The www.surfacestations.org project continues to collect new stations, though we could certainly use more help in the midwest, particularly Kansas, Nebraska, Oklahoma, and Texas.
This NOAA USHCN climate station of record #415018 in Lampasas, TX was found to be tucked between a building, and two parking lots, one with nearby vehicles. According to the surveyor, it is right next to the ACE Hardware store on the main street of town. While likely representative of the temperature for downtown Lampasas, one wonders how well it measures the climate of the region.
View looking NE
In her survey, volunteer surveyor Julie K. Stacy noted the proximity to the building and parking, which will certainly affect Tmin at night due to IR radiance. Daytime Tmax is likely affected by the large amount of asphalt and concrete in the area around the sensor. The main street of the town (28 ft from US 183) and the ACE Hardware parking lot are visible in this photo below:
View looking south
Google Earth shows just how much asphalt and buildings there are around the sensor.
According to NCDC’s MMS database, the Lampasas climate station has been at this location since 10-01-2000. Previous location was an observer residence, which appears to have been a park-like location according to MMS location map. The sensor was apparently converted to the MMTS style seen in the photo in 1986, so the move did not include an equipment change. See the complete survey album here.
But the big surprise of just how bad this location is came from the GISS plot of temperature. It clearly showed the results of the move to this location, causing a jump in temperature almost off the current graph scale. Note that before the move, the temperature trend of Lampasas was nearly flat from 1980-2000.
Click to see the full sized GISS record
Given the entropy of the current measurement environment, I have sincere doubts that anyone can create an adjustment that will ascertain an accurate trend from temperature data as badly polluted as this. In my opinion, this station’s post 2000 data needs to be removed from the climate record.
Since there has been some discussion about how well “adjustments” take care of such problems, I thought I’d show you just how well the GISS homogeneity adjustment works with this station.
Here is the GISS plot for Lampasas, TX with the GISS homogeneity applied, I’ve changed the color to red and labeled it to keep them visually separate from the raw data shown in the plot above.
click the plot to see the original plot from GISS
Now here is the GISS raw data plot with the homogeneity plot overlaid on it:
The effect is quite clear. The recent “spurious” measurement remains unchanged, and the past gets colder.
The result? An artificial warming trend for this station that is created by GISTEMP adjustments.
105 Comments
Say it ain’t so. Joe.
Given that this relocation took place after NOAA had clearly identified the importance of homogeneity in the USHCN network, this is pretty annoying.
It’s hard to figure out what the active ingredients in the adjustment are. This site is classified as “bright” so the adjustment would be based on the history at non-bright sites in the area. It will be interesting to see what they look like.
Readers also need to keep firmly in mind that CRU and NOAA do not carry out the GISS homogeneity adjustments.
Here’s a plot for nearby Blanco TX – the first nearby rural site that I plotted. It looks like there is some similarity in the two plots. It doesn’t look impossible to estimate an adjustment, so it’s interesting that the NASA adjustment is so inadequate. The various USHCN versions also need to be plotted to verify what they’ve done.
Andrew…What about this: Llano some 35 miles SW of Lampasas 1999: 19.27C
2001 17.65………….Lampasas 1999 18.61C , 2001 18.68C Let’s see …makes
a difference of 2 global warmings in 2 years…Texas not being completely
flat in this region I also checked Blanco, 55 miles SSE of Llano and 70 miles
SSW of Lampasas…Guess what: 1999: 19.47C … 2001: 18.18C…As this is
Wacko-science…I also checked Waco (pop. 120.000): 1999: 20.39C … 2001:
19.09C Lampasas pop. 5-6000, Llano some 3000, Blanco some 1500…
Any more witnesses needed??
(Just copied from my 1st post on Anthony’s site…) Addendum:
NB Waco temp difference from the rest…
Query: Is there some sort of method as to how station sites are selected?
Is there any information to suggest that this station is used by GISS in calculating US or Global averages?
Didn’t you guys notice the Prius parked next to the station instruments? Why, that alone makes up for any alleged urban heat island effect, all that sweet creamy goodness rubs off the Prius onto the measurements.
This is beyond insane, we have people recalibrating the satellites based on this level of contaminated ground data, and using data from sources like this to browbeat everyone over an alleged temp rise in tenths of a degree.
RE5 (Terry)
The main criteria is to have a human being present 7 days a week to record the temperature in a log book which is then sent to NCDC for transcription into the climate database.
Even with the new electronic MMTS sensors like the one shown, the readout is an LED display, and there is no automated logging.
Thus we see stations placed where the observer is regularly present. In this case, a radio station (with a DJ) is the operator. This is why we see many stations at fire departments, forest service offices, ag experiment stations, water treatment plants, and even sewage treatment plants.
RE4 Staffan, Thanks for pointing those out. It is clear the homogenity adjustments failed and that nearby stations are cooler the question is why?
RE2,3 Steve Mc
I wonder if this is an end point problem? In this case the station data continues upwards unabated. Perhaps the GISS algorithm can’t identify the spurious data unless it sees a spike with a down-slope also? Or perhaps simply it can’t recognize it at the endpoint because there is no “forward” data to compare to?
Re #6
deadwood,
The fact that it is included in the GISS network indicates it is used in the US and global averages. Those blue line graphs are a product of the GISS web site here:
http://data.giss.nasa.gov/cgi-bin/gistemp/gistemp_station.py?id=425722570030&data_set=0&num_neighbors=1
Please continue your interesting work Anthony. Anthony needs “help in the midwest, particularly Kansas, Nebraska, Oklahoma, and Texas”. Anybody out there?
It really comes down to an ill-thought-out decision to use the brute force method of quality control. If all you’ve got is an algorithm, then every series is a trend to be “adjusted.” In reality, every station record should be treated like an individually prepared meal at a fine restaurant rather than an (un)HappyMeal coming off the McTemperature production line. You get what you pay for and when you eat too much slop you get sick. What makes it worse is that they’ve known about this problem, warned about it, done nothing about it, and then used the bad data anyway after soaking it in ketchup.
OK, metaphor alert off.
I wonder if the presence of the Prius is a good proxy for measuring the bias of the station observer?
RE 2,3 The observer reports she also has surveyed Blanco, and Temple, will be uploading soon. Temple appears to have closed in 2003 and is a sewage treatment plant.
Even though I think there are significant errors due to changing local environments for the reporting stations, I wonder if the evidence in this case supports the assertion that the recent rise in GISS is due to a location change in October 2000. The really dramatic temperature anomalies seem to arise from the years 2004-2006, which start 3 years after the move.
Is there a GreenPeace sticker on the bumper of that Prius?
I guess it could be argued that the ambient CO2 level around the Prius is lower, evidence that the site is actually reading lower temperatures than otherwise expected. Thus the station should be adjusted upward.
Yancey: Good observation. The rise does seem to start a little late. But there is no evidence about why.
Perhaps the site change and an instrument problem both contribute. The calibration schedule and repair log from 2000 onward should be inspected. (I would hope instruments showing the greatest changes would receive the greatest scrutiny within any organization.)
RE18 “The calibration schedule and repair log from 2000 onward should be inspected.”
Good luck finding those.
#16, 18. Except that your perception is wrong. Here are plots of data calculated from USHCN (up to May 2007) – GISS only uses data to March 2006. (It’s hard not to be snarky about such laasitude.) Look particularly at the second graph showing annual differences, the difference is immediately after the move. I plotted TOBS here, but the USHCN adjusted doesn’t pick up any difference. (The 2007 values are 5-month averages only BTW so don’t overinterpret them; GISS only shows data to 2005.)
Steve,
Now that is an argument I will buy.
Steve, Yancey: Steve’s comparision to Blanco does support the premise that the 2000 relocation caused a rise a Lampasas.
But what I find puzzling is that 2001-2003 didn’t show much difference. And then 2004 and 2005 both had large jumps.
So what happened in 2004 at Lampasas? In 2006 the difference dropped. Why? Of course by 2006 we are running out of data so I think the rise in 2004-2005 is more curious.
ushcnv2. change point analysis fixes everything.
I actualy havent been able to get the papers sited on this Issue, but NOAA is NOT
going to do UHI or urban adjustments anymore given the new Change point analysis
adjustments that wiggle your data for documented and undocumented changes.
When was the satellite dish installed?
RE 23 “ushcnv2. change point analysis fixes everything.”
Mosh, I spoke with a fellow at Pielke’s conference that worked on the math for it and the said this of NOAA when I asked “what do you think of USHCN V2?”
“They did it wrong”.
re 26.
The papers cited by NOAA explaining the change point analsysis are behind the green wall
( money) so, somebody will have to free the paper.
Bob Tisdale (#25)
The C-band satellite dish craze in Texas was back about 1990-94. Programming is now mostly digital, and everyone has either Dish or DirecTV. Those 8′ dishes are available for the removal in many locations. I’d be willing to lay you odds they would give you that one if you promised to torch the mount off cleanly at the surface of the concrete.
Hmmm. Are we really going to have to exclude stations as UHI contaminated in urban areas with populations of 6,000?
If that’s the standard, there will be hardly a valid temperature record from western Europe or eastern China, among many other places.
Has anybody calculated how much asphalt/concrete/roof has to be within X feet of a sensor to affect it, or is this done by eyeball?
Not that Lampasas looks like a great spot. For one thing, it looks as if the sensor would be in shade a good part of the time, and in Texas I would expect that to impart a cooling, not a warming, bias.
Considering the NE view, my guess is that the Sun would be on the brick wall behind the station from 11am or noon on to the end of the day and the brick wall would radiate heat for hours after sundown.
Harry #28:
Surely you aren’t suggesting that Lampasas has a cool bias. I think it is time for a reality check. Just look at the temp trend post 2000.
I guess I just don’t understand your post.
Harry – I think you have your directions reversed. That MMTS is just to the SW of that building, it will be baking in that TX sun all the live long day.
And throw the dime away.
===============
Just an observation. I notice in the first photograph that there is an Air-conditioner hanging out the window within the area concerned…. Has been my experience that it gets pretty hot behind those things… ; )
I’m not a climate scientist, so’s I am probably not qualified to draw any conclusions from this simple observation…. : )
Steve and Harry:
Since Julie took the pictures in February at 1600 hrs, there is no doubt that the MMTS will be in the sun for almost the entire day. The shadow from the nearby cars is consistent with the reported time of day and time of year.
I think a UHI effect can be created very easily given that the guidelines suggest that the ground should be representative of the local environment. I am sure that someone out there has a table indicating the effects of asphalt versus grass, asphalt versus scrub, asphalt versus sand, etc. Obvioulsy the more extensive the expanse of non-representative surface the greater the effect upto some point.
The link to surfacestations.org at the top of this article seems to be dead. Is there a problem?
#4 Sorry Anthony Watts for “adjusting” your name…
For K, Harry and everybody else: If we assume assume
no replacing was made of the Lampasas station between 1929
and 1934 the natural(?) temperature rose from 17.50C to
20.28C (raw GHCN+USHCN corrections) that’s 2.78C in 5 years…
With the change in Oct 2000 the rise from 1997: 17.34 to 20.45
3.11 in 2005, “with a little help from our friends”??!!
But wait, we have more: Months surpassing 30.0C:
1901/8……..30.0
1909/7……..30.1
1910/8……..30.3
1917/7……..30.0
1918/7……..30.7
1918/8……..30.8
1924/8……..30.8
1925/7……..31.1
1933/7……..30.0
1934/7……..30.1
1934/8……..30.3
1951/7……..30.1
1951/8……..30.3
1952/8……..30.6
1954/7……..30.6
20…??? WARMER summers in SE TX I don’t think so also
Dec-Jan cooling, fewer “10.0PLUS” DEC-JAN months since then…
Yes, I also noted the 1.5 x 11 years 1918,1934,1951 of
double 30.0PLUS months…Pure coincidence until further
research…Anthony, I think I’ll spend quite a few hours
on surfacestations.org …
If only everybody thought that the Texas Hill Country is not a great spot.
FWIW, Harry, the problem with this site is not UHI effect. Lampasas has parks, rivers, and ranches surrounding it, as well as Lake Buchanan (pronounced BUCK-an-an by locals; my grandfather worked on Buchanan dam). The problem is microsite bias. As a former owner of a brick home in the Texas Hill Country (Dripping Springs), the I guarantee you that radiated heat after sundown is substantial.
Here’s something else thats disquieting about this site. During the period 1895-1999, there are only 5 months in which there is not a monthly average and no year in that 105 year period is missing more than one month. In 2000 and 2005, there were 5 missing months in each year, 3 in 2003 and 1 in 2001 and 2004. Why is this? Is the modern record representative of actual diligence and the past record inclusive of invented numbers? Or has there been a deterioration of observation practices on Tom Karl’s watch at NOAA? Wouldn’t that be ironic after what seems to be a century of nearly uninterrupted record-keeping, through peace and war and economic depression?
This can’t possibly be a USHCN station, where is the barbecue and air conditioning exhaust?
Kristen@39:
Look on the second floor above the dish and you will find the AC. Now its missing only the BBQ.
It’s unfair to blame the deterioration of quality of some COOP stations on Tom Karl. NCDC has been underfunded for years; to my mind their budget should be at least twice its present size if we’re going to expect climate data to receive the full care it deserves. The responsibility for direct oversight of the local COOP stations is assigned to the local NWS forecast offices; NCDC doesn’t have any field reps to deal with such matters.
As we become less of a small farm, stay-in-the-family-for-generations type of society (where someone was there, unpaid, to take measurements independent of peace, war, and depression), relying on long-term volunteer observations is increasingly problematic. That’s where the Climate Reference Network comes in: the first national surface network whose primary goal is to monitor climate change. While that’s been underfunded too, they’re getting close to full deployment.
hm. good points first: this at least is moving the right direction. looking closer at this Texas station might provide some information, about how a change of station type might influence the TREND at the station.
but the analysis is weak. little information about the old position. no picture, no map, no elevation information. the comparison with a random station from the region (Blanco) is a nice idea, but again rather useless without the necessary informations.
LaDochy has written a paper on a moved station in LA.
will your “analysis” ever rise beyond the “bad station, temperature increase” casting doubt level?
RE 41: we all welcome the CRN
The initial stations were sited some years ago, but the studies on things like network density
requirements have yet to be published. ie,, survived the guantlett
http://www.ncdc.noaa.gov/crn/research/netwkspatialdensity.html
http://www.ncdc.noaa.gov/crn/research/deteropitnumsta.html
Some good work is being done. I like section 7 of this paper which describes how faulty sensors
can go undetected and how CRN handles this with THREE sensors
“One sensor developed a problem leading to an
error of about 1C. Because of the redundant sensors,
the close initial calibration of the sensors, and the
hourly automated quality control review process, it was
immediately clear when this occurred and which sensor
was at fault. Repairs were quickly accomplished. In
systems with single sensors, this small temperature
error would be very difficult or impossible to detect.
Such time-dependent biases affect the fidelity of the
climate record. In the USCRN these can be detected
quickly, and thus reduce uncertainty in the quality of
the climate record provided to decision makers.
The following figure shows 8 days of temperature
difference data from three normally performing sensors.
The temperature differences rarely exceed 0.2C.”
It’s also a pleasure to note that in 8 days of test using 3 Calibrated sensors
in the same enclosure that the difference between these three sensors rarely
exceeded .2C.
And so back to the debate about whether the entire globe has warmed .18C per decade
There are two issues. One is the impact of the 2000 move. The other, potentially bigger issue is that the MMTS is (probably) now in a more dynamic environment.
A building is torn down, another goes up, asphalt is repaved, someone plants a tree, surface drainage changes, an A/C is installed or removed, the store installs an awning, the satellite dish is removed, a billboard is installed, and so forth. The likely subtle changes at this location make it difficult/impossible to detect trends in the tenths of a degree C per decade.
The Blanco,TX station is now online for anyone who is interested.
http://gallery.surfacestations.org/main.php?g2_itemId=7154
RE41 John N-G, Underfunding of NCDC is not the problem, and I’d point out that NCDC has nothing to do with the placement, maintenance, and training for COOP stations and observers. The real problem is the closing almost half of the WSFO’s in the USA in the early 90’s and eliminating quite a number of COOP manager positions. COOP managers at WSFO’s these days have twice or more of the territory and less budget than they used to. They also sometimes have other duties.
I do wholeheartedly agree with: “As we become less of a small farm, stay-in-the-family-for-generations type of society (where someone was there, unpaid, to take measurements independent of peace, war, and depression), relying on long-term volunteer observations is increasingly problematic.”
I think in today’s electronic age, I’m sure many wonder why its still being done this way, daily readings, write in a logbook, logbook page mailed to NCDC once a month, and transcription of handwritten data to database at NCDC. It’s tremendously inefficient and is open to errors at many points. I’ve designed a solution, a new replacement MMTS and I have it in trial now at one of the sites LaDochy used in his study.
You can see my new MMTS design here: http://wattsupwiththat.wordpress.com/2008/01/12/my-new-wireless-mmts/
And a simple first trial here: http://wattsupwiththat.wordpress.com/2008/01/20/a-weeks-worth-of-data-from-my-new-mmts-unit/
Putting units like this in place would ensure better continuous records and less transcription errors. The existing MMTS radiation shields could easily be retrofitted.
Re: #45 A.W.: Ken Crawford of OU worked very hard within NOAA to establish a project to modernize the COOP network by replacing it with automated sensors, high-frequency reporting, and expandable hardware, all modeled after the Oklahoma Mesonet. For climate monitoring there would have been some issues because a lot of station moves would have been necessary in the tradeoff of siting quality vs. data continuity, but ultimately the project came up empty because of (my personal opinion) lack of funding/prioritization within NOAA since it didn’t fit squarely within the NWS mission.
NCDC might be happy to take over care and feeding of COOP if they had the budget, but right now it’s the NWS field offices’ responsibility, with (agreed) uneven results.
Re: #43 S.M.: I sent in a reply that had some technical issues and got spam-filtered, so I won’t repeat what I said in case it shows up here eventually. But the main point is that the studies in your first two links are long since completed and appeared in the Journal of Climate in 2004.
RE 46. THANKS, I’ve been checking back at the CRN site to see if they posted the papers.
I’ll check the source you gave. Hopefully we can get them for free.
Thanks again for all your help.
Thanks. Vose on the number of stations needed.
http://ams.allenpress.com/perlserv/?request=get-abstract&doi=10.1175%2F1520-0442(2004)017%3C2961%3AAMTDSD%3E2.0.CO%3B2&ct=1
“A procedure is described that provides guidance in determining the number of stations required in a climate observing system deployed to capture temporal variability in the spatial mean of a climate parameter. The method entails reducing the density of an existing station network in a step-by-step fashion and quantifying subnetwork performance at each iteration. Under the assumption that the full network for the study area provides a reasonable estimate of the true spatial mean, this degradation process can be used to quantify the relationship between station density and network performance. The result is a systematic “cost–benefit” relationship that can be used in conjunction with practical constraints to determine the number of stations to deploy.
The approach is demonstrated using temperature and precipitation anomaly data from 4012 stations in the conterminous United States over the period 1971–2000. Results indicate that a U.S. climate observing system should consist of at least 25 quasi-uniformly distributed stations in order to reproduce interannual variability in temperature and precipitation because gains in the calculated performance measures begin to level off with higher station numbers. If trend detection is a high priority, then a higher density network of 135 evenly spaced stations is recommended. Through an analysis of long-term observations from the U.S. Historical Climatology Network, the 135-station solution is shown to exceed the climate monitoring goals of the U.S. Climate Reference Network.”
WOW.!!!!!!!!
That will be a fun arrow to have in the debate over the number of stations needed.
RE46 John N-G “…but ultimately the project came up empty because of (my personal opinion) lack of funding/prioritization within NOAA since it didn’t fit squarely within the NWS mission.”
Spot on. Forecasting and warning has always been at the forefront of NWS funding. Climate is the bastard child of the NWS and it shows in how they’ve dealt with the surface network, especially in recent years when technology driven solutions have givem them great strides in forecasting and warning capability. The COOP network “modernization” consisted of one poorly implemented “MMTS” system, which actually made the records worse.
One COOP manager I spoke with told me they call it the “Mickey Mouse Temperature System” in their circle. It is not hard to see why.
Steve
I had looked at the area surrounding San Antonio about a year ago, interested in the UHI in comparison to surrounding rural and non-rural stations. I had compiled the data and made plot which I could email to you for your info. Let me know where to forward to.
Regards
Roger Dueck
John-N-G#41
Another side effect of “settled science”.
gdn #50: Quite the opposite. The argument goes: “We don’t need to monitor the climate carefully, because it’s not changing.”
Oops.
So the difference between Lampasas and Blanco is not enough? Try these:
Llano
Bourne
Dublin
Luling
Those four (along with Blanco) are the closest rural stations to Lampasas (also classified as rural).
You want urban? Let’s go for urban:
Austin
Temple
Waco
San Antonio
I don’t even need the photos to tell me something’s rotten in Lampasas. The data’s screaming.
Can somebody authenticate this alleged cite? Also, should all urban areas be excluded from temp. measures?
Here is the site survey form. You could e-mail the surveyor if you doubt her veracity or competence.
With regard to urban areas being excluded, keep in mind that GISS classifies this as a rural station.
I suspect it is a complete waste of time to explain to eric how an “alleged” GISS site is identified. He seems to be practicing a particularly virulent strain of “denial” that appears to becoming more common in AGW trolls.
RE46 John N-G “…but ultimately the project came up empty because of (my personal opinion) lack of funding/prioritization within NOAA since it didn’t fit squarely within the NWS mission.
I wonder how much it would cost for a few hundred automated measurement stations. Maybe we could privately fund them. I would be willing to contribute to something like that.
RE53, The simple way to authenticate the “alleged” site is to do the following:
1) Look at this image from the survey in hires:
http://gallery.surfacestations.org/main.php?g2_itemId=34278&g2_imageViewsIndex=4
2. Note the call letters of the radio station on the sign: KCYL and KACQ
3. Look up Lampasas in NCDC’s MMS database: http://mi3.ncdc.noaa.gov/mi3qry/login.cfm
use guest login, then search for “Lampasas”, click on the results, then the Location tab, which displays:
[2006-05-01] 9999-12-31 31.071670 (31°04’18″N) -98.184720 (98°11’04″W) GROUND: 1032 FEET NNE/.7/MILES LAMPASAS 06 – EDWARDS PLATEAU CENTRAL (+6)
Location Description: RADIO STATION KCYL WITHIN AND 0.3 MILES NW OF PO AT LAMPASAS TX
4. Go to GISTEMP: http://data.giss.nasa.gov/gistemp/station_data/
Type in “Lampasas”, see results, there’s only one. Click on it, and you get this graph:
http://data.giss.nasa.gov/cgi-bin/gistemp/gistemp_station.py?id=425722570030&data_set=1&num_neighbors=1
That’s the proof of the “alleged” site.
#53 Eric, I’m the surveyor. You are free to email me. You can also call the radio station using the contact information on the survey form. NOAA’s MMS website lists Ronnie Witcher and Trista Rice, and KCYL radio as the curators. I didn’t note this in the survey, but the building in the pictures is the Witcher Building, which houses KCYL radio. If you call them and talk to them, please let me know if I have made an error.
re 55. urban cool parks. Rural hot pockets.
RE 54. Folks, eric is asking a perfectly valid audit question. If he has doubts
about the veracity of the survey, the GPS coordinates are there. He can go check.
He can become a climate audit auditor.
Hmmm. hansen could have checked these sites. Peterson could have. All the paid professionals
could have. Dang, imagine that. Visit every site. In fact, If you worked at Noaa or Nasa
you could probably get funding to do that.
What makes a man and hundreds of volunteers donate their time ?
Must be Exxon.
Sarcasm off …. argg dang switch doesnt work
I have a question about the chain link fence shown in the picture. It has the angled barbed wire on the top to prevent people frome climbing over it. Did the surveyor have to climb the fence, jail-break style, or did they just walk around it?
Also, was this station picking up any smug emmisions from the prius?
Pardon me for interrupting, everyone- I’m a lurker here, not a scientist, but have been very intrigued over time by this blog’s dialogue.
Does anyone have a sense of what proportion of temperature stations suffer from these “placement” biases? Is that a “knowable” stat?
‘I wonder how much it would cost for a few hundred automated measurement stations.’
Not a whole lot. There is a 37,000-acre sugar plantation down the hill from my house, and it has an automated, off-the-shelf meteorological data collection system. Not with hundreds of stations, but it’s scalable.
The plantation bought it from a commercial service that had been developed in the California rice country and for the same purpose: both kinds of crops are burned and they wanted to avoid trouble with folks living downwind.
++++
As for my earlier question about shade, I wasn’t suggesting that the net bias would be cooling. But I’ve been in Texas and grateful for shade. Shouldn’t monitoring stations always be in the open? Rural ones, anyhow.
RE#64, Chris, out of the nearly 500 stations surveyed thus far, here is how they break down:
Click picture for larger image
RE 66: Anthony, did you make the plot site rating v.s. T anomaly for the stations surveyed?
Eric, I am a just an interested spectator in these debates, but yes, I believe all urban temp measurements should be excluded from climate studies. Because each urban site can be effected by “urban heat islands” it would be diffecult to assertain how each site is being affected by being inside the hot zone. I would like to see a climate study that used only data from known good rural stations with out any adjustments or estimations. I dont know if this would show a warming trend, a cooling trend or no trend – but you know what? I dont think anyone else knows what such a study would show either. Thats why it should be done.
Anthony, could you plot the differences of each station (original minus homogenous)for Miami TX and Lampasas, and plot on the same graph? They look similar, and it visually looks like it is increasing. I am curious as to whether it increases incrementally or in, say, decadal steps.
One of the issues is you dont really know what stations are used by GISS.
They Ingest a Bunch from USHCN ( and others) the elimate some problem sites ( documented of course)
and then they adjust stuff.
basically what you want is this.
1. List of INPUT STATIONS.
2. List of stations Used for calculations with reasons for each dropped station.
Ideally.
sod says:
Indeed – if we had this information in the first place, Anthony’s surface station audit wouldn’t have been needed. Which brings up the questions of how we can hope to determine what adjustments are needed, and how GISS, HADCRU et al can “find” less than 1C trends in the data. But what do I know?
RE 68. Hansen 2001 attempts something close to this.
I believe there MAYBE enough UNLIT stations in the USA to sample the USA. According to NOAA
we can sample the climate trend in the USA with 135 stations.
Since I am not a denier, I accept this consenus. 135 good sites.
a good data set would be this.
1. Lights == 0. AND
2. Population =
Anthony,
Do you know if your approach to tracking the quality of US sites is being picked up by other countries? Anyone out there wondering if the data they’re compiling is suspect?
Thanks!
J
#73 Jerry.
I am wondering the same. My Wife tells me they had the same “white boxes” at schools in Japan that they would collect temperatures from for science class on a daily basis. I am surfing around to find out if that data was/is collected and if its still done that way. If that data set is still intact and has not been influenced by station/instrument changes, it might show a much different picture in surface temps. Also being Japan population/industry vs area has always been very dense, I suspect the measuring conditions to be much more static than in North America.
Max says:
Japan would be a good place to start when looking for non-US data. However, I would not assume the UHI effect is non-existent because car and air conditioner use increased exponetially in the last 30 years. En mass, these things probably add additional heat to an already crowded urban environment.
Re 75, Automobile and air conditioner use in Japan is alot older and constant in Japan than just 30 years. They pioneered much packaged/affordable a/c residential technology in the mid 60’s that we are still just starting to get here. The UHI is probably much more long term and constant. Its probably just a good place to look for surface temps from a long term industrialized country, where things have remained constant.. Well except for maybe a large temperature spike in the Hiroshima and Nagasaki regions around 1945.
The lat and long are wrong … or these are different sites:
http://www4.ncdc.noaa.gov/cgi-win/wwcgi.dll?wwDI~StnSrch~StnID~20024915
Help me out here?
http://maps.google.com/maps?f=q&hl=en&geocode=&q=31.1+n,+98.2+w&ie=UTF8&ll=31.100992,-98.199996&spn=0.003601,0.006083&t=h&z=17&iwloc=addr
Somebody, remap this site and give a thought as to what is going on.
Re #77 Hi Eric. Does the difference have to do with one being listed in degrees and minutes while the other is in degrees and fractions of a degree?
79: I do not know. However, Watts gave an alternate reason under the UFA thread. If you have an explanation along the lines you propose, please lay it out.
Clicking on “Additional Station History Data from MMS” for Lampasas gives the following:
Rounding the 31.07167 to 31.1 and the 98.18472 to 98.2 resolves the issue?
The “Additional Station History Data from MMS” also provides a map option which shows the location of the old and new station.
M. Jeff is correct. It is the rounding of the decimal degrees conversion that is causing your confusion.
Lampasas station
You shouldn’t round them though.
31.07167 and 31.1 are about 3141 meters away from each other.
98.18472 and 98.2 are about 1458 meters away from each other.
That puts the rounded points 3.5 KM away from the actual location.
I know this is climate science, but although 31.07167 to 31.1 F (nor 98.18472 to 98.2 F) isn’t a big deal in a temperature variation, being three and a half kilometers away from your target is.
BTW, that should have been -98.2 and -98.18472, but the distance is still correct of course…. Although there’s probably no station south of Jomda in mainland China. 🙂
RE80 Eric, this is exactly why when volunteer surveyors go get these stations, we ask them to get a GPS reading for an accurate location.
This Google Earth link from the GPS coordinates provided by the surveyor show the actual location of the station, which coincides with the description of the station location in the NCDC MMS database. At KCYL radio station there.
GPS lat/lon:
http://maps.google.com/maps?f=q&hl=en&geocode=&q=31.0712+-98.185&ie=UTF8&t=h&z=17&iwloc=addr
Street address for KCYL radio from their web page http://www.lampasasradio.com/contact.html 505 N. Key Avenue, Lampasas, TX
http://maps.google.com/maps?ie=UTF8&q=505+N.+Key+Avenue+Lampasas,+TX&t=h&z=17&iwloc=addr
Note that there is a slight difference, the street address map from Google Earth happens to be wrong, and points just a bit north of the GPS coordinates. But you can clearly see the Witcher Building where the USHCN is located on both Google Earth images.
Yahoo says the same thing, see the Yahoo map here
So does Live Maps seen here
Curiously, Ace Hardware at 507 North Key Street, referenced as being next door by Julie the surveyor of this station, is shows the location of the Witcher Building and USHCN station on the Google Earth and other maps, see it here
So one must assume then that the street location entry in the database used by these map companies for 505 N. Key Street in Lampasas is wrong. It is not uncommon. Maps aren’t perfect, misalignments or changes in local street numbering systems sometimes happen.
The point is, if you are looking for and expecting perfection in locating these stations from any of the databases used, you won’t find it. That is why the only way to locate them sometimes is via logic and deduction and plain old investigation and research.
If that still troubles you, then there is nothing I can do to resolve it. We’ve presented all our information, you can choose to accept it, or try to disprove it. Either way, the surfacestations.org project continues unabated using the methods that we have worked out to get around all the roundblocks and metadata entropy.
Anthony #86
Is this link useful?
A guide to co-ordinate system geodesy
I’ve been gone all day, but Anthony and the others have explained it. I’ve been using gps systems for years, and the street address database is usually somewhat off. My own home address is wrong by half a block in every street map database I’ve ever used.
It’s critical to be aware of the difference between decimal degrees and degrees-minutes-seconds and other formats. In the the volunteer biological surveys I do, we use UTM rather than lat-long. We find that a little easier to use for navigating on foot.
Based on the surface station surveys I’ve done and attempted to do, I’ve found the MMS coordinates for the current locations to be sometimes spot on, but often to be just in the ballpark–close enough for people to find the station fairly easily, but not precise enough for the purpose of this project. For example, the MMS coordinates show the Lampasas station to be right next to the ACE store, or possibly on the corner of the roof of it. But the MMTS turned out to be at the radio station which takes the observations, next door. Either location would be a poor site for a temp sensor.
The NCDC link that Eric used, with the rounded-off coordinates, is almost worthless for the site analysis that Anthony is performing. The MMS database is far more accurate.
Yes, rounding lat/lon to 1 decimal point will put you way off, as we saw earlier, a few thousand meters (obviously depending on the amount of rounding). 2 decimal points can still put you off enough to waste a lot of time looking around. If you’re on the ground looking for something, UTM (Universal Transverse Marcator, not University of Toronto Mississauga) or MGRS (Military Grid Reference System) is much easier because the numbers are distance rather than degrees.
Using a 10 digit grid coordinate in MGRS (a subset of UTM) will get you to to a 1 meter square. For example, 14R is the zone Lampasas is in, and NV is the 100,000 meter square in MGRS.
14R 0577775 3437830 is the UTM for the DD (which comes out as 31.071666272538°, -98.184715154091° lol), or, in zone 14R, 77,775 meters East of the center of 14R and 3,437,830 meters North of the equator.
So the MGRS is 14 R NV 77775 37830. Our point is 775 meters east of the 77 line and 830 meters north of the 37 line on an MGRS coordinate maps.
One thing to be careful of though: the accuracy. A 1:50,000 scale map is only accurate to about 50 meters at 90% confidence, so going to 100 or 10 meters (6 or 8 digit) is enough. If you want it more precise, you’ll have to get a map that covers less area, preferably 1:10,000 but you probably can’t find those. The 1:25,000 are okay (basically a 7.5 minute map scale), so an 8 digit would get you 10 meter squares, 90 accurate to within 25 meters of the actual location. So we see how important having the correct numbers is.
Looking it up here by UTM (14 577775E 3437830N (NAD83/WGS84)) for a 1:24,000 map (no grids, sadly), the link actually looks it up with DD: http://www.topozone.com/map.asp?lat=31.07167&lon=-98.18472&datum=nad83&u=4&layer=DRG&s=24 This gives it as a church in that location.
Conversions are a bit tricky. For example, using the decimal degrees (DD) of 31.07167 -98.18472 introduced errors here, where the decimals themselves came out wrong because the lat/lon conversion truncated the degrees minutes seconds (DMS) and recalculated themselves:
DMS 31° 4′ 18″ N, 98° 11′ 4″W
DD 31.071667°, -98.184444°
UTM 0577801 3437830 14R
However, inputting it in DMS only truncated the last digit of minutes, which resulted in the same exact DD.
31° 4′ 18.01″ N, 98° 11′ 4.99″ W
31.07167°, -98.18472°
UTM 0577775 3437830 14R
However, you’ll notice the UTM is different. In practical terms, the Northing is the same, but the Easting is off by 26 meters. However, another conversion of DMS had different numbers than those:
UTM at 0577778 3437881 or 3 meters east and 51 meters north of the above, or about 51.1 meters away.
This is the proper location. 577774.9999999895 3437830.0000001504
Get geotrans from NIMA here:
http://earth-info.nga.mil/GandG/geotrans/index.html
All you wanted to know about topographic maps
http://geology.isu.edu/geostac/Field_Exercise/topomaps/index.htm
Convert lat/lon to UTM:
http://tools.wikimedia.de/~magnus/geo/
Look up by any coordinate system:
http://nmviewogc.cr.usgs.gov/viewer.htm
I should have said that’s the proper location (577774.9999999895 3437830.0000001504 rounded, in the 14R NV, the way its usually written, NV577753437830, which is in the NV5734 1KM grid square) on an MGRS map, according to the DMS/DD given in the by Anthony:
Of course,as I said, tropozone tagged it as a church.
OK, I got a moment with the plantation manager (#65).
He pays $4K-$5K for automated weather stations, including the solar panels.
Continuous read, digitally transmitted by radio to base station, temps accurate to around 0.1 degree (didn’t get to ask C or F).
Imagine that: Instead of arguing about it, we could have a thousand sensors, running at practically no cost, installed, for around $10 million.
I would suggest such claims of accuracy be taken with a grain of salt. Automated systems using thermistors and other electronics are subject to a variety of problems ranging from sensor degradation and calibration drift to spurious thermal and electrical effects. For examples of long term experience with accuracy of automatic sensing systems, search the Internet for MESONET. These problems and the experience of various automated networks have been dicussed previously on this blog.
Attention: All of the NCDC’s systems will be down due to a scheduled power outage from
Sat, Feb 23 at 7am EST until approximately Sun, Feb 24 at 8am EST.
——————————————————————————–
[Skip to end of header ]
Phenomena Other Considerations Map Remarks Files Related
Identity Updates Location Other Party Data Products Data Programs Equipment
Station Name: LAMPASAS
Country: UNITED STATES
State/Prov: TEXAS
County: LAMPASAS
Latitude: 31.07167 (31°04’18.012″N)
Longitude: -98.18472 (98°11’04.992″W)
Elevation: 1032.00 FEET ( GROUND )
POR: 1946-09-01 => Present
Climate Div: 06 – EDWARDS PLATEAU
IDS: COOP NUMBER = 415018 NCDC STATION ID NUMBER = 20024915 NWSLI = LAMT2
( Not Locked )
No Info Src Permissons
Tab Remarks: View (0)
2006-05-01 – 9999-12-31
Station Location (click on data for more detail)
Lat / Lon
[ 31.07167 (31°04’18.012″N) /
-98.18472 (98°11’04.992″W) at primary location ]
Geographical Information
Geo Region: AMERICA, NORTH
Country: [ UNITED STATES ]
State: TEXAS
County: LAMPASAS
NWS Region: SOUTHERN
WFO:
Climate Division: EDWARDS PLATEAU
Time Zone: [ CENTRAL ]
Elevation
GROUND: [ 1032.00 FEET ]
Relocation: [ .7 MILES NNE ]
Obstructions/
Exposure: [ BLDG 010-070/100-120/10-10 ]
[ BLDG 290-010/120-100/13-13 ]
[ BLDG 190-250/75-52/24-24 ]
[ TREES 090-140-190/120-90-75/22-23-24 ]
Other Regions:
Topography: [ NEARLY LEVEL IN-TOWN. ]
Location Description: None
2006-05-01 – 9999-12-31
http://www.ncdc.noaa.gov/common/ncdcfoot.html
Can anyone translate this obstruction data?
Downloaded Saturday, 23-Feb-2008 01:34:25 EST
Last Updated Tuesday, 04-Dec-2007 11:25:52 EST
Please see the NCDC Contact Page if you have questions or comments.
——————————————————————————–
Dynamically generated by http://mi3.ncdc.noaa.gov – Created by [NCDC] – Version 5.3.4 – September 20, 2007
THe notes here indicate that the device may have been moved from a parking structure … not a rural area.
“moved from a parking structure”?
Eric, it says nothing of the sort. You can’t even follow simple metadata correctly it seems.
Here from NCDC MMS is the previous location info:
[2000-10-01] 2006-05-01 31.071670 (31°04′18″N) -98.184720 (98°11′04″W) GROUND: 1032 FEET NNE/.7/MILES LAMPASAS 06 – EDWARDS PLATEAU CENTRAL (+6)
Location Description: OBSERVER’S RESIDENCE, WITHIN AND 1 MILE SW OF PO AT LAMPASAS, TX
[1988-05-19] 2000-06-30 31.050000 (31°03′00″N) -98.183330 (98°10′59″W) GROUND: 1024 FEET — LAMPASAS 06 – EDWARDS PLATEAU CENTRAL (+6)
Location Description: OBSERVER’S RESIDENCE, WITHIN AND 1 MILE SW OF PO AT LAMPASAS, TX
The key clue here is “observers residence”.
Here from Google Earth is the location from that data (31.050000 -98.183330) on the map:
http://maps.google.com/maps?f=q&hl=en&geocode=&q=31.050000+-98.183330&ie=UTF8&ll=31.050126,-98.18342&spn=0.002606,0.003707&t=h&z=18
Find the “parking structure” How about that “parking lot” north of the big lawn and the creek. Note the buildings around the previous location lat/lon and the trees. But that’s not where it was. Most likely the location was in the field behind the house aka “observers residence” just to the east of the Google Earth marker. The house with the lane to the main road.
In another great waste of time, Eric has failed to prove anything except that he can’t interpret metadata for what it is.
What next from you? The sensor is actually in another town called Lampasas?
And here is another way to find the old location. Measure from the post office. Back in the day before GPS, in the era of USGS maps, they’d use the Post office for the town as a benchmark to measure from. From the NCDC description in the post above:
From online Yellow Pages
http://www.magicyellow.com/category/Government_Post_Offices/-State_TX.html
Here is a postcard of the old post office in 1946
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=140205972790
Note the station started at the “observers residence” on 1946-09-01 according to NCDC MMS records (see the map tab)
And here using Google Earth Plus, is the ruler applied to the town from the old post office location to the “house with the lane” near the (green) marker at the lat/lon previously described in the post above. 1.14 miles to the southwest.
Mind you, the description of “1 MI SW OF PO” is based on magnetic heading, and is an estimate of distance, not an absolute.
FYI the new post office is 1.5 miles from the lat/lon specified by the NCDC database, and it is unlikely they used that reference point.
But with these minor uncertainties, be assured this still will not be proof enough for our man Eric, that the previous location note being a “parking structure” in and “urban” setting.
Mapping and location even in the GPS age isn’t perfect, and far less than perfect years ago. Get used to it.
Yea … I did misread it.
After following this discussion for days, it finally struck me, Julie KS, where is the rain gauge? Aren’t all these stations supposed to include a rain gauge? I’ve forgotten what you said in the original post, but did you speak to someone at the radio station about the siting of the instruments?
RE98 Excellent question George. Just FYI, there is no requirment that the rain gauge be next to the temperature sensor. I have visited sites where the rain gauge may be as much as 200 feet away to get away from wind effects.
I think that the NWS places more interest in the rain data than temperature becuase the official lat/lon of the station is where the rain gauge is, not the temp sensor.
Re97 Thank you, I probably could have phrased my response better.
I went through quite a number of the surveys and observed (all across the country) that the reported coordinates largely (say 9 of 10) matched up with the survey coordinates. That jives with the fact that your site location instructions fail to even mention any data problems/discrepancies. So, I question some of the inconsistent statements made over at the Miami thread — i.e., that the NOAA/Meta data is commonly off or differs from the survey measured coordinates. Perhaps Miami is just an odd ball … but more on that later. In all events, it is clearly inconsistent to say that the NOAA/NCDC reported coordinates and site survey recorded coordinates are commonly off … or that the NOAA/NCDC metadata is commonly in error … as some strongly argued over at the Miami thread to explain away the very obvious location data discrepancies associated with that site.
Case in point:
…
RE101 Well without showing which ones you examined, and the data from them and the method, your statement means nothing.
No more comments from you Eric unit you do what you promised in delivering the set of questions to the COOP manager in Phoenix. As I said there, you seem to have plenty of time to rattle on in comments here, but can’t seem to finish the list.
That disingenuous, and you’ve demonstrated that here again. You really aren’t interested in the answer, only criticism.
So until you provide that list of questions to the NOAA Phoenix COOP manager via email, no more comments from you.
I should have questions today. Meantime, it’s still a free country …
I just noticed this thread is still active.
#98 George M, yes, that is a good question: I did not speak to anyone in the radio station. I probably should have, but when I arrived there on that Sunday afternoon, there was only one car outside, and I assumed that the person inside was pretty busy operating the station. I was kind of excited that the MMTS was right there on the street and so easy to get to, and I just started snapping pictures. My car is one of the two cars in the photos.
I was on my way home from a meeting in Fort Worth. After I left Lampasas, and about halfway to Blanco, I started thinking maybe I should have looked around more to see where the other instruments were. Then I decided that it wasn’t important enough to go back and look, since I had the location of the MMTS.
I included the contact information for the radio station on my survey form, so it would be easy for anyone who wants more info to call them and ask.
As Anthony says, the instruments aren’t always together. You can look at this survey I did in Dubois, WY:
http://gallery.surfacestations.org/main.php?g2_itemId=17594
At Dubois, the coordinates given in the MMS marked the location of the precipitation gauges, but the MMTS was quite a distance away right next to the observer’s house. This is why I think that in some cases the technician’s intent isn’t to mark the exact location of the temp sensor. It seems to me that they often mark the nearest street access, or the front door of the curator’s building, or the nearest rain gauge–whatever is convenient, just to mark the general location of the weather station. Just my opinion.
RE104 Eric you make claims but don’t present any information, such as a list of stations where you say ” that the reported coordinates largely (say 9 of 10) matched up with the survey coordinates.”
It looks like the “(say 9 of 10)” is clearly a guess on your part since you have nothing to back it up.
Without providing any information to back up your claim, such as which (say 9 of 10) stations you examined, your appearance of being only interested in criticism and being disingenuous is bolstered.
Again of the thousands of people who visit this website, you are the only one claiming that there’s a problem and that the wrong station in Miami has been identified. That despite many of those people providing clear information to the contrary, and in support of the correct station being identified. There would have been many who would have pointed out the mistake if what you assert is true.
IMHO that you continue to insist a problem but keep stalling on the questions to be submitted to the COOP manager(while posting more criticisms) seems agenda oriented rather than truth oriented.
Thus, like in the other thread, until you meet your agreed upon obligation, I’ll accept no more comments from you.
So closing comments this thread should give you less time to post comments and more time to focus on that list.
I may reopen it if new information that is relevant is presented.
In the meantime, it seems that a third party has agreed with microsite bias problems that are being pointed out by the surfacestations project. See that here