By Steve Goddard
I need help from the readers to determine if 2010 will finish ahead or below 2005 – the red line in the DMI graph.
2010 is currently tracking just below 2005, but note there was a downwards dip in mid-September, 2005. What caused this?
The PIPS video below shows what happened in September, 2005.
In mid-September, strong winds started blowing off the East Siberian and Laptev Seas, which compressed the ice towards the North Pole. This caused the dip seen in the DMI graph.
The images below show the current date in 2005 and 2010 respectively. Note that 2005 had a lot of thin/low concentration ice in the Laptev/Kara Seas which was vulnerable to being blown around by the wind in September. The ice is less extensive, but thicker in that region in 2010.
What do you think? Will 2010 beat 2005? Please explain your reasoning.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
stevengoddard says:
August 13, 2010 at 12:13 am
EFS_Junior
Sticking up for PIOMAS?
You might as well tie yourself to a boat anchor.
_____________________________________________________________
Hmm, given what you’ve posted on the subject matter to date, why of course, what with all of PIOMAS’s real numbers with actual engineering units of measure attached vs your ill defined ROI and ill defined pixel-metres and PIPS 2.0 is actual data vs the reality that PIPS 2.0 is an operational/navigation model and your specious arguments that your predictions are perfect because you’re using PIPS 2.0 model output nonsensical handwaving.
Peter Ellis says:
August 13, 2010 at 1:57 pm
Because my method of prediction is very poor and completely statistical in nature. I fit a line to the 2002-2009 JAXA data where y = minimum extent and x = extent on a given day (in this case, 08/12). I then use the generated line to predict the minimum for 2010 given that day’s extent in 2010. However, there is high uncertainty with this method because (a) the line doesn’t fit the data that well and (b) there are only 6 degrees of freedom in the fit.
I reported the 95% confidence interval as I calculated it. Remember, this is statistical, not reality. For instance, say I tried to weigh a small mass several times and got the following values:
1
0.1
2
0.5
8
1.5
The average & std dev according to Excel are 2.18 & 2.93, and the 95% confidence interval is +/- 2.34. Clearly, the mass of the object isn’t negative even though the 95% interval has negative values in its range. It’s the same here, a value greater than the current extent is physically impossible, despite the statistics.
-Scott
While not quite getting to 2006, I think that sea Ice will surpass 2002, and 2005, and 2009. making it the 6th lowest ice extent.
http://psc.apl.washington.edu/northpole/webphotos/noaa2.jpg
Meltponds are freezing up, which correlates with DMI’s 80N temperatures.
EFS_Junior says:
August 13, 2010 at 6:30 pm
This is one of the weakest arguments I’ve seen. Are you even an engineer? If so, you’d realize that plenty of engineering metrics are DIMENSIONLESS – Reynolds numbers, Peclet numbers, etc. As to relationships between pixels and distance, how else do you suggest converting between distance and image pixels? I guess you don’t use Google Earth, a GPS, or online maps because they all have relationships between pixels and distance? Wow, just wow. And in case you haven’t noticed, Steve has admitted that PIPS 2.0 is a model….just like PIOMAS.
I definitely don’t agree with Steve’s extreme confidence in his methods, but so far they’ve done a good job so far and the ad hominem attacks from the professionals show just how much he’s struck a nerve.
-Scott
Ian H says:
August 11, 2010 at 5:42 pm
Please try doing the analysis before saying that it “would show it as much closer to 2009”. You weren’t specific saying “the past few weeks”, but using July 24-Aug 13 in the JAXA record showed an R^2 for 2009 = 0.9872 and an R^2 for 2005 = 0.9895. The best fit is for 2006 (as Steve has indicated in another post…2010 is nearly parallel with 2006) at R^2 = 0.9938.
In terms of slope (2010/other year), 2005 = 0.944 and 2009 = 1.264. With respect to closeness to unity, 2005 is the best match here, followed by 2006 (2009 is the worst match).
-Scott
Sure PIOMAS and PIPS used hypothesis. PIPS works—obviously. PIOMAS is sorely wrong—-obviously. When a hypothesis gets supporting evidence it’s a good hypothesis. When a hypothesis doesn’t get supporting evidence it is bad.
PIOPMAS is a bad hypothesis. PIPS is working good. Get over it.
BTW, should I believe what PIOMAS has to say about what will happen 30 years from now if what it says about now is wrong?
As for my 1.0 Prediction ‘only the Fast Ice left’ (fastened to the Greenland & Canadian Islands) … apparently this Ice spread out, and so I was right — – only the Fast Ice is left.
But that’s an excuse. I was wrong.
But … for a reason. And those blasted Fires (remember, when I posted on the Hurrican predictions – –
1. El Ninos normally disrupt Hurricane Patterns, even if giving them more energy theoretically — as Ninos mostly start up just West of Peru which is actually due SOUTH of the Carribbean & not so far away.
SO = LESS Hurricanes than Norm.
2. BUT, this was a “Modoki” or one starting in the WEST — so = MORE (like 2005).
3. BUT, a VERY Strong El Nino also activates the Saharan Dust storms
= No Hurricanes at all (nearly).
Perhaps the soot fall kills the Arctic weather. However, in my Sea Ice Update — if Helen@ur momisuglyarcus.org doesn’t substitute something else — again — I will state it is mainly the TIME LAG when one switches Hot-to-Cold — one can get Hot water still crawling along at 4 mph, yet cold (La Nina) Air giving CLEAR SKY = SUN … plus driving Ice out of the Basin – – like 2007 – – and 2010, NOW. But not last month.
… But the MAIN MELT MONTH is July, and the “lull” hit 2010 right in the Gut.
Even so, I expect less than HALF of Last year’s Minimum ICE, by Volume. It appears to already be under 2007’s minimum of ~ 5000 km3. (14,600km3 August Mean (monthly or Aug 1 ?) with a 10,200 Anomaly on July 31 = 4400 ????) Charts on left side of page at: http://psc.apl.washington.edu/ArcticSeaiceVolume/IceVolume.php
… It’s just that 2000+km3 of 30 foot thick Ice that never affected anything before … is now spread out over the basin in chunks.
… BAD NEWS: Next time there will be no unexpected EXTRA ICE to save the day.
The Cavalry is all used up.
… GOOD NEWS: the 60-year cycle implies we have substantial ICE GAIN for several years as, for 27 more years, La Ninas will outnumber EL NINOs by 2:1 , AND be stronger. In fact, both Long & short Cycles are in “Cold” Mode at present (the short has 10 year up & 10 down , plus a variable number in between, thus its multiple names as the “22”, “24” 26, or 28 year cycle. Note NOT ALL YEARS ARE COLD – – but we can expect more-Nina-than-Nino years, from 2003-2013. The long cycle seems to affect strength more than number & it ran Cold 1947-77 and Hot 1977-2007 — changeovers were midyear & FAST). Thus we MAY get unlucky 3-5 years from Now, BUT if we avoid that, we are safe for perhaps 50 years, I think.
The other risks of Global Warming rate THOUSANDS of times less — e.g.
6 Billion lives x 45 years loss x 10% = 27 Billion Life Year Loss. Risk is for say 3-4 years every 60 years = say 1.5 Billion or 1,500 MILLIONs.
100 million Displaced by Sea level Rise over 100 years @ur momisugly 1 year of Life for the inconvenience = 27,000 times LESS than this year, 1,500 times less than average.
DMI 80N Temps are plummeting, but remember that the freezing point of seawater is 271.23 K (at the surface), not 273.15 K. The purple line in this picture is placed at the freezing point of seawater. The light blue line assumes zero salinity, which obviously is not representative of the actual conditions in the Arctic Ocean. Still, having the temperature this far below average at this time of year is unusual, especially with the Arctic Oscillation starting to go Negative again.
Temperature north of 80N went down a tad more:
http://ocean.dmi.dk/arctic/meant80n.uk.php
Steve,
This looks like a stochastic governor effect. Hit the buffers too early and you get overshoot. Look closely at the melt deceleration rates through Aug and Sept across all years; consider the latent heat involved and how the energy input is dropping off. Sept 11th seems to be the sweet spot. If it goes into freeze mode much before that, it’ll likely take a proportionate dip again to compensate (other factors being equal). 2005 and 6 were both aiming for Sept 1st from about mid August. Email me if you want the plots, etc. that suggest this. Sorry, don’t hold with this fast/slow ice malarkey; apart from ocean current delivery, the amount of energy involved is fairly finite.
Scott says:
August 13, 2010 at 7:20 pm
Peter Ellis says:
August 13, 2010 at 1:57 pm
@Scott
It’s currently using present extent to predict a final extent of 5.03e6 +/- 4.29e6 km^2 (95% confidence interval).
… how can your error bars allow a forecast minimum of >= 6.14e6 km^2 given that that’s the current value? Is time about to reverse?
Because my method of prediction is very poor and completely statistical in nature. I fit a line to the 2002-2009 JAXA data where y = minimum extent and x = extent on a given day (in this case, 08/12). I then use the generated line to predict the minimum for 2010 given that day’s extent in 2010. However, there is high uncertainty with this method because (a) the line doesn’t fit the data that well and (b) there are only 6 degrees of freedom in the fit.
I reported the 95% confidence interval as I calculated it. Remember, this is statistical, not reality. For instance, say I tried to weigh a small mass several times and got the following values:
1
0.1
2
0.5
8
1.5
The average & std dev according to Excel are 2.18 & 2.93, and the 95% confidence interval is +/- 2.34. Clearly, the mass of the object isn’t negative even though the 95% interval has negative values in its range. It’s the same here, a value greater than the current extent is physically impossible, despite the statistics.
-Scott
_____________________________________________________________
I don’t know how you’re getting such large confidence bands that are O91) with the extent predictions.
I get the following values (Date of Prediction, Predicted Extent Minima, 95% confidence (+/-) interval (Excel Student T which is for small samples and gives larger confidence bands);
7/6/2010,4.685E6 km^2, 0.481E6 km^2
7/15/2010,4.935E6 km^2, 0.423E6 km^2
7/24/2010,5.003E6 km^2, 0.361E6 km^2
8/2/2010,4.963E6 km^2, 0.320E6 km^2
8/11/2010,5.043E6 km^2, 0.239E6 km^2
8/20/2010,4.959E6 km^2, 0.135E6 km^2 (5.675E6 km^2 extent on that date)
8/29/2010,4.937E6 km^2, 0.078E6 km^2 (5.285E6 km^2 extent on that date)
Thr last two dates use a calculated extent for those future dates derived from JAXA statistical model 2003-2010 inclusive.
The predicted 2010 minima extents show remarkable consistency for the real data, 7/15/2010 through 8/11/2010 inclusive).
Note that the confidence intervals are at least O(0.1) yours, and the confidence interval collapses rapidly as it approaches the actual extent minima.
EFS_Junior says:
August 14, 2010 at 9:52 pm
Looking at the plot, it’s clear that you’re right, my 95% confidence interval is way too high. I’ll have to troubleshoot my spreadsheet to find what is incorrect. I guess I pulled a Michael Mann and chose to write my own statistical stuff instead of letting a dedicated piece of software do it.
Thanks for your help.
-Scott
Charles W.;
Funny how everyone assumes warming bad, cooling good. Historically, it’s been the exact opposite. Personally, I hope the Arctic clears out and we get a Holocene Optimum!
Scott says:
August 15, 2010 at 12:35 pm
EFS_Junior says:
August 14, 2010 at 9:52 pm
Looking at the plot, it’s clear that you’re right, my 95% confidence interval is way too high. I’ll have to troubleshoot my spreadsheet to find what is incorrect. I guess I pulled a Michael Mann and chose to write my own statistical stuff instead of letting a dedicated piece of software do it.
Thanks for your help.
-Scott
_____________________________________________________________
What I did was subtract the linear regression curve from the JAXA 2003-1009 values. That gives you a set of vertical deviations from the linear fit for those years, I placed these valuse in an adjacent column called “sigma” I then did C9 = STDEV(C1:C7) to get the standard deviation (I didn’t use STDEVP as the dataset is small), finally I got 95% confidence by setting C10 = CONFIDENCE.T(0.05,C9,7), using Excel 2010.
I also did all the calculations out longhand (in the spreadsheet without using Excel functions) to double check my understanding of Excel 2010, all longhand calculations agreed exactly with the Excel 2010 function call results.
Hope this helps.