UPDATE: We have the photo situation under control, Please don’t go to the Carefree Skyranch Airport as they’ve been getting a number of calls and visits. A follow up report is coming. – Anthony
================================
Since Steve Goddard and I posted recently on the subject of high temperatures in Arizona, this seemed like a good followup to problematic climate data and stations we noted in that post. In my previous post about record minimum high temperatures in Southern California I showed a map with all the new records plotted. But, there was a curious red dot record high temperature “anomaly” on it, 109°F in Carefree, AZ on July 8th:
I thought this was curious, especially since there were no other record high temps set in the state of Arizona in the last week. So, I decided to see what I could find out about the station.
My first visit was to the NCDC MMS Metadatabase to get the lat/lon of the station, plus any other info I could find:
I found the lat/lon, and an indication that it was at the Carefree Skyranch Airport, as seen from this approach photo from the airport webpage:
The photo above shows quite a bit of green for Arizona, I wasn’t sure if that was indicative of irrigation or a wet spring.
When I plugged the lat/lon of 33.8161, -111.9019 into Google Earth, it gave me the location of the NOAA weather station at Carefree airport. Right away something jumped out at me:
Check out the albedo difference due to the airport tarmac asphalt. Warmer there on sunny days possibly? I checked the weather for that day, Thursday, July 8th, and found it was full sun all day.
The red dot signifies the NCDC provided lat/lon. Note, that this was gathered (according to NCDC metadata) with a Lowrance GPS. However, the matchup isn’t always spot-on with mapping programs, plus that, since NWS has the most interest in rainfall data for hydrological forecast verifications, they take the GPS reading over the rain gauge, not the temperature sensor.
I determined that the Carefree station temperature sensor was an MMTS electronic type (on a pole) and that it had two rain gauges.
I also learned that this station was not a USHCN station, but was a Class A COOP station, and does report to the climatological database as indicated by the publish to CD note:
I also learned that the station had been converted from Stevenson Screen to MMTS in 1986:
And that apparently the observer had decided to switch observing times, but NOAA lost track of that info:
A close up aerial view from Bing Maps shows the location in detail. I was able to spot the rain gauges, but not the MMTS temperature sensor on the pole:
Interactive view available from Bing here.
The metadata from NCDC on station location, citing obstructions, shows three trees and a building nearby, all of which are visible in the image above. I’m certain the location is correct:
So what we have is a station near a building, in the middle of a sea of asphalt, in the summer in Arizona. I suppose I’m not surprised it was the lone high temperature record last week for Arizona.
Perhaps somebody who lives in Carefree or knows somebody who does can get a photo of the MMTS temperature sensor from 4 compass points and an overall view. It would be interesting to see where exactly it is located. It is a municipal airport, and it looks like the NOAA equipment is in full view of the public parking lot.
I’m betting it is near the rain gauges. Since one is a tipping bucket gauge, requiring a power cable (if it is a Fisher-Porter type with conical top) then the NWS could have killed two birds with one stone when laying cable fro the MMTS electronic sensor also.
This station data is used to adjust other nearby stations that have missing data in NOAA’s FILNET process, and since it is published on the Climatological CD, may also get used in climate studies of temperature.
I’ll check with my friend, former state California climatologist Jim Goodridge to see if he has the data on one of his CD’s from NOAA, and hopefully we’ll get some data from that station to help tell the story. Or, if anybody knows where to get it online, don’t hesitate to point it out.
Now here is where it gets interesting.
I surmised that the airport might read warmer due to the asphalt environment the temperature sensor is located in. The proof turned out to be pretty easy to find. Thanks to the many private weather stations that Weather Underground logs, I was able to locate a private station in Carefree, AZ just north of the airport and all its high resolution data for Thursday, July 8th, when the record of 109°F was tied at the airport. Here’s the tabular data showing it recording the high of 104.2°F at 3:22PM:
As indicated by the Weather Underground page, the station is a Davis Vantage Pro 2 PLUS model with the solar radiation sensor (notice the watts/m2 in the tabular data), a unit I’m very familiar with because I provide that model via my online business. I have no reason to doubt it being just as accurate if not more than the NOAA sensor. It has a similarly sized GILL radiation shield as the NOAA MMTS. It also has NIST traceable calibration for its sensors.
From the XML feed of observations provided by WU, I was also able to get the precise lat/lon of the private station, which appears to be in the observer’s back yard. I plugged the lat/lon into Google Earth and created the image below using the GE measuring tool and my paint program for annotations:
And people try to argue that airport siting of weather stations, or that siting in general, makes no difference.
You can homogenize rationalize just about anything.
I suggest that the NOAA high temperature record for July 8th, 2010 in Carefree, AZ may very well be erroneous, and a byproduct of location.
UPDATE: Commenter “Regg”, seems to think that the 129 feet elevation change between the two stations (that I didn’t think was large enough to be worth mentioning, since Google earth shows only a 10′ elevation change) could account for “most” of the 5°F difference. I considered this when I wrote the article.
Unfortunately, he’s wrong. Dry adiabatic lapse rate calculates out to about 0.7°F difference if we accept the 129′ difference in elevation between the two stations. – Anthony
===================================================
UPDATE2: We have the photo situation under control, Please don’t go to the Carefree Skyranch Airport as they’ve been getting a number of calls and visits. A follow up report is coming. – Anthony
Seriously –
Do any of the statisticians here have the capability to determine what the chances are for record temps – record highs, record lows, record highest daily lows and record daily lowest highs – is, given that we’ve only been keeping records for 150 years in most places in the US and many for only about 100 years?
Of course if we’d only been recording temps for, say, 10 years, we would have a LOT of records being set. But how often would we be seeing SOME record in SOME place in the US, given the brevity of our record keeping?
Of course, you do mean Desert Forest, I presume…
Maybe. Cringe. I was so busy correcting the second “r” I tried to put in Forrest!
REPLY – Better Dessert Forest than Bedford Forrest. ~ Evan
Thanks Steve Mosher for the format of query string to get Weather Underground historic data. I’ll see if I can quickly modify the climate scraper program I wrote a few months ago to grab this data as all it does is to just create a sequence of query strings for the time period desired and outputs the returned data to a series of files. It should (I think) be fairly trivial to modify my program and I’ll post the link to download it from once I’ve got it working. (Although this statement is coming from a former programmer who had every estimate of time required to finish a program multiplied by 5 by his boss at the time).
From the standpoint of producing data for use in R, is one big file of all of the readings desired or a file for every day?
I recently did a small analysis of temperature data over the last 100 years for 5 urban and 5 rural sites in arizona. I would like to send it to anthony just to see what he thought. What email do i send it to?
I call it AZ warming, and it looks at how the temperature adjustments for GISS data for the 10 sites actually accentuate global warming claims (erroneously IMO) over the past 30 years when accounting for urban heat effect.
Boris Gimbarzevsky says:
July 12, 2010 at 8:11 pm
Thanks Steve Mosher for the format of query string to get Weather Underground historic data. I’ll see if I can quickly modify the climate scraper program I wrote a few months ago to grab this data as all it does is to just create a sequence of query strings for the time period desired and outputs the returned data to a series of files. It should (I think) be fairly trivial to modify my program …
—…—…—…
Thank you. But I do doubt it will be trivial. My former boss multiplied “should cost” to “will cost” by about 4.
I was interested in your albedo-full sun-asphalt idea. If it was true then it sounds like something worth following up. Would your idea suggest that during the night the two records would be closer?
I graphed the full days temp record for two record you talk about. The records do seem to diverge around 6am and remain apart by 2 degrees. The 4 degree difference you talk about comes from the peak in the two records coming at different times (15:00 and 16:00hrs). So far so good but the following night doesn’t come back together. A 2 degree difference remains through the night.
I was wondering if you’d comment how this fits with your analysis. In fact if you could graph the full days records for comparison it might be useful. A comparison of cloudy/sunny days for the two records might be useful.
Anthony – the green stuff on the right of the photo looks like a golf course to me (therefore it’s irrigation). Here in New Zealand, it is quite common for golf courses to be built next to airports (because of building restrictions).
Don’t know if the influence of golf courses on airport weather readings would be a suitable research topic for you…
All the best.
Paul Deacon
REPLY: Yeah I’ll add that to the other kajillion projects I have so people like Tamino and Romm-bomb can complain that I’m wasting time “studying” golfing. 😉 -A
Between the airport and the private weather station there appears to be a golf course, which accounts for the green areas noted at the beginning of this post. I note that the wind at the time the private station recorded its maximum was from the SE, meaning that it was blowing towards the station from the golf course. I wonder how much irrigation that golf course has? How much cooling would the transpiration from all those greens and fairways cause?
As George Carlin once said (I think, if my memory serves me after hearing this about 35 years ago): “Currently, it’s 109 at the airport. But, who cares? Nobody lives there anyway!”
Have climate scraper program for Weather Underground data written and just testing it out. Very simple mods to my program to scrape Canadian temperature data but an extra space accidentally put in the query string took a couple of hours to find and resulted in very interesting behavior from the Weather Underground server. Wireshark was very helpful in finding this error. Will make sure program’s working properly before posting a link to the program on my website. Output file is a mess consisting of:
data_line
and is probably simplest to just use Notepad or a word processor to delete the lines. No, I’m not going to clean it up as that means writing code to find strings in the downloaded data rather than just streaming bytes to a file. Maybe someday when I’ve got more time.
Has anyone looked in detail at Weather Underground archived data? These readings are taken every 5 minutes and are usually in peoples back yards which are far better places to measure temperatures than airport runways and would be interesting to compare “official” airport temperature records in a number of cities with the closest Weather Underground station. Conceptually a very simple project but I don’t have the time to do it. Computing an average temperature from the 288 daily readings available from the Weather Underground stations is a lot more precise than the average of min + max temperatures (and can’t believe anyone would actually use that method).
@Gail I have notice 2-4F added to the daily lows for my local weather station on a routine basis. I will see 48F in the morning at dawn or just after yet the next day the “official” reading for the low is 53F. I generally check the weather everyday when I get up so I have noticed this.
I probably ought to start keeping a record.
Yes, it would be useful to have a written record. They may have “lost” theirs.
You should be careful note whether the hourly temperatures you see labeled as actual temperature readings or predicted low and high for the day and whether it is supposed to be a reading of the same apparatus as the daily log and not just a thermometer on the roof.
At least one screen shot would be good to go with the records you keep.
Maybe Anthony could set up a network of volenteers like the one checking station siting quality where people adopt a local station where hourly data can be seen and compare it to the daily low/high record.
I’m sure that there are some people who are so intent on “saving the planet” that they see nudging up a few temperatures as a case of the end justifying the means.
” Computing an average temperature from the 288 daily readings available from the Weather Underground stations is a lot more precise than the average of min + max temperatures (and can’t believe anyone would actually use that method).”
Backwards compatibility.
This way of recording temps relates to the historical use of max/min mercury thermometers. It has to be done the same way now if we are to compare to previous records. It is not supposed to accurately reflect the true mean over 24h.
“From the standpoint of producing data for use in R, is one big file of all of the readings desired or a file for every day?”
best is a long comma delimited file. just the numbers,
Day, month, year, time, temp
That way excell guys can go nuts
Photos! We wants photos!
Have made modifications to climate scraper program and put a version which works just with Weather Underground sites on my website at:
http://drgimbarzevsky.com/Downloads.html
Download link is below the one for Canadian climate data climate scraper. It’s also in zip file format so a little easier to extract.
It’s crude, but it works. Have tested it on a couple of local Weather Underground stations. Slow, but that’s just because of the amount of data; ~8.5 Mb/year. Right now can only download whole years and RSN will add ability to download smaller sections of data. Files are comma seperated values with delimiters between days.
I live close by and wanted to see for myself, and also took some pictures today around 5 pm. If you wish Anthony, I’ll send them to you. In the meantime, they are on my msn space for viewing:
http://cid-a4bbb2fe0ecf9d20.photos.live.com/browse.aspx/New%20album?uc=2
Within around 20 feet of an AC which is inside a block fence, and about 10 feet from a Coke machine. The MMTS sets on the side of the porch roof, connected by galvanized pipe fittings, and 8 feet above black asphalt parking lot.
IMO depending on the wind direction and currents created by the porch roof, walls and block fencing, the instrumental data could be significantly compromised by the heat from the AC and Coke machine, not to mention the asphalt and cement patio.
In any event, the location appears to break all or most siting requirements. Only incompetents would use this station for climate data.
REPLY: It is as I suspected, FUBAR. Thanks much – Anthony
Glenn said at July 13, 2010 at 6:58 pm:
Looking at the pictures; is there any siting rule with which it complies?
It is worse than useless.
Glenn says:
July 13, 2010 at 6:58 pm
Well, we have lots of them.
My weather station (Davis Vantage Pro) is compromised by a small lot. I used to have it mounted on a fence between two trees, but when we had the fence replaced,
I had a pole installed in a better place for it. It’s still too close to trees, but now its closer to the house and the asphalt (bitumenous concrete) parking area, and a window air conditioner.
But it’s still lots better than this site and many others producing atrocious data!
Actually, it seems to do pretty well. It’s high enough so it gets lots of sun, it’s not in the stream of the air conditioner, the driveway is mainly a problem when there’s sun and a light east wind (vary rare) and the temperature sensor is about 10’/3m off the ground. So’s the anemometer, so that’s pretty bogus.
Oh, lord. And this is COOP-A?
When they said USHCN was the “best of the best” maybe they weren’t kidding. More’s the pity.
The mind boggles.
Airports, villages & small towns need a weather station. Cities need some weather stations. Megalopolii need a bunch of weather stations. When the weather man on the evening news gives the local temp, maybe he should give the mean (after throwing the max and min readings out) and not the “Official” NOAA instrument reading. Wouldn’t hurt. Might even be ‘cooler’?
I’m only now reading the post; but it comes to mind that the entire “global warming” scare is based upon the relocation of recording stations to airport environments.