What if we wanted to use temperature data for the good old USA to show some good old American cooling? What data would we use?
If we use the whole-USA data from NOAA (actually the data for the “lower 48” states), that’s not going to be easy. The trend since 1895 (when the data begin) is warming:
The dots show annual averages, although the final one is highly uncertain since it only includes data for the first three months (the rest of the year isn’t available yet). Nonetheless, the trend is decidedly upward, as shown by the red line (a lowess smooth). The current rate (based on monthly data rather than annual averages) is over 5°F per century. Even if we use a linear trend, quite ignoring the fact that it’s warming faster now than before, we still get a rate of 1.4°F per century since 1895. Not cool.
Maybe we could do better if we focus on individual states. But which ones? Here’s a map showing the linear trend rates for all 48 states, with red circles warming, blue circles cooling, larger circles for larger rates:
Not good. Only two states (Alabama and Mississippi) show any cooling at all, and even the fastest-cooling (Alabama) is only at a paltry -0.35°F/century. Rhode Island, however, shows a linear rate since 1895 of warming at 2.9°F/century, twice the national rate.
How about we abandon the trend in daily mean temperature, and go for the daily high or low instead? We’ll probably want high, because nighttime seems to have been warming faster than daytime. Here are the rates for overnight lows:
Here they are for daily highs:
Well, we’re getting somewhere. Restricting to daily highs, Alabama is cooling at -0.7°F/century.
But it’s still not very impressive (as the map shows). Why not pay heed of the fact that winter has been warming faster than summer, and get rid of the wintertime data? Using just the summer months, and just the daily highs, we finally have something to work with:
Allright! Now that’s a nice big blue dot on Alabama; its daytime-high summer-only temperature linear trend shows cooling at -1.5°F/century. That’s something to talk about. Good thing we went with the linear trend, because the recent trend in Alabama summer-only daytime-high-only temperature is warming at +1.1°F/century.
Maybe we should even recommend that everybody who wants to study climate should focus on daily high temperature only, and summertime temperature only.
Recently John Christy and Richard McNider put together some new temperature time series to compare with climate model simulations. The only used locations in Alabama, only daytime high temperatures, and they used only data from summer months. What a coincidence.
This blog is made possible by readers like you; join others by donating at Peaseblossom’s Closet.
Open Mind 
That is brilliant as usual and the punchline is very funny. Oh dear.
They truly are a disgrace.
And to think they could have tossed in Indiana and Iowa, too. What restraint.
That Christy is a real piece of work. I hope he’s lambasted left right and centre.
This Christy & McNider (2016) paper shows all the symptoms of its authors having come up with some findings which individually are trivial but which collectively could probably/possibly be formed into something non-trivial – but unfortunately the authors have not yet managed to work out what that non-trivial something actually is. It’s an prime example of “can’t see the wood for the trees” which is ironic give their suggestion for the cause of the cooling of Alabama is re-forestation. As Christy & McNider actually live in Alabama, I surprised they didn’t think to ask if the re-forestation occurred at the same time as the cooling which their data evidently shows is confined to the period 1955-80.
John Christy and Richard McNider are very unlucky to choose such a unrepresentative sample. ;>)
How the hell did a paper like that pass peer review? Seriously?
The weak unrepresentative trend of high temperatures, summer temperatures and Alabama is not the only problem.
The high temperature and the summer temperature are most likely have the largest cooling bias due to improvements in how the thermometers are protected against (solar and infra-red) radiation).
The homogenization method extremely weak, it is home-made, untested and uses completely outdated design principles. This while the homogenize the summer monthly maximum temperatures, which are more noisy than the annual means and thus much harder to homogenize.
For details see my last post.
This paper should not have passed peer review.
You literally cannot make this stuff up….
Let’s see…
> Num_seasons Num_measures Num_states 1/(Num_states*Num_measures*Num_seasons)
[1] 0.001736111
At best, it’s a 575 :1 longshot. Actual probability is likely far more rarified. Pretty darned good shooting to hit the mark! Impressive!
Christy et al have mastered the art of How to Lie with Statistics. Here they are using the Texas Sharpshooter Fallacy.
Well done. Between this post of Tamino’s and one by Victor Venema, some key flaws of Christy & McNider have been neatly laid out. I was mildly concerned that McNider might have been a hapless graduate student, but it turns out he’s been publishing since at least 1981, and has coauthored papers with Spencer and Pielke Sr. as well.
Well, that explains everything.
Far be it from me to defend S&C, and the choice of Alabama summertime TMax is decidedly odd (and not well defended in the paper, IMO), but in the interests of fairness it should be noted that they do say this:
“While the heart of this investigation is the methodology used to find and correct for inhomogeneities in climate station data, Alabama is at the center of a rather curious region of long-term anomalous cooling in the southeast United States (Christy 2002; Rogers 2013). The reasons for this cooling have not been definitively determined, and, as demonstrated later, climate models do not replicate this negative trend. Thus, the dataset we construct may aid in defining attributes of the cooling and help us to understand its causes.”
This certainly does not excuse the selection: If the heart of the investigation is what they claim, why go off on a tangent about the causes of cooling in Alabama? How does that help with the “heart of this investigation”? Why not use the full dataset for this study and write a separate paper about weirdness in Alabama?
But at least they do admit that it’s anomalous. That’s worth at least 1 point, isn’t it?
At a guess, a peer reviewer pointed out the fact that they used unrepresentative data, so they threw this in to pretend they addressed the criticism. All it says is ‘we know this is unrepresentative, but it doesn’t matter because of Reasons’.
Sorry, not “S&C”. Christy & McNider. My bad.
Wow. That’s a big, fat cherry… irresistible to some, it seems.
Even with their acknowledgment that the decline is anomalous, there’s as market for data points like this– others can pick the cherry and omit the acknowledgement. I expect some will.
Oddly enough, C&M’s homebrewed homogenization still appears to produce a record reasonably close to that of NCDC’s homogenized series (see their Figure 9). None of their results really cast doubt on any of the official published temperature indices, Pielke Sr.’s blustering on twitter notwithstanding.
Nicely done, Tamino.
But as far as “The reasons for this cooling have not been definitively determined, and, as demonstrated later, climate models do not replicate this negative trend” goes, I think, @chrisd, that undermines any credits they get by writing “Alabama is at the center of a rather curious region of long-term anomalous cooling in the southeast United States”.
I think it’s d20 all over again.
Tamino — Pleased to see you using the degree sign. However, the SI convention includes a required space between the number and the degree sign. This helps disambiguate between temperature degrees and angular degrees, as in 90 °F versus 90° right angle.
There is something of a history of claims of cooling in the southeast US but thanks for this walk in the garden of forking paths
The Berkeley Earth analysis ends in 2013, but in monthly means there is no sign of a cooling trend in the S&C epicenter, Huntsville Alabama:
http://berkeleyearth.lbl.gov/locations/34.56N-85.62W
The trends in Huntsville are similar to those of Alabama and CONUS, as suggested by the table in the link
Not only worse than you think, worse than you could possibly think. The “warming hole” that they found is well known, commented on and investigated in a series of papers extending back to the WG1 TAR.
As Barton and Victor noted this paper should not have passed review.
More to the point, it should be retracted. Not even an undergraduate would normally be permitted to get away with junk such as this. The American Meteorological Society should be running, not walking, to see that retraction happens.
> … a required space between the number and the degree sign.
Lovely detail, as a lifelong nitpicker I can only say: Squee!
Off topic a little, but whatever happened to the Spencer/Christy paper on their new version of UAH temps? I thought Spencer said it was to be submitted last fall sometime.
According to a posting of Woy’s monthly “latest global temp. anomaly” at the planet Wattsupia here (but unreported on Woy’s original post here), as of April Fool’s Day this year “the paper describing the methodology (was) in peer review.” But don’t hold your breath. The blog post used by Woy et al. to launch his splendid new v6.0 TLT described that this paper of theirs would still be some months away from passing peer review. We’re a year down the road from that launch and they said at that time – “It will likely be close to two years before a peer reviewed paper with greater detail gets published in a scientific journal.”
So will Woy’s paper appear before RSS TLT v4.0 appears? Given RSS TLT v3.3 gives very similar trends to Woy’s new UAH TLT v6.0, and with other RSS v4.0 showing very big trend changes from their v3.3 equivalents, UAH will likely again be yielding different trends from RSS. In the past such differences could be explained because UAH v5.6 was for a much lower average altitude (UAH v5.6 – 2,500m, RSS – 4,200m). Indeed, ignore the first few years of that v5.6 record and it gave trends pretty-much the same as the HadCRUT4 surface record. With UAH TLT v6.0 averaging very close to RSS v3.3 and for such similar average altitudes (UAH v6.0 – 4,500m), altitude no longer provides a reason for any large discrepancies in trend. They will be direct equivalents.
I’m not a scientist I’m an engineer, nonetheless the place I work in is termed a lab, and it’s my lab. My staff calibrate industrial temperature, electrical/electronic and pressure equipment. There are two things I take extremely seriously, honesty and safety, almost everything else is negotiable.
If I caught one of my staff playing the sort of fundamentally dishonest lousy games Christy and McNider are playing here they’d get the b*ll*cking of their lives. If they were within 18 months of recruitment I’d fire them on the spot.
Very impressive!
The national broadcaster in Australia has a “fact check” unit that usually do a reasonable job, but the following on gun deaths in Australia following our tightening of gun laws seems a bit dodgy:
http://www.abc.net.au/news/2016-04-28/fact-check-gun-homicides-and-suicides-john-howard-port-arthur/7254880
It seems to have a bit of the climate change denial whiff to it.
Would you be willing to have a look at it?
Figure 9 of this 2013 report: http://www.aic.gov.au/media_library/publications/mr/mr23/mr23.pdf shows that percentage of deaths from knife wounds has increased slightly since 1989 while deaths from firearms have decreased. The Fact check report you cite only looks at gun data. It seems to me that comparison to all homicides is important.
Since there are not many homicides in Australia, it is difficult to get statistical significance from gun only data. Some of the data has been processed by NRA supporters in the USA.
Maybe now they have lost the satellite data
And recognising the fact that warmer nights and warmer winters seems to be a compelling indicator that it is not the sun, even for the willfully ignorant – it seems an easy concept to grasp
This is the new battleground in a managed retreat