New paper on the quality of USHCN station siting

I comne to the Watts site to get informed and read and be educated by the intelligent commentary of the many regular commentors here . I thank all of you for that and for the relatively easy to obtain education . I’ve also learned to ignore a few commenters such as lazy teenager and entropic man and w. connelly who bring deception and no valuable contribution other than the intelligent rebuttals that come from the regular commentors.

Gail Combes, thank you. When I put the two Norfolk graphs on a light box the results were interesting. From 1960 on, the two graphs showed a strong link with each other (and, approximately, with the AMO). The airport showed a consistent warming of about 0.5C compared with the city, which could well be the new runway
Before 1960 the city was about 2C warmer than the airport, despite the runway laid in 1938. The airport station seems to behave more like the rural stations.
Without detailed data on the area I would hypothesise that the main driver in the changes to the airport station was urbanisation of the northern part of the city from 1960, the old runway not having much effect on airport temperatures. Any thoughts?

From Entropic man on July 28, 2012 at 3:37 am:

For the meteoroologists this is a no win situation. If you leave the station where it is, you get degraded data because of the development.
If you move to a new site, with a new microclimate, you lose the continuous record which is important for assessing long term trends.

So bad data is better than no data at all?
If what’s being measured is trends in airplane traffic or air conditioner usage, or there are spikes from using the barbee or burning trash, how does that serve science? All that does is yield higher high temperature readings. And if that’s what you want, to show modern temperatures are higher than in the past (when an instrument didn’t have those influences), then proper siting is much less of an issue.
If you move to a new site without those influences, and get cleaner data that actually measures the real local climate, how can that not be a benefit? Call the old data corrupted and start a new record.
Why the worry about the local micro-climate? Hansen has GISS smearing temperature data up to 1200 km, more than the distance between Chicago, Illinois and Atlanta, Georgia. People are quite willing to complain about global warming without worrying about even regional temperature trends. If “experts” don’t worry about such continuous records when deciding what the long-term trends must be, how can that be an issue?

One solution would be a law banning any development liable to distort the data from a weather station, though I doubt that the general population would stand for it.

That you could even propose such says a lot about you. You would prevent a farmer from clearing a wooded area on their own land for farming, which would be a land-use change, for the sake of a thermometer? You would prevent someone from building a house, or garage, or putting in a paved driveway?
And it’ll have to be a national law for consistency. Repeated in every country supplying temperature data for the global indexes. Perhaps you want it as a UN law, so it covers Antarctica and the oceans as well?
And you think the unwashed masses, excuse me, the general population, would be the ones who wouldn’t stand for that? Fascinating.

Entropic man on July 28, 2012 at 9:07 am:

Menne et al concluded that the overall effect of environmental distortion was was neutral; Watts et al concluded some false warming. Both agreed that a large proportion of the stations suffered some +ve or -ve distortion.

Menne 2010 used an incomplete list from the surfacestations project, only 43%, that wasn’t current and wasn’t quality controlled, said data basically misappropriated from Anthony and the project in what was termed a “professional discourtesy”. Others would call it stealing.
http://wattsupwiththat.com/2010/01/27/rumours-of-my-death-have-been-greatly-exaggerated/
Based on that incompleteness, problems in methodology and what was actually being (erroneously) compared to reach their conclusions, etc, Menne 2010 should be disregarded.
The other paper that should be considered, which would not be the earlier Watts 2009 but the more-complete follow-up paper Fall et al 2011 (aka the Surfacestations Paper), should be regarded as authoritative.

If the processes generating the distortion are so complex that we cannot reliably compensate , 90% the temperature data become useless and neither side, warmist or coolist has much chance of proving a case using it.

Please feel free to convince fellow “warmists” they should stop using those records to claim dangerous global warming is occurring.
Meanwhile you can stick to the satellite record for modern times.

Before either side can prove its point, we need a more positive attitude. I would encourage Mr. Watts to stop running the record down and look for ways of salvaging the maximum amount of data from it. Any ideas?

Before you can identify how to salvage the data and how much should be salvaged, first step should be identifying how the data has been screwed up. Thus your recommendation is Anthony should stop checking and reporting how it is screwed up?

Entropic man says: @ July 28, 2012 at 9:07 am
….If the processes generating the distortion are so complex that we cannot reliably compensate , 90% the temperature data become useless and neither side, warmist or coolist has much chance of proving a case using it.
Before either side can prove its point, we need a more positive attitude. I would encourage Mr. Watts to stop running the record down and look for ways of salvaging the maximum amount of data from it. Any ideas?
_______________________________
I think you would have to examine each record very carefully. As you just saw with the two Norfolk data sets, you have the influence of the ocean, urbanization and the airport. Other records are discontinuous, the stations have been moved and Cthulu knows what else.
Frank Lancer has done some work: http://hidethedecline.eu/pages/posts/ruti-global-land-temperatures-1880-2010-part-1-244.php
And of course so have a lot of people here a WUWT: such as :link and link
AJ Strata, a NASA engineer, did an error analysis: link that sums up the situation best. We just do not have the accuracy or precision to be making the statements that are being made from the historical land based data. Joanne Nova confirms this with her article Australian temperature records shoddy, inaccurate, unreliable.

… An independent audit team has just produced a report showing that as many as 85 -95% of all Australian sites in the pre-Celsius era (before 1972) did not comply with the BOM’s own stipulations. The audit shows 20-30% of all the measurements back then were rounded or possibly truncated. Even modern electronic equipment was at times, so faulty and unmonitored that one station rounded all the readings for nearly 10 years!
Only 15% of sites were compliant with the BOM stipulations for Fahrenheit maximums, and only about 5% were compliant for Fahrenheit minima…. Around half of all sites were so bad that three out of 10 records were rounded or truncated to whole degrees.

And this is from another large chunk of land where we would expect the data to be of the best quality!
On top of that as a few commenters have mentioned the heat content of the air is not just temperature but has to also include the humidity. The relative humidity has been decreasing over the time period that the temperature was increasing (1948 – 2004) link
From my point of view the data is being treated as if it came out of a precision analytical lab with calibrated equipment and college educated lab techs when it is actually coming from field measurements using also ran equipment and who know what type of operators.
Silk and sows ears come to mind.

Do air conditioners get more powerful with time?
Based on my experience, it likely is not the air conditioner’s power that changes –but rather people’s expectations and presumed right to be comfortable. 40 years ago, air conditioners were rare in my state. 30 years ago, the well-to-do had air conditioners. 20 years ago, air conditioners were common, but were used only on really hot days — why spend money unnecessarily? 10 years ago, people adopted the attitude that they had the right to be comfortable all the time, but they still opened windows on cool evenings. Now, opening windows is too much work, and most air conditioners are now running 24/7.

LazyTeenager says:
July 27, 2012 at 5:14 pm

No.
The adjustments are intended to remove some forms of error.
While its possible that the adjustments introduce errors of their own those errors are not necessarily all biassed in the same direction and in fact are likely to cancel out to some extent.

Lazy Teenager,
You are making quite a leap of faith here.
I know adjustments SHOULD be intended to remove some form of error; but sad to say, the results of most adjustments are reinforcements of someone’s agenda. GISS has many unexplained adjustments to its database:
http://www.climate4you.com/ClimateReflections.htm#28012012: Reflections on effects on NCDC and GISS global data series by transition to GHCN version 3
In a world interested in the facts, raw databases would always be available. Adjusted versions of those databases would have detailed explanations of EVERY single adjustments. This, sadly, is not the case. We are simply informed (sometimes) that adjustments have been made to “correct errors.”
Imagine the uproar if votes in the upcoming election were ‘adjusted’ by poll monitors.
However if I’m wrong, please demonstrate how GISS has made its adjustments.

Entropic Man:
Although others have stated this, let me say it in different words:
The ‘warmist’ position is that there is an increase in temperatures during the period covered by the record (with the probable cause being increased atmospheric CO2, or at the very least primarily due to human activity).
The ‘skeptic’ position is that the record is contaminated, perhaps too much to ever recover any useful information. The only ‘human activity’ affecting the temperature record is local land use, A/C units, jet wash, paving, BBQs, and other things directly affecting the area immediately surrounding thermometers.
I’ve yet to hear anyone refer to skeptics (or myself) as a ‘coolist’, and that would be highly inaccurate. I don’t think either warming or cooling are happening other than natural cycles which are fairly easy to see in the record. None of these self-described ‘climate scientists’ have ever suggested a plausible cause for the LIA or MWP, they seem to be too busy pretending they didn’t exist.
As with others here, I’d believe there is warming (or warming worth worrying about) if anyone would demonstrate to me that data and methodologies are sound. I see no such evidence. Then again, I live in a winter climate and can see absolutely zero downside possible if there was warming. I’d much prefer if our coldest winter lows were -35C instead of -42C.

The LIA correalates strongly with the Maunder Minimum, a period of sustained low sunspot activity.
http://www.solarstorms.org/SunLikeStars.html
The MWP is harder to pin down, especially since it shows much more in the European data than worldwide.
I’ve included three links to give you a flavour of the debate.
http://www.newscientist.com/article/dn16892-natural-mechanism-for-medieval-warming-discovered.html
http://www.skepticalscience.com/medieval-warm-period.htm
http://wattsupwiththat.com/2009/11/29/the-medieval-warm-period-a-global-phenonmena-unprecedented-warming-or-unprecedented-data-manipulation/

Incidentally, a quick trawl through the station website while writing my previous post showed a mumber of stations with +ve distortion, but I cannot recall offhand seeing any stations which were underreading due to shading, frost hollows, etc.
Could anyone give me a few examples of -ve distortion stations on the site, to reassure me that the voluntary contribution method used to collect the data is not overcounting +ve stations and undercounting -ve ones.

Entropic man says:
July 29, 2012 at 6:47 am
On the basis of the discussion here…
_____________________________
You are still missing the fact that temperatures were reported as whole degrees in the past. This means any reports of less than one degree are artifacts of the calculations.
If the data resolution is not there you can’t stuff it back in using “Averaging” That only works if you are doing repeat measurements of the same exact thing like measuring the length of a board.
What is done with temperature is the same as having 30 hundred cavity machines making widgets. Measuring a widget from each machine and taking the average does not give me more measuring accuracy like measuring the same widget 30 times would. The two types of measurements are apples and oranges. Unfortunately they are considered the same by climate scientists.
The fallacy of reporting numbers to tenths and hundredths just because the computer spits out a couple extra decimal places is another reason skeptics think the data stinks. Seems significant figures are no longer taught in school any more much less statistics.

JJ says:
July 29, 2012 at 8:55 am
“Your ‘suggestion’ has no basis in reality. It is fabricated from the whole cloth of your preconceived conclusions and biases.”
I am disappointed to find that your approach is so negative. I have proposed a hypothetical solution to a problem which Mr Watts will also have considered, avoiding anything controversial. I get rudeness in return.
To demonstrate that you are capable of more than tobacco lobby tactics, perhaps you would like to suggest how the graph of my simplified interaction between a station and an air conditioner should look.

I’m not too surprised by BEST’s results. I’ve looked over a lot of historical data available at NCDC and elsewhere, and it’s pretty obvious that it’s warmed significantly since the 1800s. Before 1850, there were only scattered observations conducted by the U.S. Army Signal Corps, but they definitely support a cooler climate. This doesn’t mean the change is entirely attributable to man, but I suspect there is some correlation.
Still, global warming will NOT be a catastrophe and, in fact, is more apt to be a boon to civilization. According to BEST, it’s already warmed 1.5C since 1800. Far from a catastrophe, this has corresponded with the growth of civilization! So why should we fear another 1 or 2 degree rise over the next 50+ years? It’s the CAGW-ers we should be focusing our efforts, NOT someone like Dr. Muller, who is making an reasonable, good-faith effort. As Dr. Muller correctly points out, while there has been some increase in temperature, there is NO evidence of the catastrophic effects the CAGW crowd goes on about.

Entropic man says:
I am disappointed to find that your approach is so negative.
I am disappointed, but not at all surprised, that your approach is so willfully obstuse.
I have proposed a hypothetical solution to a problem which Mr Watts will also have considered, avoiding anything controversial. I get rudeness in return.
Your hypothetical solution has no basis in reality. It is wholey a product of your imagination. You got an honest assessment of that. What is required in this circumstance is science, not people making shit up. We have way too much of that already.
To demonstrate that you are capable of more than tobacco lobby tactics, perhaps you would like to suggest how the graph of my simplified interaction between a station and an air conditioner should look.
You are demanding an imaginary answer to an irrelevant question. Imaginary simplifications of real world circumstances that are far from simple are the problem here, not the solution.
A monitoring system has to discern between (at minimum) three intermingled quantities: the signal , the variability, and the contamination. In order to deduce one of the three, you have to know the other two. You can’t fabricate the other two, you have to know what they are. You are attempting to fabricate a contamination profile.
“Say 1C”? Why say 1C? Why not say 1.5C? Or say 0.1C? Or say 0.01C? You assume a step change, despite having been provided information that demonstrates the failure of that assumption.
Then you make up an operation profile. Flat after the start? Why? You made that up.
Increasing at some point later. When? You make that up. At what rate? You make that up.
When you don’t have the data necessary to answer the question, the proper response is to collect proper data, not to make stuff up. If you can’t get the proper data, then you admit that you cannot answer the question, you dont make shit up.

James Schrumpf (@ShroomKeppie) says:
You can get more accurate, and be sure that the length is 35 cm +/- 0.5 cm and not 36 or 37 cm, but you can’t get to +/- 0.05 cm unless you switch to a tape marked in mm. Is that not right?
That is correct, for the measurement of a single board. If what you are interested in is not the measurement of a single board, but the average of 1,000 boards, then there is the possibility that you can do better. If the remainders of the measurements (those +/- 0.5 cm bits) are evenly distributed between + and -, then they effectively cancel out over large numbers of measurments.
The question is, are the remainders evenly distributed between + and -? To know, you’d have to have more accurate measurments. If you had those more accurate measurements, you wouldn’t even be asking this question. In order to take advantage of this benefit of large N, you have to assume that the remainders average out, and you have to be correct in that assumption .
Sometimes, such assumptions are well supported. Sometimes, they aren’t. All to often, such assumptions aren’t even recognized, let lone supported.
What process produced these boards? Does it have any bias that could skew the distribution of the remainders? How do you know? Same with temps.

JJ says:
July 29, 2012 at 12:42 pm

James Schrumpf (@ShroomKeppie) says:
You can get more accurate, and be sure that the length is 35 cm +/- 0.5 cm and not 36 or 37 cm, but you can’t get to +/- 0.05 cm unless you switch to a tape marked in mm. Is that not right?

That is correct, for the measurement of a single board. If what you are interested in is not the measurement of a single board, but the average of 1,000 boards, then there is the possibility that you can do better. If the remainders of the measurements (those +/- 0.5 cm bits) are evenly distributed between + and -, then they effectively cancel out over large numbers of measurements.

How would that work? Assume there are a thousand volunteers with standardized tape measures marked in cm. out there, and every day they get a new board to measure. Every day at noon they measure their daily board with their identical Craftsman tape measures, and then call in their measurements in centimeters with +/- 0.5 cm error bars.
What would the statistical process used be if one wanted to know the average monthly length of the boards to a precision of +/- 0.05 cm? Stats was a very weak spot for me in math, and I can’t even claim total understanding of calculus. But it seems to be to be a “trick” of some sort to presume you could do some kind of statistical analysis and then say that the average board length was actually 35.3 cm +/- 0.05 cm.
If a brief explanation is possible, I’d love to hear it, but I don’t want to exacerbate anyone’s incipient carpal tunnel syndrome!

So essentially, what is happening is that if a thousand people look at their cm-marked tape measures and measure their board at somewhere between 35 and 36 cm, and they estimate 35.3 cm +/- 0.5 cm, we can assume that the +/- errors average out to zero and that the 35.3 measurement is good precision? Sounds like you’d also have to hope the human reading errors average out to zero as well as the tape measure errors.
Sounds like a lot of hoping, and it doesn’t sound justified when everyone is measuring a different length board.