Guest Post by Willis Eschenbach
This is an extension of my previous post entitled “February Fantasy Versus Reality“. Please read that to get the basic ideas. To recap, a study in Science magazine said
Despite the rapid warming that is the cardinal signature of global climate change, especially in the Arctic, where temperatures are rising much more than elsewhere in the world, the United States and other regions of the Northern Hemisphere have experienced a conspicuous and increasingly frequent number of episodes of extremely cold winter weather over the past four decades.
…
The Arctic is warming at a rate twice the global average and severe winter weather is reported to be increasing across many heavily populated mid-latitude regions, but there is no agreement on whether a physical link exists between the two phenomena.”
To test this claim of increasing “severe winter weather”, in my previous post I looked at the average February temperature of the continental US to see if it was cooling. It hasn’t been cooling.
However, a couple of commenters correctly pointed out that the issue discussed in the study was not average temperature. Instead, the authors were talking about “episodes of extremely cold winter weather” such as those Texas experienced in February of 2011 and 2021.
Looking for a more accurate measure of extremely cold winter weather, I got the daily temperature data for the Southern Great Plains from NOAA. Here’s a map of the area in question.
Figure 1. Map of the National Climate Assessment regions.
Then I calculated the standard deviation (a measure of how widely spread out the temperatures are) of the February temperatures. I reasoned that if there were short sharp cold spells, the standard deviation would be larger.
Figure 2. Standard deviations of February minimum daily temperature for the Southern Great Plains. Cold spells are indicated by an increase in the standard deviation.
In Figure 2, we can clearly see the Texas cold spells of 2011 and 2021. But is there a “conspicuous and increasingly frequent number of episodes of extremely cold winter weather over the past four decades”?
Well … in a word, no. Figure 2 shows there was a serious cold spell in 1951. And the Texas State Climatologist agrees, saying:
Jan.–Feb. 1951: Freeze. On Jan. 31.–Feb. 3 and again on Feb. 13–17, cold waves swept over the entire state, bringing snow and sleet. Heavy damage was done in the Lower Rio Grande Valley to truck and citrus crops, notably in the earlier of these northers. During the norther of Jan. 31–Feb. 3, the temperature went to –19°F in Dalhart.
However, during the thirty years after 1951, there was little in the way of “episodes of extremely cold winter weather” until the decade and a half from 1981 to 1996. During that time there were a number of cold episodes, although not as intense as in February 1951. In the coldest of these, in February of 1985, San Antonio got a rare snowfall, and they saw the coldest day ever recorded in Midland, Texas.
However, in the quarter century since 1996, there have only been the two extremely cold spells mentioned above, in 2011 and 2021.
If we divide the 72 years of the record into three 24-year periods, we have only one “episode of extremely cold winter weather” in the first period; six somewhat warmer episodes in the second period; and only two episodes in the most recent 24 years.
So no, in the Southern Great Plains, there is not a “conspicuous and increasingly frequent number of episodes of extremely cold winter weather. Nor is “severe winter weather … increasing” as they claimed. Neither of those statements is true.
Then I thought, “Well, maybe I’m looking too far south. Maybe the claimed effect is visible in the Northern Great Plains”. So it was back to the drawing board, and here’s what I found.
Figure 3. Standard deviations of February minimum daily temperature for the Northern Great Plains. Cold spells are indicated by an increase in the standard deviation
Although there is greater variation in the February minimum temperatures in the Northern Great Plains NCA region, the same situation prevails as in the Southern Great Plains—one February “episode of extremely cold winter weather” in the first 24 years, a half-dozen or so in the middle 24 years, and the two cold Februarys in 2011 and 2021 in the final 24 years. And there is no trend in the data.
Another beautiful theory runs hard aground on a reef of ugly facts.
Best to all,
w.
Data Access—I’ve put the 72 years (1951-2022) of daily Southern Great Plains temperatures, both maximums and minimums, in my Dropbox for download. It’s a fairly small file entitled Great Plains South nClimDiv.csv, 588 KB, in CSV format so it can be opened in Excel or other programs.
My Usual—I can defend my own words. I choose them very carefully. I cannot defend your (mis)understanding of my words. So please, when you comment, quote the exact words you are discussing.
How to report a null result. The weather records do go back farther than 1950, but are presumably not in searchable format. Besides, the claim was for a more recent change in severe weather.
1) If you have a link to daily regional records going back further than 1951, please post it.
2) To understand what’s going on with the weather you need to use the longest dataset you can find.
3) You are correct that they only used data from 1979 on. However, if you use that short dataset, it looks like there’s been a large decrease in extreme weather. Only with the longer dataset can we see that there’s been no overall change.
w.
The 1975 to 2015 warming that much more affected the Arctic than the tropics, reduced the temperature differential between the Arctic and the tropics in the Northern Hemisphere. The result SHOULD be better weather in the 1975 to 2015 period in the Northern Hemisphere, based on meteorology 101.
https://iopscience.iop.org/article/10.1088/1748-9326/ab4867/meta
Yes, and as expected, cold extremes are decreasing in intensity, duration, frequency, and area.
Data is to climate alarmism as Afghanistan is to empires. The rocks upon which so many are shattered.
Except Afghanistan is poor and these publishing empires of agenda science get monetary rewards from these publications via tenure and promotion and travel.
Of course there are no data for the futre climate, just predictions that are consistently wrong.
Unfortunately, the data for the past temperatures is controlled by the same people who predict much faster global warming in the future.
And their paychecks depend on such predictions.
That’s why they revise historical temperature data from raw numbers, to infilled numbers, to adjusted numbers, to homogenized numbers to pasteurized numbers, to readjusted numbers, to re-re-adjusted numbers
Here are some examples with GIF moving charts from my blog:
:NOAA US average temperature from 1920 to 2020, Raw Data vs. Adjusted Data presented to the public (science fraud) (honestclimatescience.blogspot.com)
Pre-1980 global average temperature “revisions” from 2000 to 2017 (science fraud) (honestclimatescience.blogspot.com)
Watch US climate history get changed to better support the CO2 is evil narrative (honestclimatescience.blogspot.com)
.
Earth shows net cooling for 19 years and The Rules of the Lebensraum game.
SUMMARY
1.A battle for Lebensraum, i.e. land, energy and food resources, broke out when Russia invaded Crimea.An associated covid pandemic, and global poverty and income disparity increases now threaten the UN’s Sustainable Development Goals. During the last major influenza epidemic in 1919 world population was 1.9 billion. It is now 7.8 billion+/ – an approximate four fold increase.
The IPCC and UNFCCC post modern science establishment’s “consensus” is that a modelled future increase in CO2 levels is the main threat to human civilization. This is an egregious error of scientific judgement. A Millennial Solar ” Activity” Peak in 1991 correlates with the Millennial Temperature Peak at 2003/4 with a 12/13 year delay because of the thermal inertia of the oceans. Earth has now entered a general cooling trend which will last for the next 700+/- years.
Because of the areal distribution and variability in the energy density of energy resources and the varying per capita use of energy in different countries, international power relationships have been transformed. The global free trade system and global supply chains have been disrupted.
Additionally, the worlds richest and most easily accessible key mineral deposits were mined first and the lower quality resources which remain in the 21st century are distributed without regard to national boundaries and demand. As population grows,inflation inevitably skyrockets. War between states and violent conflicts between tribes and religious groups within states are multiplying.
2 The Millennial Temperature Cycle Peak.
Latest Data (1) https://www.nsstc.uah.edu/data/msu/v6.0/tlt/uahncdc_lt_6.0.txt
Global Temp Data 2003/12 Anomaly +0.26 : 2023/01 Anomaly -0.04 Net cooling for 19 years
NH Temp Data 2004/01 Anomaly +0.37 : 2023/01 Anomaly +0.05 Net cooling for 19 years
SH Temp Data 2003/11 Anomaly +0.21: 2023/01 Anomaly -0.14 Net cooling for 19 years
Tropics Temp Data 2004/01 Anomaly +0.22 : 2023/01 Anomaly – 0.38 Net cooling for 19 years.
USA 48 Temp Data 2004/03 Anomaly +1.32 : 2023/01 Anomaly + 0.12 Net cooling for 19 years.
Arctic Temp Data 2003/10 Anomaly +0.93 : 2023/01 Anomaly – 0.72 Net cooling for 19 years
Australia Temp Data 2004/02 Anomaly +0.80 : 2023/01 Anomaly – 0.50 Net cooling for 19 years
Earth’s climate is the result of resonances and beats between the phases of natural cyclic processes of varying wavelengths and amplitudes. At all scales, including the scale of the solar planetary system, sub-sets of oscillating systems develop synchronous behaviors which then produce changing patterns of periodicities in time and space in the emergent temperature data. The periodicities pertinent to current estimates of future global temperature change fall into two main categories:
a) The orbital long wave Milankovitch eccentricity, obliquity and precession cycles. These control the glacial and interglacial periodicities and the amplitudes of the corresponding global temperature cycles.
b) Solar activity cycles with multi-millennial, millennial, centennial and decadal time scales.
The most prominent solar activity and temperature cycles are : Schwab-11+/-years ; Hale-22 +/-years ; 3 x the Jupiter/Saturn lap cycle 60 years +/- :; Gleissberg 88+/- ; de Vries – 210 years+/-; Millennial- 960-1020 +/-. (2)
The Oulu Galactic Ray Count is used in this paper as the “solar activity ” proxy which integrates changes in Solar Magnetic field strength, Total Solar Insolation , Extreme Ultra Violet radiation, Interplanetary Magnetic Field strength, Solar Wind density and velocity, Coronal Mass Ejections, proton events, ozone levels and the geomagnetic Bz sign. Changes in the GCR neutron count proxy source causes concomitant modulations in cloud cover and thus albedo. (Iris effect)
Eschenbach 2010 (3) introduced “The Thunderstorm Thermostat Hypothesis – how Clouds and Thunderstorms Control the Earth’s Temperature”.
Eschenbach 2020(4) in https://whatsupwiththat.com/2020/01/07/drying-the-sky uses empirical data from the inter- tropical buoy system to provide a description of this system of self-organized criticality. Energy flow from the sun into and then out of the ocean- water interface in the Intertropical Convergence Zone results in a convective water vapor buoyancy effect and a large increase in OLR This begins when ocean temperatures surpass the locally critical sea surface temperature to produce Rayleigh – Bernard convective heat transfer.
Short term deviations from the solar activity and temperature cycles are driven by ENSO events and volcanic activity.
Fig 1 Correlation of the last 5 Oulu neutron cycles and trends with the Hadsst3 temperature trends and the 300 mb Specific Humidity. ( 5,6 )
The Oulu Cosmic Ray count in Fig.1C shows the decrease in solar activity since the 1991/92 Millennial Solar Activity Turning Point and peak There is a significant secular drop to a lower solar activity base level post 2007+/- and a new solar activity minimum late in 2009. In Figure 1 short term temperature spikes are colored orange and are closely correlated to El Ninos. The hadsst3gl temperature anomaly at 2037 is forecast to be + 0.05.
Soooo many assumptions must be accepted to allow for the conclusion. ie:
“Earth has now entered a general cooling trend which will last for the next 700+/- years.”
See Figs 2 and 3 at http://climatesense-norpag.blogspot.com/
No one has ever made accurate long term climate predictions.
Mr. Page is very unlikely to be the first.
The abstract to the Science paper on which the Forbes article was based said they had used ‘observational analysis and modeling’ to reach their colder February results. Modeling, maybe. Observational analysis, nope. WE just provided that part.
Such study misrepresentation comprises academic misconduct. There are many other ‘climate science’ misconduct examples, including in Science. Marcott’s 2013 Science hockey stick paper is exhibit 1. I alerted then Science senior editor Marsha McNutt to it by providing written proof using Marcott’s own stuff. There was even a clear smoking gun. Her assistant acknowledged receipt, but no followup and no retraction. Details were then published as essay ‘A high stick foul’ in ebook Blowing Smoke.
Based on his famous (but dishonest) Science paper, Marcott got a tenure track assistant professorship at U Wisconsin Madison main campus.
Sicko Award to UWM
I don’t blame UWM. They thought they were getting a rising young academic star, famously published in Science. I don’t blame Science; peer review doesn’t screen for deliberate academic misconduct. I blame McNutt. She did nothing AFTER she was provided (receipt acknowledged) incontrovertible written proof from Marcott himself of academic misconduct (thesis fig4.3C versus Science fig1g is a smoking gun). She knew, yet did nothing.
“Such study misrepresentation comprises academic misconduct.”
No, you have misrepresented their paper. They said in that abstract:
” We use observational analysis to show that a lesser-known stratospheric polar vortex (SPV) disruption that involves wave reflection and stretching of the SPV is linked with extreme cold across parts of Asia and North America, including the recent February 2021 Texas cold wave, and has been increasing over the satellite era.”
They did show that SPV disruption has been increasing, and they did show that it is linked to extreme cold events. They didn’t claim to have observed colder Februaries; in fact February only came up because they mentioned the 2021 Texas event.
Nick Stokes February 1, 2023 4:23 pm
No, Nick, they did not “show that SPV disruption has been increasing”. Their simplified MODEL, complete with a Tinkertoy imaginary “slab” ocean, fed with the output of another MODEL, did not “show” but merely claimed, that SPV disruption has been increasing.
And they did claim an increase in “extreme cold events” like the Texas February cold spell … but there was no such increase in the Texas area that they identified as being affected.
Next, if the SPV is linked with extreme cold weather events, we should see more “extreme cold” events. I’ve shown that’s not true for February. Here are the other two months of meteorological winter for the South Great Plains. They show the same—no increase in “extreme cold” events. I’ve shown the trend since 1979, since that is the period of their study. It’s very slightly decreasing in all three months. Not increasing. Decreasing.
Finally, for those who don’t know Nick Stokes, for a couple of decades now he’s been known in climate circles as “Racehorse” Nick Stokes. This is in honor of a quote from Racehorse Haynes, a famous Texas lawyer. Here’s the quote, from Racehorse Haynes’ own mouth:
I’m sure you can see the resemblance. Trigger warning—do not expect Racehorse Nick to ever agree that he might have even slightly overstated his case. Never seen it happen.
w.
“No, Nick, they did not “show that SPV disruption has been increasing”. Their simplified MODEL, complete with a Tinkertoy imaginary “slab” ocean, fed with the output of another MODEL, did not “show” but merely claimed, that SPV disruption has been increasing.”
It is not the outcome of a model. They observed pressures at various levels, identified the patterns associated with the stratospheric polar vortex (SPV), and counted the days on which P4, the stretching pattern appeared.
This is laid out in a tableau in Fig 1 below. The top row are the stratospheric pressures P100, then the plot of frequencies, then pressures at P500 and sea level, and then surface temperatures. The evidence for increase of disruption is the P4 plot with the trend in red.
“And they did claim an increase in “extreme cold events” like the Texas February cold spell … but there was no such increase in the Texas area that they identified as being affected.”
No, they didn’t. They claimed that SPV events had been increasing, and gave the daily counts. And they demonstrated in some detail that the Feb 2021 event was associated with SPV stretching. They did that by analysing that winter, not by claiming a frequency increase.
Nick Stokes February 1, 2023 9:01 pm
Willis Eschenbach
Nick Stokes
First, those are not observations. They didn’t observe pressures at various levels. It is the output of a reanalysis model:
And what is MERRA-2? It’s the output of a model (emphasis mine).
“Observations”, my okole …
Second, their further results are the result of feeding the output of the model described above to a second model. From the paper:
Moving on, from the next comment:
Willis:
Nick:
Yes, they DID claim an increase in extreme cold events. From the paper:
And besides, you don’t own a dog …
w.
Willis,
w: “First, those are not observations. They didn’t observe pressures at various levels. It is the output of a reanalysis model”
Well, you can call any calculation a model. But reanalysis in no way corresponds to your description
“Their simplified MODEL, complete with a Tinkertoy imaginary “slab” ocean, fed with the output of another MODEL, did not “show” but merely claimed, that SPV disruption has been increasing”
No slab, and it isn’t fed but output of another model. Instead, as in your quote above, it is fed by observations:
“More recent updates to the model are presented in Molod et al. (2011). The GEOS-5 system actively assimilates roughly 2 × 10⁶ observations for each analysis, including about 7.5 × 10⁵ AIRS radiance data”
It is a way of transferring the observations onto a grid. Pressure observations mainly come from radiosonde.
w: “Second, their further results are the result of feeding the output of the model”
The “further results” are an analysis of the causal mechanism relating SPV and cold events.
w: “Yes, they DID claim an increase in extreme cold events. From the paper:…”
Your quote is not from the paper or its authors. It is from commentary from a Science staffer, H Jesse Smith.
As someone who enjoys having you here on the site to offer a counterpoint, Nick, I think you should quit digging the hole you’re in. It’s pretty clear what they were claiming, and Willis has shown they were wrong.
“It’s pretty clear what they were claiming”
How is it clear if no-one will actually quote what the paper said?
All Willis seems to have said is that they mentioned February 19 times.
“Your quote is not from the paper or its authors. It is from commentary from a Science staffer, H Jesse Smith.”
This is from the summary at the top of the Science page, and that sentence is the widely reported summary of the paper.
Nick appears to believe that this is an error. If so, the authors will contact Science and demand a correction.
Nick is almost certainly wrong, as usual – and there will be no correction.
“Nick appears to believe that this is an error.”
No, I don’t believe it is an error. It is something said by HJS, may well be true. But the paper did not say it.
Nick Stokes:
Oh, please. They did claim an increase in extreme cold events, even though you don’t have a dog.
The paper says, inter alia:
and
and
and
So yes, Racehorse, the study authors absolutely did claim an increase in extreme winter cold events.
w.
Willis,
The basic problem here is that you are not following your own excellent advice:
“quote the exact words you are discussing.”
If you would quote upfront what the paper actually says that you disagree with, and then produce your refutation, then we could see whether the two things match. Instead you have produced an analysis of statistics of cold in the Southern Great Plains which really doesn’t relate to whatever the paper was saying about the SPV pattern and its occurrence.
Nick, I’ve quoted their claims of increasing extreme cold spells over and over ad nauseam.
And I’ve produced statistics of cold in the Southern Great Plains that directly refute their claims of increasing cold spells.
Do try to keep up with the topic of the post, there’s a good fellow.
Next, I’ve said nothing about the SPV pattern, in part because it’s not based on observations but instead is the result of a computer model.
That is, unless you think that we have actual daily observations of 2-m surface temperatures, surface pressures, and 100 hPa, 500 hPa, and 850 hPa atmospheric temperatures, geopotential heights, and meridional heat transport over the entire Arctic region on a 0.3125-degree longitude by 0.25-degree latitude gridcell basis for the last four decades … because that’s what they say they’ve used.
Like I said. Computer model output.
w.
Willie E. gets a 10-yard penalty for getting Nick the Stroker all wound up…. This is equivalent to walking past the monkey cage at a zoo and banging your steel drinking cup on the bars of their cage. Moderator Bait. Wake up Charles!
Yeah, Richard, you’re right. I just didn’t want fools to believe him. Not sure that’s possible, though. As the Doobie Brothers said
w.
“A man hears what he wants to hear.
And disregards the rest.”
Paul Simon, from his excellent song, “The Boxer”, one of the best songs in my collection of about 25,000 songs, although the ending stretches out too long:
Simon & Garfunkel – The Boxer (2021 Remaster) – YouTube
Add to that, “greening reversed”.
https://mobile.twitter.com/aaronshem/status/1597321456152547328
Very concerning that this was published. Classic case of merging incompatible data sets among other problems. https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2022EF002788
@curryja
@mattwridley
https://www.nature.com/articles/ncomms13428
https://www.nature.com/articles/s43017-019-0001-x
https://nph.onlinelibrary.wiley.com/doi/10.1111/nph.16866
https://www.science.org/doi/10.1126/science.abg5673
https://www.science.org/doi/10.1126/science.abb7772
https://www.science.org/doi/10.1126/science.abg2947
https://essd.copernicus.org/articles/14/4811/2022/
conspicuous and increasingly … over the past four decades.
Do the authors believe Carl Sagan’s millions of years or don’t they? Every climate article ever written should have some short disclaimer about how relatively short a time how relatively small a community has been studying how chaotic a phenomenon.
” And there is no trend in the data.
Another beautiful theory runs hard aground on a reef of ugly facts.”
This discussion seems to have gone off the rails. The original paper is here. It doesn’t make any special claims about February, anywhere. The Forbes mention of that seems to be based on a single reference to the Texas freeze of 2021. In fact the paper doesn’t have much data on February at all. Fig 1 and 2 have Oct-Dec patterns. Fig 3 caption starts
“Lower-stratospheric and tropospheric trends in late fall and early winter project onto SPV stretching precursors.”. For Fig 3 and 4 the data is for October snow cover and Oct-Dec sea ice.
Nick, the paper mentions February no less than NINETEEN TIMES.
w.
Its easy to check, so I did. Downloaded the paper from Nick’s link, ran the search tool in the PDF reader, and it does indeed show 19 occurrences.
Glanced through to see what they are about. Mostly they are about a range of observations for a period of some months including February, but there are quite a few with specific observations on the particular month of February.
I don’t know if this counts as ‘special claims about February’. But what’s clear is that whether or not there are remarks which count as ‘special claims’ is immaterial.
Willis’ point is not that the paper makes claims about February that can be called ‘special’. Its that February is a significant data point in the paper which is cited and discussed and that an examination of past February temperatures are therefore relevant to the paper’s argument.
He is right. No question about it. Anyone can verify this in five minutes by doing what I just did. I don’t understand why anyone would think they could dispute this.
“there are quite a few with specific observations on the particular month of February”
In fact, those are when they analyse the specific event of Feb 2021. That accounts for 7 mentions. The rest is mostly mentioning a range like Oct-Feb. But the key thing is that they offer no specific February data, as I said. Only October snow or Oct-Dec Kara sea ice. Yet Willis concludes:
“Although there is greater variation in the February minimum temperatures in the Northern Great Plains NCA region, the same situation prevails as in the Southern Great Plains—one February “episode of extremely cold winter weather” in the first 24 years, a half-dozen or so in the middle 24 years, and the two cold Februarys in 2011 and 2021 in the final 24 years. And there is no trend in the data.
Another beautiful theory runs hard aground on a reef of ugly facts.”
All the posts in the series have been about supposed cooling in February. There is no basis for that exclusive focus in the paper.
Do you think that extreme cold winter episodes have become more common in Texas in recent years?
If so, do you think this is due to global warming, and if so, why?
Why is it that when the data is so clear once exemplary publications refuse to look at the data? Could it be that there are none so blind as they that don’t want to see? And if so, why don’t they want to see?
Sadly, Science Magazine and its parent organization, AAAS, is just empty shell of its former self in terms of ethical behavior, adherence to the principles of science when it comes to any controversial topics like climate or racial diversity in the STEM fields. They have become, similar tyo what Cliff Mass noted about UW academia on his blog post, “censorship in the furtherance of free speech” Orwellian fascists. On any of these topics they push a desired narrative without regard to facts or data.
Willis,
In a releted vein, here is some UAH monthly data updated today to show a non-warming trend over Australia for the p[ast 10 years and 9 months.
Maybe, some day Aussie school children will graduate without ever having felt global warming at home. (Not that I say this other than to mock snow forecasts by former experts).
Geoff S
http://www.geoffstuff.com/uahfeb2023.jpg
This is the first article I read Thursday, and the first one on my list of what will grow to 12 to 24 recommended climate science and energy articles today: Honest Climate Science and Energy
As usual for Willie E., it is well written with easy to read charts.
But it seems to me that the February has increasingly frequent number of episodes of extremely cold winter weather over the past four decades is one of the weakest Climate Howler claims in the past year. Everyone living in northern states knows the winters are warmer than in the 1970s with fewer periods of extremely cold weather. That’s why we love global warming here in Michigan.
Willie E. debunked that bogus data mining in his first article on the subject, and then slam dunked the claim with this article. Two great articles.
Willie E. is one of our best weapons in the battel to refute Climate Howlers. He’s a good scientist and good writer too — two qualities that are rare in one person.
I’m confident William Happer could write a good article refuting claims made by Al “the climate blimp” Gore. But would that be a valuable use of his time?
Willie E. is free to write on whatever climate science subject interests him.
I can also hope his valuable time will be focused on the best Climate Howler propaganda. This February claim was inaccurate data mining — pretty bad propaganda even compared with the usual Climate Howler BS. An easy target.
I would just ask Willie E. politely to think about refuting bigger targets than this one. I only hope for one of our best scientists shooting at bigger targets than this one. … And this post is intended as a complement, and a suggestion, not an insult.
This is statistical data. The alarmists don’t respond to statistics. It actually angers them. Remember AOC’s pronouncement that we make a mistake when we pay more attention to facts than the “truth” (paraphrase, with her point).
Not saying facts/statistics are to be ignored. But somehow we need to respond to her “truth” while using statistics rather than using statistics to refute her “truth”.
It’s like a recent argument I had with a relative about having solar panels installed on his roof.
The US government provided cost-benefit analysis of doing so with the slope of his roof. I noticed that, at Montana’s latitude, what counted was not the slope of his roof but the most efficient angle, which was 30* for a SW facing roof – because that was the angle that shed snow the best. Plus the cost didn’t include installation. Also asked how often he had to wash the panels because of dust grime, reducing their effectiveness. (I saw in the UAE that dust covered the solar panels there. Really dumb idea in places withoutwater or cleaners.)
The US government assessment was that he’d get payout in about 10 years, maybe 8. But factoring in the needed angle and dust problem, maybe 12 was payout. Lifespan was claimed to be 25 years but maybe 17 was more likely these days. And then the cost of removal and management if the toxic debris should be added into the cost.
They also claimed it would add 7% to his home value. No, it wouldn’t. Like repainting you’ll house and having bread baking in your oven during views doesn’t. Just speeds up the sale.
My point was not to not install solar panels, but to look clearly at what was the living “truth” (you don’t save money or reduce CO2 much) , not the sociopolitical AOC “truth” (the planet is under an existential threat for fosdil fuel use).
The saving money but was a persuasion technique called “thinking beyond the sale”. You forget to analyzing what you are doing today because you are focused on a sparkly possibility of tomorrow, including what your neighbors and friends might think of you.
So, how do we counter the “misinformation” of extreme cold weather events? I don’t know exactly at this moment. But stats are only the backup to persuasion efforts, not the cause.
Perhaps the focus could be: 2023! Back to the grand old ’80s! What a great climate we have! And then the stats.
Focus on the positive conclusion rather than stamping on the dumb claim with statistics.
Perhaps you are using the “wrong” baseline Willis. Instead of using the data best fit trend line, if you used the modelled regional warming trend line, then you would see an outbreak of conspicuous cold episodes on the right of the graph /sarc.
Great photo at the top. Chicago, Feb 2011 Groundhog Day 🤣🧑🎄
I tend to stay home on days like that.
(… and watch the flames in the wood stove.)
I reasoned that if there were short sharp cold spells, the standard deviation would be larger.
nope bad reasoning. people make the same mistake looking for hotspells.
and heat waves.
the standard deviation will capture frequent rapid deviations from the mean. it wont capture
more severe winters. it wasnt designed to.
but you know when you have a hammer every problem looks like a screw
Mosh, if it’s not a valid measure, please explain to us how it accurately identifies the 1951, 1981, 1985, 1989, 1996, 2011, and 2021 Texas February cold snaps.
Juju? White man’s magic? Blind luck?
You’re arguing against ACCURATE, VERIFIED RESULTS.
Foolish. Do your research before uncapping your electronic pen.
w.
I live in SE PA and work on a large virtual team. Most on the team were very alarmed by the warm winter so far in the mid-Atlantic region. They were surprised when I explained the typical La Nina weather pattern for the mid-Atlantic.. It explains the warm pattern to a T. To my surprise, no one challenged me with global warming and climate change mantras. Many were 30-50 years old with very poor weather memory and pointed to 1993 ( the perfect storm) and 1996 ( 33 inches in Philly, Post Minimum El Nino), and 2010 (Philly breaks annual snowfall record, Post Minimum El Nino) as normative for the typical PA winter They completely forgot the warm winters of 1989, 1997, 1999 and 2020. The coldest winter was 2013 with 11, 2 inch clippers, no one remembered that winter where it was sub-zero for 10 days with snow cover 1/16-4/5/2013. 2013 was the last of Bastardi’s post minimum, ENSO neutral predicted cold winters.
I am glad we have excellent statisticians contributing to WUWT… its a science blog as it should be. My point is with my anecdote is that the media shapes people’s reality and memory of their experience. When people gettogether they just echo what they have be taught to believe and remember. This is the reason WUWT must remain a science blog not an extension of the world of political twits. People can form their own understand with good info……. this was my experience 15 years ago with WUWT. If it becomes too shrill people won’t trust the info. Most people are in the middle and they are tired of the crazy. Good info is good info. Good science can be trusted.
“I am glad we have excellent statisticians contributing to WUWT” I hope you do not mean Willis and that WUWT promulgator of amateur, home-grown, non-peer reviewed statistical theory (maths free of course) “Professor” Kip! Neither have any formal qualifications in statistical methods. Being able to calculate means and standard deviations and do some basic graphics in Excel does not a professional statistician make! I havent received a response on some good-faith questions about the analysis Willis posted for this WUWT post in my comment (February 1, 2023 4:48 pm). Like what happened in his analysis at the spatial level and so how did he calculate standard deviations when the raw data consists of measurements across weather stations as well as Julian days? How did he account for non-random missing values, and non-random station drop-in and drop-out? No details; no pre-print paper with a methods section, no access to computer code used for the data processing/analyses in say Supplementary Materials (in this case Excel functions – no serious professional statistician relies only on Excel for their analyses) to verify his methods, and no metadata to give detailed descriptions of the variables in the Excel datafile he provided via DropBox! WUWT posts like this have to be viewed as unverified analyses and that should limit their credibility until that verification can be carried out which requires all of the above i.e. write a legitimate scientific manuscript even if it only gets to preprint status on ResearchGate and then summarise it on WUWT. Then we can do a proper Open Source review in Comments. Over to you…Willis and co.
Of course ignoring good-faith and technically proficient critiques is something authors can get away with on sites like WUWT when, despite its many failings, at least peer review requires these critiques to be seriously addressed and the technical merit of both the critique and the response by the author is also up for scrutiny as they should be – so power to the Open Source journals that also publish the review process.
That should read “Open Access” not “open Source” journals.
steve_showmethedata February 2, 2023, 6:40 pm
Thanks, Steve. You wouldn’t recognize a good-faith question if it bit you on the distal end of your alimentary canal.
I didn’t answer your inchoate whining for several reasons.
First, I assumed that professional statisticians were involved in creating the NOAA nClimDiv regional data. The reference I read about nClimDiv methods bore that out. And I assumed that you were familiar enough with Google to go and see if they did a good job or not. Given that you may not be that familiar, here’s the doc I consulted.
Next, I didn’t answer because I calculated standard deviations in the normal way. I took all the February days for a given year and took their standard deviation. Not sure what anyone would do differently.
Next, I didn’t answer because it was evident that the method was working—it correctly identified the February cold snaps in 1951, 1981, 1985, 1989, 1996, 2011, and 2021.
Next, your slavish worship of formal qualifications, in statistics or anywhere else, is foolish. Some of the biggest garbage I’ve seen in peer-reviewed climate science has come from folks with PhDs and qualifications up the wazoo. Instead, I judge by results. Me, I have no scientific qualifications at all … but the folks at Nature magazine didn’t whine about that when they published my peer-reviewed “Brief Communications Arising”. Like any true scientist, and unlike you, they judged me on my work and not my “qualifications”. Further details on my lack of any qualifications are in my post “It’s Not About Me“. It’s an interesting read.
Next, I do almost no work in Excel. Instead, I use the computer language R. I wrote my first computer program a half-century ago, and I’ve written programs in Algol, Datacom, C/C++, Mathematica (3 languages), Basic, Hypertalk, VectorScript, Pascal, Visual Basic for Applications, and R. How about you?
Finally, I didn’t answer because you are a professional gadfly and critic for whom nothing is ever good enough. Heck, you didn’t even have to go to the trouble to pull the data out of 72 different online files as I did—I provided you with the data … and despite your claims that I’m such an amateur, you didn’t even bother to do your own analysis of the data to see if your results were different from mine.
You did none of that. You were just looking for something to bitch about.
And you’ve done it so many times that I generally just skim your complaints to see if there’s anything worthwhile in them … and in this case, there sure wasn’t.
Best regards,
w.
PS—Yes, I have no formal training in statistics. Never took a single class in it.
However, I understand the effect of the Hurst exponent on statistical significance, and I independently invented a way to calculate that effect, a way that was first invented by a very good statistician, Demetris Koutsoyiannis. I can show you why the Koutsoyiannis method is better than the Nychka method for that purpose.
I independently invented a method of Fourier analysis that doesn’t mind gaps in the data. I found out later it’s actually called a “Date-Compensated Discrete Fourier Transform”, or DCDFT (Ferraz-Mello, S. 1981, Astron. J., 86, 619). I love independently coming up with things like those—it proves that I understand the subject well enough to make valid discoveries, regardless of whether someone found them first.
I also understand when to use standard linear regression and when to use Deming regression; how to measure skew and kurtosis; the generation of appropriate pseudorandom data for Monte Carlo analyses; how to standardize variables; the difference between white, blue, and red noise; the use of fractional Gaussian noise; when to use Student’s T-test and why it’s not the Student’s T-test; p-values and their limitations; single and multiple linear regression; cluster analysis; principle component analysis; Godel’s Incompleteness Theorem; the problems with heteroskedasticity and non-stationarity; the effects of overfitting; and a host of other statistical subjects.
In short, I’m self-educated in these matters, not uneducated.
The part of my post about you not having formal qualifications in statistical methods was a response to JC’s post (re:” I am glad we have excellent statisticians contributing to WUWT“) and not your work shown in the post. I did not let myself degenerate into personal attacks as you have done but simply reported the fact that you now confirm that you have no formal qualifications in statistical methods. So, no, both my posts directed at your article judged your article on its lack of detail on the methods you used contrary to your derogatory comments (“Like any true scientist, and unlike you, they judged me on my work and not my “qualifications”). You posted a link to a dataset with no accompanying metadata which is your responsibility to provide if you want to claim scientific credibility otherwise post links to the original source data (with metadata) and in your missing methods section describe how you processed this data further, if you did, for your analyses as required in peer-review journals (including any Supplementary Material).
“Next, I didn’t answer because I calculated standard deviations in the normal way. I took all the February days for a given year and took their standard deviation.” If your source data only has regional daily means (i.e. means across stations) then you should dig deeper since those means are not fit for purpose (to give valid unbiased means and standard deviations) if they involve, as is certainly the case, non-random within-station missing values and non-random drop-in and drop-out stations. This point you failed to acknowledge and it is important since different sets of stations contributing to different time points can inflate the variance and thus your sample standard deviation as an artefact of the sampling imbalance (technically, station random effects do not cancel out in unbalanced datasets like they do in balanced spatio-temporal datasets so marginally they contribute to the temporal variances and that process needs to be modelled).
“Not sure what anyone would do differently”. Exactly, which shows your lack of knowledge and understanding of modern statistical methods for unbalanced spatio-temporal datasets. I gave you a link to a statistical methods paper of mine dealing with data similar in nature i.e. unbalanced spatio-temporal dataset, which has now some 280 downloads and over 750 views. Bottom line is unless you use methods such as that that take into account the lack of balance in the spatio-temporal dataset there is no way of knowing if your overly-simplistic analysis method has delivered misleading results.
I did not do any analyses myself because I am taking the part of an expert reviewer (expert in statistical methods) as I have done many many times as a referee for scientific journals over my career.
I do stand corrected that you do not restrict yourself to Excel, I was misinformed on that, and I humbly withdraw that comment. I have used FORTRAN, Apl, GLIM, GenStat, S-plus, and R (all my recent work is in R) as my programming languages over the last 46 years of my career as a professional statistician.
The data analysis you did is only one aspect. You failed to mathematically demonstrate that annual sample standard deviations in minimum temperature (means across stations) and the long-term trend fitted to them was a valid method for drawing the inference you aimed to draw about ‘increasing “severe winter weather”’ (i.e. increase in frequency of extreme cold weather events). If you failed to provide that mathematically explicit demonstration I would reject your paper until you did as an expert reviewer!
Here is an example of how you explicitly link the inference of interest to a specific parameter in a deterministic plus stochastic mathematical model of the data generation process (remembering that “all models are wrong, but some models are useful”) . Determine the null distribution of the parameter of interest, estimate, test, provide confidence intervals, estimate the power to detect the alternative hypothesis given a range of nominal parameter values. This all needs to be mathematically explicit (see Appendix of this paper of mine DOI: 10.9734/CJAST/2022/v41i333946).
I have published in mainstream statistical journals (see below) but these two papers didnt fit the political correct narrative so I had to go to ARRB and CJAST Open Access journals
DOI: 10.1007/s10651-014-0306-3
DOI: 10.1198/108571102302
DOI: 10.1023/A:1009662915320
DOI: 10.2307/2533104
OK I will walk back the comments related to my statement “if they involve, as is certainly the case, non-random within-station missing values and non-random drop-in and drop-out stations.” NOAA has used other valid methods to deal with this using the raw station-level unbalanced data. The above comment about mathematically explicit statistical inference still applies though. Even though I didnt do my research on the data well enough for this type of Blogging (I should stick to publishing papers in the scientific literature, and that’s now locked in), at least I didnt resort to inane personal insults.
Thanks, Steve.
Here’s my issue. I publish something based on NOAA nClimDiv data using a type of analysis which is obviously a valid method for identifying cold snaps in Texas.
How do I know it’s valid?
Because it correctly and very visibly identifies the Texas cold snaps of 1951, 1981, 1985, 1989, 1996, 2011, and 2021.
I provide a link to the NOAA data itself, as well as doing all the work of downloading all 72 * 12 = 864 monthly datasets and posting the results up in a usable form so folks like you wouldn’t have to go through all that trouble.
And what do you do in response?
You accuse me of lacking qualifications in statistics and not having a “knowledge and understanding of modern statistical methods”; you claim I’m using some unspecified improper method of taking standard deviations; you tell us that you are a statistical wunderkind; and you accuse NOAA of using improper methods in aggregating the nClimDiv data … charming …
… and what do you NOT do in response?
You don’t provide even one scrap of evidence that your foolish accusations are correct. You don’t make the slightest effort to point out where NOAA went off the rails in producing the nClimDiv aggregate data for the Southern Great Plains. You don’t bother to detail where I went totally wrong in taking the standard deviations of the daily temperature averages … you know, the standard deviations that correctly identify the 1951, 1981, 1985, 1989, 1996, 2011, and 2021 Texas cold snaps.
In other words, you don’t do one bit of the work necessary to back up your handwaving claims that my obviously valid analysis is somehow secretly fatally flawed.
I grew up on a cattle ranch, and since this post is about Texas, I’ll resurrect a saying from my youth about folks like you.
“He’s not a cowboy—he’s all hat and no cattle”.
w.
Willis, you are even mis-representing my above mea culpa – you say that I “accuse NOAA of using improper methods in aggregating the nClimDiv data” whereas what I posted was “NOAA has used other valid methods to deal with this using the raw station-level unbalanced data”. My post is not very long so you couldnt have missed that so how do you construe that statement of mine as accusing NOAA of using improper methods?? Talk about ungracious.
You say “How do I know it’s valid?
Because it correctly and very visibly identifies the Texas cold snaps of 1951, 1981, 1985, 1989, 1996, 2011, and 2021.”
What is your mathematical definition of a “cold snap” (=extreme cold weather event, I assume)?
If you had any training (self or institutional) in statistical inference you would know that one cannot validate or justify a general method of statistical inference (i.e. standard deviations can be used to infer the frequency of “cold snaps”) using post-hoc comparisons with one or more realisations (i.e. your data) of a sampling process. You need to describe the inferential theory in mathematical terms employing a mathematical description of the sampling process, the statistical model of the process involving parameters of interest which are used to carry out the required probabilistic inferences. Its not called mathematical statistics without reason, and that was my major in my batchelor of science at the University of Sydney (1976) and my Masters and PhD were replete with mathematical statistics theories some of which were novel. Try publishing your method in a statistics or climate science journal and “it’s valid ; Because it correctly and very visibly identifies the Texas cold snaps of 1951, 1981, 1985, 1989, 1996, 2011, and 2021″ will not cut it. Its telling that you did not have a reference in the peer-review literature for your method, preferably in a statistics journal, which if you had I am sure you would have quoted since you like to promote your ability to re-discover statistical theories – so no published method papers in statistics on your CV then because you were always beaten to it??
BTW the NOAA approach is to balance the unbalanced spatio-temporal data using gridding and methods of interpolation (thin-plate regression splines) to be able to provide data products that can be validly used without needing to directly model the unbalanced raw dataset using more complex models such is linear mixed models which is the approach I have used and is common in science since the researcher is dealing directly with the raw base dataset.
“Reliance on single-day values and individual points is discouraged due to the significant uncertainty that is inherent in such a product, as a result of the spatial distribution of the underlying observations, differences in observation time between neighboring stations, and interpolation errors. Spatial and temporal averaging tends to reduce the effect of these uncertainties, and time series of such aggregated values can be shown to be suitable for climatological applications.”
https://www.ncei.noaa.gov/data/nclimgrid-daily/doc/nclimgrid-daily_v1-0-0_user-guide.pdf
steve_showmethedata February 3, 2023 3:57 pm
Steve, you opened the bidding by saying:
So yes, you DID accuse NOAA of using improper methods.
You continue:
You reply:
So your claim is that I cannot say a method works simply because it works?
OK, you can run with that. Here in the real world, we have a curious saying, viz:
But I guess in your world it’s easy to argue with success unless it also includes, what was it, a:
Right. You can wait around for that. Or you could provide it yourself.
Oh, wait, I forgot. You don’t provide anything. For example, you don’t provide a description of exactly what is wrong with my method. You just say the equivalent of
Since it works very well in practice, and every single peak in the graph corresponds with a historically verified Texas cold snap, I fear your theoretical objection is meaningless.
w.
Willis, I did ask for more detail “I have downloaded this data but its not weather station-level data but single values for each day so what are these values; means across stations?” and I hedged by saying “the raw data is most likely a highly unbalanced spatio-temporal dataset”. I was not fully informed until I checked out the source data at NOAA which showed me that they had “balanced” the dataset. It was my mistake to make the wrong assumption about the source data but it was partly because you did not have a methods section describing your source data in any detail. So at no time did I say in effect “NOAA got it wrong”.
You say “Oh, wait, I forgot. You don’t provide anything. For example, you don’t provide a description of exactly what is wrong with my method.” Your method is so poorly defined its hard to be specific except to point out what it lacks and its your proposed statistical method of inferring trends in the frequency of extreme weather events (e.g.”cold snaps”) not mine. So what’s lacking? 1. A rigorous mathematical definition of what defines a “cold snap” which you have not given. Minimum temperatures are a continuous variable, so is a “cold snap” a set number of consecutive days below a set minimum temperature? Note that involves two arbitrary values. How many of such “cold snaps” can potentially occur in the one month of February? You are somehow discretizing a continuous variable to define a type of discrete event.
The currency of statistical methods is providing the maths to show how your proposed statistic, the standard deviation, quantifies the expected frequency of extreme events (e.g.”cold snaps”). So what’s lacking? 2. The definition of an estimator, as a function of the above statistic, of the expected frequency of the extreme event for the month of February and how the sample estimates as realisations of this estimator trend across years. Note its not the trend in standard deviations that is important because that’s not the inference you are interested in, its the trend in the above estimator of the expected number of extreme events (i.e. cold snaps) in this particular month. If you can give such an estimator is it unbiased, consistent and minimum variance compared to alternative estimators?
So the lack of 1. and 2. above are what’s wrong with your method.
Your method involves a vaguely defined “cold snap”, that only you know how you define, some standard deviations that look like they are telling us something useful but we are not sure exactly what that “something” is. So many theoretical and practical questions to answer. Remember verified statistical methods used in practice have background mathematical theory behind them involving many important issues that professional statisticians consider, too numerous to consider here. Maybe you can collaborate with a professional statistician to address these shortcomings in what you have presented due to it having no theoretical underpinnings.
Addendum: since I can also think practically, how does your method discriminate between Februaries containing unusually warm “spells” to those with cold “snaps” since hypothetically both could equally inflate the standard deviation through your unspecified statistical mechanism?
Apparently, they also forgot how cold December was.
Short weather memory, decidedly.