Or perhaps that the seas aren't rising:
It's not clear in Rud's article which paper he's writing about when. He seems to go between Anthony's published paper, Fall11, and the unpublished and seriously flawed paper that Anthony announced with gusto when he turned his back on the BEST study. This was after he was "prepared to accept" it. In his still unpublished paper, Anthony forgot that time of observation bias needed to be allowed for in US records. (He's had poor Evan Jones slaving away for the past three years or more, but so far nothing more has emerged.)
One of the main takeaways from Anthony's published 2011 paper was that the trend in mean US temperature anomalies is accurate. The authors found differences between poorly sited stations and well-sited stations in maximum and minimum temperature trends. For poorly sited stations the trend in the minimum was greater, while the trend in the maximum was less than for well-sited stations. However these differences pretty well cancelled each other out when the mean was calculated. As stated in the paper:
The opposite-signed differences in maximum and minimum temperature trends at poorly sited stations compared to well-sited stations were of similar magnitude, so that average temperature trends were statistically indistinguishable across classes. For 30 year trends based on time-of-observation corrections, differences across classes were less than 0.05°C/decade, and the difference between the trend estimated using the full network and the trend estimated using the best-sited stations was less than 0.01°C/decade.
Sources of bias in land surface temperature
Rud starts out disputing Anthony's finding and wrote:
It is generally accepted that there are two major land temperature record issues: microsite problems, and urban heat island (UHI) effects. Both introduce warming biases.
That is not correct. It is incomplete. These factors can introduce both warm and cool biases. When it comes to mean temperature, there is no discernible difference once appropriate adjustments have been made. Urban heat islands do (by definition) introduce a warm bias, to a point. Without specifying microsite "problems", any bias could be to the warm side, the cool side, or could balance out equally (warmer or cooler).
In built up areas, the bias could be warm or cool. It depends on the surrounds. This paper by Lei Zhao et al in Nature last year, indicates that some built up areas in dry climates introduce a cool bias, with convection during the day decreasing the max temperature change by 1.5 ± 0.2 kelvin.
You may recall the kerfuffle about Rutherglen, too, in regard to microsites. At one point a local politician wanted to shift the weather station because it was more prone to frost than nearby sites and he thought it would drive away tourists.
Rud then wrote:
The SurfaceStations.org project manually inspected and rated 1007 of 1221 USHCN stations (82.5%) using the 2002 Climate Reference Network (CRN) classification scheme (handbook section 2.2.1). The resulting preliminary paper shows a large temperature trend difference (about 0.1C/decade) between acceptably sited stations (CRN 1 or 2) and those with material microsite problems (CRN 3, 4, 5).
If Rudd's writing about the unpublished paper, then it doesn't mean zilch, because it didn't allow for time of observation bias and could well have other major flaws as well. Given the published paper, then the positive trend difference only applies to minimum temperatures, not the mean. The maximum temperatures have an equal negative bias, which cancels out the positive bias. So there is no difference in the mean between poorly-sited and well-sited weather stations.
Urban effects
Rudd then turns his attention to urban heat adjustments. There have been other papers that demonstrate that urban heat adjustments for the USA are accurate. Menne et al (2010) came out not long before Fall11, and used Anthony's data (which he didn't like much) to show the same thing that Anthony and his colleagues did. In 2013, Zeke Hausfather and colleagues did an independent analysis of US temperature records. They found that there were differences in the raw data, but that the homogenisation process effectively removed bias, since around 1930:
Homogenization of the monthly temperature data via NCDC's PHA removes the majority of this apparent urban bias, especially over the last 50–80 years. Moreover, results from the PHA using the full set of Coop station series as reference series and using only those series from stations currently classified as rural are broadly consistent, which provides strong evidence that the reduction of the urban warming signal by homogenization is a consequence of the real elimination of an urban warming bias present in the raw data rather than a consequence of simply forcing agreement between urban and rural station trends through a spreading of the urban signal to series from nearby stations.
One other problem that I see with Rudd's 'analysis' is that he jumps from averaged temperature changes across the US to specific sites. This looks like a classic denier cherry pick. The homogenisation process is used to ensure that the record as a whole properly reflects temperature changes over time and space. GISS doesn't go to each station by station to adjust the data. That would be an onerous task with no guarantee of success. In the FAQ, GISS states:
Q. Does GISS deal directly with raw (observed) data?
A. No. GISS has neither the personnel nor the funding to visit weather stations or deal directly with data observations from weather stations. GISS relies on data collected by other organizations, specifically, NOAA/NCEI's Global Historical Climatology Network (GHCN) v3 adjusted monthly mean data as augmented by Antarctic data collated by UK Scientific Committee on Antarctic Research (SCAR) and also NOAA/NCEI's Extended Reconstructed Sea Surface Temperature (ERSST) v3b data.
Different data sets show the same thing
The main problem though is that Rud doesn't do any robust analysis. He ignores studies that do. He finished up by writing unsupported statements:
Automated homogenization algorithms like GISS use some form of a regional expectation, comparing a station to ‘neighbors’ to detect/correct ‘outliers’. BUT 92% of US stations have microsite issues. So most neighbors are artificially warm. So the GISS algorithm makes the hash illustrated above. How could it not? And by extension NCDC, BEST, Australian BOM, …
Rud alleges that all the data has a warm bias and implies that different organisations all process data in the same way. He's wrong on all counts. NCDC and GISS both use GHRCN v3 for global data but process it differently AFAIK. The paper by Hansen and co (2010) describes the homogenisation process used by GISS. Berkeley Earth (BEST) takes a very different approach. The Australian Bureau of Meteorology (BOM) takes another approach, which is described on its website. The fact is that while there are some differences, all the different approaches point to the same result:- the planet is getting hotter. Ice is melting. Seas are rising. All the science denial in the world won't change the hard facts.
 |
| Data sources: NASA GISS, UK Met Office Hadley Centre, Berkeley Earth |
From the WUWT comments
There were a lot of people talking about their local weather, it was the hottest ever summer in Seattle and a mild summer in Texas, going by the comments. Here are some other thoughts from WUWT:
knr wants to know if there's any meaningful precise value for the average temperature of Earth. Not really (see this article from NASA's GISS). That's only one reason most organisations report trends rather than actual temperature, how it's changed relative to a past period.
August 3, 2015 at 9:23 am
The question I would rise is , if you where to sit down think what would it take to scientifically come up with a meaningful value for the average temperature of the planet and given that , how well do we currently match this?
I suspect we find that the conditions required to produce this value in manner that actual has a scientific value , are not met , and that we are using a value which in reality is ‘better than nothing ‘
Experimental design 101, if you cannot take the measurements in the manner required than any value you produce is suspect and subject to errors , and if you do not know the errors its subject too, then your ‘guessing with numbers’
Now what is the actual state of our ability to produce this value in meaningful way , any one know ?
Salvatore Del Prete agrees with Rud Istvan, that the data that's been collected by people all over the world since the 1800s ought to be thrown away. I guess they don't like inconvenient data.
August 3, 2015 at 9:32 am
The GISS data is worthless and should be thrown out.
Louis Hunt wonders how BEST can make a statement based on their analysis. (Why is it that most deniers at WUWT don't ever bother linking to sources? I suppose I should be glad that some bother quoting sources. Most don't. They just make up stuff.)
August 3, 2015 at 9:43 am
Berkeley Earth (BEST) has the following comment in their FAQ:
“Our UHI paper analyzing this indicates that the urban heat island effect on our global estimate of land temperatures is indistinguishable from zero.”
How can they honestly make such a statement?Read their paper, Louis, and you'll find out.
Ronald doesn't want to toss out any data, but he doesn't want any corrections to bad data (eg calibration errors). Nor does he want the raw data adjusted to make all the records comparable across the globe (eg for time of observation bias, siting changes, UHI etc). Ronald thinks an ice age is coming, any day now.
August 3, 2015 at 10:10 am
The only good data is the raw data. Every adjustment is plain wrong. But yes I do understand that adjustments need to be maid to keep up whit the non excising global warming. So both temperatures in the past and present must be adjusted tho fit the models.
Its not good but oke what to do about it?? if you tell someone the temperature is adjusted your a skeptic who doesn’t know about climate.
The only thing we cane do is sit back relax and watch the world turn colder, colder and colder.
Science or Fiction voices the irrelevant thought that Tony Heller's "science" is as bad as if it were done by his dog. Which is about right. (Tony Heller is "Steve Goddard".)
August 3, 2015 at 5:25 pm
Tony Heller is an expert in coming up with clever tests, capable of falsifying even poorly defined theories.
That is science. And that would be science even if it was done by my dog.
References and further reading
Fall, Souleymane, Anthony Watts, John Nielsen‐Gammon, Evan Jones, Dev Niyogi, John R. Christy, and Roger A. Pielke. "Analysis of the impacts of station exposure on the US Historical Climatology Network temperatures and temperature trends." Journal of Geophysical Research: Atmospheres (1984–2012) 116, no. D14 (2011). DOI: 10.1029/2010JD015146 (open access)Menne, Matthew J., Claude N. Williams, and Michael A. Palecki. "On the reliability of the US surface temperature record." Journal of Geophysical Research: Atmospheres (1984–2012) 115, no. D11 (2010). DOI: 10.1029/2009JD013094 (open access)
Hausfather, Zeke, Matthew J. Menne, Claude N. Williams, Troy Masters, Ronald Broberg, and David Jones. "Quantifying the effect of urbanization on US Historical Climatology Network temperature records." Journal of Geophysical Research: Atmospheres 118, no. 2 (2013): 481-494. DOI: 10.1029/2012JD018509 (open access)
Muller, Richard A. "Influence of urban heating on the global temperature land average using rural sites identified from MODIS classifications." Geoinformatics & Geostatistics: An Overview (2013). doi:10.4172/2327-4581.1000104 (pdf here)
Lei Zhao, Xuhui Lee, Ronald B. Smith, Keith Oleson. "Strong contributions of local background climate to urban heat islands". Nature, 2014; 511 (7508): 216 DOI: 10.1038/nature13462 (pdf here)
- Urban heating and what drives it - HotWhopper article re above
Hansen, James, Reto Ruedy, Mki Sato, and Ken Lo. "Global surface temperature change." Reviews of Geophysics 48, no. 4 (2010)., doi:10.1029/2010RG000345. (open access)
Frequently asked questions - GISS Surface Temperature Analysis (GISTEMP)
From the HotWhopper archives
From the HotWhopper archives
- On Plain Denial: Watts Dismisses His Own Evidence That is Counter to His Viewpoint - March 2013
- Heat sinking, temperatures rising in the US of A - April 2014 - with comments by Evan Jones and Victor Venema
- Rud Istvan goes on a Gish gallop at WUWT - May 2015
Judith recently and famously best-friended BEST (along with Andy Revkin) but now BEST is coming back to severely bight her bum. It's all tuning to shit. Next thing we know Mosher will end up being a revered and highly respected climate guy. I tells ya he's on a trajectory.
ReplyDelete(And one day I'll learn how to back this up with appropriate HTML links to comments but for now you'll just have to take my word for it)
The data was only admissable when there was enough short term noise in it for deniers to pretend there was a 'pause' in warming.
ReplyDeleteI don't want to swim in the polluted waters of WUWT, but are you selecting only the comments which have terrible spelling and punctuation, or are they all like that?
ReplyDelete(Checked my message three times to find any errors. Won't find any until AFTER I post.)
It's not easy to avoid WUWT "thoughts" that contain poor spelling, bad grammar, strange punctuation, weird science, or plain crankery. (They are just comments though, not guest articles. And like here, I don't think they are editable after the fact.)
DeleteEven if most stations had problems, the homogenization algorithms pick out drastic changes in the temperature trend at one station that don't appear in the others. The existence of a given perturbation at a given year amongst a whole swath of similarly located stations would be an astounding accident - but would make sense under a hypothesis that there really was a temperature perturbation in that region. So the homogenization removes perturbations that are likely not real and keeps ones that likely are, based on a simple probability argument as above. Algorithms that can be that selective are good algorithms.
ReplyDeleteThere seems to be some of the standard WUWT confusion over anomaly calculations and offsets vs. trends here.
ReplyDeleteAs I understand the terms, "warm bias" and "warming bias" aren't the same thing. A "warm bias" is a bias in the average, while a "warming bias" is a bias in the trend. Zhao et al. show a cool bias, but not (AFAICT) a cooling bias. UHI do, by definition, entail a warm bias. Biases in the average are removed by the anomaly calculation, they don't affect the trend. Istvan is claiming that UHI introduces a warming bias in the trend, which doesn't follow from the definition--it has to be determined by observation.
That's probably a good distinction, in my understanding too. Data is collected by observing some data generating process, and a bias is anything that would change the instrument's reading from what it would be otherwise. Biases can impact the data generating process itself, or the instrument; and can also be drifts or just step-changes, i.e. changes in the stationarity or the mean values of either the process or the instrument. "-ing" biases typically name drifts, as expected just from the grammar, where we'd say "cool" to mean a step change that didn't change how stationary a series is.
DeleteOf course, multiple step biases (or even just one) can cause a change in the calculated slope when you move across the bias point. This is for instance what occurred with the buoy v. ship data in the ERSST.v4 switch, where it was the increasing fraction of buoys that caused an apparent change in the "trend". Really there was no trend change, though the *slope* did change because of a growing step bias.
With Rud, there is no there there.
ReplyDelete