Guest Post By Willis Eschenbach
Anthony pointed out the selling of overhyped claims of the “dramatic thinning” of Arctic ice here. The title of the underlying scientific study is much more prosaic, Response of ice cover on shallow lakes of the North Slope of Alaska to contemporary climate conditions (1950–2011): radar remote-sensing and numerical modeling data analysis. (PDF). To their credit, the authors make no such claims of drama in their text, which is generally quite appropriately restrained.
Here is their complete “dramatic” dataset of the lakes around Barrow, Alaska, the northernmost point in the US:
Figure 1. Percentage of lakes in the low-lying tundra around Barrow, Alaska that are partially thawed in late April, 1992-2011. Photo Source.
It’s an interesting study. They noted that partially thawed lakes look very different on radar than when the same lakes are frozen solid. As a result, they’ve collected solid data that is not affected by urban warming. So … what’s not to like in the study? Let me start with what is to like in the study.
I do like the accuracy of the measurements. It’s an interesting metric, with very objective criteria. I like that they listed the data in their paper, and showed photos for each of the years. I like that they didn’t try to project the results out to 2080.
What I didn’t like is where their study went from there. After collecting all that great data, they immediately sent out for that perennial favorite, a global climate model … not my style at all.
So rather than pointing out that their study is models all the way down, I figured I’d just show the kind of analysis that I would do if I were handed the lake thawing data.
First thing I’d need for the analysis? MORE DATA. Piles and piles of data. So I went out and I dug up two datasets—Barrow temperature, and Barrow snow depths. I started with just the temperature, but it turns out that the correlation between temperature and the lake thawing isn’t all that good. It doesn’t explain much, the best correlation is with temperatures in December, 4 months prior to the thawing, at a correlation of 0.68. However, at least it gives a good idea of what’s been going on, because we have good records clear back to 1920.
Figure 2. Winter temperatures in Point Barrow (pale blue line) and the 17 year Gaussian average of the data. Photo source http://www.panoramio.com/photo/63484316
I note in passing that Barrow has a well-documented Urban Heat Island that is at its strongest in winter … and despite that, the 1930s and 1940s both had warmer winters than the last decade. I also note in this context of winter-business-as-usual that the study says:
Climate-driven changes have significantly impacted high-latitude environments over recent decades, changes that are predicted to continue or even accelerate in the near future as projected by global climate models …
… but I digress.
So the next obvious suspect for a correlation with the lake thawingis the snow depth. It’s an odd fact of nature that snow is a good insulator. It both slows down heat transfer by insulating the surface, and it keeps the wind from contacting the ice.
So I looked at the average snow depth data (scroll down to “Custom Monthly Listing” in sidebar) … but it’s not all that good at emulating the ice thawing either—in fact it’s worse. With snow depth, the best correlation with average snow depth is only 0.51, again with December coming out on top. So, having investigated single variables to try to emulate the lake thawing, I turned to the combination of snow depth and temperature … not much luck there either. In fact, the only way I could get a good correlation was to use the combination of the Nov-Dec-Jan average temperature, and the December snow depth. This gave me a correlation of 0.81, and a p-value of 0.001 … which turns out to be just barely significant. Here’s the emulation:
Figure 3. Emulation of Barrow lake thawing. Observations (thick red line) compares well with the emulation (thin green line). Correlation is 0.81, p-value is .0010.
Now … why did I say that a p-value of 0.001 is “barely significant”, when the usual level is a p-value of 0.05? Well … because I looked at so many possibilities before finding what I sought. All up, I looked at maybe 40 possibilities before finding this one. If you want to establish significance at the level of a p-value of 0.05, and you look at 40 datasets before finding it, you need to find something with a p-value less than 1-10(LOG(0.95)/N, where N is the number of datasets you looked at. For N=40, that gives a required p-value of better than 0.0013 … so with a p-value of 0.0010, my emulation just made it under the wire.
Next, I looked at what that same emulation would look like over the whole period 1950-2013 for which we have records, and not just the period 1992-2011 of the study (the “N=20” of the title). Figure 4 shows that result.
Figure 4. Exactly as in Figure 3, but covering the entire period of record.
OK … not a lot going on there. Now, those who follow my work know that I’m quite skeptical of this kind of modeling, particularly with such a short record. What I do to test that is first to find a model with an acceptable p-value. Then I take a look at both the emulation shown above, along with the same emulation using just the first half of the data to fit the parameters, and then the same thing using just the second half of the data. Figure 5 shows that result:
Figure 5. As in Figure 4, but showing the emulation based solely on the first half of the data (light blue), and that based solely on the second half (dark blue)
As emulations go, in my experience that’s not bad. The general shape of the emulation is well maintained, and neither of the two half-data emulations go far off of the rails, as is all too common with this type of analysis.
So that’s how I’d analyze the data, at least to begin with. My conclusions?
Well, my first conclusion has nothing to do with the lakes. It has to do with Figure 2, which shows that there is nothing out of the ordinary happening to Barrow winter temperatures. So whatever you might want to blame the lake thawing on, it’s not the local temperature. It’s hasn’t much changed over almost a century, it just goes up for a while and then down for a while.
The second conclusion is that the changes in the lake thawing dates over the period of study are not “dramatic”. In fact, they are boringly mundane. The only thing “dramatic” is the press release, which is no surprise.
The third conclusion is that I wouldn’t trust my emulation of lake thawing all that far … the problem is that with N=20, we have so little data that any conclusions and any emulations will be fraught with uncertainty. Heck, look at Figure 1 … up until a few years before the end of the data there was not even much trend. It’s just too short to conclude much of anything.
Next, I wouldn’t trust their “CLIMo Lake Ice Model” much further than I’d trust my emulation above. Again, the underlying problem is lack of data … but to that you have to add the unknown performance of the CLIMo model.
Finally, while the authors were restrained in their study, they cut loose in their quotes for the press release, viz:
“We’ve found that the thickness of the ice has decreased tremendously in response to climate warming in the region,” said lead author Cristina Surdu, a PhD student of Professor Claude Duguay in Waterloo’s Department of Geography and Environmental Management. “When we saw the actual numbers we were shocked at how dramatic the change has been. It’s basically more than a foot of ice by the end of winter.”
and
“Prior to starting our analysis, we were expecting to find a decline in ice thickness and grounded ice based on our examination of temperature and precipitation records of the past five decades from the Barrow meteorological station,” said Surdu, “At the end of the analysis, when looking at trend analysis results, we were stunned to observe such a dramatic ice decline during a period of only 20 years.”
I see nothing “stunning” or “dramatic” in their results at all. Overall, it’s quite ho-hum.
My warmest regards to all, it’s bucketing down rain here after a long period of drought, life is good.
w.
AS USUAL … if you disagree with me or anyone, please quote the exact words you disagree with, and give us your objection to those words. That way, we can all be clear exactly what it is you are objecting to.
DATA AND CODE: Primary sources given above, plus it’s all in my Excel spreadsheet, Barrow Lake Thawing …
I wonder if data on cloud cover would shed more light? Is there any data on black carbon being deposited ? Either of those might have an effect on the timing and rate of ice melt.
Hi Willis, nice work. What’s your reference for the exponential correction to the p-value? If I remember correctly, the Bonferroni correction (which has come in for a lot of controversy) just divides .05 by N, so for an N of 40 it gives a value of p=0.00125 that needs to be beat to achieve significance. This is almost identical to the value you found from the exponential, so who knows, maybe that’s what Bonferroni did in the first place.
I’d investigate humidity in the area, if there’s enough data to work with.
HeHe – cloud cover would shed more light
Lance Wallace says:
February 5, 2014 at 11:51 pm
I have no authority for it at all, I derived it myself some years ago as follows, although I’m assuredly not the first person to do so.
Suppose something has one chance in 10 of occurring, call it a 10-sided die. What are the odds that you will throw a “1” if you throw it three times?
Well, the way you calculate that is to use the odds of it NOT happening. There’s a 90% chance of NOT throwing a 1 on any throw. So in two throws, the odds of NOT throwing a “1” are 90% * 90% = 81% … so the odds of getting a “1” in 2 throws are one minus that, or 1 – 81% = 19%.
Similarly, the odds of NOT throwing a 1 in 3 throws is 0.903, and NOT throwing a 1 in N throws it is 0.90N … with the odds of throwing a “1” in each case being one minus the odds of it NOT happening.
So, I’m just applying that to the question of the p-value. I’m looking for some N that will give me a p-value of 0.05. The odds of that NOT happening is 0.95. So I’m looking for a p-value such that
(1 – p)N = 0.95
Taking the log of both sides gives
N log(1 – p) = log(0.95)
Solving for log(1 – p) gives
log(1-p) = log(0.95)/N
Taking the antilog, we get
(1 – p) = 10log(0.95)/N
Which gives us the required value of p to give us p=0.05 in N tries..
Now, I’ve never even heard of Bonferroni or his eponymous correction until Dr. Robert Brown referred to it on one of my posts the other day … but I’ve calculated the odds that way for years.
Who knows … I might be wrong, it’s happened more than once …
w.
Wonder what missing years 2012 / 2013 would do with figure 1. They were both colder than the previous years, and 2013 Nenana Classic latest melt ever should give another indication.
Manfred says:
February 6, 2014 at 12:20 am
Interesting question, Manfred. I wonder myself, as it will be an interesting test of my emulation.
w.
“Figure 1. Percentage of lakes in the low-lying tundra around Barrow, Alaska that are partially thawed in late April, 1992-2011. Photo Source.”
How do they define “partially thawed”
?
I would guess early spring is the one?
Very useful. Not a strident takedown; not even a takedown; just a great job of putting this work in context. It’s too bad the authors had to fire up the spin machine for the media, but I guess that’s the way of the world now. Does every Ph.D program now include a course in Media Management?
I really do not wish to make a comment i just with to be able to make comments on the posts here i have failed to find anywhere to find a place to register
I find this pretty poor
[Reply: You do not need to register. ~ mod.]
Any correlation whatever with summer temperatures?? After all, if the peak lake temperature in summer had gone up over the 20 years, presumably, all else being equal, it will take longer to cool down, less ice will form over the winter and therefore there is less to melt in the spring, ergo it melts earlier all things being equal.
Just asking……….
How do they define “partially thawed”
Easy – radar can spot water, and it’s quite a different signature from ice. So if there is water, then it’s partially thawed.
“less ice will form over the winter and therefore there is less to melt in the spring”
Not relevant in this case, as they are only taking about the totally frozen lakes.
On the previous related thread some of us commented about the missing recent years. I was also wondering why they chose an area near Barrow.
—————
And on the other hand Alaska had a January heatwave. It’s worse than we thought! 😉
http://www.alaskadispatch.com/article/20140129/forget-polar-vortex-how-alaska-dealing-its-heat-wave
Blame it on the Jetstream
http://www.mnn.com/earth-matters/climate-weather/stories/why-is-it-warm-in-alaska-and-snowing-in-atlanta
Excellent work again, Willis.
You manage to do a wonderful job, in my opinion, by looking at the study objectively and giving both positive and negative criticism, all the while keeping it simple enough that even I can understand and follow what you’re doing first time through.
Any chance you’d attempt to publish this as a response to the paper?
Of course it’s a warm Winter in Barrow, move those people to Wyoming to get the cold Winter feel.
If you have ever stood on the bare ice of a frozen lake or river in the early winter as the air temperature is dropping quickly you will likely have experienced a heart-stopping moment as a crack races across the ice almost under your feet. Ice contracts as it cools. The booming and thumping and cracking of the ice is almost magical.
There is a very non-linear relationship between ice thickness and temperature and snow depth, and those cracks are part of the equation. It goes like this:
Suppose that early in the winter there is 5 inches of ice on the lake. The buoyancy of that ice will put the top of the ice about half an inch above where the surface of the water would normally be because about ninety percent of ice is under water.
If there is no snow the ice will freeze relatively quickly because the top surface is exposed to the air. A little bit of snow cover will insulate the ice and prevent it from freezing as quickly. But suppose that it snows a foot on top of that five inches of ice. What will happen? Snow can vary greatly in density but a rough rule of thumb is that a foot of snow equals one inch of ice. So now we have a weight of snow on the five inches of ice that pushes the surface of the ice below the water level. The water comes up through the cracks in the ice and saturates most of the snow by capillary action. Now there is a foot of slush on top of five inches of ice and that thick insulating layer of snow no longer provides much insulation.
The free-flowing water under the ice will circulate up and down in the water column based on its density, which is controlled by its temperature. Ice formation is not as fast on the bottom of a sheet of ice as it is on the top because the least dense water is at +4 degrees C. As the water temperature contacting the bottom of the ice gets down almost to the freezing point those water molecules are trying to head down out of the way so they don’t get caught in the ice.
On the other hand, the water in the snow above the ice has no ability to circulate because the snow restricts it’s movement. As the surface cools it freezes and the thick layer of slush above the ice quickly becomes a very thick layer of ice.
Now … why did I say that a p-value of 0.001 is “barely significant”, when the usual level is a p-value of 0.05? Well … because I looked at so many possibilities before finding what I sought. All up, I looked at maybe 40 possibilities before finding this one.
==============
Willis, you’ve nailed the most overlooked failure in climate science. Faulty statistics caused by researchers “cherry picking” their method. They try different methods until they finally find one that supports what they were looking for, and this is the method they publish.
What is also over looked, is that you might have accidentally chosen method 40 on your first try, and never tried the other 39 methods. From this you wold have assumed that your results were much more significant than they were. Chance tells us that any one correlation may have no meaning.
The 39 methods that didn’t correlate, these are telling you something important. They are telling you that the correlation your found with the 40th method was spurious. If every method you try gives a good correlation, then there is likely a true correlation. However, if 39 methods fail and only 1 works, then the correlation is simply accidental.
This is the problem behind the hockey stick and so many other studies in climate science. The researchers search and search for a statistical method that shows correlation, while ignoring the larger set of methods that say there is no correlation. From this they make the faulty conclusion that the correlation they find is real, when in fact the correlation is simply due to chance.
While I certainly appreciate the approach and effort Willis has taken, this study falls into the category: Again, “evidence” (spurious as it may or may not be) of warming is not evidence of AGW and certainly not evidence of CAGW. Furthermore the language of the press release intending to covey an unnatural (dramatic, surprising, unprecedented) attribute to the warming is completely unfounded based on the geologic record instead of the snippet of time captured within modern observations.
Nice work Willis, I like it. Thanks, by the way, for mentioning your adverse results. I wish everyone would.
Bill Thompson – very informative, lets hope that Surdu et al are equally well informed.
I am not that smart, but what I recall is that the data points need “independence” in some special sense. I don’t see how consecutive time series measurements have that quality. And if they did – couldn’t you just take measurements every second instead of every year and get great correlations? I will try to brush up on my stocastics. In the meantime, I recommend the looking at the 150 years of ice data for some Madison WI lakes at http://www.aos.wisc.edu/~sco/lakes/msnicesum.html.
Lance Wallace says:
February 5, 2014 at 11:51 pm
This is almost identical to the value you found from the exponential, so who knows, maybe that’s what Bonferroni did in the first place.
==============
Willis’ method appears closer to the Šidák correction.
Wikipedia has this to say:
Additionally, the results of the two methods are highly similar for conventional significance levels.
“Prior to starting our analysis, we were expecting to find a decline in ice…”
________________________
“…seek, and Ye shall find…”
There are good reasons why people traditionally put mercury in thermometers, and not ice. The phase-change is indicating heat flux, not temperature (other than being above about zero degrees Celsius).
As mentioned by posters above, humidity and cloud cover will influence melting rate. Snowfall, rainfall, wind-speed and wind direction, I’m sure people can suggest other factors without the need to wheel in the hot models.
The Model results have to be completely discounted. For the satellite data, the first year is Pinatubo, so colder than ‘normal’ with more completely frozen solid lakes. Then nothing happens until 2008. The paper itself depends on just four years. That is insufficient to establish anything.
The PR hype around a nothing finding shows what is wrong with grant seeking Mann wannabes.
There is not much more to be learned.
I’ve missed the part where the study established “… tremendously in response to climate warming in the region.” The study found that or they just said that?
Isn’t the conclusion that the thawing is inconsistent with the past temperature and thawing record and that the “tremendous” part is apparently due to other unidentified factors? A difference in measuring techniques of thawing, an unidentified thawing factor or a combination? It seems like a complete misstatement of findings by the researchers – something they feel entitled to do if the data has been carefully collected.. That does not entitle misstatement of the conclusions.
What I really appreciate about Willis’ technical posts is that he includes the data or spreadsheets for us to examine. In fact one of the reasons I come to WUWT is to improve my techniques for data analysis.
I have a question Willis, about the spreadsheet. In this and in another post you are using a “gaussian.xla” add-in. I’d like to understand how this works. Do you have any references for it? I assume it is performing a Gaussian smoothing with the number of regression points given as an argument, but it appears to have only one data point as an input. Are the coefficients calculated? Or is it a recursive implementation? etc.
Thanks
Thanks Willis. Looks like normal weather cycles. So what else is new. Good to see a study that gets off the main roads and can look at remote sites.
I have seen too many studies on melting permafrost – which was clearly due to mechanical actives by animals and human – mostly human. I have designed things to go over, on, and under the surface in permafrost areas which is sometimes a misnomer since most work is in the active layer and the active layer and its structure can have large effects on everything in permafrost areas including lakes. Sporadic permafrost extends all the way down to 40 degrees of latitude. For 15 years I worked in areas with sporadic and continuous permafrost. We even wrote a paper or two in the official multinational effort Design Manual for Cold Climate Utilities Delivery a long long time ago when we were first starting to recognize how to mitigate installations in permafrost areas (!970’s). I had the opportunity to fly over many lakes in the North West Territories and Northern Saskatchewan and observe their behaviour and beautiful fall freeze up patterns.
The approach used to study these lakes by satellite is interesting, but there are many, many variables that are obvious to someone who has worked in the north and studied multi-season ground temperature profiles and the latency involved over a period of years along with the incredible variation in freeze up in adjacent groups of lakes. It is far from uniform.
I very much liked your mathematical approach. I believe I have seen others in both northern studies and economic studies (could be wrong). A few years ago I used something similar to predict company revenues and margins based on forecasts provided from a number of sources which could be provided a probability of being over, under of close based on previous performance and current economic conditions and projected economic conditions related to our company’s business, It worked quite well as long as you screened the GI to prevent GO.
The Arctic is a beautiful and interesting place to work and live. We have tons to learn in a multidimensional world where things are not always what they seem on the surface:
http://en.wikipedia.org/wiki/File:Vertical_Temperature_Profile_in_Permafrost_(English_Text).jpg
Thanks for your discussion paper. Lots of good information there for those who can apply it.
I know you are more of a southern ocean guy but you might get a kick out of reading some of a design manual for “Cold Climate Utilities Delivery”. This one is from 1979 and I would have to say it is probably a better effort for the time than some of our climate models. but it was worked on by people from northern countries every year for many years and was based on empirical evidence. Somewhere I have a copy of a manual from 10 years later and the improvements were incredible. Someday, Climate studies will get there. Mind you, this manual was a meant to be a practical document based on the best available technology of the time and written by people much smarter than but like me, they were actually building things, and discovering what did and didn’t work well. It may be a fun peruse for a limited few:
http://nepis.epa.gov/Exe/ZyNET.exe/20008A08.txt?ZyActionD=ZyDocument&Client=EPA&Index=1976%20Thru%201980&Docs=&Query=&Time=&EndTime=&SearchMethod=1&TocRestrict=n&Toc=&TocEntry=&QField=&QFieldYear=&QFieldMonth=&QFieldDay=&UseQField=&IntQFieldOp=0&ExtQFieldOp=0&XmlQuery=&File=D%3A%5CZYFILES%5CINDEX%20DATA%5C76THRU80%5CTXT%5C00000001%5C20008A08.txt&User=ANONYMOUS&Password=anonymous&SortMethod=h%7C-&MaximumDocuments=1&FuzzyDegree=0&ImageQuality=r75g8/r75g8/x150y150g16/i425&Display=p%7Cf&DefSeekPage=x&SearchBack=ZyActionL&Back=ZyActionS&BackDesc=Results%20page&MaximumPages=1&ZyEntry=1
The EPA and Environment Canada did some good work 30 years ago.
First of all, congratulations on acknowledging the affect on probability of looking at numerous possible correlations. Lots of very smart people have made a total botch of that.– parapsychologist J.B.Rhine may be the best known example. Is your computation correct? I have no idea but if it’s wrong I doubt it’s very wrong. FWIW, I would probably have gone with p/40 instead — which, given my general record with statistical things, probably indicates you are likely to be right. I suspect that the answer may also depend on whether one assumes that all the possible correlations examined are independent of each other.
I think there may be some serious thinking going on about that in the area of medical research which seems to be far worse than climatology when it comes to institutionalizing dubious statistical notions. A lot of folks seem to be taking John Ioannidis opinions in Why Most Published Research Findings are Wrong seriously http://www.plosmedicine.org/article/info%3Adoi/10.1371/journal.pmed.0020124 If nothing else the guy certainly knows how to pick a title. If I stumble onto anything illuminating on probabilitiy adjustment in the future, I’ll send you an eMail or something.
One other point about the lake study that bothered me is that I’m sure it’s not intentional, but the Alaskan Spring breakup in 2013 was notoriously late. I can’t help but wonder what the paper would have looked like if had chanced to include 2013.
Well it certainly is a coincidence that this study focuses on Barrow when it just so happens that this location has by far the most warming since 1977 than any other location. The amount of warming since 1977 for Barrow has been 4.6F while the bulk has been in autumn at 9.9F.
The Alaska average has been 0.8F of cooling since 1977. And they accuse skeptics of cherry picking. See how Barrow stands out in this table: http://oldclimate.gi.alaska.edu/ClimTrends/Change/7712Change.html
By the way, that page was exceptionally hard to find. It is buried. A few years ago it was on the Alaska Climate Research Center main website but I guess it didn’t look good visually. By comparison, using the 1976-77 climate shift and going back a few more years highlights warming and obscures the overall cooling since 1977. Do a side by side with the link I gave above and this current one: http://oldclimate.gi.alaska.edu/ClimTrends/Change/7712Change.html
Quite a visual difference don’t you think?
Ooops, this is the second link: http://climate.gi.alaska.edu/ClimTrends/Change/TempChange.html
Willis – What were the correlation results like for March and April temps to %lakes partially thawed? Seems to me that would have been the first thing to check, not NDJ temps or snowfall amounts. Agree that n=20 is a pretty low number to base much on.
“steveta_uk says:
February 6, 2014 at 3:01 am
How do they define “partially thawed”
Easy – radar can spot water, and it’s quite a different signature from ice. So if there is water, then it’s partially thawed.
“less ice will form over the winter and therefore there is less to melt in the spring”
Not relevant in this case, as they are only taking about the totally frozen lakes.”
So if there is rainwater on the ice in the spring the radar will consider it as thawed?
Snow insulation was a good catch. Nice analysis Willis. Bill Thomson’s discussion on ice cracks and the non-linear relationship between temperature and ice formation is interesting.
My take is they took a period of rapid warming and blamed it on a correlation with several lakes not freezing to bottom. As the length of observation is increased (as Willis shows) the trend is not significant and well within natural variability. The problem is – you could never publish that in anything other than a blog.
Record ice thickness on my pond this year.
Willis
What about wind?
I’m amazed at the wind patterns around the globe.
http://earth.nullschool.net/#current/wind/isobaric/1000hPa/orthographic=-118.89,64.29,695
cn
AussiejB says:
February 6, 2014 at 12:57 am
I really do not wish to make a comment i just with to be able to make comments on the posts here i have failed to find anywhere to find a place to register
I find this pretty poor
[Reply: You do not need to register. ~ mod.]
—————————————————
That’s true he doesn’t need to register but, he could improve his tone.
cn
From “Don K says: February 6, 2014 at 8:21 am
“One other point about the lake study that bothered me is that I’m sure it’s not intentional, but the Alaskan Spring breakup in 2013 was notoriously late. I can’t help but wonder what the paper would have looked like if had chanced to include 2013.”
With a Pinatubo start date and 2013 end date omission, your confidence in being “…sure it’s not intentional” appears a bit misplaced. Add the “dramatic thinning” hype, and then delete the words “a bit” from the previous sentence!
Clay Marley says:
February 6, 2014 at 8:09 am
Thanks, Clay. To steal a line from Steven Mosher, for me, publishing the data and code with the analysis is what distinguishes science from advertising …
That’s a “user function” that I wrote a long while back to do a gaussian average. It takes as input a single data point and a value for the width of the gaussian filter (actually the (filter width -1)/2. It assumes that the list of data is in a single column, with some text above the data to prevent averaging before the start of the data.
About the only interesting thing about the function is the treatment of the end-points, which uses a method I invented and I discussed in Michael Mann, Smooth Operator.
Hang on … OK, I just posted up the “gaussian” function as a text file, it’s here.
w.
Rud Istvan says:
February 6, 2014 at 7:15 am
We get lots of folks with strange beliefs here. We have pressureheads, we have cyclomaniacs, and we have volcanicians … Rud, take a look at Figure 2. See the “W” shaped drop in temperature around 1990?
Well, the big drop was in 1989-1990, which is the year of the bottom left point of the “W”, at just cooler than -28°C. 1991 and 1992 were about the same, 1991 was a bit warmer than 1990, and 1992 just very slightly warmer than 1990.
Now, the first year of the satellite radar data was 1992, which you describe as “Pinatubo”, despite the fact that Pinatubo erupted in June of 1991.
So your claim about Pinatubo is totally faslified by the data. There is absolutely no sign of any reduction in temperature due to Pinatubo in the winter in Barrow. It’s all your fantasy, unsupported by the facts.
Volcanoes don’t have much global effect on the temperature in general, as I’ve shown many times. You’re beating a dead horse, Rud.
However, you are correct that “the paper itself depends on just four years”.
Regards,
w.
BobN says:
February 6, 2014 at 9:11 am
It’s in the spreadsheet, but here’s the short answer. This is correlation, monthly temperatures to lake thawing as measured in April:
OCT, 0.54
NOV, 0.65
DEC, 0.68
JAN, -0.08
FEB, 0.4
MAR, -0.24
APR, 0.17
And yes, of course I checked that first, as I said in the head post …
w.
ferdberple says:
February 6, 2014 at 6:39 am
Thanks, Ferd. Indeed, the method I derived independently is the Sidak correction … I do love itwhen that happens. People say “oh, pooh, that shows you didn’t invent it, it was invented before”, to which I reply “Yes, isn’t it great! It shows I’m on the right track and understand the subject pretty well …”
w.
There is another metric, less rigorous perhaps than radar reflections. It is the opening and closing dates for the ice roads that truckers use all over the north. Most of those cross lakes, and without sufficient ice, no road. It is the lakes that determine the opening and closing dates. The opening and closing dates, the loads allowed, all of this is indirect evidence as to ice thickness. It should give an interesting comparison to the barrow lakes radar data.
Unfortunately I do not have the time to follow through on this idea, I would hope that someone else does.
lh
@Alan Robertson
>>“Prior to starting our analysis, we were expecting to find a decline in ice…”
>“…seek, and Ye shall find…”
Exactly. But so what? Thinner ice can be the result of warmer water underneath. Maybe it is summer temperature-days that give the hint. There might be a drought in early winter causing thicker ice or ‘wet weather’ delivering a lot more snow than usual. Ice is notoriously unpredictable even watching the very same lake.
I just read that the Black Sea ice froze 45 feet thick one winter long ago. I find that alarming.
I do not find some periodic earlier melting of lake ice alarming at all. It just means more fish production because of better conditions. When it cycles colder production will drop again.
Willis Eschenbach said @ February 6, 2014 at 10:23 am
So… that “W”, is it the signature of warming, or is it the signature of Willis? 😉
Nice work BTW.
Willis, I admire your restraint in this critical review of their paper. I would have been a lot more negative! IMHO your acceptance of their “data” as valid and useful means you also accepted their “assumptions”. My first criticism of the data validity would have centered on the dates of each radar image taken. How can they seriously compare “percent of Barrow lakes patially thawed” (the equivalent of “average temperature anomaly”?) when the radar images from which this data is gleaned are not even taken on the same date each year. The data ranged all the way from April 8th all the way to May 6th! – a full month of variability. The very first assumption was that you can equally compare lake thaw data year to year when that data is a single data point each year spanning out over a month! Sorry, but again IMHO I call BS! The most useful description for this kind of analysis is GI=GO. Going into a study with an admitted bias – “ie. we expected to see” certain results! – can’t help but produce results that are suspect without double blinding the data. Had the pilots taken the data in a manner that pushed the date back sequentially further and further towards April 1st, I suspect they would have gotten a completely different trend, but their “model” would have included an “adjustment” that would have given them the results they expected.
On a side note: What happens to fish or any “life” for that matter in a lake that freezes all the way to the bottom? Also the water at the bottom must be extremely salty, or at the very least have an interesting annual sediment trend of the lakes dissolvable salts and metals for that year.
DonV said @ February 6, 2014 at 1:20 pm
Glycoproteins are “fish antifreeze” in Antarctic notothenioids and Arctic cod. If there are fish in them thar lakes, one imagines they also possess these proteins. It’s a fairly new area of research despite the obvious applicability for the manufacture of food, or food-like substances at least.
http://www.ncbi.nlm.nih.gov/pubmed/12653993
Behind a paywall I’m afraid.
DonV says:
February 6, 2014 at 1:20 pm
Don, I didn’t mention it, but the very first thing I did was see of there was a correlation between the dates and the ice thickness … curiously, it was backwards, with the later dates corresponding with the thicker ice, correlation = – 0.24 … with an addition problem, which was that the p-value for the relationship is 0.20, very weak.
So I didn’t deal with that, since the correlation was so poor. From memory, the original authors did the same, you might check their study …
w.
And when the water freezes all the way to the bottom or not, I first wonder whether the bottom to ice top distance has stayed the same or not.
I used to walk to school barefoot in the winter time (short pants too), and would break the ice in the puddles with my toes. Some puddles were shallow and froze to the bottom, while some were deeper, and only had a thin ice skin. The deeper puddles break easier than the shallow ones, being just water supported.
“””””……Chuck Nolan says:
February 6, 2014 at 9:42 am
AussiejB says:
February 6, 2014 at 12:57 am
I really do not wish to make a comment i just with to be able to make comments on the posts here i have failed to find anywhere to find a place to register
I find this pretty poor
[Reply: You do not need to register. ~ mod.]
—————————————————
That’s true he doesn’t need to register but, he could improve his tone.
cn……”””””
Ruffians and Barbarians; probably heathens too; we have to put up with them as neighbors; they’re all we’ve got !!
As mentioned above, there may be another metric for judging ice thickness on Arctic bodies of water. Here are some sample databases on the opening and closing dates for different Arctic roads and crossings starting in 1983-84. Others are available at the first link. Can’t you hear them begging to be analyzed?
http://www.dot.gov.nt.ca/_live/pages/wpPages/roadConditions.aspx
http://www.dot.gov.nt.ca/_live/pages/wpPages/Open_Close_Dates_Ice_Bridges.aspx
http://www.dot.gov.nt.ca/_live/pages/wpPages/Open_Close_Date_Winter_Roads.aspx
willis….
willis….
analyze me…
Willis Says
Thanks much Willis, I’ll take a look. I remember that article well. Its the one that got me turned onto LOESS curves.
Les H, I’ve looked at 1st and last date of a given year for ice bridge operation on this chart and have been unable to see any pattern or correlation one way or the other. Haven’t looked at the other links.
http://www.dot.gov.nt.ca/_live/pages/wpPages/Open_Close_Dates_Ice_Bridges.aspx
The idea is that the total number of days that the ice bridge is open for traffic gives a measure of ice extent/integrity for any given year ( the crossings are opened and closed based on a range of factors including ice thicknes, density strength etc). Each crossing would be indexed to itself (its’ own average) and then the anomaly would give some indication of the extent of “unusual” ice melt or extension in any given year. Or perhaps graph the raw numbers of days for each year and do the statistical analysis for trends and anomalies. Such analysis is beyond me, but the initial graph of one crossing looks really intriguing.
Just think of the potential title “Ice road Trucks weigh in on Climate change!”
ferdberple says:
This is a classic case of “torturing the data” — keep repeating the question until it gives the answer you want to hear and only report that one. It’s not just a figure of speech.
LesH says:
February 6, 2014 at 8:58 pm
Sadly, I just took a look and there appears to be no common signal … thanx for interesting data, tho …
w.
If N=20 is too low a number, just convert to dog years. Or imagine if our planet was in a lower or higher orbit how much little or more time a year might be.
LesH Here’s some fun with cherry picking.
With that in mind and looking at your 1983/84 to 2010/11 ice bridge chart take a look at Tsiighetchic, the last Mackenzie River crossing before the Mackenzie Delta town of Inuvik.
During the 1997/98 season, at the height of the global warming scare the ice bridge stayed open until May 17th, a date so late in the season it had nor has not been exceeded before or since. The chart doesn’t show spring 2013 numbers when a new record may have been set.
Hi again
Rick and Willis, thanks for taking a look at that info for me. The reason I thought it might be a good metric is that the closing date of those crossings has to do with the spring break-up, which is not a measure of the temperature at the site of the bridge, but a factor of the water flow out of the drainage basin. In this case, the drainage basin is a little bit bigger than the entire state of Alaska.
A metric that would reflect melting on that scale would be useful in establishing general trends.
Again, thank you for you help. I thoroughly enjoy your comments, humor and your articles. Press on.
lh
Les, I had hopes for it as a metric as well. The problem is the ice bridge opening lengths look like this:
Spreadsheet Canada Ice Roads.xlsx, thanks for the kind words.
w.
Good one Willis.
That was my chuckle of the day.
At the risk of flogging a dead horse…
If I remember right the average opening on those crossings ranged about 130-140 days, which means a lot of the points are on the edge of, or over, a 2 sigma deviation from the norm. Two things come to mind.
First, is that this kind of behavior is not strictly random, No bell curves would come out of this. My mind wanders as to the cause of such an accentuated swings in the data. I wonder if there is something in the nature of flushing in the watershed; with cold winters accentuating the ice dam effect upstream and warm winters minimizing it? If so, is there a different statistical process for such data? Of course, if this Arctic data has such accentuated swings, what other Arctic data exhibits the same tendency? but the mind wanders…
The second thing that comes to mind is that while a statistically accurate line or curve may not be able to be draw through the data; it would seem obvious (eyeballing it) that the average of that data indicates only a short decline (perhaps 5 days?) in the overall length of the ‘open road’ season over the 28 years of the data. So while a very weak warming trend may inferred, it is certainly NOT the kind of dramatic change indicated by the original paper. While this data may not positively affirm a specific trend, it does lend weight to doubting the assertion of the original paper.
regards to your tireless crew!
lh
LesH
The study of climate and weather has taught us that torturing signals from disparate data points requires an open mind. I think that you and I should jointly apply for a grant to study this ice road thing.
A grant you say?? They would want us to get published!
To get published we would:
1) need to lack relevant degrees
2) have a predilection for speculation
3) have a dramatic media presence with no sense of humor or irony.
Hmmm.. speaking for myself, 1=2 out of three ain’t bad but given the feeding frenzy, I don’t think we would make it.
You will have to find another dancing partner!
—
ps. sarc intended (just in case you fared better than me on the check list!!)