Ray has been wondering what my model will predict for the NSDIC September minimum. Well.. since it’s now September, I used three potential linear fits:
- Most recent JAXA extent report.
- CT ice area report.
- The average JAXA extent report from Sept 1-8.
My model says of these three, the best predictor is the most recent JAXA extent report with the average over September coming in on its heels. The ice area report has a high AIC relative to the other two indicating it has 1/20th the probability of being “the right” model relative to the JAXA extent report and was dropped from the analysis. My best prediction is then the weighted average of the remaining predictions with weights estimated based in AIC values. The current bets projection is =3.47 million sq. km.
Am I kicking myself that I didn’t enter 3.5 millions sq. km. as my model told me on the day bets closed? Am I kicking myself that I am tied with Carrick but he will edge me out on time because I entered my 3.6 late? The answer to these is “Yes and yes.”
Still, it does look like Carrick and I are going to be closest with him taking the Quatloos for the late bets.
The JAXA provisional figure for Sept. 7th was revised upwards by 62,656 km^2, almost the same as on the 6th, resulting in a confirmed figure of 3.665 million km^2, a change in extent of only -11,875 km ^2, and a 7 day average change of -24,531 km^2/day.
The actual figure for the 7th was 129,177 km^2 higher than the projection based on the 2008-11 average change x1.24, since June 1st.
The provisional figure for the 8th of 3.596 million km^2 suggests a fall of 68,750 km^2, and even without the usual upward adjustment, that figure would be 84,404 km^2 above the projected figure of 3.511 million km^2. If the provisional figure for the 8th is adjusted upwards by the same amount as that for the 7th, the confirmed figure for the 8th would be 3.658 million km^2, which is 6094 km^2 lower than the 7th but 147,059 km^2 more than the projected figure for the 8th.
So the actual figures continue to gain, relative to the projected figures based on the projected figures based on the 2008-11 mean x 1.24, suggesting that the estimated September average of 3.65 million km^2 based on the projection, is still likely to be too low.
Based on a rate of change of the 2008-11 mean x 1.227, the projected mean for September is now 3.762 million km^2, but if the ice recovers at only the 2008-11 mean x 1.0, the figure would be 3.740 million km^2, and at the mean x 0.5, 3.696 million km^2.
Lucia, based on the above figures, I still think that your model is underestimating the likely September average.
I personally doubt if the figure of 3.47 million km^2 is likely, since I don’t expect the minimum to fall much below 3.6 million km^2.
Are you using the latest, i.e. provisional extent figure in the model?
If so, that might be a reason for the under-estimation, since the provisional figure has so far been, on average about 56,000 km^2 too low
and is usually revised upwards in the confirmed figure.
Consequently I would suggest only using the confirmed figure in the model.
I realise I am putting my “neck on the block” in this, (probably putting too much faith in historical data) and one day of weather will no doubt prove me wrong.
Ray–
My projection model grabs whatever the values are the instant I run the script. If those are revised after I post, I would have to rerun the script. So, I used: 3595781 as the final value.
Lucia, I think it’s probably a good idea to excise that last data point.
Ray, my script does ignore the last day as unreliable.
I used a Monte Carlo approach to get the following
Day of Minimum: 13 (10-16)
Minimum Ice: 3.64 ± 0.04 Mm
Mean Ice: 3.71 ± 0.09 Mm
Weather’s still the key player.
Lucia,
When each day’s provisional figure is revised, another provisional figure will be added, which *could* be equally inaccurate, so as Carrick says, I think it would be better not to use the latest, i.e. provisional, figure.
I don’t know how much difference this will make, but using the provisional figure will tend to underestimate the ice extent.
Sorry, I should have added that as the rate of decline in the ice gets lower, the amount of underestimation will get worse, since the provisional figure seems to be getting less accurate.
Recently it has been out by the equivalent of several days.
The confirmed figure for Sept. 7th was higher than the provisional figure for Sept. 5th.
The actual figure may never get as low as the provisional figure for the 8th.
The confirmed JAXA extent figure of 3.675 million km^2 for Sept. 8th, seems to show an INCREASE of 10313 km^2 over the figure for the 7th.
However the provisional figure for the 9th shows a fall of 81,094 km ^2, so this may not yet represent the minimum extent figure for September.
On the other hand, the provisional figure for the 7th was revised upwards by 79,063 km^2.
Of course, if ice extent continues to grow from now onwards,
the actual minimum would be 3.665 million km^2 on Sept. 7th.,
but I still think we may see a lower figure on the 9th.
Ray. Where are you getting your JAXA numbers? This is the most recent ice extent on the webpage that I’m viewing. I’m assuming it is provisional.
The latest value : 3,593,750 km2 (September 9, 2012)
Ike, yes that is the latest provisional figure for the 9th, but unreliable.
I am talking about the latest confirmed figure for the 8th, which is 3,674,844 and apparently increasing.
I would put real money, not quatloos, on the confirmed figure never falling to 3,593,750, or even below 3,600,000.
I have been recording the provisional v confirmed figures since August 30th, here they are (provisional first) :
30/08/2012 3819219 3877031
31/08/2012 3801406 3836250
01/09/2012 3740781 3797969
02/09/2012 3710625 3754844
03/09/2012 3679844 3746875
04/09/2012 3683281 3726563
05/09/2012 3628125 3681094
06/09/2012 3614219 3676406
07/09/2012 3601875 3664531
08/09/2012 3595781 3674844
09/09/2012 3593750
Graph them and you will see what I mean.
We’re getting a bit of stretching according to the wind models, that could account for a couple of day bounce without it meaning we’re at the absolute minimum. Wait about 7 days, then pop the cork.
Thanks. I see the significant differences between confirmed and provisional.
Carrick,
Thanks,I know what you mean.
There is a danger of concentrating on the numbers alone, without looking a the real world.
Ray
It may or may not be more accurate to skip the provisional figure. On the one hand, it’s new data and does capture what has happened to some extent. To the extent that the new data contains some information, using it is likely to improve the forecast. As a first guess we might expect that the errors are equally likely to be high or low, so we wouldn’t expect using the new data to bias the prediction. But we don’t really have data to test the assumption that the provisional data are unbiased relative to the updates. To test that assumption we’d need to have access to historical revisions which I don’t have.
Certainly, using the provisional data in a fit based on updated data would mean my uncertainty bounds should be larger since we know the new data will be updated. But once again, I would need historical data on the magnitude of revisions to figure out how much larger they should be. So, for now, each person can come up with their own best estimate. I post mine. It’s a forecast, so I don’t think it matters all that much whether I do it Carrick’s way or my way.
If you want to have a forecast that’s different from mine, that’s fine. In fact… that’s sort of the point of everyone betting.
That seems to be a good rule.
Lucia,
Sorry, I don’t know how to do the intented quote:
“As a first guess we might expect that the errors are equally likely to be high or low, so we wouldn’t expect using the new data to bias the prediction. But we don’t really have data to test the assumption that the provisional data are unbiased relative to the updates. To test that assumption we’d need to have access to historical revisions which I don’t have.”
I have been watching this since the middle of August and to the best of my recollection, the provisional figure the errors are always positive, i.e. the provisional figures always overstate the subsequent actual figue. I have been keeping a record of the provisional figures since August 30th (because I noticed a bias) and the average positive error is 56,125 km^2 and it seems to be getting larger. Obviously this has more significant impact as the actual daily changes grow smaller. I don’t know if the error changes sign when the extent increases.
I don’t know how the provisional figures are calculated, but so far they have been worse than useless. The average daily change since August 30th, based on provisional figures, was -78,402 km^2, while the average change based on confirmed figures was -22,465 km^2.
By the way, I am not challenging our model or your prediction, only trying to help.
Sorry, I have got my signs mixed up in the above post.
The correct post is as follows:
I have been watching this since the middle of August and to the best of my recollection, in the provisional figure the errors are always NEGATIVE, i.e. the provisional figures always UNDERSTATE the subsequent actual figure. I have been keeping a record of the provisional figures since August 30th (because I noticed a bias) and the average NEGATIVE error is -56,125 km^2 and it seems to be getting larger. Obviously this has more significant impact as the actual daily changes grow smaller. I don’t know if the error changes sign when the extent increases.
I don’t know how the provisional figures are calculated, but so far they have been worse than useless. The average daily change since August 30th, based on provisional figures, was -78,402 km^2, while the average change based on confirmed figures was -22,465 km^2.
By the way, I am not challenging our model or your prediction, only trying to help.
I contacted IARC-JAXA yesterday and asked why the provisional figure appeared to be always lower than the confirmed figure.
I haven’t had a reply yet, but today, no provisional figure has been published. Coincidence?
Anyway, the confirmed figure again shows a rise, from 3.675 million km^2, to 3.695 millin km^2, an increase of 20,469 km ^2, twice that of the previous day.
The total rise is still within the normal range of falls for this time of year, so I do think it is probably too soon to say we have had the minimum. For example in 2007, there was a gain of 42,657 km^2 between Sept. 17th and 20th, so it looked like the minimum had been reached, but there was a fall of 55,782 km^2 over the next 4 days, so the actual minimum was on the 24th.
I think that we have to go at least 100,000 km^2 above the 7th figure of 3.665 million km^2, before we can be almost certain that was the minimum.
Even so, the recent increases do reduce the chances of a September mean below 3.7 million km^2.
Ray, the stretching even is still going on, but the air temperature has warmed above -2°C. This could actually make the ice more vulnerable and result in future melting. If you look at it near minimum from previous years.
At this point, ideally you’d like to verify total area is also increasing, though there are systematic errors in that that can make the reported numbers deceptive (the finite resolution of the satellite images is the issue here).
Carrick (Comment #102996)
September 10th, 2012 at 6:21 am
“Ray, the stretching even is still going on, but the air temperature has warmed above -2°C. This could actually make the ice more vulnerable and result in future melting. If you look at it near minimum from previous years.
At this point, ideally you’d like to verify total area is also increasing, though there are systematic errors in that that can make the reported numbers deceptive (the finite resolution of the satellite images is the issue here).”
“Could” is not definitive enough. “Will actually” is more appropriate.
It’s a complex system, and it’s difficult to anticipate everything that could happen. One doesn’t generally speak in absolutes in science, unless you can eliminate every possible confounding factor.
Coulda,shoulda and woulda are all speculative. A warming arctic WILL melt more ice.
Actually that depends on whether it’s above freezing or not. -1.2°C isn’t warm enough to melt the surface of the ice sheet.
Looks pretty warm to me fella!!! http://arctic.atmos.uiuc.edu/cryosphere/NEWIMAGES/arctic.seaice.color.000.png
But that graphic doesn’t show temperature!
I know it shows the result of temperature..
But it’s not temperature.
Do you even have a high school education?
Ray the wind-generated spreading (“Ekman transport”) continues, but the surface air temperature has dropped again below -2C. It stayed warm long enough to clean the lens off of the camera, so can get a view from near the pole again. As you can see the meltponds are all frozen over, and now covered with snow. Additional insulation.
Open water cools more quickly than water with ice on it, so in a way, the larger open area may end up with an accelerated freeze.
The question now is the sea temperature in the region the ice is going to get pushed towards. At the moment, the flow is towards Greenland which isn’t good, but it’s expected to reverse by the end of the week.
Carrick (Comment #103031)
September 10th, 2012 at 1:09 pm
But it’s not temperature.
Do you even have a high school education?
Don’t need one to tell the ice is melting do you?
According to Environment Canada , Nunavut Canada anomalies (from 1971-2000 baseline) spiked as high as 8C in Dec 2010.
They then plunged to almost -3 in April 2011 and have averaged around 1.5C since then.
The “D” column is Departure From Normals:
http://climate.weatheroffice.gc.ca/prods_servs/cdn_climate_summary_report_e.html?intMonth=8&intYear=2012&prov=NU&txtFormat=text&btnSubmit=Submit
Where is Nunavut?
http://en.wikipedia.org/wiki/File:Nunavut,_Canada.svg
Spikes that high occur every 8 years or so.
Jan 1940 – 9.1C anomaly as an example.
http://climate.weatheroffice.gc.ca/prods_servs/cdn_climate_summary_report_e.html?intMonth=1&intYear=1940&prov=NU&txtFormat=text&btnSubmit=Submit
10 years later, -4.4 to -9.8 anomalies
http://climate.weatheroffice.gc.ca/prods_servs/cdn_climate_summary_report_e.html?intMonth=1&intYear=1950&prov=NU&txtFormat=text&btnSubmit=Submit
It is not perhaps it is melting or maybe it is melting or could be melting. It “IS” melting….
Carrick (Comment #103036)
September 10th, 2012 at 1:35 pm
Ray the wind-generated spreading (“Ekman transportâ€) continues, but the surface air temperature has dropped again below -2C. It stayed warm long enough to clean the lens off of the camera, so can get a view from near the pole again. As you can see the meltponds are all frozen over, and now covered with snow. Additional insulation.
Open water cools more quickly than water with ice on it, so in a way, the larger open area may end up with an accelerated freeze.
The question now is the sea temperature in the region the ice is going to get pushed towards. At the moment, the flow is towards Greenland which isn’t good, but it’s expected to reverse by the end of the week.
You talk as if an ant were to piss it could possibly put out a forest fire…
It may be melting. But based on provisional data Jaxa extent is higher than it was on 9/5. So I wouldn’t be absolutely certain the minimum has not been reached.
Mind you, I doubt it’s been reached, but I wouldn’t bet my house on it.
Any Quatloo bets on a record low winter extent this year since 1979?
“Any Quatloo bets on a record low winter extent this year since 1979?”
On 14 days in Mar/Apr it hit 98.x% of the 1980s average.
I think we will see some days over the 1980s average.
Lucia,
“It may be melting. But based on provisional data Jaxa extent is higher than it was on 9/5. So I wouldn’t be absolutely certain the minimum has not been reached.”
That’s based on confirmed data, not provisional.
They seem to have stopped issuing provisional data, for at least 1 day anyway.
Today’s data was for the 9th only. Normally the confirmed data for the 9th would have been accompanied with provisional data for the 10th, but today it wasn’t.
cyclonbuster,
“It is not perhaps it is melting or maybe it is melting or could be melting. It “IS†melting….”
It may or may not be melting, but on the most recent figures, it is growing in extent.
“You talk as if an ant were to piss it could possibly put out a forest fire…”
Carrick was expressing an opinion (to me), and was explaining the possibilities. I don’t know why you take exception to that.
ANS: Because he’s a fruit.
Ant’s love fruit too……..
“It may or may not be melting, but on the most recent figures, it is growing in extent.”
But not in thickness… LOL! In fact that too “IS” at a record low. http://psc.apl.washington.edu/wordpress/wp-content/uploads/schweiger/ice_volume/BPIOMASIceVolumeAnomalyCurrentV2_CY.png
I received an answer to my question to IARC-JAXA and here it is:
“The provisional figures are tentative values determined
with the uncompleted satellite data which have spatial
data gap. So the confiremed figures are set next day using
the satellite data with full spatial coverage.
We know the setting is puzzling people. So we changed the
setting of our sea-ice data uptate system yesterday so that
the provisional figures are not shown on the web.”
Personally, I find it incredible that they were publishing figures based on only partial coverage, without including an allowance for the area not covered.
However, this explains the massive differences between provisional and confirmed figures.
Cyclonebuster,
“But not in thickness… LOL! In fact that too “IS†at a record low.”
But surely the subject of this bet is extent?
But cyclone, the Antarctic “thickness” is actually showing long term gains! LLOL!
http://vimeo.com/46429608
MrE (Comment #103102)
September 11th, 2012 at 12:15 am
But cyclone, the Antarctic “thickness†is actually showing long term gains! LLOL!
http://vimeo.com/46429608
Really better check with GRACE she begs to differ with you….
The total global ice mass lost from Greenland, Antarctica and Earth’s glaciers and ice caps during the study period was about 4.3 trillion tons (1,000 cubic miles), adding about 0.5 inches (12 millimeters) to global sea level. That’s enough ice to cover the United States 1.5 feet (0.5 meters) deep.
“Earth is losing a huge amount of ice to the ocean annually, and these new results will help us answer important questions in terms of both sea rise and how the planet’s cold regions are responding to global change,” said University of Colorado Boulder physics professor John Wahr, who helped lead the study. “The strength of GRACE is it sees all the mass in the system, even though its resolution is not high enough to allow us to determine separate contributions from each individual glacier.”
http://www.nasa.gov/topics/earth/features/grace20120208.html
Cyclone, Greenland isn’t in the Antarctic.
The confirmed extent figure for the 10th is 3.671 million km^2, a FALL of 23,907 km^2 since the 9th, and only 6,875 km^2 higher than the potential minimum on the 7th.
The cumulative mean extent for September at Sept. 10th, was 3.709 million km^2, compared to 3.713 million km^2 on the 9th.
Clearly, the current situation is finely balanced and depending on conditions, could go either way over the next few days.
cyclonebuster,
Again, I thought the subject the bet was arctic sea ice extent, the melting of which adds nothing to sea level rise.
Cyclonebuster –
Ray (#103054) said that the ice is growing in extent.
In response, you say in #103097, “But not in thickness… LOL! In fact that too “IS†at a record low.” And link to a graph of ice volume.
However, the the graph to which you linked, extends only to mid-August (perhaps the 20th by eyeball). The fact that ice volume was still decreasing in mid-August, does not contradict or mitigate the claim that ice extent is currently increasing in early September.
In fact, your comment is directly contradicted by this graph, current as 2 Sept., which shows that ice volume has “bounced” from the bottom and is currently increasing. The graph is from the same site as your reference, in fact both are shown together on this page.
Postscript to #103113 above: To clarify, the graph cited there is ice volume anomaly, not the actual ice volume. However, the mean ice volume flattens in early September. Hence, a significant increase in anomaly (greater than 1000 km^3) at this time of year implies that the ice volume itself is also increasing.
Ray:
I think upon reflection they are agreeing with you. 😉
Now we know how it was generated we know it will always be lower than the real number by an unknown factor.
The provisional data is basically meaningless, and sensibly, upon contemplation, have chosen to no longer include it in their updates.
On another topic, the large systematic corrections in GRACE I’ve never felt comfortable with.. these are larger than the effects they are trying to measure, and there is no really good way to independently confirm the model used for the systematic corrections.
CESat a newer system with fewer internal systematics, finds, unsurprisingly in contradiction to GRACE Mass Gains of the Antarctic Ice Sheet Exceed Losses. (H/T WUWT.)
It’s unsurprising to me because I’ve always felt GRACE was producing unreliable values. What it is designed for, mapping the gravitational field of the Earth, it is fine at. A lot more work needs to be done to deal with the complex issues related to subsidence and changing mass load above continental plates, IMO, before these heavily modeled results can be taken at face value.
The value posted for 9/10 is 3671406. That’s below the value for 9/6 (3676406) above the value for 9/5 (3681094) but below the value for 9/4 (3726563). So, for now, the extent is going sideways. But it’s still early. Too early for re-freezing to have any strong advantage over melting and, certainly, weather patterns will be affecting what happens over the next 2 or 3 weeks.
Ray:
Again, the concept that the sea level won’t appreciable change whether the ice is frozen and floating on the surface or melted is not something cyclone understands.
I don’t mean that in any nasty way, I am certain, at this point, it is merely true. We’re dealing with somebody with a remedial knowledge of science.
One of the web sites I look at is the archive of ice spreading done by NRL.
Files have names like: “icespddrf2012091018_2012091600_035_arcticicespddrf.001.gif”
The format for the file contains yyyymmddhh_yyyymmddhh
The first date is the actual time the file was generated, the second is modeled surface flow rates (it includes back-casting as well as forecasting. The most accurate value for a given day is probably the backcast value.)
Anyway This is what is predicted for the end of the week. Ice movement into very cold water. Unless something major crops up the minimum ice will have occurred this week.
(I tend to think of the seven or 10 day smoothed value as being a more reliable measure of the “real” mean, not all of those wiggles in the data are signal, some of them are measurement error.)
MrE (Comment #103106)
September 11th, 2012 at 2:49 am
Cyclone, Greenland isn’t in the Antarctic.
Fact “IS” you missed what the NOAA article said about Antarctica……. Comprende?
I’m pretty sure I just linked scientific evidence that, as expected from the global climate models, the ice mass in the interior of Antarctica is growing.
Do you like looking clueless?
Carrick (Comment #103127)
September 11th, 2012 at 9:20 am
I’m pretty sure I just linked scientific evidence that, as expected from the global climate models, the ice mass in the interior of Antarctica is growing.
Do you like looking clueless?
You” ARE” clueless……. Big difference from looking and being……
http://www.skepticalscience.com/pics/ArcticEscalatorv2.gif
LOL, cyclone links to an advocacy group run by advocates … to counter peer reviewed scientific research.
And the best part, the article I linked to related to Antarctica interior ice mass, and he points us to an image of … wait for it…. Arctic sea ice extent, accompanied by a st*pid puppet show.
Carrick,
Aren’t the large GRACE corrections made to account for GIA? This latest estimate uses satellite altimetry data, so presumably that also needs GIA correction(?)
I found a nice review paper written by the lead author of your linked article. This extract:
…if I’m reading it right suggests my presumption was correct but that altimeter GIA (ERS being a forerunner to ICESat, AIUI) corrections are slightly smaller.
Paul S, as I understand it, it’s an easier correction to make because you’re just measuring surface height changes (and there are corresponding radar measurements of ice thickness so you have ground truth), the correction is smaller, the effects less complicated, and because the effect of GIA runs in the opposite direction to the observed effect (an increase in elevation of the ice sheet over time), it’s harder to manufacture an increase in ice mass.
Also, by a little smaller, isn’t the correction for ICEcap about a factor of three smaller? Put another way, the corrections for ICEcap are smaller than the final numbers in their analysis, for GRACE this is reversed.
I believe the ICEcap results are also largely consistent with what is expected from modeling (from the abstract):
so we have a degree of coherence seen here between model and data, not observed with GRACE. Unexplained variability can be real, but it can also be due to improperly modeled systematic error.
It’s be interesting to see what happens to GRACE’s result if you include their gravinometric measurements together with the radar and laser soundings.
BTW, thanks for the link. I’ll read it when I get some down time. If you find anything factually wrong with what I’m saying, I’m happy to eat humble pie, this isn’t my area, but I’m very interested in having my understanding “right”.
I agree. The information content in the wiggles is mostly that it’s wiggling up and down now. Happens near a minimum– but of course, weather being what it is, it’s no guarantee.
Carrick:
“(I tend to think of the seven or 10 day smoothed value as being a more reliable measure of the “real†mean, not all of those wiggles in the data are signal, some of them are measurement error.)”
I’m not entirely sure what point your making. Do you mean more reliable than the daily figures? If so, I agree, but according to the IARC-JAXA website, as far as I can tell, the daily figure is actually an average of the last 2 days, so is smoothed to an extent.
Anyway, since the current September mean is now over 10 days, it should be fairly reliable.
Ray, yes I mean more reliable in the sense that you’ve reduced measurement related uncertainty. As I said not every wiggle in data is associated with the quantity you’re trying to measure. There’s uncertainty introduced into these measurements e.g. by cloud coverage generated by storms. Some of the wiggles are from these, and from the pixellation of the measurements themselves (spatial under-sampling).
Just an FYI, here’s a paper discussing uncertainty in measuring arctic sea ice extent and its impact.
I’m not arguing that we’re not near the minimum, btw. I was just saying that the day with the lowest value in JAXA isn’t a true measure of the actual minimum ice extent due to measurement error and that you needed use a smoothed version in order to do so. (Of course there will be uncertainty in this too, due to throwing away the high frequency component of the signal).
Carric
I agree. We can’t know for sure the date when the minimum occurred, and I think we’d likely get a better estimate of the right day using centered 7 day averages.
Carrick,
Also, by a little smaller, isn’t the correction for ICEcap about a factor of three smaller?
Could be that I’ve misunderstood. I can’t get to the paper right now but I thought there were a couple of figures suggesting the difference wasn’t that big.
so we have a degree of coherence seen here between model and data, not observed with GRACE.
I don’t think that necessarily follows. Estimates suggesting large net mass balance loss are also coherant with model predictions of increasing snowfall if you allow for more dynamic ice loss.
That said, the Zwally review paper sets out a good case for believing the larger estimates to be erroneous.
Thanks Paul. I am going to go back and review the numbers. In my memory at least, the altimetry measurements are more tightly constrained and suffer from “significantly less” systematic biases that need to be corrected for than GRACE, relative to their respective measured quantities.
Regarding greater dynamic ice loss—I don’t think that’s expected from the models, so I believe what I said is true: ICEcap is close to what is expected from the models. That doesn’t make either right, my only real point is you get a tad more confidence in a measurement if it lines up with model output (unless you really don’t trust the models, then everything is on the table).
Carrick,
It’s my impression that scientists involved have very little confidence in numerical models of ice sheet dynamics, particularly at spatial scales where important surface details can’t possibly be included due to grid resolution. Hence projections of sea level change contribution from ice sheet dynamics have tended to derive from extraploations of currently observed behaviour.
What model output are you referring to? I don’t think any of the CMIP3 models inclued dynamic ice sheet capability. Not sure about CMIP5.
Paul, my impressions have been gleaned mostly from reading from reading IPCC AR4 [and literature reviews]. Do you really need GCMs to model ice sheet dynamics? It’s my impression they do that sort of thing with regional scale modeling.
It’s my understanding you also don’t need anything beyond regional scale modeling to predict increased precipitation in the interior in a warming climate with constant RH. That’s part of why I thought that was a fairly robust prediction.
[In going back through the literature, the quick answer is “no”. It’s a legitimate question as to “which models” predict what. That will take some time to review. If you know of quantitative models that predict larger than generally anticipated dynamic losses, I’d be happy to have the link.]
Well, now we know for sure, and all uncertainties are removed!
On Sept. 11th, the extent as fallen by 80,156 km^2, to 3.591 million km^2, so the figure of 3.665 million km^2 on Sept. 7th, was not the minimum.
As a result of this, the running September mean is down from 3.709 million km^2, to 3.698 million km^2.
Such large daily falls are unusual, but not unheard of, in September. In 2003, there was a daily fall of 96,719 km^ on September 15th and in that year, the daily minimum was reached on Sept. 18th.
Carrick,
I don’t know of any model runs at all that have included “complete” ice sheet physics using historical or projected forcing data in a transient mode. Are there any?
By the way, I’m entirely on board with the expectation of increased precipitation causing a positive mass balance change. In terms of the expected overall net change though, you’d need to take into account melt, ablation and dynamic ice sheet flows. At least for the latter of these I don’t think there have been any concrete model expectations published at the multi-decadal scale of immediate interest.
I think comparing models with outputs in this case requires a separation of sinks/sources, i.e. comparing model snow accumulation with observed accumulation, model melt and dynamics with observed melt and dynamics. Can GRACE or ICESat be used to draw inferences about the accumulation rate in isolation, and are there significant differences between them?
Ray,
“Well, now we know for sure, and all uncertainties are removed!”
If you’re referring to the JAXA fall of 80,000 sq km (approx), it’s not for sure. JAXA numbers are often revised next day, mostly upward. The previous day melt of 24,000 is safer.
Personally, I think in inferring melt from the difference of daily totals, you have to expect a lot of noise.
Nick,
If you read my previous posts, you will see that I am aware that in the past, the “provisional” figures are always revised upwards but that is no longer the case.
As a result of me sending an e-mail to IARC-JAXA they are no longer publishing “provisional” figures, because they were based on partial data.
Their exact reply to me was as follows:
“The provisional figures are tentative values determined
with the uncompleted satellite data which have spatial
data gap. So the confiremed figures are set next day using
the satellite data with full spatial coverage.
We know the setting is puzzling people. So we changed the
setting of our sea-ice data uptate system yesterday so that
the provisional figures are not shown on the web.”
I must admit, when I saw today’s figures, I thought they might have reverted to “provisional” figures, but I don’t think that is the case.
The difference between “provisional” figures and the confirmed ones, was not “noise”, but simply missing data.
Of course, we still can’t be sure that the daily figures are 100% accurate, but they are all we have to go on!
I probably exaggerated when I said “all uncertainties have been removed”, but I think it is now fairly certain that Sept. 7th was not the minimum.
Paul S yes there are some. They involve mantle physics so (from my perspective) are rather beautiful things to behold on their own.
Excellent question. Not sure I know the answer to this one. It might be fun to direct your questions to one or both groups and see if they respond.
Nick:
It’s Ray’s fault. He asked them why they included provisional numbers when they don’t mean anything, and they basically say doh! and quit publishing them. Dang bloggers. 😉
Carrick,
“He asked them why they included provisional numbers when they don’t mean anything, and they basically say doh! and quit publishing them.”
Not *strictly* correct.
I passed no judgment.
I only asked why the confirmed figures were usually lower than the provisional figures and if that was always the case, whether or not the extent was decreasing or increasing.
They must have come to the conclusion that they were meaningless by themselves.
Just to make a liar out of me, IARC-JAXA seem to have upwardly revised the figure for the 11th, and appear to have published a provisional figure for the 12th.
The confirmed figure for he 11th is 3.603 million km^2, still quite a large change of -68,125 km^2 from the 10th, and the provisional figure for the 12th is 3.585 million km^2, a change of -17,500 km ^2 compared to the 11th.
Based on recent revisions, the latter change figure could end up being positive, but I don’t know if that is a valid assumption. It may be that they have found a more accurate method of estimating the provisional figure and that previous revisions are no longer a useful guide.
Based on the confirmed figures only, the running September mean for the 11th is 3.699 million km^2, compared to 3.709 million km^2 on the 10th.
Further explanation of the figures from IARC-JAXA, in response to another e-mail from myself:
“Sometimes a part of satellite data were transferred to our data
processing system after the confirmed figures was fixed (data delay).
In such case, the confirmed figures are revised on next day (but not so large I think). ”
Just when I thought I had it sussed!
Thanks for looking at this Ray. I’m guessing this is covered ground on Neven’s ice blog, but he’s so prolific that unless you are a regular reader it’s hard to keep up with him.
Carrick,
I am not familiar with Neven’s ice blog, although I suppose I should be, but there just isn’t enough time to read everthing, even on this blog, let alone others.
OTOH, if it had been covered, would the “provisional” figures not have been removed before now?
The “confirmed” extent figure for the 12th was unchanged at 3.586 million km^2, a change of -17,500 km^2, compared to the 11th.
The published figure for the 13th, is 3.570 million km^2, a change of -16,093 km^2 compared to the 12th, so the extent appears to continue to fall, albeit at a relatively slow rate.
This leaves the running mean figure for September at 3.681 million km^2, at the 13th., compared to 3.690 million km^2 for the 12th.
Based on the 2003-11 mean, the extent figures don’t normally show a sustained increase until September 18th.
The 7 day mean daily change is currently -15,245 km^2/day, compared to the 2003-11 average 7 day mean daily change of -9,521 km^2/day for the same period.
The ice extent figure for the 13th is confirmed at 3.57 million km^2, and the figure for the 14th is 3.542 million km^2, a change of -27,969 km^ compared to the 13th.
As a result, the running September mean for the 14th, was 3.671 million km^2, compared to 3.681 million km^2 on the 13th, a change of -9,924 km^2.
The 7 day mean daily change at the 14th, was -17,545 km^2/day, compared to the 2003-11 average 7 day mean daily change of -4,983 km^2/day.
The actual extent figure for the 14th was 55,065 km^2 higher than the projection based on the 2008-11 mean daily change since June 1st, having reached a figure of 201,615 km^2 higher on September 9th. So while the extent is catching up on that projection, a final September mean of over 3.6 million km^2 remains a possibility.
Even if the extent continues to fall at the same rate as the last 7 days, the final September mean would be 3.52 million km^2, but that would imply no increase in extent during September. While that would be highly unusual (unprecedented?), it has been a very unusual year so far.
The rate of decline in IARC-JAXA seems to be increasing again, with a change of -65,938 km^2 on the 15th, leaving the extent figure at 3.476 million km^2.
As a result, the rolling mean for September is now 3.658 million km^2, compared to 3.671 million km^2 on the 14th.
The 7 day mean daily change was -28,437 km^2/day, compared o the 2003-11 average for the same period of -5,037 km^2/day.
The figure of +55,065 km^2 above the projection based on the 2008-11 mean change, which I quoted yesterday, was actually a day out, and the actual figure for the 14th was only +262 km ^2, and on the 15th it was -66,936 km^2, so for the first time since August 29th, the actual figure is lower than the projection.
With only 3 days to go until the 2003-11 average date on which a sustained increase in extent starts, it now seems unlikely that that will occur this year.
Ray. I saw that. I need to tweak my script because it threw an error owing to the “NA” in the JAXA record for this data in 2002. Otherwise, I’d have blogged it. I’ll tweak Monday.
No JAXA data for the 16th yet published on the website.
I notice that it is a puplic holiday in Japan, so that may be why.
It is apparently “Respect-for-the-Aged Day”.
Just wondering, does that mean that it is o.k. NOT to show respect for the aged on all of the other days of the year?
Ray–
We celebrate Thanksgiving but it’s ok to give thanks everyday. 🙂
The buoys are registering above 0°C right now, and the melt ponds are back.
If I’m reading it right, the day of minimum extent in 2007 was 9/24.
As we’ve been saying upstream, this is all about weather now.
The figure for Sept. 15th has now been revised upwards to 3.519 million km^2, so the actual change on that day was only -23,125 km^2.
The figure for Sept. 16th. was 3.489 million km^2, a daily change of -29,531 km^2, and that for Sept. 17th. is 3.537 millon km^2, daily change of +48,125 km^2, leaving the latest figure slightly higher than that for the 15th.
The figure for the 17th may be revised, but on past experience is unlikely to be revised downwards.
As a result of the upward revision of the figure for the 15th, the rolling Sept. mean at the 17th was 3.643 million km^2, i.e. still above the 3.6 million mark.
The rolling 7 day change is now -19,174 km^2 and the extent figure is now 31,862 km^2 above the projection based on the 2008-11 mean daily change, since June 1st., and that projection continues to be a fairly accurate guide to the daily extent figures.
Of course, it is too soon to say if the figure of 3.489 million km^2 on the 16th represents the Sept. minimum, and I think we need at least 7 days of increases, before we can be absolutely sure.
The above freezing period continues, at least at the buoy location, and we’re seeing widespread melting there now. Solar exposure is getting pretty low now, we’re only four days away from the equinox, so radiation-driven ice melt shouldn’t be very important now. It’s just the warm water under the surface I think.
A bit of a confusing picture with today’s figures.
The figure for the 15th has been revised upwards again, leaving the daily change at -20,938 km^2.
The figure for the 16th is unchanged, but that for the 17th is revised downwards, leaving the change at only +16,093 km^2 (proving me wrong when I said it was unlikely to be revised downwards).
The figure for the 18th (I suppose we have to consider it provisional), is 3.554 million km^2, a change of +49,063 km^2, compared to the revised figure for the 17th., but only +17,031 km^2 compared to the original figure.
The net result is the second successive daily gain and a change of +65,156 km^2 since the possible minimum on the 16th.
The rolling Sept. mean is now 3.637 million km^2 and the 7 day change is -7,009 km^2.
Having been a mere 70 km^2 below the projection based on the 2008-11 mean daily change on the 17th., the actual figure is 48,509 km^ above it on the 18th.
Not much ambiguity about the figures for the 19th.
The daily extent is 3.674 million km^2, up 119,844 km^2 on an unchanged figure for the 18th.
This means that the extent figure is now 185,000 km^2 above the figure of 3.489 million km^2 for the 16th, and subject to further adjustments, I think it looks fairly certain that was the minimum for the season.
The rolling September mean at the 19th was 3.639 million km^2, so it also seems possible that the figure of 3.637 million km^2 for the 18th was the minimum mean figure for September.
The 7 day rolling daily change is now +12,612 km^2, compared to the 2008-11 mean daily change of -1,758 km^2 at the same date, and the extent figure is now 145,393 km^2 higher than projected using the 2008-11 average daily change since June 1st.
Actually, on reflection, the size of the daily gain for the 19th is so large that it makes me slightly suspicious, and I wouldn’t be surprised at a downward revision tomorrow.
The daily gain figure for the 19th is the largest recorded for that date since 2003, although not unusual for early October.
At this rate, the September mean could go anywhere!
Ray–Well… probably up anyway.
No change in the extent figure for the 19th, and a further increase of 105,937 km^2 for the 20th, brings the latest extent figure to 3.78 million km^2.
This brings the rolling Sept. mean to 3.646 million km^2 and the 7 day mean change figure to +30,044 km^2/day, compared to the 2003-11 daily mean change of -1,253 km^2/day for the same date.
The actual extent figure is now 216,940 km^2 higher than the projected extent based on the 2008-11 mean daily change since June 1st.
Having fallen to a record low, the ice now seems to be recovering at a much faster rate than normal for this time of the month.
Even the daily change for the remainder of the month were equal the current 7 day mean change, the September mean extent figure would be 3.746 million km^2.
Using the 2003-11 daily mean change figure, the Sept. mean would be an almost identical 3.745 million km^2.
A small fall in the extent figure for the 21st, to 3.765 million km^2, down 15,312 km^2.
As a result, the cumulative September mean is now 3.651 million km^2, up 5,667 km^2 and the 7 day rolling mean change is +31,853 km^2/day, compared to the 2003-11 average of -1,307 km^2/day.
The 21st is the last day on which the 2003-11 7 day mean daily change is negative.
The latest extent figure is now 173,437 km^2 higher than the projection based on the 2008-11 mean daily change since June 1st.
If the extent were to continue to increase at the rate of the current 7 day mean, the September average would be 3.733 million km^2.
Based on the 2003-11 mean daily change, it would be 3.738 million km^2 and based on the 2008-11 mean daily change, it would be 3.746 million km^2.
An increase in of 11,406 km^2 on the 22nd, bringing the extent figure to 3.776 million km^2, still slightly lower than the figure on the 20th.
As a result, the cumulative September mean increased by 5,670 km^2 to 3.657 million km^2 and the 7 day mean change increased to 36,473 km^2/day, compared to the 2003-11 7 day average change of +3,423 km^2/day.
The extent figure for the 22nd was 135,388 km^2 higher than the projection based on the 2008-11 mean daily change since June 1st.
Estimated final September mean figures, based on various assumptions of future changes are as follows:
Based on 7 day mean = 3.733 million km^2
Based on 2003-11 mean daily change = 3.736 million km^2
Based on 2008-11 mean daily change = 3.736 million km^2
Those estimates would currently make DRUK the closest bet to the actual.
The 2012 daily change has been conforming pretty well to the 2nd order polynomial trends of both the 2003-11 and 2008-11 mean daily change figures, but is currently below that those trends, so a further “correction”, similar to those on Sept. 19th & 20th might be expected before the end of the month.
Whether this trend continues, depends on whether we get wind-driven compaction or not. Given how thin the ice is, it could drive numbers down to the mid 3.6’s if it happened this week.
Everybody’s called the minimum by this point. No way to get below 3.6 without another major polar storm (about 1/2 of the ice is still so fragile that a major storm, while not predicted, could cause real havoc).
Carrick,
Do you mean the average when you say 3.6 million?
I wondered if the 225,000 km^ or so, which “appeared” on the 19th and 20th might have been near the 15% concentration threshold and a very small increase in concentration just tipped it over. If that was the case, it presumably could “disappear” just as quickly.
Ray, I don’t think the average will ever reach 3.6 Mm2 without say a series of nuclear explosions or one huge asteroid.
I just mean that the daily value could in principle dip below 3.6 Mm2 before going into “full recovery”, but if it did, it probably would require another major polar vortex.
The extent figure for the 23rd is 3.823 milion km^2, an increase of 46,406 km^2.
As a result, the cumulative September mean is now 3.664 million km^2 and the 7 day mean change is +47,634 km^2/day, compared to the 2003-11 7 day average of +10,469 km^2/day.
The extent figure at the 23rd was 139,751 km^2 higher than the projection based on the 2008-11 mean daily change since June 1st.
Estimated final September mean figures, based on various assumptions of future changes are as follows:
Based on 7 day mean = 3.746 million km^2
Based on 2003-11 mean daily change = 3.740 million km^2
Based on 2008-11 mean daily change = 3.741 million km^2
The extent figure for Sept. 24th is 3.792 millon km^2, an increase of 49,688 km^2 compared with the 23rd.
As a result, the cumulative September mean is now 3.673 million km^2 and the 7 day mean change is +52,433 km^2/day, compared with the 2003-11 average 7 day change of +17,316 km^2/day.
The extent figure is now 165,075 km^2 above the projection based on the 2008-11 mean daily change since June 1st.
Estimated final September mean figures, based on various assumptions of future changes are now as follows:
Based on 7 day mean = 3.749 million km^2
Based on 2003-11 mean daily change = 3.741 million km^2
Based on 2008-11 mean daily change = 3.741 million km^2
I seem to have got my fingers in a twist in the above post.
The correct figure for the 24th is 3.872 million km^2.
A small fall in the extent on the 25th of 4,219 km^2, to 3.868 million km^2.
As a result, the cumulative Sept. mean has increased by 7,804 km^2, to 3.681 million km^2 and the 7 day rolling average change has fallen by 7,612 km^2 to +44,821 km^2/day, compared with the 2003-11 average 7 day change of +22,817 km^2/day.
When the large increases on Sept. 19th and 20th fall out of the 7 day mean, there will be a fall to closer to the 2003-11 average and the current 5 day mean, excluding those figures, is +17,594 km^2/day.
The actual extent figure was 113,048 km^2 (3%) higher than the projected extent based on the 2008-11 mean change x 1.24 since June 1st.
Estimated final September mean figures, based on various assumptions of future changes are now as follows:
Based on latest 7 day mean daily change = 3.734 million km^2
Based on 2003-11 mean daily change = 3.733 million km^2
Based on 2008-11 mean daily change = 3.737 million km^2
Based on last 5 day mean daily change = 3.721 million km^2
Based on a zero change for the remainder of Sept. = 3.712 million km^2
Consequently, a mean figure of over 3.7 million km^2 now seems almost certain, barring further declines in the extent, although that is not impossible.
The extent figure for Sept. 26th is 3.882 million km^2, a change of +13,906 km^2 compared with the 25th.
This brings the cumulative Sept. mean to 3.688 million km^2, an increase of 7,738 km^2, and the 7 day mean change to +29,867 km^2/day, compared with the 2003-11 7 day mean daily change of +26,594 km^2/day.
The actual extent is now 51,052 km^2 (1.3%) higher than the projected extent based on the 2008-11 mean daily change since June 1st.
Estimated final September mean figures, based on various assumptions of future changes are now as follows:
Based on latest 7 day mean daily change = 3.724 million km^2
Based on 2003-11 mean daily change = 3.730 million km^2
Based on 2008-11 mean daily change = 3.731 million km^2
Based on last 5 day mean daily change = 3.722 million km^2
Based on a zero change for the remainder of Sept. = 3.714 million km^2