With 33% of the USHCN weather station network now surveyed, the site quality rating is now applied, see the USHCN Station Master List file in HTML and XLS format.
The rating system for site quality was borrowed verbatim from the new Climate Reference Network being put into operation by NCDC and NOAA to ensure quality data. Their siting criteria can be found here.
I welcome input on this work in progress. The site rating will now be a running total in the spreadsheet and always available online as new stations are added to the survey. What is important to note is that the majority of stations that have a rating of 4 are MMTS/Nimbus equipped stations, which according to NCDC’s MMS equipment lists, make up 71% of the USHCN network. It appears that cable issues with the electronic sensors have forced them closer to buildings, roads, etc because NOAA COOP managers don’t often have the budget, time, or tools to trench under roads, sidewalks etc to reach the site where Stevenson Screens once stood. While this isn’t always the case, a pattern is emerging.
For background, see this first: Conference presentation given at CIRES/UCAR on 8/29/07 describing this project and the methods used to assign station site quality ratings, along with examples of many site issues seen thus far.
Click to view the slideshow I presented at UCAR
Immediately after the conference, a senior official at NCDC requested a copy of the above slide show, which I provided to him on CDROM. After receiving it, in a follow up email he inquired as to distribution rights which I granted within NCDC and NOAA for the purpose of review. That was last week. Thus far no issues have been raised with the presentation content. Since no issues were raised at the conference or in the two weeks afterwards (two weeks as of today) I have decided to release it publicly.
Note that of the 33% surveyed, only 13% meet the CRN site criteria (Rating of 1 and 2)for an acceptable location to accurately measure long term climate change free of localized influences.
Climate Reference Network Rating Guide – Class 1 and 2 are considered best, 5 is the worst.
Class 1 – Flat and horizontal ground surrounded by a clear surface with a slope below 1/3 (<19deg). Grass/low vegetation ground cover <10 centimeters high. Sensors located at least 100 meters from artificial heating or reflecting surfaces, such as buildings, concrete surfaces, and parking lots. Far from large bodies of water, except if it is representative of the area, and then located at least 100 meters away. No shading when the sun elevation >3 degrees.
Class 2 – Same as Class 1 with the following differences. Surrounding Vegetation <25 centimeters. No artificial heating sources within 30m. No shading for a sun elevation >5deg.
Class 3 (error 1C) – Same as Class 2, except no artificial heating sources within 10 meters.
Class 4 (error >= 2C) – Artificial heating sources <10 meters.
Class 5 (error >= 5C) – Temperature sensor located next to/above an artificial heating source, such a building, roof top, parking lot, or concrete surface.”
Climate Audit 
202 Comments
Anthony I just sent you and Mc an email. After my review I got a brainstorm of sorts.
Have a look, It might make an interesting project for someone here who has decent R skills.
Can you make a statement yet about temp trends when only class 1 and 2 stations are considered versus trends when class 1-5 are considered? Most CA readers would believe trends of class 1+2 subset to be less, while most RC readers would believe trends of class 1-5 would be the same as class 1-2 subset.
To clarify:
Wascally wabbits are going to be in a real snit about this!
Oops, I didn’t see that on the main page. Please snip #3.
Matt asks a really good question. There might not be a whole lot of class 1 and 2 stations to work with, but it would be interesting to see what their numbers are.
Anthony,
What do you know about station histories? Are the Class 1-2 of today also qualified as Class 1-2 over an extended period? It would seem to me that to get an accurate assessment of temperature trends, one would better use a smaller (even much smaller) sample but of higher quality. If Mann et al. can reconstruct the entire hemisphere with a few dozen proxies, and claim 0.1C accuracy, surely a few hundred stations with good location and known histories would provide a relatively accurate estimate. IMO it would be better than trying to figure out the Hansen code!… But ultimately, it’s really not clear that surface temperatures are a good indicator of “global” warming. SST’s would be better, or more accurately, ocean heat content.
They need to be upgrading to wireless, specifically the zigbee type system, specifically designed for sensors. (www.zigbee.org) Its rediculous that with cheap good sensors and the new technologies available to link them, that they are still using old land line systems, especially when the results of bad data is an impact to the economy of hundreds of billions…
RE 2.
See my comment in #1. Anthony was kind enough to slip me an advance copy of the XLS
and I had fun looking at Class1 and Class 2 last night.. couldnt sleep
Let’s just say, I think there is a story there. I’ve downloaded 25 or so of the class1 &2
and did some eyeballing.. Anthny & SteveMc have my Xls… It would be an easy matter to post a tab
delimited file for all the class1&2..
Later, I have to go earn money
Anthony:
It was worth the wait. The FLIR should be detecting red faces in various government agencies I would guess. Your presentation may have helped shaken loose the code as well. Congratulations to you and all those intrepid documenters of facts. Presumably Eli will be gracious enough to offer his congratulations, even while he disputes the significance of the results for the temperature record, i.e., the errors are randomly distributed.
I’m curious where those class error numbers came from. I strongly suspect they were guesstimated.
RE11, Where did the ratings come from? They came from Michel Leroy’s 1998 study, and CRN adopted them with Karl’s blessings.
I had nothing to do with choosing the error values associated with the ratings.
RE 2,7 A couple of caveats when looking at this survey:
1- This is the current state of the network, no history considerations have been made at this point
2- I have not established any trends for any particular station rating sets. My goal was to present the siting information first.
RE 11.
Anthony has been VERY clear about this. Let me give you background. When AW started posting sites
Eli Rabbett came on and blathered about the CRN. I went to the CRN site and found the siting ranking
guideline written by Leroy.
I hate playing the credential card… but here’s the flop
http://www.wmo.ch/pages/prog/www/IMOP/publications/IOM-94-TECO2006/PROGRAMME.HTML
Go check out Leroy.
Anyway, Leroy has a categorical ranking system. He assigned a “bias” figure to his
ranking system. As Anthony has noted, as CRN has noted, Leroy’s “bias” is an ESTIMATE.
For a while I have Been trying to source Leroy’s paper. However, it was delivered
at an AMS symposium. Eventually I will track it down. However, Leroy’s work
dovetails nicely with Geiger’s work on Microclimate. Have you read Geiger?
Ah well.. anyway
Here is a BET.
According to Published Global temperature anomalies, the peroid of 1975 to 2006 showed
a near linear increase.
That looks like .7C since 1975. And its damn near linear.
SO HERE is the bet.
I take all of the class 1 & 2 sites and plot their trend from 1975 to 2005.
I take all of the class 5 sites and plot their trend from 1975 to 2005.
( crap I havent done that)
Which dataset ( class1&2) or Class5.. will match the global trend better.
Care to bet?
12, but what exactly does it mean when the class 3 error is +/- 1C? Was that calculated? Do they know if the error is symmetrical around the mean? I have serious doubts that we even know the error on the error, if you know what I mean.
A more critical question than what the error is is is it symmetrical around the mean? If it is, averaging enough of them will reduce the error. If not, it won’t. And I don’t know if anyone has any way to address that question.
My own personal opinion is that data has to be presumed bad until proven good. Thus, this questionable data should simply be excluded. I don’t think that it can be salvaged by Hansenmatics. But I don’t know what the error estimates bring to the table.
Thoughts?
Larry:
I am with you: Unknown direction and non-trivial error, throw the bums out!!
re 15
“12, but what exactly does it mean when the class 3 error is +/- 1C? Was that calculated?
Do they know if the error is symmetrical around the mean?
I have serious doubts that we even know the error on the error, if you know what I mean. ”
Without Leroy’s paper and the studies that back it up I would be guessing. The point is not
CORRECTING this bias. The point is AVOIDING this bias. Hansen happily lives with the microsite
bias because he BELIEVES the bias is normal with mean zero. There are three options
1. characterize all microsite bias and remove it.
2. Utilize only those sites that meet guidelines
3. Assume the microsite bias averages to 0.
#1. Characterize and remove. This would require extensive microsite modelling OR dual site
studies.
#2. The number of sites will go down; data quality will go up.
With 33% of the sites surveyed 50-60 sites meet guidelines.
Assuming the other 66% yeild accordingly we are talking 180
sites for the US that are class 1 or 2.
A. Gavin Schmidt claims that the ENTIRE NRTHERN HEMISPHERE can be sampled by 60 sites
B. at 180 sites we would have 30 times the number of sites that sample Brazil
C. The Climate Reference Network is planning roughly 100 sites.
Bottom line. We do not have to engage in navel examining studies to figure
out how much bias Volcanic cinder has versus Asphalt.
#3. ALL of the studies of microclimate bias indicate that the bias is a warm bias. ALL.
Larry wrote
“is it symmetrical around the mean? If it is, averaging enough of them will reduce the error.
If not, it wont. And I dont know if anyone has any way to address that question. ”
Lots of ways to address this issue. The factors realting to BIAS are almost all warming Biases.
If you like I can walk you through it. All studies indicate the microclimate bias is a warming
one.
Here is a thought experiment. You are standing in a feild holding a ball. The sun illuminates
it. See that spot the sun is shining on? Now, tell me how you propose to cool that spot.
1. Shade it.
2. Blow on it.
3. Paint it white.
4. Put water on it.
Now let me do things to make it hot.
1. I paint it black.
2. I change the material under the spot to be a heat sink and store energy all day long
3. I put up wind breaks so you cant cool it with breeze
4. I build reflectors that concentrate the IR on the spot
5. I make the surface so it cant breathe
6. I drain water away.
7. I Put heat sources ( buildings.. people) all around the spot.
8. I cut trees down
Making a spot on the ball hot is easy.
Basically all you have to fight Microsite UHI is shade.
Tarmac and concrete surfaces presumably raise temperatures on hot days, shading lowers temperatures on sunny days, etc. But the issue must be the introduction of variable errors, either because the bias changes temperatures in different kinds of days or because local site effects have changed over time, each affecting the anomaly data. Ignoring the biases and changes means the data is not worth anything; adjusting for the biases and changes and the final data depends as much on the assumed adjustments as on the original data.
17, two things:
1. The characterize option presumes that the site issues have always been there. Not a safe assumption.
2. Those things that cool the instrument don’t really cool it below the temperature of interest, because the temperature of interest is the air temperature a.k.a shade temperature. They actually correct the temperature. The only thing that can actually make a thermometer read below the true air temperature is lawn irrigation, which can actually cool the air. Everything else heats. So your point is even stronger than you realize. At the risk of a snip, the fact that heating is a lot easier than cooling is due to the second law.
That “1” is best and “5” worst is clear from the site criteria. Where does it say that anything over a “2”
is not acceptable?
RE 19, Leon Palmer is doing some work with USGS photos for some sites to try to recreate the station history.
What is really needed though, are the B44 forms from NCDC which show station history each time a COOP manager updates the sketch and notes items around the station. Unfortunately, the B44 forms are held within NCDC because they obtain the observer name and address. Perhaps that info can be redacted so that they can be published for research.
Anthony and all his team: my hat’s off to you again. Great work. It will truly add to the knowledge of temperature changes and measurements. 14, SteveMo, my bet’s on Class 5, of course.
RE: #14 – Wull, let’s see …. 0.7 deg C is well within the Class 5 error bar …. so, let me make a counter bet. I bet I can “fit an elephant” – as it were! LOL!
#17:
Gavin Schmidt says that the northern hemisphere’s temperatures can be accurately sampled with 60 stations, but I would surely hope this isn’t a figure we would take for granted. It sounds a bit low.
What about the UHI effect? Are there now interactions between the quality of the station and the UHI adjustment? Example: Hansen is adjusting an urban site with the data from a nearby rural site, but the urban site data will be corrupted not only by its own poor quality, but also by the station used to adjust it. Granted, it is far less likely that the rural station will be of poor quality, but there are undoubtedly still a large amount of them out there.
RE 23.
I think UC or willis or someone similiarly brilliant tracked down Gavins 60 site claim
to a paper written by Shen..Search around on CA…shen94 or shen98..
The papers are not for the weak of synapse.
Having said that. The key Point is this: 60 sites WOULD suffice if they were pristine.
I would agree with that, but it’s a hypothetical “if”.
Re: 17
Don’t forget that the minimum temperature (night) is at least as important as the max temp.
In doing a variance of the annual mean changes, winter minimums going up (some evidence to date, but a thorough look at a few long term, quality stations would help) could alone be enough to drive +2 deg F/century. Just looking at the annual means doesn’t work to explain just where the increases come from, nor what the consequences might be.
#23
I see no reason why microsite effects would not apply equally to urban and rural locations. In urban locations it is probably more like noise to the dominant UHI effect – but they are still there. In rural locations, especially Arctic rural locations the microsite effects may be significant. Anthony has been there yet but I suspect that we will find odd things occurring in places like Cambridge Bay that confound some of the winter warming effects.
Hansen et al would be aware that there is another revision of the US temperature record coming, thanks to this work. My guess is that it will show that the current day is 0.6 degrees colder than the 1930s high. It is going to be hard to froth at the mouth about AGW when the world is shown to have cooled.
RE #20
One observer (Academy, SD) had a publication that had a relatively complete history of “his” station. It moved around quite a bit between say 1916 and 1948 (where most NOAA database records seem to start). The info is likely out there for many stations, but would typically require more extensive investigation locally of the “old timers”.
My general take is that ANY location or equipment change will introduce a bias. The introduction of the MMTS/Nimbus seems to have driven plenty of location and equipment change right around the time of interest (the end of 70’s cooling and the beginning of the current warmer trend).
I would like to take this opportunity to thank Dr. Joshua B. Halpern, Distinguished Professor of Chemistry at Howard University; Washington D. C. whose perverted comments motivated KBSF volunteers to complete active stations in New England.
Re #29
I think rural microsite effects may be much worse becasue of the rural sites impact on the adjustments to urban sites.
RE 23.
Let me schematize the argument for you and others.
1. 50 years of climate science shows that UHI is real.
2. Hansen, Jones, and Parker compared Rural to Urban and FOUND NO DIFFERENCE.
They concluded this:
3. URBAN SITES must be in cool parks.
In short:
1. UHI is real.
2. We found no difference between Rural and urban
3. URBAN SITES MUST BE LIKE RURAL SITES, LCATED IN COOL PARKS
Now, you all need to understand the basis of that argument. THE BASIS is the science that
shows UHI is real. Urban SHOULD BE warmer than rural. Hence, they need to explain
why they didn’t find this.
They COULD HAVE concluded this;
3′. RURAL SITES are contaminated by UHI-like phenomena.
That is.
1. UHI is real
2. We find no difference between Rural and Urban
3A. RURAL SITES must be contaminated by urban like features.
When you look at the arguments structure the choices are clear.
So, Now comes the question. Cool parks? or non compliant rural sites?
How do you tell if 3 ( cool parks) is true? Visit the site and observe.
If you find the site on the roof of a building, then its not in a cool urban park.
How do you tell if 3A ( bad rural sites) is true? Visit the site and observe.
If you find rural sites that are not really “rural”, then you begin to undertstand.
You go google UHI studies. UHI Atlanta, UHI new york
You will see the science is settled. The urban enviroment is unnaturally warm.
Hansen, Parker, and Peterson, all wondered if this bias found its way into
the temperature record. They found It didn’t. WHY NOT? .
A. The urban stations are in cool zones
B. The rural stations are corrupt.
Peterson, Parker and Hansen ALL picked A. But they never LOOKED AT THE SITES. Surafacestations has.
Cool urban parks: ZERO
Corrupt Rural: Dozens if not hundreds
Conclusions?
Let’s not forget that these numbers include all the photos of the airport ASOS stations that were not actually visited but pictures were taken from NOAA pages, so this sample will be a little on the higher quality side.
What fraction of the “warming” occurs in the Winter? The Spring? At night? On the Winter Solstice? On the day after a hot day? a cold day? After it rains?
RE 28.
Everything I have read says this.
1. Its really hard to drive up TMAX. ( think magnifying glass)
2. Its easier to drive up Tmin ( think heat storage)
I give you a simple example. Las Vegas. Last 10 years. TMAX: FLAT. TMIN: climbing
Now, could some physics genius at RC explain to me how this happens?
I get the TMAX thing. The sun heats the ball. And since Insolation hasnt changed
( sorry sun nuts) TMAX is pretty dang constant. But TMIN goes up? How’s that?
Everything we build or stores and gives off heat. During the day our impact is
lost in the noise of insolation. we come out at night, and warm the planet then.
There’s a vampire thing here, but I havent figured it out.
Re #35
So in essence the good sites (ASOS) have been “surveyed” which is the opposite of the insults thrown at Anthony et al which accused him of cherry picking bad sites. Some idiot even accused him of Photoshoping the pictures. The ASOS sites have their own problems as Steve Mc has noted. ASOS is used for aviation purposes and is sited for use by aircraft/ATC, not for climate. How much local heating is cause by runways and jets is unclear.
RE 29.
Yes. As I look at the causes of UHI and MicroUHI I see this.
You can only mess up a site so much. Mathematically, the UHI bias/Microsite bias
IS LIKELY TO BE logistic in nature. there is a limit to hw badly you can screw things
up.
“there is a limit to h(o)w badly you can screw things up”
Does Mosher’s Law apply to GISS or is GISS a law-free zone?
RE37 Barry, lets not be so quick on the “good sites” label for ASOS stations in USHCN. I used to think the same, and by the distance criteria used in the CRN classification scheme that may be true.
But then we have the HO-83 hygrothermometer, which has a flawed design and gave 4 years of new record high readings for the Tucson airport.
While I was at Pielke’s conference, a site survey from New Orleans came in, and surprise, its not an ASOS per se, but does have an HO-83…an early model, which doesn’t even have the air deflection skirt on the bottom to deflect downward air from the “mushroom” cap so that the air isn’t recycled back into the HO-83.
see it here: http://gallery.surfacestations.org/main.php?g2_itemId=23786
then the newer model with the skirt here at Mount Shasta, CA:
http://www.norcalblogs.com/watts/images/HO-83.jpg
The big flaw in the HO-83 is that they cheaped out in the design, and used a single aspirated chamber to get both air temp and dewpoint. The dewpoint sensor uses a Peltier chip which is hot one side, cold the other, and cools a mirror to the dewpoint so an optical sensor can detect it fogging up…then it cycles again.
Only problem is that as the aspiration fan ages, it gets less efficient, leaving some of the heat from the Peltier chip unvented from the chamber, hence biasing the next air temp reading. Combine that with recycled air from a design that allows the mushroom cap to force air back down the side of the chamber for re-ingestion, and you have a perfect little convection oven.
Until we know the equipment history on the ASOS network, they are all suspect due to this equipment issue.
re 32.
Kristen when he maligned you I popped all my rivets. It wasnt that you were a female.
Heck, I’ve said nasty things to Dr. Curry. For me it was this. You are starting
down the road of education. ( Ok I taught at university) and it pissed me off to
see a learner treated that way. I better stop now.
RE 40.
Good one. You know of curse… COURSE that the “o” key is messed
up on my keybard and unless I bang on it like a mnkey it wont show
up.
arrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrg.
Re #40
Anthony,
My post was somewhat tounge-in-cheek. The ASOS sites may look better than they are, but they may be some of the best you’ll see, at least from a visual perspective. As I said airport sites aren’t necessarily the best locations for climate research, let alone having equpment problems. The primary user want’s to set altimeters, ensure safe aircraft takeoff weights, and visibility. They probably don’t care if the temperature sensor is alittle off, only if it’s way off.
I seem to remember that NASA treated station moves to airports as being a move to a cooler site (from the city proper), which I think is open to argument (depends on the specific move) and without before and after site surveys you don’t know.
Oh, and congrats on you’re talk.
RE31 I’d just like to add a note thanking Kristen and her folks, and her volunteers for getting a huge number of stations in New England. It just goes to show that anger can be a good thing when it results in motivation to overcome such crass commentary from Rabbet/Halpern/Bunny.
Some of the sites she did were used in the UCAR presntation, like the rooftop one and the one with the nonstandard sensor on a metal building.
The thing that bothers me the most about the sites I’ve surveyed and read about are those near vegetation.
Large bushes and trees affect the view of the sky, they affect wind and mixing and they transpire. How all this plays together, as the vegetation grows, is complex but real.
The Port Gibson MS site has a young magnolia tree planted twenty feet away. Magnolia trees become quite big. While it does not affect the MMTS today, it will eventually affect it by blocking sunlight and nighttime IR radiation, by reducing local wind and mixing and eventually by cooling nearby air through transpiration. What a mess.
RE: #45 – Another thing associated with magnolias – massive bird droppings. Birds just love to nest – perch in magnolias.
Barry/Anthony,
I was only commenting on microsite effects. The other problems you bring up are valid as well.
If half the warming in the last few decades is due to poor station sighting, then I agree, this is a big story.
Assuming a big portion of the warming is a result of poor station sighting, you need to prepare for HOW you share this info with the world. This site does a great job of finding a lot of “AH-HA” issues, but then moves on to the next curiosity, leaving the casual reader to attempt to tie it all together themselves. It’s a shame, because those with technical backgrounds can do that if they have the time. But those without the background can’t. It might sound crazy, but look for a PR person sympathetic to your cause. Engage them. Help them understand the details. They will work wonders helping you get the message out to folks. They will help you create simplified collateral material, sound bites, and “rude QA” that will explain things in ways you never would have imagined needed explaination.
RE48, Matt is right, and I’ve noted this as well. We have a bit of attention deficit. Time to stop and write the report. Then we can be the ones to say “RTFR”. 😉
Anthony,
I am truly blown away by your excellent slide show. You do not have to be a climatologist, an engineer, or even a scientist for that matter to understand the meaning of your work. I forward your slideshow to a few non-scientists AGW friends of mine. We regularly deliberate how it is scientists know that the planet has gotten warmer during the past century. Now we know. It got warmer from substandard temperature sites from heavily biased locations.
I am particularly thankful you left the sublime IR imagining to the end. No more twaddle from AGWers about the innate subtly between a few inches of gravel under a Stevenson Screen surrounded by a sea of asphalt, the fact of the matter is this, large parking lots in the general vicinity of a temperature sensor affects the true air temperature, period. Case closed!
Ian
RE50 Ian, Thank you. As a TV meteorologist for 25 years, my job was to take complex science, be it meteorology, climatology, or astronomy and make it understandable to the average viewer in under 3 minutes. Your post on understanding this new issue was the highest compliment I could be given.
Anthony,
Many thanks for posting your presentation and congratulations once again! It would have been great to hear it as well. We were very tickled to see that a few of the sites that we surveyed while on vacation made it to you in time for your slide show.
Surfacestation.org should speak volumes of Hansens High Quality Data to say the least. Thanks for the opportunity to participate.
#36 Steven,
So would that be a vampire in a nut shell because I think that pretty much nails it!
Anthony,
I too would like to congratulate you on a fine presentation! The powerpoint slides were very powerful in that the audience could see all of the non-compliant stations for themselves. And, like Ian, I really liked those FLIR images – you could really see the large change in surface temperature going from the grass to the asphalt.
One point to stress is that many of buildings, parking lots, A/C units, and other man-made structures seen in the photos were *not* there 50 or 100 years ago. Can you really compare a temperature reading from 1910, when the thermometer was in an open field, with the temperature reading today, where the sensors are now surrounded by parking lots, brick buildings, vents, and other UHI-producing features?
I also think you should collaborate with someone (perhaps Dr. Roger Pielke?) to publish a paper in a suitable journal. That would give this work a broader visibility to the scientific community.
49 You should start preparing for a press release.
Just wanted to pass on my congratulations to Anthony and the team. This is a mamoth undertaking and goes to the very core of understanding climate.
If the initial data is a mixture of both good and tarnished data, then any studies arising from that data will most likely lead to the incorrect conclusions.
I think in years to come, when the entire survey is complete, people will look back at this moment, as a watershed in climate analysis (starting in the USA).
When near enough, almost, approximately, and probably right no longer acceptable in climate information sourcing.
i would not be surprised that the entire USHCN network was overhauled, and very clear mandates are put in place to ensure the “purity” of climate data.
It will no longer be just the “weather station”, which nobody cares about, but an important part of the countries infrastructure where true accuracy becomes paramount.
UK Waldo only appeared in 1987 Quote from From UK Met Office Climate study.Climate Memorandum No21 June 2006.
As expected, the most significant trends (at the 1% level) are for increasing temperature, with mean temperature increases varying from over 0.8 °C for the Midlands and East Anglia since 1914, down to 0.4 °C in North Scotland. However, there was virtually no trend in mean temperature between 1914 and 1987, and it is only since 1987 that the temperature has notably started to increase. The winter is the only season which has not seen significant increases during the 1914 2004 period.
Winter and autumn sunshine has increased very significantly for northern, central and south-east England since 1929, which is likely to be linked to reduced air pollution brought about by the Clean Air Acts. Most of the increase has taken place since the late 1960s. North Scotland has had a 6% decrease in annual sunshine, which is significant at the 5% level
So its got warmer because its got sunnier? Is that it ?.
Compelling presentation — even at 33% sampled, the USHCN network deficiencies are clear to even those with a minimum of science! Your well-structured repetition of each flaw, parried any “cherry-picking” thrust by defenders.
Great work Anthony!
Regarding the USHCN Station Master List: Would it be possible to add a column for Hansen’s Urban/Rural indicator?
I was hoping to make a breakdown of surveyed station counts and quality for Hansen “Urban/Rural” like this:
CRN Station Quality Breakdown by Hansen “Urban/Rural”
### (33%) of stations surveyed
Hansen Hansen
CRN Urban Rural Total
——- ————— —————- ————-
1 ### ### ####
2 ### ### ####
3 ### ### ####
4 ### ### ####
5 ### ### ####
Un-Surv. ### ### ####
——- ——- ————-
Total #### #### ####
Someone could get much fancier than this with percentages etc.
I would love to see an answer to the question: “How good are the “Rural” stations?”.
#51
Anthony,
Your presentation is simply superb and I have not enough words to compliment you. I’m sure it will get a well deserved success.
RE: 24
Actually I believe the study Gavin @ RC is quoting from is: Estimation of Spatial Degrees of Freedom of a Climate Field by Xiaochun Wanga and Samuel S. Shenb. What Gavin quotes is very misleading. Now having read it and read the supporting papers, one major points come to light. First, the underlying assumption is made that the best-fit distribution represents the ‘true’ data well enough, so that the effects of noise in the simulated data and in the real data are the same. The impact of these assumptions is that while they say stations in the paper, they are referring to cells. As a further note, the cell size is quite large (5×5 degrees) so basically they are saying if you have a station that can represent a cell, only 95 cells are needed to measure a trend. Currently NOAA says that it takes 16 – 24 stations to measure a cell this size.
So my understanding is that using the MCM they proposed only 60(winter)-90(summer) cells +/- 5 cells are needed to determine what the temperature trend, they do not say what the precision of the the results will be, nor can they be definition know what cells. This would indicate that every cell has to be measured at a high level of precision, and that while they discuss ‘stations’ they are actually talking about cells.
Oh, and I almost forgot, since your using so few observations (cells), then you loose the Law of Large Numbers (LLN) so each cell has to have a higher precision. The NOAA stations per cell gets the precision to 0.1 degree C per cell. That is not addressed in the DoF argument either.
I therefore say that Gavin’s claim that 60 ‘optimally’ placed stations could measure the whole NH is not supported by this paper.
A class one station could still be in the middle of a big city, or in an area that developed a lot over the last few decades.
What we need are truely rural stations that are also class one stations. If there are any left.
Vernon,
Same Shen. This paper is online as well (that’s the way things should be) ,
Click to access wang_jclim_1999.pdf
it takes us back to Shen94
Now, let’s just find those 60 (rural, Class 1) stations, then there’s no need for complicated UHI adjustments. Simple average (no variance adjustments anymore, please), and that’s it.
mosher: 6:23pm –
pristine, and evenly distributed.
mosher – 6:56pm –
Allow me to expand upon your remarks.
Even though parks can be cooler than the surrounding city, unless the park is HUGE, it will still be affected by the surrounding city, and hence be warmer than a truely rural setting.
Some of those stations are so bad, that merely labeling them a class 5 doesn’t seem like enough. Maybe we need a class 6 +/- 5C and a class 7 +/- 10C
RE: 62 Actually, in the 1999 paper it was shown that the earlier estimates for global climate trends dof were under estimated. That was one of the major points of the paper. 106 +/- 5, so 111 cells world wide would need to have stations with an optimal weight for each cell.
The authors, are in my opinion, pretty sloppy and switch from station to cell, which are not the same thing. Further, in developing their dof, they exclude the Arctic, Antarctic, and much of South America, and Africa.
re 48.
You might divide the last century into 3 periods.
From 1910 to 1940 the global temps go up abut .4C in those 30 years.
I Would expect the class 1&2 sites to track that pretty well.
From 1950 to 1975 you have a roller coaster of sorts.
Frm 1975 to 2005 you have a .7C increase.
Now that rate is double ( almst kinda) the early century rate.
QUestion: what do the Class1 & Class 2 sites show?
I don’t believe I’ve found a site in Anthony’s class1 or class list ( I’ve looked at half)
That shows a .7C warming from 1975 to present. So, interested folks can have a look and do
a complete job. I am probably going to go look at class 5 sites during this period. I have the
ADD real bad.
I DONT think you are going to see half of this .7C go away. I think 10-20% is a better guess.
Muscle Shoals
Fort Valley
Tombstone Berkeley Electra Fairmnt Independence Orland Quincy Susanville cheyenne wells Stamford(U) APALACHICOLA Fort Meyers(U) pensecola (U) Tallahasse(U) Savahhana(U) Priest River Angola Crawfordsville(S) Goshen (S) Greencastle Huntington(S) Washington Baton Rouge(U) Prtland me(U)
Hey were can I buy a vowel? my “o” key is getting worserer and wrserer everyday
Re: 62
“It takes us back to Shen94
Shen et al. (1994) showed that the global average annual mean surface temperature can be accurately estimated by using around 60 stations, well distributed on the globe, with an optimal weight for each station.
—
In this scenario, I wonder who gets to choose what “well distributed” means and what the “optimal weights” for each station are? Not us “jesters”, methinks…
RE68 Mosh, Fairmont, CA is one of the very best sites, because it is isolated and undisturbed, with only one short distnace station move long ago. It’s at a Los Angeles Water District reservoir way off from the city. Before 9/11/2001 I could have walked in there, now its secured because the reservoir is an easy mark for terrorists with a bottle of whatever. Fortunately I was able to do a survey from by “outside the gate” visit and sat photos since I could clearly locate the Stevenson Screen and measure using the photo analysis tool.
Another site which I’ve tried twice now to get which also is likely to be a well isolated 1 with a long and solidly uninterrupted history is Rancho Tejon, about 50 miles west of Fairmont. I got within 500 feet but was turned away. Sat photos aren’t good enough to do a survey since I cannot pick out the Stevenson Screen.
Anybody have an “in” with the Tejon Ranch cattle company?
How long will it take to collate the temperature records by the 5 different site rankings?
Very good presentation Anthony.
It seems you have found a forcing the IPCC has missed: Data forcing.
Excellent job and thanks to all the volunteers.
Steve #33
Well said. I would add that the definitions of rural and urban are too loose and the basis for allocating a site to one or other of these categories are also suspect as we found in looking at Brazil. The lit/unlit differentiations looks equally problematic. In reality the more we can quantify microsite contamination the more it makes sense to create a single continuum for non-climate sources of heating and cooling. This continuum should be granular enough to reflect changes to the site.
#67
Shen 94:
gridded UK vs. global station-by-station, not sure if it matters, but seems quite a large step. Need to read Wang99, maybe it clarifies..
Anthony, per MMS, these are the managing parties: HNX , JOE RYAN , VICKI ROGERS. Have you tried contacting these people? Not sure what HNX is – HNX the airport is in Wyoming.
Anthony, and Volunteers:
Superb work and excellent presentation. Even I could understand the Pie-Chart, and not for the first time when following this debate, my response to the nonsense we have been fed on the subject amounted to a red mist of anger. Only Matt’s comment at #72 brought down my blood pressure with laughter.
John Lang at #72 is right about collation. I for one would be fascinated to see each class of station given a separate line on the same graph.
re 71
It shuld be easy BUT I’d hesitate to do or say anything with finality at this time.
1. the data at GISS is kinda squirrely at present..
2. A prper analysis would go back to the surce USHCN data
3. A defensible apprach needs to be defined.. That is, do we use Hansens method or something else
So, be patient
Re#41 – Not just a “learner treated that way,” but treated that way by someone making a career in the highest profession of education!
Of course, when I first found out about Halpern, I came across a website with student reviews of one of his classes…didn’t seem like much of an educator. I don’t feel the need to pile-on and dig-up that link now (nor do I have the time).
re 77.
Yes I recall that. That is what made me all the more incensed. Luckily Dr Curry
was on thread that day and made a nice offer to Kristen about visiting Georgia Tech.
In the end suspect it would take more than Halpern’s antics to shunt Kristen’s enthusiasm
to ground.
Anthony,
Great job. I’m curious what the impact will be for the number of Type 1 and 2 stations as the remaining 66% stations are surveyed.
What is the ratio of urban to rural sites thus far surveyed?
I would expect that most of the sites thus far surveyed are more urban then rural, if for no other reason than they are far easier to get to.
RE#77, see unthreaded, found it rather quickly along with another interesting tidbit.
Anthony’s work is quickly beginning to show that UHI adjustment can’t be based purely upon population densities. It has to be based upon some combination of population density and site scoring because:
1) a station may exist in the middle of a large urban area and be scored a class 1 or 2 and still be suffering from UHI
2) a rural station may exist in a low-density rural area, but still be sited such that it is scored a 4 or 5 and still suffer UHI-like effects.
Only if we begin to do UHI adjustments considering both population density and site scoring will we be correctly accounting for warm bias in station observations.
Re 81,
wouldn’t the site scoring negate the population density? Both are proxies measuring the same thing: bias. If one proxy is superior, why use the other?
Re 82,
It could be some sort of 2-dimensional table adjustment that considers both population density and site score. Stations which have both a high population density and a bad site score (4 or 5) would have the highest downward adjustment for UHI. Stations which have a high pop density but a good site score would get the pop-density-only adjustment. Stations which have a low pop-density and a bad site score would get the site-score adjustment only, and those that have both a low pop-density and a good site score would get no adjustment.
Just musing.
Population density is a poor proxy for non-climate heating/cooling effects – a measure of energy and gasoline consumption per sq meter plus population density plus % green space would all get us closer to a net UHI effect. For example, my guess is that today London may have less of a UHI effect than
New York given the lower use of air conditioning — but in the future the London UHI trend may be significant as energy consumption increases with new more functional building designs. What is odd is how poorly specified such pronounced effect is, which makes it hard to evaluate likely changes in that effect.
Steve, et al:
A gracious winner makes allies. A gracious loser learns from the loss and becomes a winner. An ungracious winner makes enemies. An ungracious loser finds oblivion.
There have been some remarks on this site in the last couple of days that are ungracious. The remarks re the NOAA scientists have been particularly ungracious. Presuming Steves work makes him a winner over Hansen, trust me, you all should be working to win over the NOAA scientists and make them your allies.
Ive worked with scientists for 30 years. As a rule they have integrity and honesty. However (despite their dispassionate facades), they are all very, very passionate about their work. Most scientists really do want to make a better world. If Steves science proves to be correct, be gracious and give the NOAA scientists a chance. They are NOT stupid, they do care, and they will come around eventually. If you are ungracious, they will dig in and fight to the death. There are a couple of you who should probably issue apologies.
Occasionally a scientist falls prey to his/her passion, believing that his/her theory is the only one, and manages to sell it to the public. Bad move, as most theories get holes poked in them. Therefore if wrong, the scientist will eventually be publicly crucified. You, as a group, should be cautious about making this mistake or being ungracious if you are right.
It used to take 20, 30, even 50 years to see an accepted but wrong theory finally crash and burn (think eugenics). Steves, and others, demonstrated use of the internet has the potential to shorten this process significantly. This is a wonderful development in science. Ive been a lurker on this site and the Real Climate site. Ive been fascinated by the debate. What blows me away is how quickly Steves challenge took root. If he proves to be correct, you will see more scientists develop the courage to openly challenge poorly done science. Can you imagine the lives saved if we had this capability to fight the DDT scare?
Whether Steve proves right or Hansen proves right, what Steve has started via the internet will have a huge impact on science that I believe will be positive but only if the winner remains gracious.
The other phenomenon, which I see could have a huge impact on science, is what Anthony Watts is doing by developing a grass roots movement to examine scientific results. Has this ever occurred in history? Not that I know of. Ill bet that his presentation at UCAR raised more than a few NOAA scientists eyebrows and Ill bet he made allies – with a side benefit of getting temperature sites decently upgraded which is good for climate science.
Right or wrong in their efforts, my hat is off to both Steve and Anthony for their courage and vision. It remains to be seen what will come of their efforts, but it sure has been fun watching these historic processes develop. Thanks. Graciously yours, Larry Grimm
RE: #84 – London – two things I’ve noticed. Firstly, lighting there causes noticable heating of indoor spaces, window displays, etc. Secondly, during the colder months, the shop owners often keep their front doors open and have small but powerful heaters mounted on the cieling just inside the door threshold. Walking down the street you can feel all that heat going out into the air.
>> There are a couple of you who should probably issue apologies.
Just curious, will Hansen be issuing refunds to norwegian citizens who are paying $466 each?
RE: #85 – The climate community at NOAA seems to be factionalized and diverse – a simple microcosm of the world at large. Whereas, the one at NASA seems to be more dominated by the “killer AGW” faction. Just my two pence, YMMV.
Nothing, but if you’re looking at tenths of a degree over long periods of time, why use stations that have an offset 1 degree or greateer, if you can help it? As many have mentioned, excellently sited and well distributed stations are prefered. If we want to know what it is we’re really looking at.
Hi Bernie,
You’re right. And my point was not that population density is the best metric for determining whether to apply UHI adjustments. I only used that as an example because I understand that that is currently the metric. I agree, it’s not a good one. I think it would be better if total asphalt area, and/or concrete, and/or car/truck traffic, or other metrics were used. You’re suggestion of energy use is also a good one.
#85: if you look at the continual ungracious comments of Mann, Hansen, Juckes et al, all these years, even still after proved to be wrong, it’s very hard to be gracious to them, but I do see your point. I wonder, if, when Hansen will enter this forum to defend his calculations like Juckes did, well… tried to do… That’ll be the day…
Re: 81-84
I haven’t seen the Barrow, AK, station site, but it’s at/in the vicinity of the airport. I wouldn’t be surprised to see it score a 1 or a 2. I also suspect that population density at Barrow is low. Yet Barrow has an AHI bubble in winter (Hinkel, et al.). The bubble is evident on low-wind days which must be fairly frequent since the winter temps within it average 2 degrees above the surrounding area. The bubble extends over the airport. So here we may have good siting per CRN and low-population yet a significant bias.
#85. Fair enough, but please only hold me accountable for what I say, rather than what every poster at this site says. On many occasions, I’ve urged readers to be less angry. The tone that I try to achieve is more one of irony and, if I lose that tone, please feel free to remind me. In particular, if there are any “ungracious” remarks that I am personally responsible for, please draw them to my attention and I’ll review them.
By the same token, you should also be aware that I’ve gotten used to exceedingly ungracious comments from the Hockey Team. I haven’t noticed many climate scientists objecting to such comments. For example, you’ve posted the above advice here – fair enough – but you haven’t urged temperateness in comments at realclimate operated by NASA spokesman Gavin Schmidt or at the Rabett blog, said to be operated by NASA GISS supporter Josh Halpern, or even noted in your comment here that you take any exception to what Hansen said about me.
One of the few climate scientists to take exception to HAnsen’s comment was Andrew Dressler , who objected not so much because he thought that they were offensive to me (although his comments about me are pleasant enough), but because he thought that they weren’t good tactics (which they aren’t).
Anthony, regarding your desire to get into the Tejon Rancho to observe a site, why don’t you get a letter from one of your new acquaintances at NCDC to validate your purpose and use that to convince owners/officials to admit you?
Dr. Dressler makes a good point here also. Well except for the fact there’s a little dig after, it seems. Although it is pretty humorous….
RE94 John, excellent idea. Worst they can say is no.
#85. Agree whole-heartedly. And yes, as long as I have been visiting this site, Steve has been a gentleman, only loosing his temper maybe…. twice. 🙂
Steve is a gentleman. The rest of us do and say as we please. Steve’s impact on climate science has been and will continue to be positive, regardless of what the public and the media chooses to do the the royal climate orthodoxy that has scammed the public for so long.
Larry, you may as well take your guilt-inducing message to the oceanside and tell it to the eternal surf.
😉
Congrats Anthony on your presentation. Let me be the first to say that your slide assembly and template choice was first rate. Quite visually appealing, and believe me, having sat through many Govt slide presentations, they likely appreciated that as well. The one thing I missed though was the graphs for Marysville and Orland. Did you leave them off as outside the scope, or perhaps too heavy-handed? I found the juxtoposition of the Class 1 vs Class 5 temp graphs to be breathtaking myself. Again, well done sir and thanks to all volunteers.
#85
Larry, I also agree totally with your sentiment. Civility and common courtesy are essential. I believe that most NOAA scientists are objective, thugh some are clearly letting their values interfere with their science. I would hasten to add that in the blogosphere this is one of the more polite and civil sites – both with respect to acts of commission and omission. At RC, by contrast, different opinions are not only pulverized or they are excluded. For example, no mention yet has been made that Hansen has released the code and no mention yet has been made of Anthony’s presentation.
Larry Grimm says:
Yes.
I don’t think so, and here’s why:
Those are in conflict. You either have integrity and honesty, or you are out to change the world. You can’t have both. The former is the hallmark of a scientist, the latter is the hallmark of an ideologue. You can be one or the other, but you can no more be an ideological scientist than you can be an evangelical pimp. We’ve all made our choices, and it’s evident to me who’s chosen which.
Steve: Re my post (#85) and your comment. Oh dear, I meant no criticism of you. You HAVE been a stellar example of the best in scientific comportment. I was really urging contributors to your site to follow your lead. I offer the advice as constructive criticism to a group that I think would take it in the spirit offered – contructive. I have doubts that other groups would accept the advice as constructive.
Trebor,
That would only be true if the two variables were dependant. They are not. A site with both UHI and microsite issues is worse than a stie with just UHI or microsite issues alone.
Trebor,
Let me make an example. Tree rings.
Both water and temperature can make trees grow faster.
According to you, this means we would be safe in tracking one factor and ignoring the other.
As we have shown in the various tree ring threads. You can’t do this.
#102. Fair enough. Thanks.
To readers, there’s no need to be angry about things. Sure some of the stuff makes you mad; but generally there’s a funny side to it as well, a side that lends itself to – shall we say – jesting.
Of course, if you jest, Hansen won’t joust with you, but there’s a price for everything, I guess.
Will Hansen be issuing apologies for the “jester” comments?
#106. Mark, again there’s no need to worry about such things – Hansen’s going to wear the “jester” comment for a long time. Did you see John Brignell’s take on this? Priceless.
I spent some time this morning looking at the stations with CRN=1.
I wrote a little program to read the GHCNv2 monthly means data file, parse a subset of the stations, and dump the output to a CSV file for Excel.
I then took the CSV file, loaded into into Excel, and averaged the station temperatures every month. The plots below show the 1-year and 5-year averages of the station monthly means:
The 1976-2006 trend is 7e-5 degC / day, or 0.26 degC/decade.
I can make the program available if anyone can offer hosting.
I forgot to include the corresponding GISTEMP plot:
(My plots only go back to 1900 because Excel can not handle dates prior to 1900).
In regards to UHI proxies.
So far we have population, population density, energy use and wind velocity.
It seems to me that the strongest proxy would be some kind of population density, factored by distance.
On top of that would be an energy use per capita factor. Perhaps as a multiplier for the first factor.
Windiness is much more problematic. You would have to factor in wind direction as well as wind speed, plus the average population density in the direction the wind is blowing from. Eeek. Hopefully the first two factors would be enough by themselves so that we wouldn’t have to open up the wind “can of worms”.
MarkW
As someone else mentioned there needs to be a factor that reflects the actual land use – asphalt and concrete versus trees, grass and dirt
Sorry for so many posts — I should have mentioned that I used the un-adjusted GHCN data.
# 108
Well done John V. Quick work with interesting results. And kudos for offering up your work for independent inspection.
The remarkable aspect is not only the slope 1970 onwards but the similar slope 1902-1920. This looks like the result of some underlying cyclic phenomena with an approximately 70 year period. Just to be contentious I would hazard a guess that OCO could be ruled out and that something else is involved. It may be fruitful for climate science to consider some of the other known influences.
John V, That’s interesting. It does appear that a bit of bias has been removed from the most recent data.
What’s the significance of the line drawn through the blue part in the second chart? If it’s to show a trend, you could just as easily draw a line from 1933-1968 and show an even more alarming cooling trend. Is that supposed to indicate something significant?
John V,
Could you do a plot of the 5 year average temperature vs CO2?
Moreover, can you get the dew point temperature from the same sites and plot the dew point vs. average temp?
RE 85.
Good advice. After SteveMc announced that Dr. Hansen had freed the code, I urged
the readers here to thank the team at Nasa, using the Real Climate site. I posted
my thank you. It was rejected.
John V,
I take it that some of the GHCN data files indicate which stations are CRN? How many stations are CRN and what GHCN files contain this CRN flag?
Re #112
Better not to make adjustments until their is a justifiable reason for implementing them.
I’ve tried to understand how the Hadley Centre does them from the references at their site. I thought I was reasonably good at logic and mathematics but I must say I retired gracefully in the face of the effort involved in relating it to basic scientific principles. If memory serves me right there was an explanation of the sdjustments needed when an observatory moved location so that when at the new location you were actually creating the temperatures you would have seen at the original one. The logic seemed somewhat obtuse. If one of my students had produced that (when I was working in experimental physics) my suggestion would have been for them to go away and through just precisely what their objectives were.
John V,
One thing that puzzles me about your graphs of CRN stations is how the 1998 El Nino effect appears to be absent. In fact, your graph makes it look like a cooling period during the mid ’90s, a time when the AMO flipped to its positive phase.
#114 Larry:
The trendline from 1976 onwards was the result of a previous conversation on a different thread (unthreaded #19 I think). Somebody suggested that the best period to look at is 1976 onwards because satellites were available and the solar trends appear to deviate after 1976. Also, the only apparent trend in GISTEMP US data is from the mid-1970s.
#115 DocMartyn:
A 5-year trend vs CO2 would not say much because of the time constants involved.
#117 Aaron Wells:
I used the CRN ratings from the Excel file posted by Anthony Watts at the top of this thread. There were 17 stations reporting for almost all of the period from 1900 onward.
This was a quick first attempt. I have not looked at the spatial distribution of the stations or any of a number of other factors.
If anybody wants to run with this I can make the code available.
I can’t afford to put much more time into this. I can make the program available.
119, good catch. It could be that the dates are off, but if not, there’s a serious issue with the locations. I would particularly expect to see a spike in the 1-year averages, that doesn’t appear evident at all.
RE: #110 – Wind. In some places, especially where there is not a strongly dominent wind direction, I seriously doubt that wind matters much at all. When it’s windy, heat will be moved a bit faster. As a result, I would expect elongation and elevation of a somewhat “dilluted” UHI impacted volume. On the average, with no dominent wind direction, the elongation becomes simple radial increase of the volume and dillution of it. Integrate over time (including non windy days), and all you do is slightly lower the average long term “density” of the UHI. The outliers would be places that have a highly directional wind (such as here – where there is strong bias toward NW wind) and places with very little wind. But in the grand scheme of things, I think it is a minor impact. Average across all places impacted by UHI and wind becomes almost nothing in terms of impact. If anything, it simply means a minor decrease in urban and suburban impacts, and an increase in small UHI effects on hinterlands, which, with no wind, might not be affected at all.
I can’t seem to leave this alone…
Here’s a plot of the CRN=1 stations with GISTEMP. The difference is GISTEMP minus CRN1
The comments above about issues in the correlation between GISTEMP and CRN=1 prompted me to include the CRN=2 stations:
There are between 35 and 53 stations reporting every month.
(I really will try to stop posting after this)
Re: 113 The Gleissberg cycle?? Murray
Hansen has already come down and stated that the Y2K bug that “adjusted” the US readings didn’t affect the ROW readings because the US readings are only 2% of the world’s surface.
What will his answer be when only 9% of the surveyed stations in the US are acceptable? That puts the US reading with even LESS relevency.
On top of that, there are a couple of other points I predict:
1. When the audit is done, he can say that our results don’t match his because we “modified” his original code.
or
2. He’ll try to claim the “modified” code as his own.
RE 110 and 122.
Click to access IAUC0002.pdf
Click to access indexCD.pdf
Click to access PPG_2005_SouchGrimmond.pdf
The last may be the jackpot find.
RE: #127 – Notice someone missing from the references of the last one? Not that it means anything, however ….. 😉
re 127.
What stuns me stupid is that people still make this “cool urban park” argument.
Talk about urban legends. I didnt see Peterson, Parker, or his royal highness.
Great work, John V
It would also be interesting to see plots of CRN=4 and 5. Is there a strong bias in the poor stations, in a particular direction?
I just had a look at the locations of the CRN=1 and CRN=2 stations:
– CRN=1 station locations are biased to the south east
– CRN=2 station locations are biases to the north
– Combination is fairly well distributed except for the midwest
Before I make more plots I need to look into a better way of averaging. The simple arithmetic average has some issues because of the location biases.
Also, I have not looked into which sites are urban. If Anthony Watts (or anybody else) could provide me with a list of rural sites with CRN=1 or CRN=2 I will re-run the analysis on those sites only.
First: let’s leave Hansen alone, and try to see what we can learn from the data. I have no love lost for Hansen, but he isn’t that important.
On biases: seems to me that when calculating trends – it’s not the stations bias itself that is important – a consistent bias wouldn’t affect the trend. What matters is CHANGE in bias – such as: move of station location, change (or recalibration) of instruments at station, change of reading or reporting times, change in the environment (such as new buildings, new air conditioning equipment added over time, etc.).
I’m not sure that even the best stations (level 1 ) are free of these bias-changing events, and it’s also probable that at least some of these events haven’t been properly recorded. So I would say that even level 1 stations cannot be considered free of error margins.
Lamentably, this is the data we have, we must work with it, but keep in mind the error margins.
I’m looking for some help in determining the geographic extent of the lower-48 so that I can properly average the station temperatures.
Simplistically, this is what I’m looking for:
A program that will take a latitude and longitude and return true if it is in the USA, false if not.
Can anybody point me to something like that?
RE 133.
Well from anthony’s excell yu get this.
Lat Min: 24.55
Lat Max: 49.00
Lon min: -67.00
Lon Max: 124.37
So that will probably include some illegal immigrants from canada and mexico, but it’s the
Best I got.
RE 134:
That’s a start, but it includes a lot of ocean that should be excluded.
I may be able to approximate using scan lines from Anthony’s Excel file of station locations, but there must be a better way.
Google Maps has the info, USGS has the info, MapQuest has the info — where did they get it? And is it available in a useful form?
RE 131. I found the same thing.
Especially if I fiddled with start years.. 1975..74.. 76… 65..
I did not like the R2 I was seeing and the residuals were not pretty.
I had suggested this piecewise comparision ( 1910-1940; 1950-1975; 1975-2000)
in the hopes of better fits, but its turtles all the way down Mr Mandelbrot
I averaged 1&2s ( I played with Urban/Rural/Small town) and then I figured
That I was data mining.
So that will probably include some illegal immigrants from canada and mexico, but it’s the
Best I got.
re 132:
The only reason Hansen is important is that he is paid by tax payers. If he were funded by a private company, he would truly be irrelevant.
RE: #129 – Significantly, no Parker. No value add? I reckon…. 😉
John V. When faced with data that shows that temps have been as high, or higher in the fairly recent past, some Warmers say that it’s not the absolute temp that matters, but that the recent 1976 onwards rate of change is “unprecedented”. But, looking at your graphs, the 1912-1935 ish rate of change, leading to a peak in the 30’s is very similar to the recent rate of change and peak. So the recent (1976 on)temp rate of increase, and peak is not “unprecedented”.
CO2 cannot be the primary driver of temperature change, as according to “everyone” there was a lot less CO2 in the atmosphere in the 30’s.
[snip]
Steve Mc: first time, long time
133
John V, it sounds like your are after something from mapinfo a very handy though expensive program that is made for mapping data. There are various user groups that may be in a position to advise.
RE140 MarkR
Please lets stop this Hansen and Gavin sniping and bashing and get on with the analysis. I really don’t care what anybody thinks of them, or me, or Rabett or any persona.
What matters is the data and what we can learn from it. Opinions about personas don’t make science. Data and data analysis does.
Client: “What is 2+2?”
Geologist: “Well, based on empirical and observational data, somewhere between 3 and 5.”
Engineer: [who pulls out an HP calculator with RPN] “It is 4.00000000000”
Climatologist: “What do you want it to be?”
OK, for the original joke replace climatologist with geophysicist (it was an often deserved rip of my own field)
#142 Congrats Anthony.
In answer to #85, I agree. But I would like to point out that there are problems or shall we say differnces of approaches. Steven Mosher and I apparently were excited by Anthony’s work when he started with the screen experiment. Ray Ladbury, Tamino, Boris, BigcityLib
144, Egad…
Correct.
The monkey crap coffee bean theory?
Umm, Gavin, how do you know that?
That must be the “celebrate diversity” school of statistics. Whether or not it’s foolish kinda depends on your objectives, doesn’t it?
133
mapquest has:
http://www.mapquest.com/maps/latlong.adp
I guess if it shows a completely blue map you’ve hit the ocean.
And:
Er, specifically WHAT additional lines of evidence?
The above quote of #144 was from
http://www.realclimate.org/index.php/archives/2007/07/no-man-is-an-urban-heat-island
But we (Hansen et al)can throw out UHI as measured by EPA; We can throw out LIA and MWP and now SA, Africa, and NA; but we ARE NOT ARBITRARILY THROWING OUT DATA (just the data we don’t like, lol). Ladbury et al, (lol) maintained that essentially it was a high quality network.
One of my favorites from Ray Ladbury
I worder what he thinks now that it looks like we can delete 90% of the data assuming that if the anticipated error is as great or greater than hypothesized phenomena, it should be discarded.
So, I guess the challenge is, what will NOT pointing out their errors and high-handedness get us? I and others here I am sure don’t mind paying for what we get. But just what will we get? I personally and professionally have found that adversarial roles help more than “consensus”. Apparently NASA used to , too. At least when they had their “tiger teams”. I would like to think that the effort we have made, most especially Steve McI, Anthony, and others (John V, Not Sure, etc, etc, lately to name just a few) were as the ‘tiger team” that AGW needed, but decided they could do without.
So thank you Anthony with your IR slide that shows the potential error. And thanks to all the volunteers.
So for #85, I know that several of us are happy to see what we thought was a reasonable hypothesis offered by Anthony prove to still be viable. But, I have also learned in life that those who show no mercy should not expect mercy themselves. We do not need to pile on here, but rather be as careful as surgeons with our “jests”. If you are on Hansen’s mail list or have read some of his “unofficial” comments, most people on this site appreciate humor, not the chainsaw appraoch Hansen seems to have for his critics.
RE 139.
As I look at records I keep seeing this pattern of cooling the past… Not so much
Warming the figures from 1975 to the present… But cooling the past..
Kind of a hockey stick get rid of the MWP strategy applied to the 20th century.
John V.
I apologize if I missed this buried in the comments, but what version of data did you use for the individual sites for your CRN = 1 composite plot? Is it from GISS? If it is, is it the raw, combined or homegenized version? If it is homegenized, the good sites are probably influenced by their bad neighbors.
#133. What you need is a reverse geocoder. Here’s one:
This returns the country code of the lat/lon, e.g., US
If you need more details, i.e., state, there are lots more, just google for reverse geocoder.
http://ws.geonames.org/countrycode?lat=29.7&lng=-98.1
#133 John V.
The quickest and easiest way would be to import a text file with the lat/long coordinates into a GIS program and then select only those points which fall within the US boundary. You could then export those points back out so that anything outside the US is eliminated from the data. This is relatively easy to do but you need access to GIS software. If you don’t have access, I can do it for you if you want to send me the file.
John V
Nice plots. I don’t know if I am reading this correctly but it appears that GISTEMP(Full) is essentially the same as GISTEMP(CRN=1 subset). Therefore might we conclude that quality of the site stations does not materially impact the results. Maybe it would be interesting to compare a random subset with the CRN=1 subset.
If we are more concerned with trends rather then absolute temperature then recent site quality is likely less of an impact on trends then changes in site quality over the years.
The trends do appear unmistakable (~.26C/Decade as John V pointed out). The trends do seem to be corroborated by Balloon and satellite data, although analysis of both data sets were adjusted after they initially did not show a warming.
So why spend time trying to pick apart one mans analysis of surface station records when other efforts HADCRUT3 show similar recent trends along with satellite and balloon data?
Just what does it mean to be CRN=1? Or CRN=2? These do not equate to the site scoring 1-5 does it?
Aaron Wells:
Yes, the CRN=1,2 etc does relate to the scoring. I should have made that more clear.
Jeff C:
I used the raw, unadjusted GHCNv2 monthly data from here: ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/v2 (file v2_mean.z)
Curve Shape:
For everyone discussing the shape of the curve — that’s not the point. The GISTEMP temperature profile for the USA is well known (peak in the 1930s, low in the 1970s, rising now), and it is different than the trend for the world:
GISTEMP USA: http://data.giss.nasa.gov/gistemp/graphs/Fig.D_lrg.gif
GISTEMP World: http://data.giss.nasa.gov/gistemp/graphs/Fig.A2.lrg.gif
I am attempting to compare the GISTEMP results with my own analysis done using only the best stations (the 4% with a CRN site quality rating of 1). There are some weaknesses in my current analysis that I will attempt to fix tonight. In particular, I am currently using a straight average of all stations without considering their locations.
More plots to come if the next few hours go well…
Ok, I’ve got some new results…
My previous plots were using a simple average of all of the stations with a CRN site quality of 1 or 2. The problem with that approach was that the stations were not evenly distributed.
=====
Procedure:
So, to get around that I created a grid of data points to cover the lower 48. I did this by tracing the perimeter in Google Earth, exporting the perimeter points to a data file, importing the data file into my program, and checking every point against the shape defined by the perimeter (whew!). I gridded with a dimension of 0.5deg in latitude and longitude, which equates to approximately 55km by 32km. There were 3308 grid points to define the lower 48.
Since the goal is to compare the GISTEMP results using all sites against the results using only good stations, I tried to match the GISTEMP algorithm for calculating the overall average. I did the following for every month:
1. Create a list of all stations with readings and their locations
2. Calculate the temp at each grid point (3308 points) by averaging all stations within 1000km, using a linear reduction in the weight of each station to 0 at 1000km
3. Average the temp at each grid point to get the overall average for the month
I then averaged the monthly averages to get the yearly average.
Note:
– I did not estimate any values
– It was not necessary to merge any station records
– Using only CRN=1 stations, approximately 55 cells were more than 1000km from all stations
=====
Plots:
Each of the plots uses the following colours:
GISTEMP (dark blue)
My Calculations using only CRN=1 stations (orange)
My calculations using only CRN=2 stations (red)
Plot 1 shows the 1-year and 5-year averages:
Plot 2 shows the differences between the GISTEMP results and my results on the same scale as plot 1:
Plot 3 shows the correlation between GISTEMP and my results, using all data from 1920 (their are obvious problems in the CRN=2 stations prior to 1920):
Plot 4 shows the correlation between GISTEMP and my results using all data from 1970 (for a modern perspective):
=====
Conclusions:
It seems to me that the GISTEMP program (with all its flaws and including bad stations) gives results that are very close to my results using only the best stations. This is particularly true from about 1970 onwards.
I suspect that some of the remaining differences will be reduced when the complete network of good stations (CRN=1 or 2) is available.
Since GISTEMP compares well to the results with good stations in the lower 48, is it fair to assume that the worldwide GISTEMP results are also quite accurate?
Re#113, Dr. Marvin:
The remarkable aspect is not only the slope 1970 onwards but the similar slope 1902-1920. This looks like the result of some underlying cyclic phenomena with an approximately 70 year period.
How about old good PDO (from Wiki, sorry):
1905: After a strong swing, PDO changed to a “warm” phase.
1946: PDO changed to a “cool” phase.
1977: PDO changed to a “warm” phase.
1998: PDO index showed several years of “cool” values, but has not remained in that pattern.
http://en.wikipedia.org/wiki/Pacific_decadal_oscillation
#157
Interesting. What happens if you overlay the adjusted data?
Re. 107
It seems that Canada found jesters to be sufficiently useful to fund as late as the 1960s!
See: http://considerthis.onlinedemocracy.ca/index.php?name=News&file=article&sid=345
Maybe this would be a good way to fund your activities. Apply to the Canada Council for a grant – $3,500 in the 1960s must correspond to a tidy sum now
Re#160, sc:
Sorry, I would rather not to risk my hide to be burned at the stake:
http://books.google.com/books?id=DJxlzuOdK2IC&pg=PA109&lpg=PA109&dq=global+warming+witches+dark+ages&source=web&ots=vYjUD6wPZJ&sig=k_F-oi-I-JfUrXTQ4_Kt5dfBMEY#PPP1,M1
P.S.: must read to Steve Sadlov.
Also, I have yet to finish some courses to have credentials to apply for governmental grant in consensus climate science:
http://www.theage.com.au/articles/2007/07/18/1184559843317.html
Woops, sorry.
I did not recognize that #160 addressed Steve McI., not me.
Still, I somehow think that Steve would like the answer
Re#159:
Influence of PDO on global climate is still unquantified. The only thing suggesting that Pacific SST has serious effect on global temperatures is well documented effects of El/La Nino/a (e.g. 1998) on global climate.
157,
If I correctly understand your plot 1, despite there is almost perfect much for period 1975 onwards, there is a large difference between your and NASA calculations for period 1922-1950. It looks like they carried out substantial adjustments downward for that period.
#157: Nice John! If you find time, please try the following modifications to your algortihm:
1) Use first differences (FD). That is, take (month by month) the difference between the previous year and the current year. Work with that data. Finally, when everything else is done, you cumulatively sum (over months) from the most recent value backwards. This gives you the temperature change with the reference (the current) being 0. Of course you can then rescale the final series with respect, e.g., 1951-1980, reference period.
2) Instead of using weighted average (mean), use weighted median.
This should be more robust approach.
John V,
Thanks for your fine graphs.
CRN1 is ok with GISTEMP in the early 1900s but not in the 30s.
CRN1,2 is the reverse. All in all, uncertainties due to sampling difference is considerable.
Besides, it is a verifiable fact the IPCC has lowered temperatures in the early 1900s in its successive reports. Thanks to this revisionism, they announced a warming of 0.6°C/century in the 2001 TAR then and 0.74°C/century is their 2007 report even if recent temperatures have stabilized. A scientifically meaningless change because of uncertainties but a PR spectacular warming.
Have you compared linear trends per century between GISTEMP and CRN1,2 ?
Hello John V.
Thanks so much for doing this analysis, your detailed effort is appreciated.
I had planned on doing something similar, and I know Steve McIntyre is also working on something along these lines.
Before I comment further, I’m wondering if you’d be able to run the same analysis method on max temps only, then min temps only, discarding calculating any mean or average of max and min.
Per Demesure,#164 “linear trends per century between GISTEMP and CRN1,2” certainly would be interesting as well if you have the time and inclination.
And might I suggest trying a different data set source, just for kicks and giggles. I visited the ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/v2 link you provided, and noted that many data sets show dates corresponding with today. I’m not certain if it is SOP to do a daily update there or not. Perhaps SteveM has some older versions from CDIAC he can lend you.
Given that I have seen “raw” data change online (GISS Walhalla, SC for example) in the past couple of days, I think it is prudent to look back a couple of months and run the analysis again.
bernie,
I believe that population density would be an adequate proxie for things like concrete and asphault. Those things tend to be fairly linear with population. More people, more concrete. Not perfectly linear, but not far from it.
SteveSadlov: 122:
In those instances where the station is near the edge of town, wind from one direction could provide cooling, while wind from the opposite direction could provide warming. You would have to do a study for each location based on average wind speeds and wind directions.
You may be right and population might work well for US cities, but I do not think it will work as well for cities in South America. Sao Paolo for example has 10 million people and Buenos Aires has 15 million. These cities feel very different from New York and Los Angeles.
JohnV:
Intriguing analysis. What makes for the apparent significant difference between CRN(1) and CRN(2) stations?
bernie,
From the pictures I have seen of Sao Paolo and Buenos Aires, they both have lots of concrete and asphault. Being poorer than the US, I wouldn’t be surprised to find out that there energy useage per capita is a lot less than ours.
There may be need for a per capita income modifier to the population density proxy.
Up to a point, wealthy countries lay more concrete per capita.
I never said this would be easy.
It could be that building construction may make a difference. Definitely the type of roofing material will make a difference.
Argh, another potential modifier.
At this rate we may just have to do a complete study for every sensor in order to guess what the UHI correction would need to be.
Even after we make the adjustment for the present, trying to make back adjustments to the historical record would be further excercises in ray guess work.
Let’s just throw the whole thing out and start again from scratch.
I do not have any evidence to back this up but I suspect that urbanization (population density) would affect the station class. By that I mean that using the CRN class 1-5, which are accurate for ‘rural’ environments, that population density might have an impact on the distances listed in the class definitions. For example, putting a station in the middle of central park might achieve the requirements of a class 1 station but due to the population density, instead of 100 meters, it may be 1000 meters (or more) to meet class 1 requirements.
Anyway, that is just an idea that I have been bouncing around but I do not see away to prove it at this point.
John V. #157
Wouldn’t it be more precise to say that you used a “judgement sample selected from the best stations” rather than “only the best stations”?
What is your basis for this conclusion? If station quality does not impact the trend and if your sample of good stations is unbiased,
wouldn’t the logical conclusion be that the differences will remain about the same?
No. The set of problems associated with the worldwide network (sparse spatial coverage, large time gaps, impact of World Wars and
sporadic development) is much larger than the set of problems associated US data.
And finally, if you’re only going to compute trend lines starting in 1976, why bother with all this other work? And quite seriously, what is your basis, scientific or otherwise, for computing the decadal trend based on just 3 decades?
Hi I’m new here but I thought I’d post to see if anyone could give me suggestions for a project I’m working on. I have imported all of the stations that have been surveyed by the surfacestations effort into ArcGIS. I am also importing data layers that I think may influence the readings of a station. I will then analyze the layers for relationships between CRN 1,2,3,4,5 and the attributes of the various data layers. Right now I have a layer for “Nighttime lights of the U.S”, and I’m retrieving LANDSAT7 land cover data, and population density data. Can anyone suggest other things I might look at?
JohnV
Is the variance among the different CRN categories a way of estimating the error in GISS readings?
re 175.
Use of coal for eletrical generation. Aerosal induced cooling.
#175
Distance from the sea, altitude ?
#175
Distance and direction from the population center. Might shed some light on UHI effects.
#175 and 178 Thanks for the response.
I’m sorry, I misspoke. Not “influence readings of a station” but influence the CRN rating. What I want to do is create a model, based on the currently surveyed stations and their relations to the datasets that I use, that gives an idea of what kind of shape the rest of the US stations are in.
Your suggestions would certainly be good for looking at things that influence the readings of the stations and I can try to look at those things if you can suggest something specific that you want me to analyze that might be useful. Like I said, I’m very new to the issues you guys are working with here so it’s difficult for me to specify a particular spatial analysis. The first part just seemed like something obvious that might be useful.
Mike,
I always thought altitude is a (generally) good indicator of population. Since it’s readily available, it may be worth examining: temperature should rise more over the last century with low altitude (meaning faster urbanisation). Well, just a thought. It would be surprising nobody hasn’t had this idea before me.
Demesure,
Interesting, I’ll grab elevation data and look at that to. It’s easy enough to get.
135 John V. says: September 13th, 2007 at 4:20 pm RE 134: “Thats a start, but it includes a lot of ocean that should be excluded.”
That’s the problem with dealing with the edges of the US: the land at 5×5, the overlap into Mexico or Canada, then the mess of converting the 2×2 ocean to 5×5 land-sized and combining the squares based upon ratio. Just take a look at the Los Angeles grid square. (30-35N, 115-120W) Ugh, what a mess of calculations and methods and explanations.
Can anyone here point me to a reference for how stations were classified as lights=0 or lights=1? This seems overly simplistic to me. The satellite image of lights in the US assigns a value from 1-63 according to the amount of light in a given cell. I assume the determined a cutoff point say “55”. And every cell with a value above 55 is classified as light=1 or urban every cell with a value below 55 is light=0 or rural. Does this sound right? Thanks for putting up with a newbie.
RE 184..
Hansen uses three levels of classification. See Hansen 2001.
I’ll excerpt the relivent passages.
Also, look at pop density over time down to the zip code of the site.
Hansen 2001
Hansen et al. [1999] attempted to minimize urban influence on the analyzed temperature change by
identifying urban stations and adjusting their record such that the long-term trend was the same as the mean of rural
neighboring stations. Urban stations were identified from local population data provided as metadata in the GHCN
records. Problems with this approach include the fact that the population data were typically two decades old, so it
could not describe accurately recent urban development. Also, the effective spatial resolution was poor, as it was
not possible to tell whether a station was located in the city center, suburbs, or outskirts of the region with specified
population.
As an alternative approach to identifying stations subject to human influence, we test in this paper the use
of satellite observations of nighttime light emissions. Specifically, we use observations from a United States
Defense Meteorological Satellite taken with a highly sensitive photomultiplier tube [Imhoff et al., 1997].
Observations employed are generally those taken under a new moon to minimize reflected moonlight. A composite
of many images is used to eliminate ephemeral light sources such as lightning and fires. The observations were
acquired in 1995, so they do a good job of describing current urban development. The same data have been used to
quantify the effect of urban development on primary productivity [Imhoff et al., 2000]. The spatial resolution of the
data used here is about 2.7 km.
Plate 1 illustrates the night light data. The percent of brightness refers to the fraction of the area-time at
which light was detected, i.e., the percent of cloud-screened observations that triggered the sensor. These data are
then summarized into three categories (0-8, 8-88, and 88-100%). From empirical studies in several regions of the
United States, Imhoff et al. associate the brightest regions (which we designate as bright or urban) with
population densities of about 10 persons/ha or greater and the darkest (unlit or rural) regions with population
densities of about 0.1 persons/ha or less.As is apparent from Plate 1b, the intermediate brightness category (dim
3
or periurban) may be a small town or the fringe of an urban area. Some of the regions defined as periurban may
be a consequence of reflected light from urban areas, bleeding between detectors, navigation errors, and other effects
that spread the urban influence [Imhoff et al., 1997]. However, these problems do not prevent us from using the
periurban brightness category to identify areas where the likelihood of human influence is greater than in the unlit
regions but less than in the bright regions. As is apparent from Plate 1b, the intermediate brightness category (dim
3
or periurban) may be a small town or the fringe of an urban area. Some of the regions defined as periurban may
be a consequence of reflected light from urban areas, bleeding between detectors, navigation errors, and other effects
that spread the urban influence [Imhoff et al., 1997]. However, these problems do not prevent us from using the
periurban brightness category to identify areas where the likelihood of human influence is greater than in the unlit
regions but less than in the bright regions. The average population density in the periurban class is 1 person/ha.
Demesure, #181:
It’s possible it’s a good indicator, but the exception might be deserts. Not a lot of people living in Death Valley.
RE 181.
Population and population density change over time so altitude would be
a lousy proxy, unless the people were really fat and caved the earth in.
#187
Steven:
Absolutely, we need proxies that allow for the measurement of change over time. The old UHI formulas inlude a log(pop) value. I would like to see a more formal specification of the measurement of UHI that has been empirically tested across a range of sizes, densities and income levels.
Steven,
Thanks, that’s exactly what I was looking for.
Regarding pop change over time: I think you would want to analyze this at the finest temporal resolution possible rather than say, the change from 1900 to 2007, correct? My thinking is if an area underwent a boom at some point in the last century and experienced subsequent pop loss, the infrastucture would probably still remain, impacting the integrity of the station. You wouldn’t be able to detect this by just analyzing start point A and end point B. Is this sensible?
#157 John V. I’m very impressed with passing strangers, who with their first posts on another thread seek to deflect with discussion of CO2, and then turn out to be wizard with stats, and have all the software and everything, and really bust a gut to get graphs up first, before the author, and then conclude:
When as anyone can see, the data from 1900 to 1965 diverge markedly, and one of the charges against GISS at the moment is that they have downplayed early 20th century temps to make 1975 onward temps look high, then based on the data, your conclusions are a gross distortion.
It looks as though your purpose is to hijack, and “spoil” the thread. As you obviously know stats, has it not occurred to you that a true representation of the temperature record is in the interest of all right thinking people.
Why do you seek to protect GISS, who you must surely see, seem to have got it all wrong?
MarkR, It would help me if you showed that there was a flaw in what JohnV has done. To date, CA has focused on producing accurate and replicable work. You may be right, but at this juncture your assertions have no empirical basis. I am confident that if JohnV is misrepresenting the data that will emerge in relatively quick order.
#191 Bernie. Certainly, see #123 “Heres a plot of the CRN=1 stations with GISTEMP. The difference is GISTEMP minus CRN1”. See the difference in green. That is the amount GISS appears to be adjusted away from real temperatures, over .05C in the Thirties. The adjustments are as always in favour of making recent times seem very hot.
What follows, including his choice of #157 Plot 4 is pure disinformation and deceipt. Why not choose 1909-1929 for a measurement of rate of change and corelation?
Answer: Because it would show that the rate of increase, and max temp for that period was similar to the recent period he chose.
#192 oops, typo alert, “That is the amount GISS appears to be adjusted away from real temperatures, over .05C in the Thirties.” Sorry that should read 0.5C… half a degree Centigrade.(That’s mucho by Warmer standards).
Re: #157
Your plots show some significant differences between station categories and GISS over significant lengths of time. That they have a reasonable R^2 should not surprise or, as a matter of fact, be used as an indicator that all is well in Hansenville. With the claimed uncertainties of GISS of their temperature measurements, would not these differences be considered to be out of the 2 sigma limits? Have you made that calculation?
I would think that while trends may not change greatly with better station data, that the uncertainty levels may have to be increased significantly. It is the uncertainty that is often underestimated by the producers of time series like GISS.
MarkR # 190
Youre right, its another bait and switch. Hes using USHCN v2 data. GISS also uses USHCN v2 data except they add other stations. What John V. did was take a small sample of the GISS data (already adjusted) and compared it to the total of GISS data, also adjusted.
Well.
Here is something.
If you look at the Class 1&2, you will find a high percentage of ASOS
If you look at class 5 you find no ASOS
I think Class 1&2 criteria miss ASOS issues.
First: A CRN site versus ASOS, I’ve posted this before
Click to access Sun.pdf
Second:
http://adsabs.harvard.edu/abs/2000PhDT……..25K
third: Geography and technology. Brunn, Cutter , Harrington
See page 465 for Issues with ASOS.. Intrument etc. See page 466 On the CABLE LENGTH ISSUE
ANTHONY!
http://books.google.com/books?id=z7noWM7vAz4C&pg=RA1-PA465&lpg=RA1-PA465&dq=asos+bias&source=web&ots=AqKJlvZtbM&sig=8g3ma2kK6P1BQXX9_save69OQDM#PRA1-PA464,M1
This book might be a good purchase Anthony, plus it references studies n ASOS and the
Cable length issue
SteveMc.. Giss ingests USHCN..
Are these changes
Click to access Brown.pdf
made before hansen touches the data?
Re #197,
steve,
I have not looked at that pdf, but “official” USHCN adjustments are made at
NOAA’s NCDC before GISS gets them.
From: John V. September 14th, 2007 at 1:33 am
To: JS September 15th, 2007 at 12:07 pm
That would be about 34 1/2 hours from start to definitive finish.
Good detective work!
RE196 Mosh pg466 is not available in preview, do you have it?
I see it here
http://books.google.com/books?id=z7noWM7vAz4C&pg=RA1-PA465&lpg=RA1-PA465&dq=asos+bias&source=web&ots=AqKJlvZtbM&sig=8g3ma2kK6P1BQXX9_save69OQDM#PRA1-PA466,M1
Cut and paste doesnt work
Basically.. Covers the H0-83 Issues.
Big issues between 1930 and 1950 When sites moved to Airports
The On Pg 466, Cites Qualye et al and talks abut the issues of MMTS and the cable length
causing sites to move close to buildings.
http://ams.allenpress.com/perlserv/?request=get-abstract&doi=10.1175%2F1520-0477(1991)072%3C1718%3AEORTCI%3E2.0.CO%3B2
Might be it
Hey, Tom Karl used to be a skeptic.. kinda
http://adsabs.harvard.edu/abs/1991GeoRL..18.2253K
East west..hot versus cold since 1895
http://ams.allenpress.com/perlserv/?request=get-abstract&doi=10.1175%2F1520-0493(1980)108%3C0249:TCOTUS%3E2.0.CO%3B2
The referenced study ( Qualye
Anthony,
Google Qualye .. He has lots of good stuff
http://ams.allenpress.com/perlserv/?request=get-abstract&doi=10.1175%2F1520-0426(1990)007%3C0334%3AAROCTD%3E2.0.CO%3B2
And some stuff on SST and ships logs… Where have all the data gone
http://md1.csa.com/partners/viewrecord.php?requester=gs&collection=TRD&recid=200144001574MT&q=%22author%3ARG+author%3AQuayle%22&uid=791297055&setcookie=yes
2 Trackbacks
[…] 2009-Sep-09 (Wednesday) JohnV Leave a comment Go to comments The original article “USHCN Survey Results based on 33% of the network” was posted by Anthony Watts on September 12th, 2007. At that time there was lots of […]
[…] quality assessment of the USHCN stations here. John V presented some graphics in the comments thread here and below is my first pass – this comment is not intended to exhaust all possible cross-cuts […]