In his blog post supposedly showing that the recent Pacific Northwest heat wave had no relation to global warming, Cliff Mass displayed this chart of "Average number of days with temperatures above 99F in" Washington or Oregon and claims it has "no trend":
Yeah, right. Do you think maybe the rest of his post might suffer from such sloppy thinking? You have to wonder.
Gavin Schmidt caught Cliff on this on a comment on the RealClimate blog here. Tamino also gives it to Cliff here. After that Cliff gives up. Clearly there was some pretty sloppy thinking in his blog post.
There is indeed a trend in the data, as there is in virtually every single real dataset.
ReplyDeleteHowever, there is no statistically significant trend in the data.
Without adjusting for autocorrelation, the p-value of the trend is 0.12, not significant.
After adjusting for autocorrelation (using the method of Koutsoyiannis), the p-value is 0.40, far from significant.
So no, the claim is not false, unless you incorrectly consider every trend to be significant ... in which case there's no such thing as a trendless dataset.
Regards,
w.
Where is the data, Willis?
ReplyDeleteWhere can I download it and calculate it for myself?
Why didn't Cliff Mass do this?
David said:
ReplyDeleteWhere is the data, Willis?
Where can I download it and calculate it for myself?
You have claimed that there is a significant trend in the data. Foolishly it seems, I assumed that you actually used the data to determine that. It appears you just squinted at the data and made a snap judgment.
Me, I simply digitized the graph.
Why didn't Cliff Mass do this?
Seriously? You are asking me why Cliff Mass, someone I've never met, does or doesn't do things?
Sorry, but unlike the claimed talents of many people on the web, I'm unable to read minds or determine motives from afar. Heck, part of the time I couldn't even say why I do some things ... how on earth would I know why Cliff does what he does?
Best regards,
w.
Taming explained his analysis that found statistical significance: https://www.realclimate.org/index.php/archives/2021/07/rapid-attribution-of-pnw-heatwave/#comment-792796
ReplyDelete"You have claimed that there is a significant trend in the data."
ReplyDeleteNot in this post?
This comment has been removed by the author.
ReplyDeleteThis comment has been removed by the author.
ReplyDeleteThis comment has been removed by the author.
ReplyDeleteThis comment has been removed by the author.
ReplyDeleteDavid....there is NO SIGNIFICANT TREND. I ran the numbers. And look at the last fifty years, when global warming is significant. Do you see a trend? Of course not.
ReplyDeleteDo you have any other criticisms of my blog? Is this it? Agree with the rest?...cliff
Cliff Mass: Tamino "ran the numbers" and found significance. Tamino explained his method. What was your method?
ReplyDeleteTom Dayton, Tamino may have indeed "run the numbers", but I find nothing in his post about autocorrelation. Nature is naturally trendy, so there are many apparent trends which are totally meaningless after including autocorrelation.
ReplyDeleteBest regards,
w.
Willis Eschenbach, obviously you do not understand the methods that Tamino used. Read up.
ReplyDeleteCliff I see a trend.
ReplyDeleteTom, actually I do understand Tamino's methods. What I don't understand is whether or not he allowed for autocorrelation ... and you don't know either.
ReplyDeleteI also note that Tamino is not using using the same dataset as Cliff. In fact, he's not using real data at all. Instead, he's using the output of a reanalysis climate model, which is a very different beast. So he's comparing apples to oranges, and thus he cannot possibly falsify Cliff's claim.
Best to all,
w.
Since Willis brought up autocorrelation. What is the odds there is also a selection bias with Cliff selecting this particular graph among many possible just because it doesn't show a statistically significant trend?
ReplyDeleteWillis, if you've digitized the data, why don't you share them here?
ReplyDeleteWhy didn't Cliff Mass give a link to the data? That's de rigueur in the blogosphere. You know that. One can't take him seriously with that.
Yes Cliff, I do see a trend. So does Gavin Schmidt.
ReplyDeleteLink to your data so we can all calculate the trend. Don't be shy.
Why hide it Cliff?
Blogger David Appell said...
ReplyDeleteWillis, if you've digitized the data, why don't you share them here?
Glad to:
1911-1920, 15
1921-1930, 18
1931-1940, 18
1941-1950, 12
1951-1960, 13
1961-1970, 17
1971-1980, 20
1981-1990, 17
1991-2000, 17
2001-2010, 21
2011-2020, 19
Why didn't Cliff Mass give a link to the data? That's de rigueur in the blogosphere. You know that. One can't take him seriously with that.
You just wrote a whole post making claims that Cliff is wrong about the trend in his data when you don't even have the data ... and now you want to ask me about Cliff's reasons for not linking to his data?
1) How would I possibly know Cliff's reasons?
2) Seriously?
Medice, cura te ipsum.
w.
This is the silliest claim on the page IMHO: "As shown below, there IS NO INCREASING TREND for more record high temperatures over our region during the past century."
ReplyDeleteOf course, it's harder to set a record as time goes on. Each new record has a higher hurdle to jump. We should expect a logarithmic trend down, but what he shows is a flat trend. So what does he think he's showing here?
It's a pretty incredible claim though, that the rising temperature is not affecting the frequency of record high temps. Even if the dice are falling in a way that prevents a statistically significant trend, it's foolish to think the odds haven't changed. As David pointed out in a previous article, last weeks temps were virtually impossible without global warming.
Ah, deja vu. My friend John Phillips may have to sue Willis for plagiarism. He got the whole statistical significance thing from him.
ReplyDeleteBack in January Mr E posted a filler piece at WUWT on the trend in US temperatures and he made a big noise about the trend in some very carefully selected BEST data showing cooling.
"Millions of US citizens have been firmly convinced that the US has been warming in the 21st century, when in fact it has been cooling." he moaned.
https://wattsupwiththat.com/2021/04/01/is-america-burning/
But when John repointed out that the confidence intervals (which he did not and has not posted, bad scientist) likely included the warming he claimed to have disproven he conceded:
"John, you are 100% correct that I did not publish the uncertainties of the answer. In fact, it is NOT statistically significant. However, that wasn’t my point. My point was that millions of folks are convinced that the US has been warming rapidly in the 21st Century, and we don’t know that at all. Not only do we not know that, but the Berkeley Earth and CERES data both show cooling, and although it is not statistically significant cooling … it’s still cooling."
So there you have it. On Planet Eschenberg statistical significance is significant, except when it is insignificant, in which case when we can get a whole blog post out of an insignificant trend. To paraphrase Willis, 'the trend in Maximum Oregon and Washington temperatures both show warming, and although it is not statistically significant warming... it's still warming.
Struggling to think of a clearer double standard.....
I was a little unfair to Mr Eschenberg. I said I was struggling to think of a clearer example of double standards than his highly flexible insistence on statistical significance.
ReplyDeleteOn reflection, I can come up with two better ones.
Thomas Fuller was a coauthor with Steve Mosher of a trashy cash-in-quick book on the Climategate nontroversy. Now Mr Fuller entered into a correspondence with Tim Lambert of the Deltoid blog, a tiny portion of which Tim reproduced, which prompted Fuller to post:
"I actually don't believe men of honour publish correspondence without permission. Nor do I believe men of honour would select portions of the email that don't correspond to the entire message."
https://web.archive.org/web/20091108165631/https://scienceblogs.com/deltoid/2009/11/tom_fuller_and_senator_inhofe.php
Good one, huh?
The second example that comes to mind derives from Willis's imprimateur, Anthony Watts' 'Watts Up With That' (WUWT) website, the self-described 'world's most viewed site on global warming and climate change'. Anthony's site policy deplores sockpuppetry and also has this:
"Internet phantoms who have cryptic handles, no name, and no real email address get no respect here. If you think your opinion or idea is important, elevate your status by being open and honest."
For many years one of the most outspoken commenters and defenders of the WUWT party line was 'Smokey'. A crude Google search for the name finds over 2,500 hits at WUWT. But then on August 5th 2012 Smokey carelessly linked his username to the gravatar profile of David Stealy (aka dbs, dboehm, dbstealey and probably a few more) who it turns out was a WUWT site moderator.
That's right, for years Anthony Watts allowed one of his moderators to sockpuppet as a commentator, while the mod was all along editing, deleting or delaying arguments that might contradict the party line. And while Watts posed as a champion of 'open and honest debate'.
You know what they say about birds of a feather.
Source: https://www.webcitation.org/6AW1RKUjq 6:14am
Sorry, Eschenbach, not Eschenberg. Apologies.
ReplyDeleteWillis, thanks for providing your data.
ReplyDeleteI finally had a chance to do this calculation. For the trend of these points I get
0.42 C/decade
Without autocorrelation I an uncertainty of
0.24 C/decade
Using the most rigorous method of including autocorrelation I know, that of Foster and Rahmstorf, I get an uncertainty of
0.53 C/decade
Even with the latter, the probability the trend is positive is, I think, 79%.
Is that what you find?
If true, not much to quibble over.
I assume that you are giving one sigma uncertainty. If so, to be considered significant at the p-value = 0.05 level (the normal level in climate science), the trend needs to be greater than 1.96 times the one sigma uncertainty.
ReplyDeleteThis is NOT true in either case, with or without autocorrelation. So no, it is NOT a statistically significant trend at the 0.05 level.
Thanks for running the calculations,
w.
So we can expect a WUWT post any time now on this insignificant trend. That's what you do right?
ReplyDeleteThat's the decadal data, I would just love to see an analysis of the yearly or even monthly - but I will not hold my breath.
Here is a prediction: these heat waves will become more extreme and more frequent. About the only enjoyable thing about them will be the posters at WUWT bending themsleves preztel-shaped as they try to claim that there's nothing to unusual going on here, and even if there is, it cannot possibly have anything to do with the gigatonnes of CO2 being pumped into the atmosphere.
Sleep tight.
Willis, so we're going to bet the world's future on the fact that the trend is only 79% significant and not 95% significant?
ReplyDeleteYou can't be serious.
PS: There's nothing magical about p=0.05. It's just a number traditionally agreed upon. You have to be smart about how you interpret numbers, not according to some blind rule.
ReplyDeleteThere is a clear trend in the data.
PPS: Someone please post this trend at Cliff's site. He's blocked me for a long time.
ReplyDeleteAgreed Phil. There's a lot of bending going on over this heat wave....
ReplyDeleteTamino posted a thorough explanation of his analysis.
ReplyDeleteDavid Appell said...
ReplyDeleteWillis, so we're going to bet the world's future on the fact that the trend is only 79% significant and not 95% significant?
You can't be serious.
David, first, your previous claim that this WAS significant was total bullshit. How about you start by acknowledging that?
Second, you are claiming that the future of the world depends on whether this trend of one heatwave in one tiny part of the planet is significant or not?
You can't be serious.
w.
https://wattsupwiththat.com/2021/04/25/wheres-the-emergency/
It is indeed hard to take Willis seriously. The link is to a WUWT post rhetorically asking 'Where is the Climate Emergency?'
ReplyDeleteWhat follows is a list of irrrelevancies, errors and cherry-picks. Mr Watts shadow-banned me from his site some time ago, so if Mr Appell indulges me I will post a few sample debunkings here.
First exhibit is a graph of 'climate related deaths' falling sharply. The text say 'Deaths from climate-related phenomena are at an all-time low'. But look at the caption, it is climate deaths per million. Now population rose exponentially during the period of the graph so even if the absolute number of people killed remained static or even rose this graph would still fall sharply. Basically dishonest. Wrong metric. Fail.
We're told there is no increase in storminess, and shown a graph of Accumulated Cyclone Energy (ACE). But ACE is only one metric, many specialists prefer PDI, the US EPA say
"According to the total annual ACE Index, cyclone intensity has risen noticeably over the past 20 years, and six of the 10 most active years since 1950 have occurred since the mid-1990s (see Figure 2). Relatively high levels of cyclone activity were also seen during the 1950s and 1960s.
The PDI (see Figure 3) shows fluctuating cyclone intensity for most of the mid- to late 20th century, followed by a noticeable increase since 1995 (similar to the ACE Index). These trends are shown with associated variations in sea surface temperature in the tropical North Atlantic for comparison (see Figure 3).
We're shown a graph from Roy Spencer that is meant to show the models overestimate warming of sea surface temperatures. The graph is comically bad, the axis labelled modelled sea surface temperatures is actually modelled air temperatures, among other scientific snafus. Correct these flaws and the discrepancy disappears.
We're shown a graph showing diminishing wildfires in Canada, we are not told that area burnt by wildfire in the US has doubled since the 1990s. It is true that global area burned has declined, but you need to dig into the data. 70% of wildfires are savannah fires, and it is these that have declined, due to changes in wind and rainfall patterns. But the savannahs are sparsely populated, that decline needs to be seen in the context of more fires in populated areas. Try telling people in Western Australia, California and parts of Europe and Siberia that wildfires are 'nothing to see here'.
We're told there is no acceleration in sea level rise, and pointed to a deeply inaccurate analysis published at WUWT by, er, Willis. Talk about marking your own homework. We are not told about a response by a team of professional scientists that describes this analysis as 'disastrously wrong'.
We are treated to an apparently damning quote from IPCC co-chair Otmar Edenhofer, but it is manufactured, or shall we say 'creatively' translated from the original German. Even the sentence order has been reversed.
I could go on but there is a character limit. You get the picture. Not a credible source. Apart from the factual errors, made up quote and irrelevant metrics, the question is just a lame Straw Man, similar to the 'Catastrophic AGW' meme.
There is arguably, right now, no ongoing climate emergency with global reach, however the wildfires in California, Australia and Siberia surely count as local emergencies and were made more likely by raised temperatures and prolonged dry spells, in turn attributable to climate change. Heat waves would be another example, as are European floods, Atlantic hurricanes and there are many more. It is entirely possible that we have no global emergency now, but we are heading towards one if we do nothing. The sea level acceleration, denied in the WUWT post, is real and unchecked will likely lead to a sea level rise of at least 65cm by 2100. 50cm would be enough for the Maldives to lose 80% of its land area.
Sources:
ReplyDeletehttps://climatefeedback.org/claimre...t-claim-of-no-acceleration-in-sea-level-rise/
https://www.epa.gov/climate-indicators/climate-change-indicators-tropical-cyclone-activity#3
https://www.carbonbrief.org/factcheck-how-global-warming-has-increased-us-wildfires
https://www.nifc.gov/fire-information/statistics/wildfires
https://www.nzz.ch/klimapolitik_verteilt_das_weltvermoegen_neu-1.8373227?reduced=true
https://www.desmog.com/2017/11/07/how-climate-science-deniers-manufacture-quotes-convince-you-united-nations-one-big-socialist-plot/
Ooops. Corrected URL
ReplyDeletehttps://climatefeedback.org/claimreview/global-data-contradict-claim-of-no-acceleration-in-sea-level-rise/
"One of the more stupid debating tricks of the "skeptics" is to oscillate between Ha ha, you believe in Catastrophic Anthropogenic Global Warming which is obviously not happening so you're very silly, and when told that CAGW is a strawman that they've invented they switch to if it isn't catastrophic we've got nothing to worry about, have we?
ReplyDeleteTo which the answer is always some variant of if you can't imagine anything between "catastrophic" and "nothing to worry about" then you're not thinking."
William Connolley.
https://scienceblogs.com/stoat/2014/06/24/if-it-isnt-catastrophic-weve-got-nothing-to-worry-about-have-we
https://www.realclimate.org/index.php/archives/2021/07/rapid-attribution-of-pnw-heatwave/comment-page-2/#comment-792908
ReplyDeleteTamino makes the point that using 10 year averages reduces the number of data points and hence the degrees of freedom, making it unlikely such a small dataset will achieve significance, and suggests instead a Poisson regression...
"When I do this with the ERA5 data averaged over the study region outlined in the attribution paper, I do get statistical significance, even using 10-year averages."
Here is a prediction: these heat waves will become more extreme and more frequent.
ReplyDeleteWell, that didn't take long
https://www.theguardian.com/world/2021/jul/15/heatwave-us-west-canada-wildfires
Turns out heatwaves are like buses.
"Surface temperature is projected to rise over the 21st century under all assessed emission scenarios. It is very likely that heat waves will occur more often and last longer"
- IPCC Synthesis Report.
It’s happening. Again. For the fourth time in five weeks, a punishing heat wave is set to bake the West and adjacent western Canada. This time, the most exceptional heat is expected to focus in the central and northern Rockies, developing this weekend and peaking around Monday.
ReplyDeleteThe heat wave is forecast to bring triple-digit temperatures to at least 16 million people, challenging and breaking records into Canada. It’s also targeting an area where numerous wildfires have flared up and a smoky haze fills the skies. The arrival of even hotter, drier conditions could compound the situation.
Preliminary forecasts suggest highs in the northern United States and southern Canada could reach 20 to 30 degrees above average values for mid-July, which in many places is already historically the warmest time of year.
- Washington Post.
This is exactly as forecast by the IPCC. But that's hardly a cause for celebration. Maybe we can look forward to some free entertainment from Judy, Cliff, Willis, Anthony and the rest of the gang as the heat waves roll in and they are obliged to explain, case by case, why none of this is down to AGW. A Bittersweet experience.