Arctic Sea Ice

As we approach the final stretch...

For years we've talked about one of the most fascinating phenomena of Arctic sea ice loss as it progresses from year to year: The moment when the ice disappears regardless of the weather. We've seen some of that, back in July 2012, when extent and area numbers kept going down steadily, even though weather conditions weren't conducive to melt and should have caused a marked slowdown on the graphs (like they did in the preceding years). Some time later I concluded that this had to do with what I now call melting momentum.

The idea is that if during May and July the ice gets hit by lot of sunshine, lots of melt ponds form and they soak up a lot of heat. This doesn't immediately cause the ice to melt, but starts to make itself felt towards the later stage of the melting season. The built-up momentum keeps the melting going. For this reason I've been focusing on things like compactness, that tells us something about the amount of water within the ice pack, whether open water or melt ponds (satellite sensors can't see the difference).

But does this still hold true? Will there be a moment when the ice disappears regardless of melting momentum as well? We saw in 2016 that there wasn't much melting momentum built up during May and June, but despite the cloudiness, a record warm winter and persistent high air temperatures were enough to make the ice vulnerable enough for a big cyclone and massive dipole to have 2016 come in second on almost all graphs (see this overview). Yes, melting momentum is important, but there are other things at play as well, of course.

NSIDC Compactness (are divided by extent) as of July 29th

Now, this year has been extremely interesting so far. For the third time in a row, the winter preceding the melting season was relatively mild, the maximum was low (second lowest on record), and the trend line on many a graph went low from the get-go. Not as low as 2016, but low. However, as in previous years, clouds moved in and things slowed down to a crawl, taking the trend line as high up as 11th lowest on record on the JAXA sea ice extent chart last week. But then this happened in the last couple of days, suggesting spectacular melting on the Pacific side of the Arctic:

So, what's going on?

Melting momentum: May 2017

Here's a quick blog post, which is mostly a copy of a comment I just wrote on the 2017 melting season thread on the Arctic Sea Ice Forum. For those who want to know more about what melting momentum means, read these blog posts from 2015 and 2016.

I've been in contact with Dr David Schröder from the University of Reading and he has confirmed (or rather his model) that this year (again) there is less melt pond formation than in years with low minimums. Below is a comparison of 2012, 2013, 2016 and this year:

Caveat: This is a model result, and so the distribution of melt ponds doesn't necessarily reflect reality.

As you can see, according to the model, there are even fewer melt ponds than last year. In fact, it looks very similar to 2013. Dr. Schröder wrote to me in an e-mail:

We predict the September Arctic sea ice extent 2017 to be slightly larger than last year: 5.0 +/- 0.5 mill. km2. In spite of the lowest Arctic ice volume in April, the May melt pond fraction in May is the second lowest of the last 10 years in our simulation with the sea ice model CICE. Anomaly figures with respect to 2006 to 2015 are shown for May 2017, 2016, 2012 (highest pond fraction) and 2013 (lowest pond fraction). Stronger melting and pond formation only occurred north of Canada and some parts of the Beaufort Sea. Otherwise the pond fraction is very similar to May 2013. Below average air temperatures (with respect to 2008 to 2015) and increased snow depth seem to have had a stronger impact in reducing pond formation than the thinner ice in increasing pond formation.

What's nice about this, is that it confirms our suspicions back in February when we discussed what could be the reason there was such a divergence between PIOMAS and CryoSat (as mentioned in the March PIOMAS update). The number one suspect was snow cover, mostly due to the relentless series of storms coming in from the Atlantic. Persisting snow on the ice is the best explanation for the fact that we haven't seen much melt ponding (ice turning blue on satellite images) so far, despite ideal weather conditions during the past two weeks.

As I wrote on the ASIF back in February:

This is pretty big, IMO, as it tells us something about snow depth on the sea ice which may have consequences for the state in which the ice pack enters the melting season.

Maybe we're in the process of getting the answer to that question. The last time there was such a divergence between PIOMAS and CryoSat was during the 2012/2013 winter. Remember, 2013 was cold and cyclonic, causing an unexpected rebound after 2012's spectacle. I don't think this year will be the same, if only for the fact that volume is at a record low because of a record mild winter, but 2017 could be 2016's rebound.

As always, everything depends on weather conditions, but also on sea surface temperatures, as I suspect this was a very important factor last year, causing the September minimum to end up as low as it did, despite a lack of melting momentum.

Back to the sabbatical now (this was too important not to mention).

2016 melting momentum, part 3

Melting momentum, it's what I call the absorption of heat and solar radiation during May and June that does not directly lead to melt and a reduction in ice cover, but rather comes into play during July and August (I had a more wordy explanation last year).

This is part 3 in the series.

As we saw last month, the melt pond fraction during May wasn't particularly high compared to previous years, and quite below that of 2012, effectively cancelling out the advantage that was built up on the extent charts. Back then it already looked highly unlikely that the 2012 record low minimum will be broken in September. This trend continued during June.

Dr David Schröder has been so kind as to send me the June melt pond fraction maps he uses for his SIPN 2016 Sea Ice Outlook prediction (blog post on June Report here). These results are based on the melt pond fraction simulation model developed by CPOM researchers at the University of Reading (more info). The comparison below shows melt pond anomalies for the years 2010, 2012, 2015 and 2016 compared to the average for the last 10 years:

Caveat: This is a model result, and so the distribution of melt ponds doesn't necessarily reflect reality.

The difference with 2010 and 2012 is clearly quite large. Just like last year May and June saw very little melting momentum being built up. Last year then caught up a bit because of one of the hottest Julys on record, eventually ending in third place with a September average of 4.63 million km2 (just like 2011, with 2007 and 2012 coming in at 4.30 and 3.63 million km2 respectively).

Dr. Schröder writes:

2016 melting momentum, part 2

Happy Solstice, everyone!

Credit: NASA/Kathryn Hansen

The previous post on melting momentum was running long, so here's an addition dealing with compactness, the final piece of information we have to assess the amount of melting momentum. Just before the start of July, the month of big melt. First, a couple of paragraphs to explain what compactness is and what it can tell us.

Not only melt ponds, but also open water between ice floes can absorb solar radiation, heating up the water, and then speeding up the melt process. We can get an idea of how compact an ice pack is by dividing sea ice area numbers (SIA) with sea ice extent numbers (SIE), the products of two different ways of measuring total sea ice cover.

The idea is simple, really. The Arctic is divided into grid cells, and each grid cell has a certain sea ice concentration. On the right there's an example of one such grid cell, with white representing ice and blue representing open water/melt ponds. This grid cell measures 10 x 10 km, and thus has a total area of 100 km2. Sea ice concentration is 80% in this grid cell, which means that 80 km2 is covered with ice. This is the number that represents sea ice area.

However, for sea ice extent there's a so-called threshold of 15%, meaning that if there's 15% of sea ice concentration or more, the entire grid cell is considered ice-covered. In other words sea ice extent for this particular grid cell is 100 km2. The reason for this different way of assessing total ice cover is to reduce the influence of melt ponds, that can fool satellite sensors into 'thinking' they're seeing open water.

2016 melting momentum, part 1

Melting momentum, it's what I call the absorption of heat and solar radiation during May and June that does not directly lead to melt and a reduction in ice cover, but rather comes into play during July and August (I had a more wordy explanation last year).

As important as it seems to be for the final outcome of the melting season, it is one of those aspects of Arctic sea ice that has proven to be really difficult to measure. Observations based on satellite images have been developed in the recent past, but for some strange reason, this research didn't receive further funding.

I will discuss some indirect and experimental measurements further below, but start off with the best tool we have by far to get a feel for how much melting momentum is being built up during this crucial period in the melting season: the melt pond fraction simulation model developed by CPOM researchers at the University of Reading (more info).

Again this year, Dr David Schröder has been so kind as to send me the May melt pond fraction maps he uses for his SIPN 2016 Sea Ice Outlook prediction, showing melt pond anomalies for the years 2012, 2014, 2015 and 2016 compared to the average for the last 10 years:

Caveat: This is a model result, and so the distribution of melt ponds doesn't necessarily reflect reality.

This past May has been similar to last year's, but nowhere near May 2012. Here are two more maps that show the difference in melt pond fraction with May 2012 and 2015 (red = more melt ponds now, blue = less melt ponds now):

Beaufort under relentless (high) pressure

When I wrote the Beaufort under early pressure blog post last month, I didn't quite expect this unprecedented pressure (timing and magnitude) to keep up for so long. High pressure areas tend to be short-lived in the Arctic, but this one was still going strong when the Beaufort quick update was published. Now, almost three weeks later, things still haven't completely calmed down, and this animation of satellite images showing weekly changes from March 30th to May 4th provides an astonishing sight:

Astonishing, mindblowing, frightening even. May is the month when melt ponds start to form on the ice, pools of water that lower the albedo and cause the ice to melt faster, leading to open water that absorbs solar radiation and warms up. The open water is already there in the Beaufort Sea. Look at all that open water!

No other year comes even close to it, as can be seen on the ASIG comparison of sea ice concentration maps, or on this sea ice extent graph (based on AMSR2 data, provided by Wipneus):

Jim Hunt provides this close-up view on the resources page of his Great White Con website (also be sure to read his latest blog post on early wave activity in all this open water):

Jenga July 2015

And thus the tradition of the ASIB alliteration of months continues. This one is a bit corny, but kind of appropriate too, so bear with me. Melt Pond May refers to the onset of melt pond formation, and Junction June is about the possible paths the melting season may follow after these two months of preconditioning. The ice pack might remain relatively stable because of a lack of preconditioning, but that balance can still be upset during July, just like a game of Jenga. It has to be an exceptional July, though, and this year's July was exceptional.

To reiterate why preconditioning is important: if the ice pack gets covered by melt ponds early on in the melting season, more solar radiation will be soaked up because melt ponds are much less reflective than ice. That way something which I call melting momentum is built up, a combination of thinned ice and warmer waters on, under and around the ice floes. This momentum can sustain relatively high melting rates, even when weather conditions aren't optimal, as we've seen during the 2011 and 2012 melting seasons. The reverse is also true. Little preconditioning, as happened in the 2013 and 2014 rebound melting seasons, causes trend lines to simply stall at some point, even when weather conditions are conducive to melting.

This melting season didn't undergo any serious preconditioning to speak of. All in all more melt ponds formed on the ice pack than in 2013 or 2014, but it started late and wasn't comparable to years with lots of preconditioning, like 2010 and 2012. It was clear early on that 2015 wouldn't be breaking any records, and the main question seemed to be whether it would end up below 2013 and 2014, or become the third rebound year in a row.

But then July came, and July was hot, as can be seen on this temperature graph made by the NSIDC's Andrew Slater (hat-tip to Chris Reynolds and his latest PIOMAS analysis):

To be able to assess the impact of this relatively high temperature anomaly we need to compare it to other years, in this case the big melters since 2006, and the two recent rebound years. I've divided July up into two halves and here's the first map for July 1st to 15th:

Junction June 2015

Melt pond May alliterates well and the name conveys what it's about: the time when melt ponds first start to form. Luckily, a couple of weeks ago, someone on the forum (forgot who, but thanks!) helped me find a perfect alliteration for June: Junction June. This name refers to the month's importance for the rest of the melting season.

The idea is simple: the fraction of the ice pack's area that is covered by melt ponds during May and June, preconditions the ice for the rest of the melting season, because melt ponds lower the pack's albedo and thus soak up more solar radiation. This in turn increases melting momentum, which can sustain a high decrease during the second half of the melting season, even if weather conditions aren't perfect for melting, compaction and transport. Conversely, if melting momentum is low, chances of new September records are minimal, unless weather conditions during July and August favour huge losses. Hence the word 'junction'. The trajectory for the melting season is largely determined during May and June.

Just like last month, I'll analyse the melt pond situation by comparing this year's weather conditions with those of previous years that generated records (2007, 2011 and 2012) as well as rebounds (2013 and 2014). But first I want to look at the melt pond distribution maps that were sent to me by David Schröder from the University of Reading and are based on his research published last year. A model,  using NCEP reanalysis data for atmospheric forcing, simulates the melt pond distribution across the Arctic, and although pond area in May is only around 1% of the total melting season area, this melt pond cover fraction according to Schröder et al seems to have the strongest impact on the sea-ice state in September.

This correlation has been contested in a recent study by Jiping Liu et al, published in Environmental Research Letters back in May. The paper states that "a significantly strong relationship (high predictability) first emerges as the melt pond fraction is integrated from early May to late June, with a persistent strong relationship only occurring after late July." Either way, late June is now well behind us, so here are the melt pond distribution maps for 2012, 2014 and this year (click for a larger version):

Mind you, these maps are model results and don't display the exact locations of melt ponds.

Eyeballing the maps it's clear that 2015 has more melt ponds than 2014, but doesn't come even close to 2012. Dr. Schröder writes:

ASI 2015 update 4: massive heat

During the melting season I'm writing (bi-)weekly updates on the current situation with regards to Arctic sea ice (ASI). Central to these updates are the daily Cryosphere Today sea ice area (SIA) and IJIS sea ice extent (SIE) numbers, which I compare to data from the 2005-2014 period (NSIDC has a good explanation of sea ice extent and area in their FAQ). I also look at other things like regional sea ice area, compactness, temperature and weather forecasts, anything of particular interest.

The animation on the right consists of NSIDC
sea ice concentration maps, one for each ASI update.

Check out the Arctic Sea Ice Graphs website (ASIG)
for daily updated graphs, maps, live webcam images and
the Arctic Sea Ice Forum (ASIF) for detailed discussions.

July 7th 2015

I wanted to wait for the latest PIOMAS update and do another melt pond analysis for June, but given the current circumstances and the forecast I've decided to squeeze this ASI update in first.

As we've seen in previous ASI updates this melting season hasn't had a flying start, but it hasn't been as slow as 2013 and 2014 either. With extent going low real fast, but area lagging behind, the compactness numbers indicated little melt pond formation and no divergence to speak of. Melt ponds - and to a lesser extent openings within the ice pack - are important because they lower the ice pack's albedo and soak up a lot of sunlight, preconditioning the ice for further melting. This melting momentum, as I call it, can play a decisive role in the latter half of the melting season, trumping weather conditions.

Of course, weather still plays a very significant role in the final outcome of a melting season. This year hasn't seen any substantial preconditioning and is more similar to rebound years 2013 and 2014 than record years 2012, 2007 and 2011. In fact, this year has been very much like last year, in the sense that there is hardly any movement. Very little ice is transported through Fram Strait to the North Atlantic ice graveyard, and there is no compaction or divergence to speak of because of a lack of strong winds. Most of the extent/area decrease is caused by so-called in situ melting, or melting on the spot.

However, there is one big difference compared to last year and that's heat. Despite a very cold start, there have been several outbreaks of warm air over the ice, slowly but radically shifting the balance between extent and area data. The impact is felt on the surface of the ice pack, but doesn't translate directly into a decrease. Not yet. In theory, it should percolate through after a while, especially if the heat persists. And right now the Arctic sea ice pack is undergoing a massive heat wave which shows no signs of letting up.

More on what has happened and what is going to happen below.

Sea ice area (SIA)

Here's the latest Cryosphere Today SIA chart, to which I've added preliminary data as calculated by Wipneus:

Big drops have been alternating with small upticks, keeping the trend line near the upper half of the pack, but below that of 2013 and 2014.

ASI 2015 update 3: what's it going to be?

During the melting season I'm writing (bi-)weekly updates on the current situation with regards to Arctic sea ice (ASI). Central to these updates are the daily Cryosphere Today sea ice area (SIA) and IJIS sea ice extent (SIE) numbers, which I compare to data from the 2005-2014 period (NSIDC has a good explanation of sea ice extent and area in their FAQ). I also look at other things like regional sea ice area, compactness, temperature and weather forecasts, anything of particular interest.

The animation on the right consists of NSIDC
sea ice concentration maps, one for each ASI update.

Check out the Arctic Sea Ice Graphs website (ASIG)
for daily updated graphs, maps, live webcam images and
the Arctic Sea Ice Forum (ASIF) for detailed discussions.

June 21st 2015

Today is the day of summer solstice, when the Sun reaches its highest point as seen from the North Pole, beating down 24 hours per day on the sea ice below (when not blocked by clouds). The period preceding and following this event is very important for how the melting season plays out, as it is during this time that large parts of the Arctic sea ice pack are preconditioned through melt ponds.

This preconditioning in turn influences melting momentum. A lot of momentum almost guarantees high melt rates, whereas little momentum does the opposite, depending on weather conditions during July and August as well, of course. Last week I wrote a blog post in which I tried to assess the amount of melt ponding through May, by comparing this year to previous years.

The first half of May was extremely cold, with the second half of May showing almost the exact opposite. Still, indications were that relatively few melt ponds have shown up during May and the first half of June. This is now changing, as can be seen on the animation at the upper right hand corner of this blog post, where sea ice concentration is showing a very large blue patch of lower concentration on the Pacific side of the Arctic. That's because weather conditions conducive to melting shifted from the American side to the Siberian side of the Arctic, as was forecast two weeks ago.

Below an assessment of what's been happening in the past two weeks, and what is going to happen next week.

Sea ice area (SIA)

Here's the latest Cryosphere Today SIA chart, to which I've added preliminary data as calculated by Wipneus:

CT SIA recorded a huge 340K drop in one day. The last time a triple century break was recorded was in 2008. This brings the 2015 trend line right in the middle of the highest and lowest trend line in the 2006-2015 period.