I’m a bit burnt out, so this is a just a few notes to quench some speculations about Steve McIntyre’s role and to help everyone understand what this week has been like.
- Evan and I have been working on this since June 2011, complete redo of all station ratings…huge amount of work. Evan deserves a huge a amount of credit. After Muller could not find strong signal that we knew must be there by physics of heat sinks…and neither could we in Fall et al 2011, we went looking, and discovered the new Leroy 2010 classification system and WMO ISO approval. We knew it would take a lot of work to get old metadatabase into shape. And so it began.
- Started on paper in Spring 2012, but some of the team of people onboard had no vested interest, and with their academic burdens and no budget to pay them anything they could only devote small bits of time for reviews and writings. No fault of theirs, but like herding cats when there’s no funding and all is pro bono.
- Evan and I decided to go ahead anyways and I started writing, steep learning curve as this was my first stint as lead author.
- About a week ago I learned Muller was going to release and do the media blitz, thought he’d be at EPW Senate hearing on August 1st too. (turns out he was passed over, John Christy will be there though.). IPCC deadline coming up too. Added anxiety.
- Tried to get stats guy to the stars Matt Briggs onboard early last week (he was on list of original authors) to help with significance tests, last big hurdle. Most graphs and analysis was done.
- Turns out Briggs was on vacation camping, no fault of his, it is summer…so I figured only way I was going to get this done was to shut down WUWT and stay home from short vacation with wife and kids in Yellowstone. They went on with grandparents and I went on authoring blitz with Evan and with Dr. Pielke Sr. helping edits. Christy provided support too and I helped him craft his EPW section on this.
- So made announcement Friday. Figured on Sunday at noon so WUWT could provide peer review, and dumped my plane tickets in trash. Admittedly I was a bit overwrought when I wrote it. I’m truly sorry if anyone was mislead. Dialed it back. Went on crash self taught stats diet…not my thing, but capable of learning. and being a broadcaster, deadline pressure is a huge motivator. You learn to get it done. On-air waits for nobody. Careers die when you miss deadlines.
- In his post Friday, Steve McIntyre truly didn’t know what this was about. He was out of the loop.
- Steve McIntyre, being the classic gentleman he is, emailed me and said “anything I can do to help, I’m here”. I took him up on the offer and he did all the stats tests from Friday afternoon to Saturday night, then polished last bit of text/graphs early Sunday morning. I owe him a huge debt of gratitude. He is a true gentleman and a scholar.
- Joe D’Aleo and Willis helped with editing/proofing too. Gary Boden solved an Excel map issue for us. Evan came up with powerpoints and helped editing. He was a machine. Pielke Sr. helped with edits and citations. Bob Phelan helped with some PR language. Thanks to all.
- And the result is what you see in the press release today.
- Finally got to take a shower today about 2PM. Prior to that, Kenji was offended.
- Now on to final polish thanks to WUWT peer review and submission.
Thanks everyone for your support and patience! – Anthony
David 1 – Goliath 0
Interesting to see if Watts et al. 2012 gets picked up for next IPCC report. Somehow, why do I get the feeling IPCC will end up including Muller’s study and simply ignore Watts et al. 2012?
Anthony, you’ve proven again that you are the much better chess player, not only in terms of strategy, but also in terms of integrity and innovation. Muller, second BEST again. LOL
Wait a minute…
Anthony. In your paper you state the following…
…The USHCNv2 monthly temperature data set is described by Menne et al. (2009). The raw and unadjusted data provided by NCDC has undergone the standard quality-control screening for errors in recording and transcription by NCDC as part of their normal ingest process but is otherwise unaltered. The intermediate (TOB) data has been adjusted for changes in time of observation such that earlier observations are consistent with current observational practice at each station. The fully adjusted data has been processed by the algorithm described by Menne et al. (2009) to remove apparent inhomogeneities where changes in the daily temperature record at a station differs significantly from neighboring stations….
Then near the end of the paper you say….
…We have shown that the site-classification value is a clear factor in the calculation of the trend magnitude. We are investigating other factors such as Time-Of-Observation changes which for the adjusted USHCNv2 is the dominant adjustment factor during 1979-2008.
Correct me if I am wrong here but your paper seems to essentially be looking at raw vs fully adjusted data, then looking at the impact of station categories, the two classification schemes, airports etc. But all the analysis seems to be raw vs fully adjusted. With a strong leaning towards ascribing the result you see to station siting issues.
But the elephant in the room here is TOB (Time of OBservation) changes. How has your study evaluated TOB factors to rule them out or at least understand them as a factor in this? Particularly as you point out, although not until the end of the paper, that TOB adjustments are the dominant factor in the USHCN adjustments. As they should be.
TOB has nothing to do with measurement quality, site rating factors, instrumentation etc. Changing the time of day that measurements are taken or even what constitutes a 24 hour period can have a huge impact on the data and then results. Until you have analysed TOB factors, what confidence can one place in these results.
Perhaps a better approach, if you didn’t want to explore TOB issues, was to have done the analysis based on the GHCN data in 3 stages. Raw data only based on Site Classification; TOB adjusted data only based on Site Classification; Fully adjusted data only based on Site Classification only.
Otherwise, how can you possibly know you are comparing apples with apples?
Here are some examples of past research suggesting how big the TOB factor might be:
http://journals.ametsoc.org/doi/abs/10.1175/1520-0450%281977%29016%3C0215%3ATOOTBA%3E2.0.CO%3B2
Time of Observation Temperature Bias and “Climatic Change”
Lawrence A. Schaal and and Robert F. Dale
Department of Agronomy, Purdue University, West Lafayette, Ind. 47907
Received: August 27, 1976; Accepted: February 14, 1977
Abstract
Historical changes in time of once daily maximum and minimum temperature observations at cooperative climatological stations from 1905 to 1975 have introduced a systematic bias in mean temperatures. Unless corrected, this bias may be interpreted incorrectly as climatic “cooling” and may also affect the assessment of agricultural production potential and fossil fuel needs. Maximum and minimum temperature data for two years from the National Weather Service station at Indianapolis International Airport were used to evaluate the differences between mean temperatures obtained by terminating the 24 h period at the midnight observation and the mean temperatures obtained by terminating the 24 h period at 0700 and 1900 hours, typical observation times for AM and PM observing stations. The greatest mean temperature bias occurs in March when a 1900 observation day yields a monthly mean temperature 1.3°F above a midnight observation, and a 0700 observational day gives a −1.3°F bias. Since the number of AM observing stations in Indiana have increased from 10% of the total number of temperature stations in 1925 to 55% in 1975, the March mean temperature shows a decrease of 1.2°F in the last 40 years, solely because of the change in substation observational times. Unless the time of observation bias is considered, the mixture of AM and PM observations complicates interpretation of areal temperature anomaly patterns. This bias is accumulated in monthly, seasonal or annual values of the mean temperature-derived variables-heating degree days, cooling degree days and growing degree days—and may provide misleading information for applications in industry and agriculture.
http://www.agci.org/dB/PDFs/05S3_DEasterling_Uncertain%20Record_0722.pdf
Issues Related to Uncertainty in the Observed Climate Record
…The U.S. Cooperative Observing Network (COOP) uses an observing system that records both the highest (maximum), and lowest (minimum) temperature in the previous 24h period, then these values are recorded by the observer at a set time each day (e.g. 7 am or 5 pm), and the thermometers are reset. Figure 1 shows the bias introduced into March temperatures by a change in the observing time from late afternoon to morning, which has been the trend in the COOP network over the past 20 years. This change has resulted in an artificial cooling in the time series due to a step change to cooler monthly averaged temperatures….
Figure 1 from Easterling’s paper (follow the link) shows -1.0 to -1.5 C impact on station temperatures across most of the continental US from TOB changes. Certainly enough to explain most of the difference between the raw and fully adjusted GHCN data. Leaving very little scope for site specific factors to be significant.
Until you incorporate TOB changes into your analysis, how can you actually make any meaningful claims about anything?
Thanks for the backstory, Anthony and well done, “The Team 2”!
Of course, this work was restricted to the US so what chance is there of global extension?
Interesting to see if Watts et al. 2012 gets picked up for next IPCC report. Somehow, why do I get the feeling IPCC will end up including Muller’s study and simply ignore Watts et al. 2012?
Beautiful smokescreen by all. McIntyre and Motl playing their parts perfectly, sending the alarmists off on a wild goose chase. Result: Muller’s pseudo-scientific claims killed off within a day of publication.
Anthony, you’ve proven again that you are the much better chess player, not only in terms of strategy, but also in terms of integrity and innovation. Muller, second BEST again. LOL
Delingpole at the Telegraph among the first to report.
Global Warming? Yeah, right
By James Delingpole July 29th, 2012
http://blogs.telegraph.co.uk/news/jamesdelingpole/100173174/global-warming-yeah-right/
Have a look at this chart. It tells you pretty much all you need to know about the much-anticipated scoop by Anthony Watts of Watts Up With That?
[…]
1013 comments so far
Congratulations to you and your crew, Anthony. If this is what you guys can do with no budget, Lord only knows what you could do with a fraction of the megabucks routinely thrown at global warming research.
Pointman
But the elephant in the room here is TOB (Time of OBservation) changes. How has your study evaluated TOB factors to rule them out or at least understand them as a factor in this? Particularly as you point out, although not until the end of the paper, that TOB adjustments are the dominant factor in the USHCN adjustments. As they should be.
You make an important and valid point. I commented in the main thread that TOBS will be the main way in which this paper will be attacked. I’ll have to through the paper in detail to understand what implications this has for Karl’s TOBS adjustment.
Anthony , pity you are not a Brit , such work would get you on the ladder to becoming ‘Sir’ Anthony Watts . My cap is doffed to you sir ! Well done you and your team . Cheers Tony
Excellent work, well done all.
Did you notice ChiefIO has been doing this –
http://chiefio.wordpress.com/2012/07/26/salt-lake-city-airport/
and that Steven Goddard had this on his site –
http://stevengoddard.wordpress.com/2012/07/29/another-eureka-moment/
You and the team have pulled a lot of loose ends together with this piece.
What will the warmistas do now (?) as even the raw data from a lot of urban and airports sites is now looking very suspect.
And again well done
TOB is important but perhaps not that important.
Anthony’s paper compares various classes of met. station. TOB is handled the same for all stations. The paper shows that ‘good’ stations display much less warming than ‘poor’ stations. Unless there is a systematic difference in the way TOB is handled for the different classes of stations, TOB shouldn’t much matter in the overall result. ie. ‘poor’ stations showed much more warming than ‘good’ stations.
When you all have rested a bit….
Please enjoy some of the fun we had with speculations 😉
http://suyts.wordpress.com/2012/07/27/wuwt-speculation-thread/
Again, Congratulations and a heartfelt thank you.
As Pointman has pointed out “If this is what you guys can do with no budget, Lord only knows what you could do with a fraction of the megabucks routinely thrown at global warming research.”
Hell hath no fury like a scorned weatherman.
Great job, Anthony. Get yourself a white hat. You deserve it.
There’s nothing to attack in this paper wrt TOBS. Watts 2012 included the adjustment. What they did was use Leroy 2010 in SHAP instead of Leroy 1999. Leroy 2010 is the new WMO- ISO gold standard so I don’t believe that leaves much room for attack.
Someone asked we don’t know how much trend TOBS is responsible for. Nonsense. NOAA provided a lovely graphs of the effect step. I’ve posted a link to the graph a great many times.
http://www.ncdc.noaa.gov/img/climate/research/ushcn/ts.ushcn_anom25_diffs_pg.gif
TOBS and SHAP were each responsible for about half the warming trend. Watts 2012 puts paid to SHAP. TOBS is next up for scrutiny. As far as I can determine TOBS is sound enough from a theoretical standpoint but there’s a long row to hoe between theory and application. In other words the meta-data that determines where, when, and how TOBS adjustment is applied to invididual stations may be a problem. The reason I’ve been suspicious of these adjustments since I learned of them a few years ago is that they produce ALL the warming trend from a raw dataset that has no warming trend in it. Adjustments to datasets like this nominally improve the data on the margins making it marginally more accurate. These adjustments don’t improve the data on the margins they introduce a trend that didn’t exist before the adjustments. This is suspicious because data with errors usually has errors in both directions and in large datasets the errors cancel out. Systemic error largely in one direction only, as was the case with both SHAP and TOBS is huge red flag. Watts saw the flag waving and it took him and his cohorts years to figure out what was waving it. Good on him for having the perspicacity to see and the perseverance to eventually ferret it out.
As with any good sceptic, I will not just automatically believe this due to it supporting my own bias, but I will await further peer review and then, so long as it remains credible I will congratulate you.
I do heartily congratulate you upon your achievement so far. But I refuse to get carried away too soon. If the conclusions of this impressive piece of work are true, then it will survive close and rigorous scrutiny and will be found to be true.
I can wait for that, for it will be well worth waiting for.
Thank you Anthony for your persistence and your dedication to the pursuit of TRUTH, no matter where the data lead you. As a real, dedicated supporter of serious, evidence based science, I thank you and salute your work.
So this is what the Maya meant when they said that (Watts)2012 would be the end of the (CAGW)world? 😉
OK, the 60,000 foot overview from a non-scientist, non-statistician:
1) There are significant differences between properly-sited and improperly-sited stations in the subject set, with improperly-sited stations showing greater warming.
2) There are significant differences between urban and rural stations in both properly-sited and improperly-sited stations in the subject set, with urban stations showing greater warming.
3) There are significant differences in airport and non-airport stations in both properly-sited and improperly-sited stations in the subject set, with airport stations showing greater warming.
4) The above three results hold true for both major station instrument types.
5) The differences in 1-3 above largely disappear after NOAA adjustments, which in every case increase the amount of reported warming by factors of 2 or more.
Items 1-4 are really just confirming that long-asserted station siting biases are real. To me item (5) is the real zinger. One would think some (or even most) adjustments would be slightly negative to compensate for siting bias; to have all the adjustments up so significantly suggests very faulty methodology.
I’d be really curious to know the intersection between the station set in this study and the one used for BEST — what percentage of the BEST stations were used here?
Keep in mind what we refer to as “improperly sited” stations may still be perfectly usable for their original purpose. I’m thinking of airport stations here: what matters in flight operations planning is the temperature (and therefore air density) at the airport. That the same aircraft would get more lift over a corn field three miles away is irrelevant.
The importance of station siting bias studies is to guide the selection of ones that are useful for documenting climate trends. Not all stations are.
@David Ross
“True for a single weather station. But we are dealing with multiple biases introduced at different points in time collectively to a large collection of stations (as well as to individual stations). Collectively, the introduction of those biases can result in a trend and there are many reasons to expect that would be upwards.”
Yes, but the discontinuities induced by abrupt change in microsite environement (new road, new building, etc…) did not occur simultaneously in all stations: it is the principle of homogenization.
“Urbanization and economic growth are well established trends. Energy consumption has risen -more air-conditioners are put in place. Car ownership has risen -more asphalt laid for roads and parking lots. Etc.”
Well established ? I can also imagine that some stations experienced improvement of their environement.. And there is also a cooling bias induced by urbanization: shadows of new buildings.
““the discontinuity is corrected by the homgenization procedure””
Could you be explicit, how is it corrected?”
When a discontinuity is detected (a large change in temperature mean), you can see in the neighbouring stations if the same jump exists. If no, you must adjust for the jump…
Thanks, Anthony, et al. for not giving up on this. It is personally very rewarding to have played a small part in this. Echoing Russ Steele, thank you for acknowledging the efforts of your volunteers.
Chris Dunn
@Glenn Skankey Tamblyn
I fully agree with your comment.
The main claim of the Watts et al. article is “Not only does the NOAA USCHNv2 adjustment process fail to adjust poorly sited stations downward to match the well sited stations, but actually adjusts the well sited stations upwards to match the poorly sited stations.”. IMHO, this is not well supported by the results, because, beside good/bad sitting, other factors (TOBS, instrument changes) play a role.
The solution seem straightforfard. Take the raw data, discard poorly sited stations, and then homogenize the new dataset. Compare the “official” trend with the trend of the homogeneized with only good stations: if the second one is well below the official trend, well, Watts is right: it is a contamination of poorly sited stations to well sited stations.
I’d comment this way, as pointed out if TOBS is treated uniformly then the effects should average, but even if not, Anthony has now established a new benchmark for adjusting temperature which now will require that
A. The area of heat sinks surrounding the thermometers must be accounted according to Leroy 2010, and
B. The warming bias of badly sited stations, be adjusted down to match prime sites, rather than the prime sites adjusted up to match the majority (of poorly sited thermometers)
This work will now need to be done, if done honestly this will end up with climate data sets adjusted to have much lower UHI influence. It doesn’t matter now what attacks come, the new benchmark IS now established. Anthony should be congratulated for achieving that even if the paper doesn’t stand.
This is going to be very tricky for the Global warmists as it’s certain that this will be the first downward revision the data sets will see. I can see more, there was an Australian study that showed if the average temperature was calculated from hourly temperature rather than min-max the trend is much less, pronounced. Now if we could combine hourly obs, with Anthony’s siting analysis I wonder what falls out?
Congratulations Anthony.
Bob
Congratuations on your first lead authorship, Anthony. I’ve been there a few times and it’s a wrenching experience, so well done to see it through. Agree with Leif on the merits of a concise abstract — the historical tracing can go into the introduction.
you will never know how much the world owes you.
well done.
you will never know how much the world owes you.
well done.
wish I could afford to help you.
[REPLY: Good wishes are noted and gratefully accepted. Help takes many forms, not just monetary. Informed commentary both here and at other sites, including in the MSM, is assistance beyond price. -REP]
One more point that maybe the paper could make (IE a takeaway). The significant difference between Class 1/2 Airport sites and Class 1/2 Rural sites suggests that the Leroy 2010 procedure does not remove all the UHI bias in the record. Leroy 2010 needs to be improved upon.
This seems intuitive in that Leroy 2010 deals with heat storage but not really with direct atmospheric heating due to humans dumping GW of heat energy straight into the atmosphere. Here in in Brisbane for example this means that while the fringe rural areas are experiencing frost, Brisbane does not because of a layer of atmosphere being directly heated by the waste heat energy of millions or radiators, this affects winter results acting over many kilometers. Mind you the carbon tax may well lower winter temperature in cities, since many will no longer be able to afford their heaters any more and those pesky pensioners will just die of hypothermia instead.