Matthew R Marler:
Tough luck Steven Mosher, you goofed on that one.
Good luck ever getting him to admit this.
It’s pretty pathetic BEST is having this sort of behavior be part of its public outreach. “You criticized us? We’ll completely misrepresent what you say in obvious ways to insult you! Because that’s how we deal with legitimate concerns!”
1.6 billion records.
I’m on it.
But you realize that the brandons of the world would remain skeptical. You know there is uncertainty over exactly how many died in the holocaust.
Should be “The [GHCN] adjustments make a 2 C difference just by 1940.”
Yes Paul, with GHCN we are dealing with one level of adjustments. With GISS we can be dealing with one level (GHCN) or two levels (GHCN + GISS).
It’s nightmarish. At least with BEST there’s only one level but they still don’t adjust for UHI/Sheltering e.g. Auckland Albert Park, and overdo it in Australia compared to BOM e.g. Rutherglen 1980. If they did correct for UHI/Sheltering BEST would corroborate the NZCSC Auckland series but not NIWA’s.
Where BEST comes unstuck in NZ is kriging from far-flung places when there is data at hand e.g. Hamilton in the Waikato province is compiled from data in the Auckland province (e.g. Albert Park) and Bay of Plenty Province (e.g. Tauranga). Neither of which is similar to Waikato climatically because Waikato is hinterland whereas Tauranga is on the Pacific coast and Albert Park on an isthmus between harbours on Tasman Sea side and Pacific ocean side of the North Island NZ. Oddly, BEST overlooks NIWA’s Ruakura Research station right at Hamilton which forms part of NIWA’s Eleven Station Series (11SS).
Probably obvious but BEST’s Hamilton series is nothing like the Ruakura Reseach series situated in Hamilton.
Let me add to my reply because it is worse than that.
Suppose you are correct and maintaining the Stevenson screens did lead to the breakpoints you allude to. A Stevenson screen which is bare of paint or lacking in ventilation or otherwise in bad condition usually records a spuriously high temperature. Such issues develop slowly as the instrument deteriorates. Fixing the problem on the other hand causes a sudden correction, a temperature drop which will be detected as a breakpoint. So what should we do about sucha record?
What breakpoint correction procedures do is shift the two parts of the temperature record to align them and eliminate the ‘break’. But remember that the break in this case is actually the correction caused by fixing the instrument. The bias caused by the slow deterioration of the instrument is not detected as a breakpoint and the warming that this causes is preserved.
Breakpoint correction in this situation is a systematic procedure for undoing all the good caused by actually maintaining the instruments.
Matthew R Marler:
There are a few adjustments as large as 0.2 degrees, but they have little effect on the estimates of the global trends.
Suppose I am correct that the timing of the effect of the homogenization algorithm is (at least in part) an artifact of their choice of baseline period when calculating their climatology fields. If I am, it would mean BEST could have its modern temperatures change by as much as several tenths of a degree just by using a different baseline period. I think that’s something people might care about, even if that didn’t affect global trends (and I have no particular reason to think it wouldn’t).
This is especially true given the fact BEST released a report for the media discussing whether or not 2014 was the hottest year, a report which gave significant attention to differences less than a tenth of a degree. BEST apparently cares enough about differences of ~1% to release a report on it to get media attention. I don’t see how it can do that then dismiss concerns which affect their results by more than 10 times as much.
converting 0.2 – 0.3 to 15%-20% does little to add to your point: what is the base?
I’m not sure what you mean when you say “base” here. I was tempted to just be snarky and say, “10.” Instead, I’ll explain the reasoning. Looking at the BEST figure, I estimated it shows ~1.5C of warming since 1850. I then estimated that number would have been ~1.2C if they hadn’t calculated their “empirical breakpoints.”
I think it is useful for people to know ~20% of the warming shown in the figure is due to this particular homogenization step. Doing so lets them know the total amount of warming, the amount of warming caused by these adjustments and how the two relate to one another.
After reading the post and comments, I am fairly satisfied that the people who are working on BEST data set are sincere and their statements of why they did what they say they did have traceable logic behind those efforts.
The Urban Heat Island effect seems to me to be problematic as with the increase in global population, there has been a population movement from rural to urban that impacts the UHI effect. Adjustment for Mumbai may not be the appropriate adjustment for Rio de Janeiro because all things are not equal.
This leads me to Rob Ellison’s comment:
” Climate models at best define a probability space – there are no unique, deterministic solutions.”
Actually Ian, homogenization seems to do a reasonably good job of dealing with UHI, at least in the U.S.
I wrote a paper with the NCDC folks that looked at how well homogenization does at removing UHI in the U.S. record. We looked at various definitions of urbanity based on things like lights visible at night from space, population density, population growth, and impermeable surfaces. We also ran the homogenization algorithm both with all stations and with only rural stations used to detect breakpoints, to avoid any risk of falsely adjusting rural stations upwards. We found that there is a sizable UHI signal in the raw data, and that most of it is effectively removed by homogenization even when only rural stations are used to homogenize, across all the different urbanity definitions examined. ftp://ftp.ncdc.noaa.gov/pub/data/ushcn/papers/hausfather-etal2013.pdf
My take is that none of it matters because sea levels and overall temps have done nothing but swing for millennia and it’s hard to find anything odd in the utterly commonplace.
But, regarding UHI, it’s interesting to reflect that most of Sydney’s (Obs) hottest years are indeed clustered after 2000, while two long-record coastal+rural stations I just happened to glance at this morning for comparison have their top temps clustered well back in the past.
Yamba Pilot Station’s 95th percentile of hottest years were between 1884 and 1896. Kempsey, opened later and further from the coast, had its entire 95th percentile in a straight run from 1910 to 1919 (and that’s within BoM’s safer screening era). Similar story for the 90th percentile at these stations. Other and more mixed stories from elsewhere, of course.
A history of max temps tends also to be an history of cloud behaviour, land clearing, re-vegetation, screens, UHI etc so it’s hard to care much. (Contrary to what has been cleverly half-implied, much of Australia was drier for the half-century before 1950.) If people want to turn this vague stuff into graphs and if those graphs indicate some global cooling or pausing or warming in various lines or cycles…well, it’s good to know the world is still the world! We still haven’t flatlined after all these millennia! Still got it!
Of course, a degree of heat in Sydney is far more distinguished than a degree of heat here in the boonies. If a degree falls in the forest…
@PE
An interesting series of posts on this topic, to date. The issue of “what to do” is politically fraught – Judith C’s position of “wickedness” – but it does bring out the crazies
In my view, a hard empirical measurement of the difficulty people have with this issue as compared with issues that allow arm waving without accountability is the number of comments accruing to this post and the succeeding post (on the perennial “temperature adjustment” armflap):
Both topics were posted the same day. This one attracted about 200 comments, the succeeding one has already over 320
My conclusion ? Keeping the mega-cities from collapse while “transitioning” to renewabubbles is just way too hard for most people. It requires far too much accountability. I’ve observed this disparity for over 30 years now at all public levels (the uneven comment numbers here are just a minor example)
Matthew R Marler:
No you can’t. All you can show without knowing the true field perfectly is that one smoother is more or less smooth than another (or no smooth at all.)
Since you guys are using the extreme of “knowing the true field perfectly,” let’s take this to the extreme and suppose we smoothed the temperature data so much the value was the same across the entire globe. That would be indistinguishable from simply using the global temperature record at every part of the globe.
According to you guys, we can’t say that is bad. According to you guys, we need to know “the true field perfectly” to be able to make judgments like, “There should be some amount of spatial variation.” Just about everyone can agree temperatures in Antarctica are changing at a different rate than temperatures at the equator. If I saw results smoothed so much they found the same rate of change in Antarctica as at the equator, I’d say that is “too smooth.”
According to you guys, I’d be wrong. According to you guys, if we don’t know “the true field perfectly,” we can’t make any judgments based upon it at all. I’m sorry, but this is just ridiculous. You guys wouldn’t accept it on any other issue. You guys would scoff at anyone who said we need to have perfect information in order to draw any conclusions.
Hi Hweale78,
Post homogenization rural stations have “a slight but not significantly different from zero reduction in warming over the 1960–2010 period”. See section 4.2 and Figure 2 in our paper: ftp://ftp.ncdc.noaa.gov/pub/data/ushcn/papers/hausfather-etal2013.pdf
Yet another lazy response from SM, a pompous grunt as it were. There are no minimums in sampling theory, so your grunt is meaningless.
I’m with Tony. BEST and the rest have produced a temp series that, whatever it reflects, it doesn’t reflect history. The Northwest passage was open and sailed in a single season by Henry Larsen in 1944. There was less ice in the Arctic in 1944 than now. It was obviously warmer in the Arctic then than now. We are told constantly that disappearing Arctic ice is an indicator of a warming world. You would not know that 1944 was hotter than now from BEST or the rest. I know BEST uses 16 data series, I know BEST uses raw data and only adjusts for breaks, measures against nearby stations when there are gaps and blah blah blah, whatever. BEST missed the 1940s heat. Something’s not right.
Mosomoso
I suspect that the rural stations you quoted did not have Stevenson screens and therefore their data will be discounted byBOM who are removing records set pre Stevenson.
Tonyb
There is a pretty substantial disparity between land and satellite records over Alaska and Northern Canada (and I would guess, also Siberia, etc….). I noticed it when working over some Alaska data, and what came to mind is that the satellites can’t see down to the level of the winter inversions. Given that cold air should preferentially warm (because the wavelengths where there is water/carbon dioxide overlap are more saturated in a wetter (warmer) atmosphere, so the effect of co2 is not very logarithmically damped), that would mean that the satellite would not be picking up the very low-level warming and hence could be a source–not the only one–of the disparity. I ran this by Christy when I was looking at this in 2011 and he agreed.
Robert Rohde, Zeke Hausfather, Steve Mosher – I am pleased you are presenting this in-depth analysis and discussion. However, when will you (and BEST) perform an equivalent analysis of maximum and minimum temperature trends?
Regards Roger Sr.
Good points and we know the paleoproxies have been abused to make a case for warming that just does not exist, a more recent example being Mann’s ‘hockey stick’ science. “Overall,” says Wegman when testifying before congress, “our committee believes that Mann’s assessments that the decade of the 1990s was the hottest decade of the millennium and that 1998 was the hottest year of the millennium cannot be supported by his analysis.”
So what? Warm it a bit, cool it a bit, that’s what actually happened in the 1960s and 1970s until we got into the 1980s and it started to get warmer still.
I’m with Mike, history doesn’t cancel out. BEST and the rest can adjust and homogenise all they like.
Good grief Mosher, talk about Exhibit A evidence for climate science’s utter contempt for locality and history.
For all their self-aggrandizing, the data manipulators of the global warming movement have become the Brian Williams of science.
