Thursday, March 31, 2016

Eli Heads North

One of the things a smart bunny learns is that if you see something new to you in an area you really don't follow, everybunny else who follows the dots has seen it, but you have not seen them seeing it.  Yet, since on rare occasion they have not seen it because it is so obvious it is worth asking about and one might even learn something

Worse, in Eli's case, this is something that finally percolated through because of nonsense that Andrew Montford at Bishop Hill wrote trying to handwave the weird ice coverage this winter up north (yes, Eli knows everybunny and his brother in law is racing south to watch the Antarctic ice shelves collapse, but this is Rabett Run, Eli and Brian follow their own pipers).  Montford opined

As usual on these occasions, I take a quick look at the Cryosphere Today anomaly page, where I find the sea ice apparently still stuck firmly in "pause" mode.
As fate would have it Eli had been looking at this chart for many a year, and there, as discussed below, were things about it that well stood out to anybunny who spent their life looking at charts, things that were. . .and are . . . maybe. . . .interesting.

For example, it is clear that the nature of the anomalies changed somewhere around 2006.  The pattern of the anomaly between 2007 and 2016 is very different to that previously.  With a thirty day smooth, the post 2007 pattern is hard to miss

There is a clear minimum in September, not very exciting.  One can see this in 2013 and 2014, although not as clearly in the other years, but it is there.  However, there are also secondary minima in May/June in all of these years. 

Equally interesting is the period 1996-2006

Rather than the winter/summer pattern seen in the 2007-2016 period in the anomaly there is a roll off in all seasons, about the same loss of sea ice year round (e.g. the decrease each month was about the same or at least the month to month variation was smaller.  But what about before 1996

Eli will not engage in cycling.  HIt is true that because anomalies compare like with like, e.g. the ice area in the same month against some reference period (1979-2008 in the examples above) looking  just at anomalies can be tricky.  The boys from Bremen have a useful comparison of September minimum extent between 2012 and 1979-83

and if one looks at the period between 1979 and 2015 on the maps at Cryosphere Today the changing pattern in the September ice area is obvious. 

Oh yeah, Cryosphere Today has a useful reminder of the ability of an anomaly to hide the decline

There has been a pretty steady decline in arctic sea ice since ~ 1955.


Anonymous said...

I tried to make the point there, that is related to the point you're making here. Judging things on anomalies can be tricky, given that these are absolute, not relative. Hence a large anomaly in a month when you'd expect the sea ice extent to be large is not necessarily as significant as a large anomaly in a month when you'd expect the sea ice extent to be small.

Kevin O'Neill said...

Chris Reynolds at his Dosbat blog has written numerous articles on sea ice analyzing the data by breaking it down into regions and showing the loss/gain contribution. You may wish to review some of his past articles.

E. Swanson said...

The first graph of sea-ice anomaly is probably a plot based on the variable called "sea-ice area". I suppose it's the companion to the graph of area also shown on the Cryosphere Today page. Trouble is, the passive microwave data is calculated from an average over each pixel, which counts melt ponds as open water. As warming may produce more melt ponding and cause them to appear earlier during the melt season, the area variable may be giving a false indication of reduction in the actual area of sea-ice. Your observation of a minimum early in the melt season may be a reflection of this process, which would certainly be of importance for the understanding of the changes in the Arctic.

Also, not knowing what algorithm was used to produce your smoothed graphs, I would caution that using a moving average distorts the resulting time series. There are several filtering algorithms available which provide much better results and your analysis would be much improved if you used one of these instead of a moving average. Other research has been directed at the use of satellite data for measuring the melt ponding on top of sea-ice, so your results might provide valuable information to add to these efforts. Hey, lets do a paper!!...:-)

EliRabett said...

EFS: You can see all of the features Eli pointed to in the daily data, somewhat better if you spread it out, which for the purposes of blog postings means that a simple moving average is enough. Sometime just looking is better than righteous statistics.

Howard said...

This is most likely a result of the sea ice declines reaching a diminishing stage where further declines in sea ice require more and more warming. When the target becomes smaller, the pieces you can attrit become smaller as well. We see this type of behavior in earth systems all the time. In any event, it's fun to watch both sides make their silly little ClimateBall points to reassure the peanut gallery that your side is on the side of the angels.

I wonder why they call it the natural logarithm?

Napier's e

E. Swanson said...

Eli, As it happens, Sou has a couple of recent posts about sea-ice. She posted a link to Meier et al. 2012, which led me to an earlier paper, Meier & Stroeve 2007. In the 2007 paper, they looked at monthly anomalies and calculated trends, which did show early season melting in certain areas covered by sea-ice. The first areas to melt historically also were the ones which exhibited the largest negative trends, mol. They used sea-ice extent, not concentration. Both reports have numerous references to tempt one's curiosity...

EliRabett said...

Eric, thanks for the pointer. The 2007 paper is very useful. After years of looking at ice maps, Eli has long thought that it would be useful to monitor sea ice are north of 70 N and maybe with the pie slice btw 15W and 15E cut out. That and a separate series for everything else such as Hudson's Bay, the Bering Sea, the St. Lawrence, the Sea of Okhotsk.

You know that all of the everything else will pretty much melt out in the summer, but the winter extent is a marker of ice formation. The former is a marker of the summer melt, but, pretty much will fill in during the winter, well until this winter.

Kevin O'Neill said...

Eli - I will reiterate, Chris Reynolds at his Dosbat blog and Wipneus on the arctic sea ice forum have broken down volume, extent, and area by sectors already.

Chris' current post is on maximum extent at the sector contributions to the loss/gain.

Unknown said...

Eli - The work of "Wipneus", Chris Reynolds et al. is gathered together for your convenience over in the Great White Con Arctic resources section. By way of example, you may wish to peruse the graphs of Arctic basin sea ice area:

I'd endeavour to try and explain the significance, but bitter experience suggests it's impossible to embed images in Blogger comments?

cRR Kampen said...

The pattern change is characteristic of a forced, imbalanced system.
First thing it proves is that by the climate in existence since 2007 equilibrium is given by total summer absence of ice and that is where we are going.
The process is non-linear (cf catastrophe theory).

Barton Paul Levenson said...

BTW, Eli, Judy Hopps sends her warmest regards, and says no, she and Nick are just "very good friends."

Dano said...

Equally interesting is the period 1996-2006

We're taught in ecology courses that this pattern is a signal that the system under scrutiny is becoming unstable. Many natural systems at large scales change gradually - shocks and disturbance notwithstanding - and when the signal starts changing like that graph you should pay attention.




Dano said...

I failed to read cRR Kampen's comment before posting, apologies for the repetition.



Aaron said...

The anomalies remind us that it is a nonlinear feedback system, with "controls". The system tends to remain within a range (in control), until it is forced out of a particular set of controls.

The plots of extent remind us that the system has been forced from one set of controls to the next range of controls every few years. (e.g., punctuated equilibrium)

However, since this is the first time we have forced this system, this far, we do not know what the range of the next set of controls is; or, even if there is another range of equilibrium. We may simply lose feedback on one side of the system's equilibrium range. For example, water vapor or CH4 could reduce heat loss by radiation, resulting in rapid ice loss, or a storm could mix up warm water to the surface, or warm water at the surface could melt the ice AND increase water vapor in the atmosphere!

I have no doubt that all will happen. The questions are how many more equilibrium ranges does the Arctic have? And, what weather effect(s) will get credit for eating the last of the Arctic sea ice? (I would not consider standard climate models useful on these questions!)

Unknown said...

Eli and his bunny buddies may be interested to learn that I followed his link over to The Bishop’s Hill?

I discovered that graphical comments are also impossible over there, and Mark 1 eyeballs seem to be in short supply too!

cRR Kampen said...

Dano, it is a kind of insight that isn't really that 'main stream'. More a thing for mathematicians/mathematical physicists (knowledge used by engineers of course).
Repetition of interpreting this kind of signal of a stressed system is more than welcome by me.

Bernard J. said...

I was about to make the comment that there is in the trajectory an implication that the system is destabilising, only to see on reading the comments that both cRR and Dano had already made the same point.

Bunnies might ponder on what such an instability might mean for the global warren...

cRR Kampen said...

Ponder a supernova or something.

It is the system of systems that is getting ffed with global warming. It is the Arctic, it is ecologies in seas and on land and on continents, it is societies in general - it is everything.

As if humanity lives in a flat and all are eating at the walls and ceilings and floors at the same time - the flat will respond very nonlinearily indeed at some point but up to that point, we eat and feast.

Unknown said...

Bear in mind, however, that if you consider the physics a non linear "long tail" for the Arctic is also conceivable:,933.0.html