climate change.

Drag slider

to compare photos

Previous studies have shown that the observed decrease of Arctic sea ice cannot be explained only by internal variability. That means that the decline of Arctic sea ice is one of the strongest indicators of climate change. Having said that, the total decrease cannot be attributed to climate change. Several model studies revealed that 30 to 50% of the decrease could be caused by internal variability. Thus, it is possible that the Arctic summer sea ice extent might not decrease further during the next 10 years. Indeed, there might be years with more sea ice than during last summer.

On the other hand, the Arctic sea ice has become thinner and more vulnerable. An extreme year in which several factors are combined – advection of warmer water into the North Atlantic (increase of bottom and lateral melt), a mild winter (less ice to start with), a sunny and warm period in May and June (formation of melt ponds) and strong Arctic cyclones during summer (increasing the open water fraction) – could already melt most of the summer sea ice under current climate conditions. This is not very likely, but cannot be excluded.

Altogether, this makes it impossible to predict the precise year when the Arctic will be ice-free in summer (defined as an ice extent below one million km2) even for a "perfect" climate model. However, seasonal predictions are possible due the described feedback mechanisms.

In April, we will be able to make our first prediction of how much sea ice we can expect this summer. Ongoing research indicates that it will soon be possible to make regional predictions – to forecast a couple of months in advance when and where in the Arctic the sea ice will disappear.

David Schroeder is a Senior Research Fellow at the University of Reading Meteorology department