Cryospheric Sciences | Highlighted Paper – The ice factories of the Arctic Ocean

Sea ice forming rapidly in a lead during springtime. [Credit: Stefan Thiele]

Each year, the Arctic sea ice goes through a cycle of melting and freezing. From March to September, sea ice gradually melts and becomes thinner, and from October to March, the water freezes again. In our warming climate, we see that more and more ice melts each year. One would expect that the ice would also freeze less, but we have observed that the ice growth – or ice production – has increased over the last decades. Yes, you read it right: despite warming, there is more ice produced during winter than before!

Ice growth-thickness feedback

How can that be even though the Arctic is warming much faster in winter than in summer (Bintanja & van der Linden, 2013)? The main reason for this apparent contradiction is that the ice nowadays is much thinner, and thinner ice can grow much faster than thicker ice. Think of a lake in winter: once the lake is cool enough, the ice takes only a few days to grow 5-10 cm, but once the ice has grown a bit, it takes much longer to grow another 5-10 cm. The sea ice grows (see also this and this blog post) from below, but the ice is cooled from above by the atmosphere. This means that all the heat released during freezing (or the ‘cold’ necessary for freezing) must travel through the ice, and these heat fluxes get smaller the thicker the ice is. This effect is a negative feedback, as it counteracts the general sea ice loss caused by warming; this is the growth-thickness feedback (Bitz & Roe, 2004).

One way the growth-thickness feedback manifests, is through the fact that nowadays, there are larger areas of the Arctic Ocean with no ice (i.e., open water) in late summer, and ice grows the fastest once it just starts freezing, since it is very thin at that point. So, the increase in open water area at the end of summer is also part of the growth feedback.

Other factors influencing ice production

But the growth-thickness feedback is not the only effect which has increased the winter ice production in the Arctic. The sea ice is usually covered by snow, and snow further protects the ice from the cold atmosphere. So, the more snow there is on the ice, the slower it will grow. Snow on sea ice is difficult to measure, but we think that the snow cover will become thinner in the future, especially in the regions of the Arctic where the sea ice melts away each summer. Part of the increase in ice production could therefore also be due to a thinner snow cover.

Another effect which is important for ice production are strong winds which break up the ice during winter and form leads or polynyas, which are areas of open water within the ice pack. In this open water, ice can grow very quickly. As the ice is becoming thinner, it is also moving faster due to winds and leads form more easily. We typically measure how much the ice is broken apart by the divergence. Divergence is a measure of how quickly things spread apart. Increasing divergence during winter could be a reason why ice production has increased because more ice is spread apart, which creating leads where new ice can be produced fast.

The tug-of-war between temperature and feedbacks

What will happen in the future? Will ice production increase further (see also this and this blog post)? Climate models indicate that at some point, the ice production will hit a peak and will decrease afterwards. The main reason for this is that the increasing temperatures will eventually overcome all the feedbacks I mentioned. First, as it gets warmer, the winters become shorter, and the ice has less time to freeze — this is already happening. Second, higher temperatures during winter will also slow down the growth itself. So, inevitably, the ice production will decrease in the future. The question is: when will that be? This depends on how long the effects I described above can fight against the increasing temperatures. How long until those negative feedbacks are overcome by global warming itself? And which of the effects will give up first?

In our recent study, we built a simple linear model that can explain changes in ice production and separate the different effects of the mechanisms: the snow depth, the available open water area at the start of winter, the divergence during winter, the ocean temperature at the beginning of the winter, and the air temperature during winter. We focused on a specific region in the Arctic Ocean: the Kara and Laptev Seas, sometimes called the ‘ice factories’ in the Arctic Ocean, because a lot of ice is produced there during winter. In Figure 1, you can see that a lot of ice is produced in this region, because ice is often pushed away from the coast by the winds, which creates open water (remember: divergence), where ice can growth quickly.

Figure 1: a) Winter ice production in a climate model and b) mean winds from satellite observations. Both are averages over 1979-2020. The blue line surrounds the study area: The Kara and Laptev Seas. [Credit: Cornish et al., 2022]

We trained our linear model on a complex climate model, where we know each of the terms exactly, and we also know the amount of ice produced each winter. The linear model does a good job of explaining how the ice production changes in that model (Figure 2). In the climate model we used, the ice production increases at first, up until 2020. We see that the most important of the negative feedbacks is the available open water (i.e. the growth-thickness feedback) and the snow depth change (green and blue lines in Figure 2). These two feedbacks ‘fight’ the most against the increasing temperatures. However, in the climate model, the negative feedbacks eventually lose to the temperature increase around 2020: from then on, the ice production decreases.

Figure 2: a) Total winter ice production in the Kara and Laptev Seas in the climate model and in the simple linear model. b) The contribution of the different mechanisms and feedbacks in the simple model to the total ice production in the climate model. [Credit: Cornish et al., 2022]

Figure 3: a) Total ice production (relative to 1982/83) estimated from observations and the contributions of the different mechanisms and feedbacks in the simple model. b) as a), but smoothed curves to see the differences more clearly. [Credit: Cornish et al., 2022]

We also used the trained model on observations over the last decades (Figure 3). We see that the most important factor in the observations is the increase in divergence (orange and grey lines in Figure 3). The second most important factor is that the snow has become thinner, which has increased ice production (blue line), just as in the climate model. In the observations, we saw that the ice production was increasing up until around 2005 but has since started declining, earlier than what the climate model predicted. We think therefore, that ice production may have also passed its peak, and the negative feedbacks have already been overwhelmed by the temperature increase in the Kara and Laptev Seas. In other words, the ice factories have begun to slowly shut down their ice production.

To summarize, there are several negative feedbacks that have managed to increase the ice production in the Arctic, despite Arctic warming and overall sea ice decline. However, these feedbacks are slowly losing the fight with increasing temperatures, which means that ice production will start to decrease in the future.