Sunday, 29 July 2012

Blog review of the Watts et al. (2012) manuscript on surface temperature trends

UPDATE: Skeptical Science has written an extensive review of the Watts et al. manuscript: "As it currently stands, the issues we discuss below appear to entirely compromise the conclusions of the paper." They mention all the important issues, except maybe for the selection bias mentioned below. Thus my fast preliminary review below can now be considered outdated. Have fun.

Anthony Watts put his blog on hold for two days because he had to work on an urgent project.
Something’s happened. From now until Sunday July 29th, around Noon PST, WUWT will be suspending publishing. At that time, there will be a major announcement that I’m sure will attract a broad global interest due to its controversial and unprecedented nature.

What has happened? Anthony Watts, President of IntelliWeather has co-written a manuscript and a press release! As Mr. Watts is a fan of review by bloggers, here is my first reaction after looking through the figures and the abstract.

Press release from WUWT:
The new improved assessment, for the years 1979 to 2008, yields a trend of +0.155C per decade from the high quality sites, a +0.248 C per decade trend for poorly sited locations

In his press release, Anthony Watts does not explicitly state that these trends are for raw data. The manuscript does state this important "detail". The poorly sited locations are likely in cities where it is more difficult to find good locations. Thus what he found is that the Urban Heat Island (UHI) effect exists. I did not know that this was controversial.

UPDATE: Maybe it does not have to be due to trends from the UHI. There are two other factors. A fine alternative explanation would be that the location quality ratings were performed at the end and may thus be biased. Bad stations at the end of the period may have been better before and thus cooler. And Good stations at the end, may be been worse before and thus show a smaller trend in the raw data.
Rabett Run points out that the stronger time of observation bias for the rural stations is probably the main cause of the difference in the trends in the raw data. Steve McIntyre, co-author of the Watts et al. manuscript and blogger at Climate Audit, sees neglecting the TOB as a serious problem.

Good news is that the study finds that after homogenization, the station quality is no longer a problem for the mean temperature. As you can see in Figure 17 of the paper below, after homogenization (adjustment) the trends for all stations show about the same trend, whether they are in urban, semi-urban or a rural environment.



Figure. Top left panel of Figure 17 16 (h/t Brandon Shollenberger) of Watts et al. (2012). The first, third and fifth columns show the raw data, the second, fourth and sixth column the climatologically important homogenized (adj.) data. The first two columns are the mean temperature, the second two columns the maximum temperature and the last two columns the minimum temperature.

And from the full Figure 17 16, you can generate the table below, which shows that also the classification of the stations does not matter any more after homogenization. Stations in category 1 are well sited and in category 5 are badly sited. Results for class 5 are probably a bit noisy as the number of stations in this class only represent 12 % of the classified stations.


Table. The mean temperature trends from Figure 17 16 of Watts et al. (2012) in °C per decade.

Class 1/2 Class 3/4/5 Class 3 Class 4 Class 5
Urban 0.302 0.294 0.318 0.299 0.218
Semi-urban 0.341 0.311 0.327 0.325 0.249
Rural 0.314 0.321 0.327 0.316 0.319
Also in Falls et al. (2011), which was co-authored by Anthony Watts, it was found that the homogenized mean temperatures had about the same trend for all quality classes (no significant differences were found). Also in Falls et al. this trend was about 0.3 °C per decade. Also in Falls et al. the tend in the raw data was 0.1 °C per decade smaller. Thus I cannot see this manuscript as unprecedented. Leroy (2012) will be happy that his new siting quality classification seems to work better as judged by the larger difference in the trends between the categories. That seems to be the main novelty. This result is worth a paper, I am not sure if it worth a press release.

In the press release it is also emphasised that the temperature trend after homogenization is stronger than in the raw data. Maybe Mr Watts thinks this is new, but, e.g., Menne et al. (2009) already stated that the introduction of automatic weather stations (the transition from Liquid in Glass thermometers to the maximum–minimum temperature system) caused a temperature decrease in the raw data of 0.3 to 0.4 °C. This temperature jump has to be and was removed by homogenization.

Had the new study found clear differences in the temperature trend in the homogenized data, the study would have been interesting for the general public. Because it is the homogenized data that used to compute large scale trends in the real climate. If the homogenized data would still be partially polluted by the urban heat island effect that would have been an error. The aim of homogenization is exactly to remove artificial changes from the raw data. It seems to do so successfully, now acknowledged by WUWT the second time.

If I were reviewer of this manuscript my main question would be to clarify the statement in the abstract that "[u]sing the new Leroy (2010) classification system on the older siting metadata used by Fall et al. (2011), Menne et al. (2010), and Muller et al. (2012), yields dramatically different results." If this relates to the climatologically important homogenized temperature trends, this statement does not seem to fit with the results. If this statement only relates to the raw data, this is an important disclaimer that should not be missing in an abstract.

Style

Putting a manuscript on your homepage and inviting colleagues to review it is a good idea and can improve the manuscript. Sometimes press interest before publication is unavoidable. For example when CERN finds that the speed of light can be exceeded or finds a God particle. I am not sure whether the finding that the weather station classification scheme of Leroy (2010) is better than Leroy (1999) is of this category.

The press release does not explain why there was suddenly such a hurry, why WUWT had to be interrupted for two days and a press release had to be released on a Sunday, a day on which journalists will find it difficult to get a second opinion from a scientist. I hope my judgement of the manuscript is fair, I had only little time.

Update: Maybe I was too fast to conclude that the paper shows that trends are due to the UHI. This is the part of the homogenization literature that I am not so familiar with yet; it is frighteningly huge. A fine alternative explanation would be that the location quality ratings were performed at the end and may thus be biased to stations where the location became worse.

More posts on homogenisation

Statistical homogenisation for dummies
A primer on statistical homogenisation with many pictures.
Homogenisation of monthly and annual data from surface stations
A short description of the causes of inhomogeneities in climate data (non-climatic variability) and how to remove it using the relative homogenisation approach.
New article: Benchmarking homogenisation algorithms for monthly data
Raw climate records contain changes due to non-climatic factors, such as relocations of stations or changes in instrumentation. This post introduces an article that tested how well such non-climatic factors can be removed.
A short introduction to the time of observation bias and its correction
The time of observation bias is an important cause of inhomogeneities in temperature data.
HUME: Homogenisation, Uncertainty Measures and Extreme weather
Proposal for future research in homogenisation of climate network data.

Related external articles

Initial thoughts on the Watts et al draft
A very good summary of the main problems found in Watts et al. up to now.
This Has Become Farce
Appraisal of blog review (Quark Soup by David Appell).
Bunny bait
Rabett Run points out that the known stronger time of observation bias for the rural may be the main cause of the difference in the trends in the raw data.
Two climate papers get hyped first, reviewed later. Isn’t that a bad idea?
A well-written piece in the Washington Post, about the speed of light, Anthony Watts and Richard Muller.
Watts et al: Clunk
The sound of the drama queen's preprint hitting the Internet (Quark Soup by David Appell).
Comments on the game changer new paper by Watts et al. 2012
Roger Pielke Sr. rightly praises his friend Anthony Watts for his work on SurfaceStations.org.

References

Fall, S., Watts, A., Nielsen‐Gammon, J. Jones, E. Niyogi, D. Christy, J. and Pielke, R.A. Sr., 2011, Analysis of the impacts of station exposure on the U.S. Historical Climatology Network temperatures and temperature trends, Journal of Geophysical Research, 116, D14120, doi: 10.1029/2010JD015146, 2011.

Leroy, M., 1999: Classification d’un site. Note Technique no. 35. Direction des Systèmes d’Observation, Météo-France, 12 pp.

Leroy, M., 2010: Siting Classification for Surface Observing Stations on Land, Climate, and Upper-air Observations JMA/WMO Workshop on Quality Management in Surface, Tokyo, Japan 27-30 July 2010 http://www.jma.go.jp/jma/en/Activities/qmws_2010/CountryReport/CS202_Leroy.pdf

Menne, Matthew J., Claude N. Williams, Russell S. Vose, 2009: The U.S. Historical Climatology Network Monthly Temperature Data, Version 2. Bull. Amer. Meteor. Soc., 90, 993–1007. doi: 10.1175/2008BAMS2613.1.

Watts, Anthony, Evan Jones, Stephen McIntyre, John R. Christy, [plus additional co-authors that will be named at the time of submission to the journal]. An area and distance weighted analysis of the impacts of station 1 exposure on the U.S. Historical Climatology Network temperatures and 2 temperature trends. To be submitted, 2012.

123 comments:

dhogaza said...

"In the press release it is also emphasised that the temperature trend after homogenization is stronger than in the raw data. Maybe Mr Watts thinks this is new..."

No, Watts believes that the "supposed" trend is a result of homogenization, and that it's not really warming nearly as much as is claimed by professionals. This is a common theme in the denialsphere, that "homogenization" is just a technique to generate the trend that scientists want to see.

Total bullshit, but Watts really believes it.

Anonymous said...

1. Your Style:

"Anthony Watts, President of IntelliWeather has co-written a manuscript and a press release!"

Anthony Watts is a physicist and his co-authors are well known climate scientists.

2. Your comment

"Good news is that the study finds that after homogenization, the station quality is no longer a problem for the mean temperature."

And this is only possible, if UHI is smeared among all stations through homogenization, what is a problem, of course.

Anonymous said...

"Thus what he found is that the Urban Heat Island effect exists. I did not know that this was controversial."

BEST concluded, that there is no UHI effect on trends. Al temperature trends outlets do not correct for UHI.

Victor Venema said...

In his manuscript Anthony Watts writes as his affiliation: President, IntelliWeather, Chico, CA, USA. I myself would also have written something a bit more formal.

The fairy tail that homogenization smears or smoothed data does not become true by repeating it often. A short introduction into how homogenization algorithms really work is given in this post:
http://variable-variability.blogspot.com/2012/01/homogenization-of-monthly-and-annual.html

Victor Venema said...

I wonder whether BEST was thinking of the trend in the anomalies in average over all stations and not the trend of the raw data and especially the raw data of single stations.

Furthermore, I only now realised that part of the difference is not due to the UHI, but due to the sampling of the stations. In the group of stations that are currently not well sited, there are more stations that are now worse as in 1979 and for that reason show a warming trend in the raw data. Had the selection been performed in 1979 or had the classification been an average over the study period, the differences between the categories would likely be smaller.

As economist like to say, poor countries are characterised by a history of poor growth. But predicting from the information available in the year 1900 which developing countries are poor now is much more difficult.

dhogaza said...

"Anthony Watts is a physicist".

Anthony Watts is a high school graduate and retired TV weather reader.

He is not a physicist. He doesn't even have a B.S.

Ask him yourself.

If he had a graduate degree in physics you can be sure he'd list it after his name, on the paper.

Mogumbo Gono said...

This result is worth a paper, I am not sure if it worth a press release.

Then you are not aware of the backstory. Richard Muller violated trust. Do a search for Muller on:

http://wattsupwiththat.com

You will see that not all scientists are honest and trustworthy. Some climate scientists do not believe in the Scientific Method. And most mainstream climate scientists such as Michael Mann and Gavin Schmidt fight tooth and nail against requests for scientific transparency.

dhogazaa said...

"And most mainstream climate scientists such as Michael Mann and Gavin Schmidt fight tooth and nail against requests for scientific transparency."

Gavin Schmidt is a leader of the GISS Model E project. In what way does he fight tooth and nail against requests for scientific transparency? The entire source code to Model E is available online, along with a handy-dandy Fortran browser to make navigating through the code easy. The algorithms are documented in a variety of academic papers, all referenced in the Model E sources.

The only sense in which Muller "violated trust" is that he didn't come up with the result that Watts expected. Watts was certain that Muller's group would prove what he's insisted on for a very long time, that most of the warming in the US can be attributed to badly sited stations that magically all have a warming bias.

That's what he's out to prove here. The paper does not such thing (indeed, it mostly says that homogenization works, and if you want to attack homogenization, then you must attack the algorithms such as documented in the papers our host has referenced - it's not sufficient to say that homogenization leads to data that differs from the error-filled raw data, it's *supposed* to).

For Victor more than anyone - Watts is going to trumpet this paper of being "more evidence", or perhaps "final evidence" - that much/most of the warming trend observed in the CONUS is due to UHI and siting errors.

But the paper doesn't show this. Christy then has cover, he can point to what the paper *does* say (not much). As can McIntyre. Meanwhile the denialsphere will vehemently push the theme that the real warming trend in the US is 1/2 of the trend found by everyone else, including BEST.

Anonymous said...

Climate scientist Roger Peilke weighs in:

It certainly appears that Richard Muller is an attention-getter, which he has succeeded at, but, unfortunately, he has demonstrated a remarkable lack of knowledge concerning the uncertainties in quantifying the actual long-term surface temperature trend, as well as a seriously incomplete knowledge of the climate system.

The proper way to complete a research study is provided in the Watts et al 2012 article. This article, a culmination of outstanding volunteer support under Anthony’s leadership, shows that Anthony Watts clearly understands the research process in climate science. As a result of his, and of and his colleagues, rigorous dedication to the scientific method, he has led a much more robust study than performed by Richard Muller in the BEST project.

Finally, on Andy Revkin’s well-respected and influential weblog Dot Earth, in a comment with respect to his post

‘Converted’ Skeptic: Humans Driving Recent Warming

he writes

Muller’s database will hold up as a powerful added tool for assessing land-side climate patterns, but his confidence level on the human element in recent climate change will not. I’d be happy to be proved wrong, mind you.

Andy’s assumption that “Muller’s database will hold up as a powerful added tool for assessing land-side climate patterns” is now shown as incorrect.

The new Watts et al 2012 paper shows that Muller’s data base is really not a significant new addition for assessing land-side climate patterns, at least until further analyses are performed on the siting quality of the stations he uses in the BEST assessment.

Anthony Watt’s new paper shows that a major correction is needed Muller’s BEST study. Anthony also has shown what dedicated scientists can do with even limited financial support. Despite the large quantities of funds spent on the BEST study, it is Anthony Watts and his team who have actually significantly advanced our understanding of this aspect of the climate system. Well done Anthony!

http://pielkeclimatesci.wordpress.com/2012/07/29/comments-on-the-game-changer-new-paper-an-area-and-distance-weighted-analysis-of-the-impacts-of-station-exposure-on-the-u-s-historical-climatology-network-temperatures-and-temperature-trends-by-w/

Anonymous said...

Watt's raw data trends suggest an unaccounted UHI effect of approx. 0.1 deg/decade in the US. It may be higher as 1+2 rated station were combined to obtain a more meaningful ensemble. I would expect the effect to be higher outside the US, particularly in the developping world.

dhogaza said...

Pielke Sr has been wrong before (often embarrassingly), will be wrong again, and is wrong now.

Given that it was Pielke Sr. who came up with the surface stations project, along with Watts, and has long argued that the UHI etc has exaggerated recent temperature trends, do you really expect anything else?

To bad those pesky satellites don't suffer from the UHI or siting problems and that UAH US48 shows a warming trend 50% higher than Anthony's obviously preferred raw station class 1 and 2 trend.

Anonymous said...

Firstly, the comparison is meaningless for raw data, as some adjustments are merited.

Secondly, tropospheric data is expected to have higher trends.

Thirdly, the opposite is a valid point. Previous data sets ran warmer than satellite data, though they should not due to basic principles of physics.

Anonymous said...

"The fairy tail that homogenization smears or smoothed data does not become true by repeating it often."

Non compliant station have significantly higher trends before homogenization due to UHI. After homogenization trends are about the same for all stations. Where would you think UHI has gone if not smeared into all data sets ?

dhogaza said...

"Secondly, tropospheric data is expected to have higher trends."

It's not quite that simple. True for global averages, with most of the difference occuring in the tropics.

Got a cite that supports the notion that the US48 lower trop trend is exptected to be 1.5 times higher than the CONUS surface temp trend? Or are you just extrapolating from what's expected *globally*?

Anonymous said...

The raw data Watts et al used also lack time of observation (TOB) corrections - which will introduce a cool bias into the raw data. See Karl et al 1986 (http://journals.ametsoc.org/doi/pdf/10.1175/1520-0450%281986%29025%3C0145%3AAMTETT%3E2.0.CO%3B2) and Vose et al 2003 (http://www.agu.org/pubs/crossref/2003/2003GL018111.shtml).

Apples and oranges. Not a good comparison if they exclude known biases.

Anonymous said...

The purpose of the Watts preprint article and its timing is clear.
It justifies the press release which has been the case with 'Trojan' papers before is intended to carry the real load.

It is a spoiler for the Dr R Muller recant of his skepticism. Although we now learn Muller was never a 'true' skeptic.

It is timed to be within the grey literature limit for the next IPCC report, the fact that it may never get printed in its present form peer reviewed anywhere credible will not stop future complaints that it was ignored.

But most significant it implies that USHCN, NOAA and GISS have manipulated the raw data from the best quality stations to claim that the trend for the CONUS has been 'spuriously' doubled.
Already the insinuation is abraord that if the US data with the 'best' of observations is so wrong then the global data must be even more suspect...
And surely such an error could not be simply accidental?

izen

Anonymous said...

Read the paper, if you are looking for the purpose. Never has UHI been quantified so beautifully and consistently. The results are consistent in each blocks laid over the US. Look at figure 23. There are ample consistent results, how UHI particularly increases Tmin. There is additional valuable information about sensor types, never even looked at by highly paid climate scientists. And last but not least, there is disturbing evidence of a new warming bias introduced with the massively increased use of airports.

Anonymous said...

dhogaza, that's "BSc" -- Watts has plenty of BS.

Ben said...
This comment has been removed by a blog administrator.
Anonymous said...

The Watts paper mentions the the adjustments that it is nessecary to make to the raw USHCN data to correct for changes in instrument type as found by Menne et al 2010 around page 7, line 118. He also mentions the adjusted USCHNv2 data is adjusted for TOB - time of observation variations which have caused a cooling bias in the data. But reveals he is using the unadjusted raw data with only the minimal error correction.

The paper then selects admits that stations at airports are often compliant with the Leroy classification, and are less prone to CRS/MMTS instrument changes and TOB variation, but rejects them because they show a high trend rate compared to compliant rural stations where the raw data may be more biased from changes in instrument and TOBs.

The paper does claim to examine the differences between CRS and MMTS instrumentation, but uses 1995 as the date to classify a station as one or the other. a station may have changed method before or after that date but is still of one type in the analysis.

I am sure more perceptive and informed observers will discover other problems with the paper. But the preference shown for the sites which give the lowest trend when uncorrected raw data is used which is then used as some sort of 'gold standard' to critique the trend found after adjustments made for the instrumentation and TOB he acknowledges as required seems to indicate confirmation bias.
He even seems to prefer the lower trend from MMT stations, apparently assuming the MMT data are in need of less adjustment BECAUSE they show a lower trend!

But the chances of this making any significant contribution to the science are slim to none. as the paper also acknowledges the satellite record constrains the possible trend to AT LEAST the minimum trend found in the raw data of the compliant stations, and probably somewhat higher. The issue of how to adjust a historical weather station network to provide climate trends, a purpose really beyond its remit, will continue. I predict this article will have very little long term influence on that.

However the insinuations of the press release will echo around the doubt purveyors for quite a while.

izen

dhogaza said...

Yes, of course. Watts theme, for years now, is that such corrections are the source of modern "warming", therefore must be wrong.

His work doesn't support it, but they'll scream it wide and far regardless.

Steve Bloom said...

Other than the wingnut press and blogs, where it will be a brief flash in the pan, this error-riddled "paper" will go immediately to the dustbin of history.

Notice how, despite all the past self-righteous squawking about openness and replicability, it will be quite some time before anyone else will be able to check the work.

capitalclimate said...

He also has M.S. and Piles it Higher and Deeper.

Anonymous said...

And even lower than the UAH satellite record!

Oh, oops...

Marco

Anonymous said...

Dear Variable Variability,
As I cannot find any references to you and you do not state your credentials are in your about page, can you please tell us all what are your credentials?

Especially in the the fields of climatology, physics, paleolithics, mathematics, statistics, etc etc.

This should be known before I make any decisions on your findings of this very important release.

Anonymous said...

Oh....Steve Bloom, this paper has less errors in it than anything that has gone to the IPCC since 1991.

SF said...

As I cannot find any references to you and you do not state your credentials are in your about page, can you please tell us all what are your credentials?

Er, can't you read or follow links?

Victor Venema said...
This comment has been removed by the author.
Victor Venema said...

I had hoped that the arguments would be sufficient.

On the about page, you can find a link to my professional page.

I studied physics, did my PhD in electrical engineering. Since I work at the Meteorological Institute in Bonn, mainly on clouds and radiative transfer, for over a decade. Homogenisation is currently a side topic for me. It is a beautiful scientific problem. Best use your own brain and never rely on one one source, especially if it is WUWT.

Anonymous said...

Maybe you should go back to your 'side topics'.

[quote]
Update: Maybe I was too fast to conclude that the paper shows that trends are due to the UHI. This is the part of the homogenization literature that I am not so familiar with yet; it is frighteningly huge. A fine alternative explanation would be that the location quality ratings were performed at the end and may thus be biased to stations where the location became worse.

Anonymous said...

To all those who pick at details and suggest that the Watts paper might not be taken seriously I would just say that the basic approach and the results are what matter.

It is now in the public domain that there is a much better site assessment procedure which has not previously been methodically applied.

Also, that when it is applied, the difference in trend between sites of differing qualities becomes apparent. The most important point of Muller’s work was that there were no significant trend differences between sites of differing qualities.

The science has moved on such that the earlier assertions of Muller and the entire climate establishment are now out of date. They should graciously acknowledge that fact.

Leroy 2010 has been a time bomb waiting to go off and this paper has lit the fuse.

All else is chaff.

That is not to deny that warming has occurred but it does reduce it substantially from what we have been led to believe.

In the meantime natural variability is being shown to have a greater influence than previously recognised.

Those two factors combine to squeeze AGW into insignificance for policy purposes.

Anonymous said...

Dear Anonymous ignoramus,

Can you tell us how the field of paleolithics [sic] will have any bearing on decisions you "make" regarding Dr. Venema's findings of "this very important release" and why anyone should care what you "decide" about them?

Taylor B

Victor Venema said...

Let's not exaggerate. BEST (Muller) may have to extend one of their 5 papers on the Berkeley dataset: "Earth Atmospheric Land Surface Temperature and Station Quality in the United States". The manuscript was very short anyway.

"That is not to deny that warming has occurred but it does reduce it substantially from what we have been led to believe."

I did not see any indication that the estimate of the temperature trend is wrong. These trends are estimated on the best quality data we have, the homogenized data, and in this data not even Watts et al. could find a difference in the trends; I hope you excuse me for assuming that Watts is not completely open minded in listening to the what the data has to say.

"In the meantime natural variability is being shown to have a greater influence than previously recognised."

May I ask which study you are referring to?

Jim Pettit said...

It's nearly funny that Pielke has the gall to call Muller an "attention-getter", when the vaey obvious reason Watts felt compelled to release his poorly-written, "crowd-sourced", heavily self-referential, error-laden draft yesterday was so he could grab attention from Muller's actual science work.

Pot, meet kettle.

The Watts et al paper as it stands has absolutely *no* chance of being accepted or published in a credible scientific journal. Its conclusion are in no way supported by the data, which is cause alone for rejection. But it also contains numerous errors of the grammatical, spelling, statistical, logical, and physical. Good luck with that, Anthony...

Jim Pettit said...

I think Anthony is in for yet another scientific beatdown of the first order; he's going to shortly discover that having a draft article find favor among his army of obsequious WattsBots is far removed from actually shepherding that article through the peer-review process, where it will be taken apart syllable-by-syllable by many very smart and very educated and very experienced people who know what they're doing and whose BS detectors are set to an extremely sensitive level.

dhogaza said...

"The paper then selects admits that stations at airports are often compliant with the Leroy classification, and are less prone to CRS/MMTS instrument changes and TOB variation, but rejects them because they show a high trend rate compared to compliant rural stations where the raw data may be more biased from changes in instrument and TOBs."

Gosh ... how ... arbitrary!

I'm amazed Christy signed off on this ...

Anonymous said...

Exactly.

I hope that some red-pen PDF copies of critical reviews start appearing soon, so that people might see exactly how suitably experienced reviewers dismantle a manuscript and its underlying work.

In any research department in which I've worked, the best a draft of this nature would attract is "major re-analysis and rewriting required". A simple "Not of publication standard" would be most likely from many of my more impatient collegues...

Bernard J.

Rattus Norvegicus said...

Jim,

There's always E&E!

Joel said...

There seems to be an amusing habit in the AGW "skeptic" community to mistakenly claim people are physicists. I have seen several occasions where it was claimed that Nikolov and Zeller (the authors of this piece of pseudoscientific nonsense: wattsupwiththat.com/2011/12/29/unified-theory-of-climate ) were physicists even though it takes a simple look at their resumes online at the government lab they work at to see that neither of them have any degree in physics (although at least in this case they do have advanced degrees in technical fields). As a physicist, I suppose I should take it as a compliment that this is apparently a standard that others hold in such high regard, but I think that, with a few cranks (such as Gerlich and Tscheuschner) already in our ranks, I am loath to take on more.

Tristan Gall said...

Dr Venema

Kudos for your succinct and sensible review. I wish more of those in the climate blogosphere were able to maintain a level of civility.

Anonymous said...

Your main criticism is that the homogenised data is all similar, but what do you expect when the best data gets put through the meatgrinder with the worst data? Why does NOAA use data that they know or should know is suspect? Is this the practice in the rest of the world?

Let's wait to see what the result is of homogenising the best data, where everyone agrees that the siting is not going to affect the measurements. The finalisation of this paper was rushed and while the raw result is interesting, it's too soon to come to a conclusion.

Victor Venema said...

Statistical homogenization algorithms are not meat grinders. Please inform yourself on how the work. That they can successfully remove measurement artifacts is demonstrated, for example, in a recent blind validation study of mine.

Your suggestion for a follow-up study is a good one. That sounds more interesting as Watts et al. With the old classification this would not have made much sense, but with the new one it may be interesting. As far as I know the current algorithm to homogenize the USHCN dataset already explicitly excludes the use urban stations as reference for the others.

lderry said...

yes, this notion that the troposphere "should" be warming faster than the surface is a complete red herring. It is primarily a phenomenon of deep convection in the tropics, and so has some global impact because the tropics are important globally. But, last time I checked , the good ol' USA wasn't in the tropics. The Klotzbach paper they refer to doesn't make a lot of sense, and I'm actually surprised it was published. They claim that increased CO2 and H2O (or clouds) wouldn't allow the surface to cool as much at night , and this introduces "bias?" No, it's called greenhouse warming. Jeesh. Seems like Watts suffers suffer from "siting errors"? This explanation reflects either serious ignorance or deliberate deception, coming from some of the same crew who used to argue vociferously that the troposphere was cooling. Gotta admire their intellectual "flexibility".

Anonymous said...

With respect, you said:

"This is the part of the homogenization literature that I am not so familiar with yet; it is frighteningly huge."

So are you advocating throwing bad data at a method you're unfamiliar with? The answer is no but I think you're overreaching.

All I'm advocating is that taking the best data, with the best methods is preferable. I am not against homogenization. Anthony is pointing out that NOAA hasn't done this.

Anonymous said...

This paper is already having its intended effect. The comment thread for the press release on WUWT is full of people leaping up to claim that it proves no warming is happening, there is no AGW, that any data saying otherwise is the result of cooking the books, calls for mainstream climate scientists to be put on trial because of it, that finally the data are looking as they "should," and so on.
While the moderator 'REP' is asking some commentators to "please read the paper" about questions they have, almost none of those notes are applied to the type of comments I mentioned above (the notable exception is in reply to this post by John Coleman, but that's very far down the page). There is next to zero effort to curb or correct the wildest misconceptions and fervor even on Watts' own blog. Those misconceptions are the real point of this announcement.
In fact, REP seems to think that this study 'invalidates all the major data sets' so that's likely a case of confirmation bias. Speaking of bias, someone in the comments raised the issue of moneyed interests and REP responded that 'No one funds WUWT and results and opinions are not dictated by non-existent "funders". Read the paper for yourself and decide if the purported results are supported by the purported facts. If you have a political problem, get lost.' This was in the midst of REP thanking several posters for donating to the Tip Jar and directed another one to it. Looks very like they're catering to a certain audience that's willing to pay for hearing what they like to hear.

-WheelsOC

dhogaza said...

"Anthony is pointing out that NOAA hasn't done this."

And exactly what qualifies Anthony to make such a claim? His high school education? His belief that changing a baseline will change the slope of a trend line, a misconception a middle school class in algebra would fix?

Victor Venema said...

To prevent this erroneous message from spreading too much, I tried to write this review as fast as possible. At least I did not see anything in any serious news papers or websites yet.

Some of the cheerers on WUWT are truly amazing. Especially that they are cheering again and again, day after day. Apparently, they so much want AGW not to be true.

Victor Venema said...

I am familiar with homogenization methods, just not yet with the literature on the Urban Heat Island. I the blog-o-sphere this is seen as an important topic and dominates the discussions about homogenization, but in the climatological community it is seen as a minor factor. That there are so many papers on this topic is not because of its perceived scientific importance, but because the general public is interested.

It is difficult to respond to this anonymous comment: one of us seems to have problems with reading comprehension.

Anonymous said...

I'm really glad that someone is pointing out the real implications of this paper and how it doesn't quite jibe with the sales pitch. If the findings don't alter the trends for CONUS much after all the necessary processing steps are made (Time of Observation Bias, rural vs. urban siting taken into account, gridding, etc.) then it shouldn't be pawned off as if there was a major difference in the final data, implicitly or otherwise. In the wake of the press release, pointing out that homogenization already removes most data quality problems that result from siting issues (if I'm reading you correctly) is a message that needs to be spread. There has been too much effort spent in misleading the public regarding data handling in climate science.

I don't want AGW to be true either. For me, there are too many awesome living things that would be stressed or extirpated with a warming climate, not to mention predictions that it'll exacerbate drought in my part of the country. I'd also like telekinesis to be real. What I want to be true has very little bearing on how the world actually works. I often forget that lesson, but science is there when I need to be reminded.

-WheelsOC

Gregor Vertacnik said...

I've just read the Watt's paper. As a climatologist experienced in monitoring, data quality control and also homogenisation I must say this is one of the worst papers about climate change I've ever read.

The role of station relocation, sensor change and time-of-observation bias has already been discussed in the Victor’s blog so I will concentrate at only one example.

Let's look at the USHCN v2 database. The list of stations is here: ftp://ftp.ncdc.noaa.gov/pub/data/ushcn/v2/monthly/ushcn-v2-stations.txt

Take stations with IDs 262780, 265168, 266779 – three medium elevation sites from Nevada (the last one is Reno). You can download the raw data from: ftp://ftp.ncdc.noaa.gov/pub/data/ushcn/v2/monthly/

If you look at the annual mean temperature time series of the three stations, you would notice a huge discrepancy. It is obvious such data cannot be used directly to study climate change …

I also don’t understand why all the trends in the paper are written with 3 decimal digits without any error estimation. Are the authors so confident in the results or is it just the lack of mathematical background?

dhogaza said...

"I also don’t understand why all the trends in the paper are written with 3 decimal digits without any error estimation. Are the authors so confident in the results or is it just the lack of mathematical background?"

John Christy's name is on the paper, surely he has the background. Also McIntyre, who endlessly *claims* to have the background.

Anonymous said...

Somebody should ask mcintyre for the data. I bet watts will not produce it

Anonymous said...

have some fun. write an email to all the authors of the paper.
request the data. You will recall and should note that McIntyre asked for data as a reviewer. Watts will never consent to releasing the data.

Anonymous said...

Of course he will, as soon as this charade has served its purpose and the free market has once again been protected.

Victor Venema said...

Part of the difference in the trends is due to the UHI, but another part is probably due to the selection bias mentioned in the update. It would need a careful study to estimate which term is more important.

The problem with bad siting exists everywhere, but the situation in the US is likely not the best example. They say that when automatic weather stations (AWS) were introduced in the US, the technician had one day to install the instrument, including laying electricity and data cables from the building to the AWS. Consequently, many AWS were installed close to buildings. Having learned from this experience other countries probably took a bit more time. Can someone from the US confirm this story?

Gregor Vertacnik said...

The main paper conclusion is based upon the Class 1 and 2 station data. However, there is nothing written about the site quality of each of these stations throughout the period 1979–2008 nor about relocations nor time of observation change (if there was any) nor the date of station type change (CRS/MMTS).

It would be nice if the authors listed all stations that didn't changed location, time of observation, and station type while being in the top two classes all the time. I believe there are at most few such stations. Raw temperature data (and quality controlled) from these stations could be directly used to estimate the CONUS trend. And even this is not enough for a firm conclusion about the trend as there may be other inhomogenieties present.

I wonder if there is any chance that Watts&colleagues manage to do this task?

Anonymous said...

There are error bars of 0.08 on the Satellite data (0.23 - .08 = 0.15) and I thought there was also supposed to be a hot spot, so the satellite trend does not have to be identical to that on land only.

And yes I know the difference between a trend and an offset. I am willing to wait for peer review and also to wait 15 years to see what happens with the climate. By then many questions will be answered as the sun's activity either will or won't change much, the 25-40 year weather cycle will have had an opportunity to go down or not, etc.

No reason for anyone to jump to any conclusions in either direction. Just wait 10-15 years. The planet will not be destroyed in that short time if temp goes up 0.2 C

Anonymous said...

Wow you've already read and analyzed the data before he even released the data? You sound just like half the folks at WUWT who just say atta boy Anthony.

He is doing the same thing he criticized Muller for and he admits it up front. There is even more excuse for a skeptic to try to get some publicity. The other side does it freely, even exploiting local weather events such as heat waves or last years tornoados when they themselves know that weather is not climate.

Anonymous said...

As WheelsOC notes, and as I and many others have noted elsewhere, this 'paper' is a last-ditch dog-whistle to rouse the public minority that consists of extreme deniers.

Stir this lot up and there is a knock-on through the more general community, and such a knock-on could have momentum for perhaps a year or more. I'm not sure whether it would dramatically affect the US elections, but here in Australia this nudging of FUD could be enough to help arch-right-winger Tony Abbbott continue his mocking of a carbon price, and his promise to repeal it if and when he wins the next federal election. Such a result would in turn profoundly affect the whole world's response to global warming for years to come.

This paper might be arrant nonsense, but its effect could well be serious, and even if nothing further happens with its publication. It's good to see these quick responses to the paper's release - in instances such as this laces always need to be tied quickly.

Bernard J.

Anonymous said...

Muller's BEST paper has already been REJECTED from the JGR.

http://www.rossmckitrick.com/ one of the papers reviewers.

[Update July 30: JGR told me "This paper was rejected and the editor recommended that the author resubmit it as a new paper."]

dhogaza said...

"Muller's BEST paper has already been REJECTED from the JGR."

Given that there are five, rather than one, papers, this comment indicates a rather basic misunderstanding what might have been rejected (assuming McK's telling the truth) and what the BEST people have produced.

dhogaza said...

Besides which, what would rejection of all of BEST's work have to do with the current thread, which is an examination of Watts' silliness?

dhogaza said...

Besides which, what evidence does McK have that the rejection was based on his review? The professional climate science reaction to BEST's efforts have been "boring, shows nothing new", it's entirely possible that mainstream reviewers also called for rejection.

Only the editor knows. You don't. McK doesn't.

Note that 'JGR told me "This paper was rejected and the editor recommended that the author resubmit it as a new paper."' doesn't say "because of your review".

Marcel Kincaid said...

Here you are, "Mogumbo", anonymously slandering people left and right and you have the gall to talk of violating trust? You make these charges but can't even be bothered to provide a specific link, saying "Do a search"? Why should any sensible person believe a word you say?

We're all actually well aware of the sickening "back story" of denialism and people like you and Watts.

Anonymous said...

I see a fair amount of attack here but little actual response to what is supposedly wrong with Watts' work?

As a layman, but with an analytic background - seems some simple questions are relevant.

1. Is Leroy 2010 an improvement?

As I understand it - by adding the thermal mass of the heat source or sink to the site quality equation - instead of only using distance as in Leroy 1999, common sense says you significantly improve site quality. At a fixed distance a match, a bonfire and a forest fire will have markedly different thermal qualities.

2. As a pilot I am well aware of the siting issues with airport stations - by their nature they are NOT representative of surrounding area weather. Their purpose is to tell me the conditions AT the runway - including the radiated heat etc.

Why would a airport station ever be used for monitoring climate data? Not only do they purposely capture UHI (or AHI - Airport/Runway Heat Island) but the variance from area climate data can be from zero to significant depending on temps, sunlight etc.

3. I've read many criticisms already on Watts work, and a single issue stands out. There has been no rebuttal I've seen about the claim that the homogenization by NOAA is to INCREASE the temps on the high quality rural sites to match the poor quality urban sites. And also to do the same with the newer MMTS sites which seem to typically read cooler than older technology.

IF the highest quality rural sites and the newer tech MMTS both seems to be in some agreement - both showing lower temps than the poor quality sites - then what possible reason is there to choose to increase the temps on the high quality sites to match the poorest quality sites data?

Last - why the ridicule? Anthony Watts as far as I can tell has put an incredible amount of unpaid effort in to his work. The scientific process is supposed to be about challenge. "Consensus" is not science - if it was we'd all still be members of the Flat Earth Society.

Why is it so hard to respect the efforts - and simply address the work. If his work is as full of holes as some claim then it should be simple to replicate it and point out the SPECIFIC errors and flaws, instead of the largely ad hominem attacks.

I want to know the real answers. I started a believer and became a skeptic based on the data. No one has been able to show me why 30 or 100 or 200 years is relevant to "climate change" when even the worst case predictions are largely within the natural variability of the last 11,000 years.

I am far more concerned at why we have plateaued. Why the normal an inevitable inter-glacial warm period spike seems to have been arrested and put on hold.

We know - pretty much beyond doubt - that a climate cycle is appx 100,000 years, the 'predominant' state during a climate cycle is glacial, and that we are overdue for the inevitable sharp decent in to a glacial period.

It is not anthropogenic global warming causing that - not too many soccer moms, coal plants or SUV's 11,000 years ago. Yet our temps have been remarkably stable that entire time - trending up and down within a narrow band - withe recent warming a fraction of that natural variability.

That answer is the one I think important.

Nick said...

Gish-galloping,I'm afraid. Point 3 is addressed in the post,and some of the comments...you obviously have read neither.

Victor Venema said...
This comment has been removed by the author.
Victor Venema said...

Anonymous: "I see a fair amount of attack here but little actual response to what is supposedly wrong with Watts' work?"

Could you be more specific? My post mentions what is good about Watt et al and where the study does not match the statements in the press release.

Anonymous: "1. Is Leroy 2010 an improvement?"

Did you read my post? I wrote: " Leroy (2012) will be happy that his new siting quality classification seems to work better as judged by the larger difference in the trends between the categories. That seems to be the main novelty. This result is worth a paper, I am not sure if it worth a press release."

Anonymous: "2. As a pilot I am well aware of the siting issues with airport stations - by their nature they are NOT representative of surrounding area weather. Their purpose is to tell me the conditions AT the runway - including the radiated heat etc."

The airports I know are typically surrounded my much green as no one wants to life close to all that noise. Thus the surrounding is more often than not good. A runway does not cause an UHI effect. As far as I understood, one of the main reasons for the UHI is that the surface cannot cool well at night because the surface only sees a little part of the cold sky. If you stand next to a runway you may feel the heat radiation from the runway and thus get the impression that that the temperature is higher. A temperature measurement is protected against radiation, meteorologists want to know the temperature of the air and do not want this to be wrong due to the heat.

Even if the temperature at an airport would be higher, this would only be a problem to determine the temperature trend, if this effect is getting stronger in time. When there is more and more development (especially of high buildings) at the airport. If this happens, homogenization is there to remove this artificial temperature increase.

Anonymous: "3. I've read many criticisms already on Watts work, and a single issue stands out. There has been no rebuttal I've seen about the claim that the homogenization by NOAA is to INCREASE the temps on the high quality rural sites to match the poor quality urban sites. And also to do the same with the newer MMTS sites which seem to typically read cooler than older technology."

Again did you actually read the post? I wrote: "In the press release it is also emphasised that the temperature trend after homogenization is stronger than in the raw data. Maybe Mr Watts thinks this is new, but, e.g., Menne et al. (2009) already stated that the introduction of automatic weather stations (the transition from Liquid in Glass thermometers to the maximum–minimum temperature system) caused a temperature decrease in the raw data of 0.3 to 0.4 °C. This temperature jump has to be and was removed by homogenization."

The increase in the temperature trend is thus not due to adjustment of stations with a low trend to the ones with a strong trend, but due to the change in the way the temperature is measured, the transition from LiG to MMTS and also probably due to a change in the time of observation. Homogenization removes these artificial jumps and because they caused artificial cooler temperatures, the homogenized data shows a stronger trend. There is no evidence in Watts et al. that the good stations are adjusted to the bad ones. Watts et al. does not even study how homogenization algorithms function.

Victor Venema said...

Anonymous: "Last - why the ridicule? Anthony Watts as far as I can tell has put an incredible amount of unpaid effort in to his work. The scientific process is supposed to be about challenge. "Consensus" is not science - if it was we'd all still be members of the Flat Earth Society."

Again where did you find ridicule? Asking this question is almost provoking this as there would be sufficient reasons for ridicule. I hope the conversation at this blog stays on the topic.

As far as I can tell, Watts lives very nicely from this blog. To call this unpaid is strange. (There is nothing wrong with being paid.)

The scientific process is about being critical of your own work and the work of your peers. Making controversial claims out of ignorance or due to sloppy thinking do not bring scientific progress and may even hinder it. That is why scientific articles are nowadays peer reviewed, to reduce the amount of erroneous statements, which would only confuse a reader that is not an 100 % expert.

Anonymous: "Why is it so hard to respect the efforts - and simply address the work. If his work is as full of holes as some claim then it should be simple to replicate it and point out the SPECIFIC errors and flaws, instead of the largely ad hominem attacks."

Again, did you read the post? It details some of the specific errors and argumentative flaws of Watts et al. Where is there an ad hominem attack in the post? It would have been easy. I hope my readers are smart enough, not to be provoked. Do you prefer a mud battle over a argumentative discourse?

The rest of your anonymous remarks are outside of my expertise.

Gregor Vertacnik said...

Nick, I agree with your opinion.

Airports may not be useful for climate analysis if the surrouding area isn't changing and the station isn't relocated too many times. As far as I know most airport stations are located far from trees, which can also worsen the quality of time series.

Once again, Watts work is more or less useless to study climate change, as he didn't remove or quantify all the known influences from the series (relocations, time of observation bias, instrument change etc.)

The most visible difference between anthropogenic and previos natural climate changes is the rate of change. Interglacial-glacial cycles required thousands of years to achieve a few degres C change, while right now we're warming the Earth at least (probably much more) ten times faster. The rate itself is more challenging for adaption than the size of climate change.

Anonymous said...

Victor

I should have been more clear ... I was primarily responding to comments from others here. You are one of few who seems to have made any real effort to address the issues raised.

My comments on airport based sites are accurate in my personal experience. As a pilot airports sites will almost always have higher temp readings than surrounding areas. The stations are intentionally designed to reflect the actual temps including any heat island effect as calculating density altitudes accurately is very important.

As to homogenization ... if you have high quality well sited rural stations - why would you adjust them to match lesser quality, inferior located sites.

Gregor Vertacnik said...

One mistake in my previous post. It should read that way:

Airports may be useful for climate analysis if the surrouding area isn't changing and the station isn't relocated too many times

Anonymous said...

"The most visible difference between anthropogenic and previos natural climate changes is the rate of change. Interglacial-glacial cycles required thousands of years to achieve a few degres C change, while right now we're warming the Earth at least (probably much more) ten times faster. The rate itself is more challenging for adaption than the size of climate change."

This is simply not remotely correct.

Many periods - even within the last 15,000 years - have seen temps rising far faster than at present:

http://3.bp.blogspot.com/--Pkh3YnMDwY/TyPtQaLsRUI/AAAAAAAAAXI/aHtN0trPdvw/s1600/Natural_global_warming%252Bover%252Blast%252B10%252B000%252Byears.jpg

http://www.foresight.org/nanodot/wp-content/uploads/2009/12/histo1.png

Our temperature has remained remarkably stable over the last 11,000 or so years ... trading up and down in a complicatedly small range, with the recent warming well within that natural variability.

Victor Venema said...

Dear anonymous, to compute the air density you need the air temperature itself. If the temperature is wrong due to a radiation error, you would get the wrong air density. Do you have any scientific studies that show that the UHI effect is especially strong at airports? It is not my area of expertise, but I would be highly surprised if that were the case.

The idea of homogenization is not to adjust high quality stations to inferior ones. That is just WUWT lore. The surrounding stations are used to detect jumps and periods with trends that only occur in single station. Please have a look at my recent post on homogenization methods and their validation.

Gras Albert said...

Victor

Am I mistaken, I was under the impression that USCN Surface Stations are categorised using a substantially simplified form of Leroy 1999 (CRN, detailed in U.S. Climate Reference Network Site Information Handbook, NOAA/NCDC, 15/1/2003) which, though based Leroy 1999 only considers distance to nearest heat source, without compensating for the size of that source. The CRN simplification thus unnecessarily disqualifies many good stations that have nearby walkways, small buildings etc.

Perhaps you could provide a reference to my error?

Victor Venema said...

You are right, a very strict version of the Leroy (1999) was used. In the article "On the reliability of the U.S. surface temperature record", Menne and colleagues (2010) write: "The exposure characteristics of a subset of USHCN stations have been classified and posted to the Web by the organization surfacestations.org based on rating factors specified for the USCRN [Climate Reference Network, 2002; Leroy, 1999]. Note that the rating system used for the USCRN and retrospectively applied to the USHCN is more restrictive than long‐accepted standards used in the siting of U.S. Cooperative Observer Network stations (and therefore the USHCN), especially in terms of the allowable distance to a building or other obstruction. For this reason, a reasonably well‐sited station by Cooperative Observer standards may be assigned a moderately poor rating according to USCRN standards. Nevertheless, to evaluate the potential impact of exposure on station siting, we formed two subsets from the five possible USCRN exposure types assigned to the USHCN stations by surfacestations.org, and reclassified the sites into the broader categories of “good” (USCRN ratings of 1 or 2) or “poor” exposure (USCRN ratings of 3, 4 or 5)."

Fall et al. (2011) use the same classification: "We make use of the subset of USHCNv2 data from stations whose sites were initially classified by Watts [2009] and further refined in quality control reviews led by two of us (Jones and Watts), using the USCRN site selection classification scheme for temperature and humidity measurements [NOAA and NESDIS, 2002], originally developed by Leroy [1999] (Table 1)."

As far as I know NOAA validated the classification performed by Jones and Watts on a sample of the stations and found it to be reasonable.

Climate Reference Network (2002), Site information handbook, NOAA/NESDIS CRN Ser. X030, CRN Rep. NOAA‐CRN/OSD‐2002‐
0002R0UD0, Natl. Climatic Data Cent., Asheville, N. C. (Available at ftp://ftp.ncdc.noaa.gov/pub/data/uscrn/documentation/program/X030FullDocumentD0.pdf)

Fall, S., Watts, A., Nielsen‐Gammon, J. Jones, E. Niyogi, D. Christy, J. and Pielke, R.A. Sr., 2011, Analysis of the impacts of station exposure on the U.S. Historical Climatology Network temperatures and temperature trends, Journal of Geophysical Research, 116, D14120, doi: 10.1029/2010JD015146, 2011.

Leroy, M., 1999: Classification d’un site. Note Technique no. 35. Direction des Systèmes d’Observation, Météo-France, 12 pp.

Menne, M.J, C.N. Williams Jr., and M.A. Palecki. On the reliability of the U.S. surface temperature record. J. Geophys Res., 115, D11108, doi: 10.1029/2009JD013094, 2010.

Victor Venema said...

Dear anonymous. If you find the comments here lacking in substance and full of personal attacks, may I ask what you feel about the comments at WUWT?

"Kudos! I can hear the stuttering and sputtering from the Usual Suspects already."

"the overstated increase is just an honest mistake??"

"And nice job, Anthony. I’ve been on pins and needles for the last two days waiting to find out what was going on…and I don’t mind admitting that I was afraid you’d sold WUWT, or had gone over to the ‘dark side’."

"“Poorly sited station trends are adjusted sharply upward” — adjusted by who?"

"Garbage in, garbage out, BEST. This is what you should have been working on, had you been honest brokers. Anthony, well done, BRAVO!"

"Oh, this should fuel the fires for quite a while!"

Aeroponicus said...

All that need be done to stop any follower of the Al Gore school of 'install my policies in spite of the election or balmy weather might kill you all!' is ask them

'Why hasn't the Infrared Astronomy field simply come forth with the photos of the atmosphere 100, 50, 20, 10 years ago, and show the world the rising level of infrared in the atmosphere?'

Because
it is not there.

Victor Venema said...

Dear Aeroponicus,

"Because it is not there"

Do you mean because with this data you can show that there is no temperature trend? In this case could you post the reference?

Or do you mean because this study has not been performed yet?

In the second case. Why don't you do this study yourself? You seem to be the one who thinks that such a study can be done. I would expect that astronomical photos do not have sufficient consistency (same technology, frequency and bandwidth) to see a small trend of less than one percent (in Kelvin, which is the appropriate scale for IR measurements). But feel free to prove me wrong. If it works, it could be a valuable independent information source.

Gras Albert said...

I see, so the paragraph in your 'Style' summary above

I am not sure whether the finding that the weather station classification scheme of Leroy (2010) is better than Leroy (1999) is of this category.

might require some revision!

Are you suggesting that the crude approach employed in CRN does not significantly influence results by marking what are in fact good quality stations as 'bad'?, if so perhaps you could provide a reference to the methodology, data and code which supports this argument, by comparing CRN to Leroy 2010 categorisation and the consequential downstream effects on 'adjustments'

If so, you may of course be correct, but it would be nice to see something more than hand waving in support

Gregor Vertacnik said...

This is simply correct, as I was talking about the globe and not a single region. Your links show only central Greenland time series and we know this region is not representative for the globe on short time scales ...

Please post the graph for the whole globe showing the rate of change similar to the current one (last ~100 years).

Anonymous said...

@- Aero

The equipment did not exist 100 years ago to measure downwelling IR with the required accuracy.

But you might want to check into the ERBE program and the network of ground stations that measure downwelling IR and have detected a change on the spectra and amount .
Try here to see what is being done. -


www.bsrn.awI.de/

Izen

EliRabett said...

Hi,

It looks like the answer is that the US rural stations have a well known negative bias in the trend because of the time of observation varies over time. This is the price we pay for the benefits of a volunteer network. Since Watts et al. draw their conclusions from the raw data they missed this.

Eli was only able to figure this out after reading your review, because it sharpened the question enough to go look at the homogeneity corrections and specifically for the ones that affected the rural stations the most.

Best
Eli Rabett

dhogaza said...

And McIntyre kicks himself for not noticing that Watts was ignoring TOBS issues of the sort Eli points out:

"When I had done my own initial assessment of this a few years ago, I had used TOBS versions and am annoyed with myself for not properly considering this factor. I should have noticed it immediately. That will teach me to keep to my practices of not rushing."

Context is he'd done some playing around with the surface temp record a few years ago, properly taking TOBS into consideration.

But when asked by Watts for help with the stats - at the last minute (thus the reference to rushing despite his better judgement) - he failed to notice that the paper ignores TOBS issues.

From other things he's saying it's obvious he was surprised to be listed as an author, as all he did was help with stats a bit, as a favor to Watts.

This is not turning out all that well for Watts ...

"Since Watts et al. draw their conclusions from the raw data they missed this."

No, Watts didn't miss it, he thinks TOBS changes really do falsely inflate the trend. He's said so many times regarding homogenization, and has not discriminated between the various problems with the station data that are sorted out by homogenization. As far as he's concerned, adjusting for changes in TOBS is just another trick used by mainstream scientists to falsely inflate temp trends.

Victor Venema said...

No Nobel price for Anthony Watts?

An experienced colleague, knowledgeable about the US network gave me the tip to look into the time of observation bias (TOB). Thus this may well explain much of the differences in the trends of the raw data.

If this is really an important effect, I do not see it as an excuse that Anthony Watts is not an academic insider. This is something one should check before publishing and I would see this as a lack of rigor. That there is an TOB in the US network is no internal secret, but known from the literature, for example, studied in Vose et al. (2003).

Thus we now have three reasons, why the technical problems may cause a difference in the trends of the raw data:
1. Time of observation bias stronger in rural stations.
2. More problems due to the UHI in the bad stations.
3. Selection bias (bad/good stations at the end of the period may have been better/worse before)

Sounds like the first two problems can be solved by homogenization. And the third problem is only a problem for this study, but not for the global temperature trend.

Time for the Team Watts to start analyzing their data a bit more.

Russell S. Vose, Claude N. Williams Jr., Thomas C. Peterson, Thomas R. Karl, and David R. Easterling. An evaluation of the time of observation bias adjustment in the U.S. Historical Climatology Network. J. Geophys. Res., VOL. 30, NO. 20, 2046, doi: 10.1029/2003GL018111, 2003.

dhogaza said...

And McIntyre is saying that now he'll just have to do the TOBS adjustments himself.

I foresee a soap opera developing: McIntyre properly corrects Watts' results based on incorporating TOBS adjustments. This will sink Watts' conclusion that 50% of the observed trend is "spurious".

How will this soap opera conclude? Watts backing off his proclamation and publishing a paper that says, hey, yeah, just as everyone else understands, there's nothing wrong with the NOAA temp trend cacluations? Watts ignoring McIntyre and trying to publish as is? Does McIntyre get his name off the paper in this case?

We're already seeing hissy fits over BEST. Mosher, once one of the most reliable denialists around (and one of the more intelligent) helped the BEST team out with R wrangling and the organizing and storage of the raw data. He's now being villified by the Wattsonian clique and the Currians. Even his co-auther Tom Fuller has said unkind things about BEST. Mosher called the stats work in Watts' paper - McIntyre's contribution - "hurried and sloppy" (which apparently McIntyre doesn't entirely disagree with).

Popcorn, people, popcorn.

Anonymous said...

If the Time of Observation Bias is the cause of the differences in the trends, I still see one way out for Anthony Watts. He can send the manuscript to a statistics journal.

Victor Venema said...

Dear Albert Gras, So you see a better weather station classification scheme as something of similar importance as finding the Higgs boson (or even more important)?

Are you suggesting that this result, which does not even change the estimate of the temperature trend for the US, is important enough to bypass peer review and to immediately run to the press (who cannot judge the rigor of a scientific study) before it has been judged for obvious errors by other scientists?

The how-good-stations-turn-bad meme does not become true by repetition. Please educate yourself on how homogenization actually works.

Gregor Vertacnik said...

One can estimate the TOBS effect on the basis of Menne et al. 2010. Table 1 shows the difference between unadjusted and TOBS adjusted data for several types of stations/siting/Tmax/Tmin. Corrections span from 0.04 up to 0.10 and on average around 0.07 or 0.08 °C/decade. This changes Watts conclusion significantly, no doubt about that. The other that important factor is sensor change (Figure 3 (e) in Menne et al. 2010). This paper also states clearly:

"We note also that only about 30% of the good exposure sites currently have the newer MMTS‐type sensors compared to about 75% of the poor exposure locations."

Just the issues mentioned (apart from others) make conclusions from the Watts et al. article very doubtful.

Menne, M. J., C. N. Williams Jr., and M. A. Palecki (2010), On the reliability of the U.S. surface temperature record, J. Geophys. Res., 115, D11108, doi:10.1029/2009JD013094. (http://www1.ncdc.noaa.gov/pub/data/ushcn/v2/monthly/menne-etal2010.pdf)

Victor Venema said...

However, the poor stations for this classification are different from the new classification scheme. We will have to wait what the new analysis (by McIntre) will show. It is not likely, but Watts could be lucky.

"We note also that only about 30% of the good exposure sites currently have the newer MMTS‐type sensors compared to about 75% of the poor exposure locations."

That makes sense, as the installation of the MMTS often made the exposure worse. As the MMTS readings are colder than the older Stevenson screens (Cotton region shelters), this would mean that the heating effect of the poor exposure is even larger.

Anonymous said...

bull ...

Watts specifically noted the TOBS issue and need for further review in the paper:

" the Conclusions section of the paper does note the TOBS issue is one that needs more investigation:

“We are investigating other factors such as Time-Of-Observation changes which for the adjusted USHCNv2 is the dominant adjustment factor during 1979-2008.”

Seems to make it clear that it was not overlooked – the issue was acknowledged and noted for future review?"

And what do you expect from Mosher ... he is deeply involved with the paper Watt's work is set up to refute.

There is little evidence the work that was done is "sloppy" or inaccurate. There are comments, acknowledged in the paper, that further review is need regarding TOBS.

Your posts here are a perfect example of ad hominem attack while offering no cogent contribution to the discussion.

Victor Venema said...

Why is leaving out the dominant adjustment factor not sloppy?
But you are right, that sentence is really in the manuscript. You may write something like that about an extension of the work or a factor that is likely minor, but a lot of work.
How can you publish a manuscript, even if it is only on the web, knowing that "a dominant adjustment" is not taken into account? And especially when it would be so easy to do.
I am baffled. Was this manuscript never intended to be submitted to a scientific journal?

ligne said...

presumably he's taken a leaf from South Dakota: http://genome.fieldofscience.com/2010/02/south-dakota-legislature-declares-that.html

Mark Shead said...

Interesting thread. I appreciate the discussion and most especially the lack of ranting, thank you!

A question has come to mind...

It appears that NOAA has concluded that station temperature measurements are either "correct" or "too low" far, far more often than they are "too high". Does anyone know why?

For instance, has NOAA been directing that readings be taken at night, when temperatures would be lower than during daylight hours? If so, the overwhelmingly upward adjustments would certainly make sense... but why would routine sampling that would result in abnormally low readings which routinely require upward adjustment be desired?

JB said...

Hi Victor,
This seems to be the take away message that Watts supporters are claiming is the big issue.

"· Poorly sited station trends are adjusted sharply upward, and well sited stations are adjusted upward to match the already-adjusted poor stations."

"to match"

comment

dhogaza said...

"It appears that NOAA has concluded that station temperature measurements are either "correct" or "too low" far, far more often than they are "too high". Does anyone know why?"

Victor links to three papers that address homogenization issues in his original post. So clearly, yes, people know why. You could take a look at the papers ...

Here's a summary of a talk by Thomas Karl that addresses it, though only briefly:

http://agwobserver.wordpress.com/2010/06/10/thomas-karl-a-lecture-on-noaa-surface-temperature-analysis/

Short summary, time of observation changes have most frequently been in the direction that leads to a cooling bias. There's also an issue with changing of sensor types, which has also been shown to introduce a cooling bias.

"shown" as in observed by measurements, in both cases.

Mark Shead said...

Thank you dhogaza. I'll do some further reading.

My initial thought from imagining myself in the position of a volunteer collecting station measurement info, is that I would elect to go at a consistent time each day for daily monitoring. Due to my work schedule, the only consistent time of day available would be between 7am and 8am each morning which may indeed result in temperatures that would routinely require an upward adjustment to reflect accurate temps for "the day" as opposed to the relatively cool "morning of the day".

Does anyone have a URI to any official NOAA instructions / guidance for volunteers regarding time-of-observation?

Mark Shead said...

I think I found it at http://www.nws.noaa.gov/om/coop/

Although I could only find reference to "precipitation stations" rather than anything specifically mentioning temperature monitoring stations. (Why is that difficult to find exactly?)

Specifically this document: "Cooperative Station Observations"
http://www.nws.noaa.gov/directives/sym/pd01013015curr.pdf

"5. Preferred Time for Taking Observations. Observations at precipitation stations should be taken at 7 a.m. local time, however any time between 6am and 8am is usually acceptable if coordinated and approved by the NWSREP. Observations should be taken at the same time every day throughout the year if at all possible using standard time in the winter and daylight saving time in the summer if applicable.

While NCDC would prefer climatological observations be taken at midnight each day, this is not feasible for the observers. Therefore, stations reporting both precipitation and temperature should report at a time agreed to with the NWSREP. Unless otherwise directed, precipitation and temperature should be observed at the same time each day. Evaporation and soil temperature stations should observe all elements in the morning. "

So NOAA is recommending observations be taken at the same time each day and at a time between the hours of midnight and 8am which explains (to me) why most of the raw measurements would require upward adjustment.

I'm satisfied with that although I would prefer to read this in an official NOAA document for temperature station observers.

Rattus Norvegicus said...

As the discussion of the effect of the TOBS adjustments here has shown, applying TOBS results in consistently cooler temps in the past. This results in higher trends in stations which require it (and most do).

As they say, RTFR (read the FINE reports).

Mark Shead said...

I returned to post again and re-read my prior post to find that my choice of wording was somewhat confusing. I meant, "I would prefer to read an official NOAA document which states Preferred Time for Taking Observations of temperature stations specifically (as opposed to precipitation stations which may or may not also be temperature stations.)"

I had assumed that time of observation adjustments were done to normalize temperature readings to a specific time-of-day-temperature across all stations. However from what I am reading it appears that it is only used to normalize time-of-observation-temperatures for a specific station when time-of-observation has changed for that particular station? That doesn't seem correct but I can't find anything to refute it.

In other words, a specific station's data is recorded at 7am daily for 10 years then it changes to 8am daily for ten years and TOBS corrects for the different time used in one of the 10 year periods but the adjustment is only for that specific station and no TOBS adjustment is made to compensate for a different station which was always read at 9pm despite 9pm being significantly different than 7 or 8am?

And I'm most curious of all about why a web search for "how to adjust temperature for time of observation" does not immediately lead to a clear, step-by-step list of instructions as to how TOBS adjustments are (or should be) calculated? Is there any official (or widely accepted) procedure?

Carrick said...

Rattus, my guess is you're putting too much faith in TOBS here. There are other issues as well, one of these (a change in mean latitude over time of the reporting stations) leads to an effect with the opposite bias.

When you are dealing with complex phenomena, it's best not to put all your eggs in one basket. [It's also BEST not to rush your papers out to meet the deadline for AR5.]

Anonymous said...

"...my guess is you're putting too much faith in TOBS here"

Why guess?

There's work that's been done to justify accounts for time-of-observation bias. Why not assess it, and if you detect a flaw, counter it?

Or are you just trying to shoot at the side of the barn?


Bernard J.

JB said...

Found it as one of your comments, thanks Victor

The increase in the temperature trend is thus not due to adjustment of stations with a low trend to the ones with a strong trend, but due to the change in the way the temperature is measured, the transition from LiG to MMTS and also probably due to a change in the time of observation

Gregor Vertacnik said...

There is one example, how much TOBS matters. I calculated daily lows and highs for Ljubljana, central European city with temperate climate.

Mean highs and lows (°C) from the automated station in the period of 2002-2011 are as follows (end time of 24-hour period is given):

7 a.m. local (solar) time: 16.1 / 6.7
12 a.m.: 16.5 / 7.2
7 p.m.: 16.4 / 7.2
9 p.m.: 16.3 / 7.1
12 p.m.: 16.2 / 6.9

9 p.m. time of observation is used in Slovenian climate network, observers read both minimum and maximum daily extremes at that time. If we join observer's data series of Tmin and Tmax with the automated station series, we need to be careful to properly account for possible TOBS change.

Corrections of TOBS bias are also site-specific as the diurnal range may be very different from location to location.

Unknown said...

Fascinating arguments. I found this through a link on Climateprediction.net Out of interest does anyone know if these data sets go into the climate models that my computer is busy crunching?

Dave

Anonymous said...

The USHCN Version 2 Serial Monthly Dataset page:

http://www.ncdc.noaa.gov/oa/climate/research/ushcn/

Victor Venema said...

Dave, you will have to ask at ClimatePrediction. As far as I know they are studying the sensitivity of Climate models to perturbations in the model constants and structure. For that you do not need any empirical measurements. For validation of the models you do and to determine the initial atmospheric state in case you want to make a prediction.

Victor Venema said...

Good you found it. That would be exactly my answer.

J Bowers said...

Courtesy of AGW Observer: Papers on time of observation bias (going back to 1890).

Victor Venema said...

Thank you, I just wanted to add AGW observer link to this page.

For those of you that can read German, there are two studies going back to 1848 on the time of observation bias. Well before climate change became a political issue.

Kreil K, 1854a. Mehrjährige Beobachtungen in Wien vom Jahre 1775 bis 1850. Jahrbücher der k.k. Central-Anstalt für Meteorologie und Erdmagnetismus. I. Band – Jg 1848 und 1849, 35-74.

Kreil K, 1854b. Mehrjährige Beobachtungen in Mailand vom Jahre 1763 bis 1850. Jahrbücher der k.k. Central-Anstalt für Meteorologie und Erdmagnetismus. I. Band – Jg 1848 und 1849, 75-114.

Victor Venema said...

Yesterday Steve McIntre of ClimateAudit and co-author of Watts et al. admitted that his contribution was made in a hurry and that the missing time of observation bias is a serious problem.

People have quite reasonably asked about my connection with the surface stations article, given my puzzlement at Anthony’s announcement last week. Anthony described my last-minute involvement here.

Whenever I’m working on my own material, I avoid arbitrary deadlines and like to mull things over for a few days. Unfortunately that didn’t happen in this case. There is a confounding interaction with TOBS that needs to be allowed for, as has been quickly and correctly pointed out.

When I had done my own initial assessment of this a few years ago, I had used TOBS versions and am annoyed with myself for not properly considering this factor. I should have noticed it immediately. That will teach me to keep to my practices of not rushing. Anyway, now that I’m drawn into this, I’ll have carry out the TOBS analysis, which I’ll do in the next few days (at the expense of some interesting analysis of Esper et al.)


Now also Roger Pielke Sr. distances himself from Watts et al.

UPDATE #2: To make sure everyone clearly recognizes my involvement with both papers, I provided Anthony suggested text and references for his article [I am not a co-author of the Watts et al paper], and am a co-author on the McNider et al paper.

UPDATE: There has been discussion as to whether the Time of Observation Bias (TOB) could affect the conclusions reached in Watts et al (2012). This is a valid concern. Thus the “Game Changing” finding of whether the trends are actually different for well- and poorly-sited locations is tenative until it is shown whether or not TOB alters the conclusions. The issue, however, is not easy to resolve. In our paper


Also first author Anthony Watts admits that the study was not of high quality.

Thanks to everyone who has provided widespread review of our draft paper. There have been hundreds of suggestions and corrections, and for that I am very grateful.

I would be ashamed if I had produced a manuscript that needed hundreds of suggestions and corrections. I would take a holiday, analyze how this could happen and whether I should stay in science.

Victor Venema said...

It's okay to be wrong, but then you should fail in an interesting way that helps many scientists to understand the topic better.

Windchasers said...

No error bars, either, which are quite important in determining rate-of-change.

Also, the data in this graph cut off in 1950, so, no, it doesn't show the modern warming. This is described in the original paper, which I recommend you check:
"Temperature, accumulation, and ice sheet elevation in central Greenland through the last deglacial transition", by Cuffey and Clow, 1997, JGR.

The chart is based on a modification by Don Easterbrook from that paper. Easterbrook's modification can be found here:
http://www.globalresearch.ca/index.php?context=va&aid=10783

And as Gregor said, Greenland temperatures aren't representative of the entire globe.

True skepticism requires carefully checking your sources, the data, and actively looking for ways to disprove your hypothesis. I recommend a bit more skepticism; don't be so quick to accept the first thing you read on the internet.

Mark Shead said...

Thanks anon. That didn't offer a step-by-step but I found this which explains the/a process for doing so.

http://www.john-daly.com/tob/TOBSUM.HTM

JB said...

OH my Anthony

Speaking of statistics, Watts states that he started teaching himself statistics on Friday afternoon and posted the paper on Sunday afternoon. Overturning a big chunk of climate science in the process.

it is becoming painful, but due thanks to Victor for his early and valued appraisal.

will follow your blog from now on. John Byatt

JB said...
This comment has been removed by a blog administrator.
Victor Venema said...

Removed double comment.

JB said...

Horatio does it again

http://tamino.wordpress.com/2012/08/01/much-ado-about-nothing/#comment-64997

Victor Venema said...

And now Anthony Watts does it again and promised that the revised manuscript is finished "in the next day or two".

It would be better if he first finished the analysis, you never know which problems turn up during this. When the analysis is solid, you have read and described all relevant literature and you only want to work on the clarity of the text a bit more, that it the moment you may announce to put a manuscript on the web in a few days. Even in that case, I would wait at least a week, to be able to reread and edit my own manuscript with a fresh mind.

Anthony Watts does not seem to be able to learn from past experience.

JB said...

Just in case anyone missed tamino's great respont to victor's comment as above at open mind

Response: Analyze first, announce later? Take your time and be careful? Devote lots of thought to your results? How radical of you.

Perhaps your process is different because your goal is different.]

Victor Venema said...

So, Mogumbo Gono, you are saying this study was published on the web because of the rivalry between Watts and Muller? Interesting that a "sceptic" himself states that the main aim of the manuscript was not a better understanding of the climate system.

Victor Venema said...

I am not so sure whether I find the "satellite argument" that convincing. Satellites have their own inhomogeneity problems, new satellites or even new types of satellites go in and out of the record. And there is not much redundancy (multiple satellites measuring at the same moment) as in the surface record, which helps you to find the inhomogeneities.

Furthermore the satellites also do not measure temperature directly, but only retrieve it based on the radiation the satellite receives. Changes in, for example, clouds and aerosols could produce an artificial trend in the satellite record.

If there were a difference between the surface network trend and the satellite trend, I would intuitively start search for problems with the satellite data.

The satellites are great for investigating problems around temporal sampling and to aid in studies around the Urban Heat Island effect. You can also use them to study regional differences in climate variability and trends.

Maybe I am biased, as I know the surface network and its processing better. However, when it comes to the long term temporal behaviour of atmospheric temperatures, I tend to trust the surface record much more.

Victor Venema said...

Not making any TOB corrections is like an epidemiological study on the relationship between diet and cancer, which would not correct for smoking or age. Just like we know that smokers typically have a poorer diet and it is thus easy to get spurious correlations, we have reasons to expect that the TOB correction correlates with station quality. For example, the voluntary stations tend to be rural, the professional ones (staffed around the clock without TOB problem) tend to be in cities.

Even if by coincidence it will turn out that the TOB corrections not to create any problem, you cannot call something a scientific study and leave out such a confounding factor. This start of Anthony Watts scientific career is deeply flawed.