Happy Solstice, everyone!
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.
That is why sea ice area will always be lower than sea ice extent. If we now divide SIA by SIE, we get a percentage that gives us an idea how much open water/melt ponding there is between and on ice floes. The grid cell's compactness ratio in this example is 80/100=0.8. This, of course, can be done for the entire Arctic.
Different data sets can be used to calculate compactness. For years I have used Cryosphere Today SIA data and divided it by IJIS sea ice data (CAPIE = Cryosphere today Area Per IJIS Extent) and used these graphs for the ASI updates. As IJIS no longer reports SIE data from the Japan Aerospace Exploration Agency, or JAXA, the name has changed to CAJAX. Here's the latest graph:
Just a few weeks back, 2016 had the lowest CAJAX ratio, but now it's firmly in the middle of the pack. Based on what we know from the previous Melting momentum blog post and the latest ASI update, we can safely say that a lack of melt ponds is being compensated by open water between ice floes, for instance in the Beaufort Sea (source: Environment Canada):
As there are more data sets out there, I've decided to try something else as well as an experiment. Bear with me.
Grid cell size can vary from data set to data set. This is called resolution. The higher the resolution, the smaller grid cells are, the more accurate SIA or SIE is. I figured that if I used SIA at a low resolution, grid cells will be big and thus melt ponds and leads will have a larger impact on the numbers. If I then combine this with high-resolution extent that does the opposite - reduce the impact of melt ponds and open water between floes - I get a much more pronounced compactness ratio.
This is all a bit counterintuitive and I'm not going for the accuracy award here, as I want changes to be as exaggerated as possible. In fact, it's probably the completely wrong thing to do, as data sets not only have different resolutions, but also use different sensors, land masks, algorithms, and so on. But what the heck.
I used Cryosphere Today SIA (25 km resolution, as calculated by Wipneus until calibrated data is published by Cryosphere Today) and divided it by MASIE sea ice extent numbers (4 km resolution). Yes, I'm even combining a passive microwave product with an operational analysis product (difference explained here)! Like I always say: If you're already breaking the law by jaywalking, you might as well rob the bank too. ;-)
The good thing about MASIE is that it's one of the most precise data sets out there as it is based on the IMS sea ice product, which is updated daily by analysts from the National Ice Center - people, not automated scripts - who look at a plethora of satellite imagery to determine where the exact ice edge is. The downside of that is that there's a subjective component (different analysts) and changes in available satellite data over time. That's why in principle MASIE isn't deemed consistent enough for long-term trends.
I have found, however, that MASIE seems reasonably consistent after 2010 or 2011 (a change in data sources around that time causes years prior to have a low bias for most of the year), so I present to you CAMAS:
You can see that the spread between trend lines is much larger after day 210, and 2012 goes extremely low compared to the rest. That's because the Great Arctic Cyclone caused the ice pack to disperse so much that Cryosphere Today SIA went nuts, whereas MASIE of course kept 'seeing' all the ice. This is what I mean with exaggerated changes in compactness.
Either way, 2016 is somewhat lower here, relatively speaking, but far from the lowest (and 2007 isn't even included in this graph).
If you're a purist and want to see compactness graphs that actually deserve the moniker because both SIA and SIE data come from the same source (meaning everything else is the same), you can check out this graph that Wipneus updates every day (see ASIG Regional Graphs page), based on data from NSIDC (25 km resolution), JAXA (10 km resolution) and Uni Hamburg (3.125 km resolution):
Here, 2016 currently has the highest compactness ratio in all three different segments, indicating that the ice pack is relatively compact (few melt ponds) and little melting momentum is being built up to boost melting during July and August.
Long story short: Even though 2016 has been breaking records all year so far, as things currently stand, it will take special weather conditions for it to break any records near the end of the melting season.