Shift in timing of ice minimum over the years

[KenB]

The 2020 melting thread has some discussion about when the minimum will be.  Some of that isn't 2020-specific, so I'm starting this thread.  Apologies if there's some older thread on this subject; I looked back a couple years and didn't see one.

Regardless of what happens from here on out, 2020 has carved out a deep new low range for Arctic sea ice for part of the year and will be one for the books.  In my appraisal this means it is already a very bad year for the ice.  Keep setting new lows for different days and sooner or later one of those days will be September 16th.  We definitely will never carve out new highs for any day of the year henceforth.

Looks like, yes. Just one little detail: September 16th is not very likely for the minimum. Rather, something like September 25th...30th, more like. The reason is all the extra heat in the system likely to extend melting season a bit. The trend of it is well obvious, if one would plot a graph of 5-year running-means from 1980s all the way to 2019. It slowly extends, obviously with lots of noise outta weather effects this or that particular season. But this one, with the GAAC and remaining effect of cleaner air due to less air travel and industries much of the spring and summer due to all the quarantines around the NH - one would expect the lengthening of the season to be especially likely to be big. Which, of course, would lead to lower minimum than would "normally" be expected - more time to melt things, more things melt out in the end.

It's melting momentum, i call it. The bigger it gets, the longer it takes to exhaust its potential with everything else being "usual". And right now, it's big.

Here's a version of the plot referred to above using NSIDC extent data and including a linear trend line. Note that I didn't adjust for leap years, the Y axis is just the day of the month in September.

 

[GoSouthYoungins]

I think this is worth discussion. But it is important to keep in mind that minimum is usually barely lower than 2 weeks prior to minimum. September is a really really flat month for arctic ice.

[igs]

The 2020 melting thread has some discussion about when the minimum will be.  Some of that isn't 2020-specific, so I'm starting this thread.  Apologies if there's some older thread on this subject; I looked back a couple years and didn't see one.
Here's a version of the plot referred to above using NSIDC extent data and including a linear trend line. Note that I didn't adjust for leap years, the Y axis is just the day of the month in September.

Thanks, well done.

Always good to know than to guess.

Such small topics at times can cause heated depated due to lack of facts, now we have them for now.

 

[KenB]

I think this is worth discussion. But it is important to keep in mind that minimum is usually barely lower than 2 weeks prior to minimum. September is a really really flat month for arctic ice.

Yes, I see what you're saying.  Although some years are much flatter than others, looking at https://nsidc.org/arcticseaicenews/charctic-interactive-sea-ice-graph/.  In those flatter years, taking the single minimum date seems kind of arbitrary.  I have an idea about how to work around this; I'll see if I can implement it.

[oren]

Very good thread.
A graph showing September trajectories for selected years or all years post 2000 can shed more light on behavior around the minimum.

There is also the question of extent or area, and NSIDC or JAXA. There is also the somewhat separate matter of volume. I assume you mean NSIDC extent minimum in this discussion and the chart above?

[Tor Bejnar]

...
I keep meaning to check whether the date of minimum has shifted later. On the one hand I'd expect the extra heat in the Arctic Ocean to prolong melting later, but on the other hand the reduction in ice area at the minimum means that the edge of the sea ice is further north, where you'd expect melting to end earlier.

The two effects may currently cancel.

I like Jim Petit's graphs, particularly this one which shows that while certain 'milestones' get earlier, the extent minimum date doesn't 'go' anywhere.  (A click or "View image" will enlarge.) (Curious:  the graph updates every day, my 'preview' shows a 5-day old version...  Click on "Jim Petit" and scroll down to the Extent section for the current one ...  The "7.0M"s may last only 8 days!  This would be a new record for this span.)

[Stephan]

I think that the timing of the minimum also depends on the distribution of the ice around the minimum. If areas at high latitudes (say > 85°N) have open waters by mid Sep they likely freeze over soon (→ early minimum), whereas a more evenly distribtued ice cover mid Sept would allow ice packs further south still to melt in the autumn sun (→ later minimum). Also the weather and the wind patterns play an important role.

[KenB]

Very good thread.
A graph showing September trajectories for selected years or all years post 2000 can shed more light on behavior around the minimum.

There is also the question of extent or area, and NSIDC or JAXA. There is also the somewhat separate matter of volume. I assume you mean NSIDC extent minimum in this discussion and the chart above?

For the chart above, yes, that's NSIDC extent data.  For the discussion, I think we should be open to anything.

[KenB]

I like Jim Petit's graphs, particularly this one which shows that while certain 'milestones' get earlier, the extent minimum date doesn't 'go' anywhere. 

Yes, I like those graphs a lot, and that one in particular.  FWIW, I checked and my chart above and Jim's graph are of the same data.  So although the trend is subtle, particularly WRT the seasonal variation, it does seem to be there.  Not that I'm claiming to have done any careful analysis of significance (yet).

[be cause]

a bit weird , clicking on a melt season thread post and being delivered here .. b.c.

[oren]

I guess it's because I've moved some posts from there to here.

[ajouis]

That could be useful to the discussion, it describes a season lengthening of 5 days per decade
https://scholar.google.com/scholar?hl=en&as_sdt=2005&sciodt=0%2C5&cites=15249873421211437254&scipsc=&q=arctic+weather+melt&btnG=#d=gs_qabs&u=%23p%3DNuuFIcY9ME8J

[Tor Bejnar]

The graph with 'best fit' line (above) suggests almost 5 days later minimum in 40 years.  That's a spot more than a day/decade.  This is very slow, I think, considering how fast other things are changing in the Arctic.

[KenB]

The graph with 'best fit' line (above) suggests almost 5 days later minimum in 40 years.  That's a spot more than a day/decade.  This is very slow, I think, considering how fast other things are changing in the Arctic.

So this brought to mind two things:  first, it would also be interesting to see how the maximum point(s) have been moving.  5 days / decade in total length would match the graph above if the maximum has moved earlier by 15 days over 40 years; though that seems like a lot of movement.  Second, I think using a line here is probably not well-chosen (it was, however, very simple).  But it would make sense to me that any movement would slow down considerably as it got close to the fall equinox.  Certainly the insolation doesn't fall off linearly (I have no idea how it falls off, actually, maybe some expert could weigh in on that?). 

[oren]

Insolation in the High Arctic falls precipitously towards the Equinox.

[Glen Koehler]

   Previoius chart is solar radiation at top of atmosphere.  Tealight posted a chart of insolation at surface.  The low sun angle in the Arctic increases the amount of atmosphere through which solar radiation must pass which reduces the level reaching the surface.  Chart posted by Tealight shows values for latitude in 10 degree increments so you don't have to guess where 80N falls between 60 and 90.
    Based on comment in 2020 Melt thread by GoSouthYoungins
https://forum.arctic-sea-ice.net/index.php/topic,3017.msg274806.html#msg274806
that insolation driven melt essentially ends in last week of August (which I have not verified from other sources, but seems reasonable) I added estimated dates for start and end of insolation melt season.  Residual melt from warm water melt would account for the remainder of melt season until September minimum.
    Caveat:  I don't know any of this stuff, just making back of the envelope estimates.  Corrections encouraged.

[vox_mundi]

United States Navy Arctic Roadmap for 2014 to 2030
https://www.navy.mil/docs/USN_arctic_roadmap.pdf

Fig. 4 gives a projection of shift of open water minima between 2014-2030

[KenB]

A graph showing September trajectories for selected years or all years post 2000 can shed more light on behavior around the minimum.

Here are graphs showing September NSIDC trajectories for 1980-2019.  Some years are indeed quite flat, while others show a pretty well-defined minimum.  Plotting these all on the same y scale leads to a lot of vertical compression but shows the steady progression toward lower extents much better (top plot).  Allowing each year to have its own scale makes the minima much clearer (bottom plot).

[oren]

Great post Ken.
I certainly see drops in September, especially for the last 10-15 years. I would say the timing and shape of the minimum can have an effect of +/-100k km2 on the final outcome. Not huge, but often decisive when years are hard to tell apart. Big impact at ranks 2-5, no impact on #1 (unless this year comes neck and neck with 2012).

[Sailaway]

Great post Ken.
I certainly see drops in September, especially for the last 10-15 years. I would say the timing and shape of the minimum can have an effect of +/-100k km2 on the final outcome. Not huge, but often decisive when years are hard to tell apart. Big impact at ranks 2-5, no impact on #1 (unless this year comes neck and neck with 2012).

The location of the ice that remains in September could hold some unpleasant surprises this year. If we look at extent and volume metrics for the Greenland sea for example. If i remember correctly, July had a higher than average export to the region - but that it would have had a sunnier than average summer holidays prior to its arrival.

[KenB]

Another plot, this time comparing the date of the minimum (always in Sept. but with adjustments for leap years) with the minimum extent.  I assumed these would be negatively correlated (date of minimum going up means minimum extent going down).  They are, but not as much as one might expect:  the Pearson's coefficient is -0.293.

Histograms of both minimum date and minimum extents are included because I was learning some new plotting tricks :-)

 

[glennbuck]