Mike Smith, a meteorologist for Weatherdata Services Inc. surveyed Tuttle Creek Lake near Manhattan, KS. It is COOP-A station (#148259) at the US Army Corp of Engineers Office for the reservoir there. He had a little trouble getting photos:
Click for a larger image
He writes:
“There was a silly level of security. Required to be escorted even though it is public land with a museum on site. Required that I show two ID’s.”
Funny thing though, you can see the MMTS Temperature Sensor from US Highway 24 on the Google Maps Street View that just happened to scan the entire front of the Corp of Engineers facility:
Click to see the live Google Maps Street View
So much for government photo security.
The downside of this site, like many is that it was moved from it’s original location where the Stevenson Screen once stood. The MMTS is now on the other side of the building to the left of the open gate as seen in this photo:
Click for a larger image
The temperature measurement was moved to a place within feet of an asphalt road, where the MMTS cable can easily be brought into the office building without having to trench under the asphalt road:
Click for a larger image
The NCDC Google Map engine shows the station getting progressively closer to the administration building over time. According to the NCDC MMS database, this station was converted from Stevenson Screen Mercury Max-Min thermometers to MMTS on 10-03-1985:
This placement of the MMTS temperature sensor closer to the building that has the electronic display for it due to cabling issues is a theme we see repeated again and again in the USHCN network. With that I’ll point out that this station is not a USHCN, but is a COOP station. Even so, this station is in the COOP-A network, which reports climate data to NCDC.
Unfortunately, I don’t have access to the data for this station to plot it for discussion here. If anyone knows how to access this station data, I welcome a note and/or link.
It is my opnion that the regular sensor moves closer to buildings and domiciles alone could account for as much as .5°C warming since 1985 when the MMTS started to be introduced.
I certainly hope that isn’t an active anemometer sitting on that palette visible in the third photo. At about 2 feet above ground it wouldn’t be measuring anything much more than wake effects off passing trucks…
Turns out that these COOP stations have quite a bit of available data.
Checkout:
http://hprcc1.unl.edu/cgi-bin/cli_perl_lib/cliMAIN.pl?ks8259
http://www.hprcc.unl.edu/stations/index.php?action=metadata&network_station_id=148259
http://mcc.sws.uiuc.edu/resources_links/links_climate2.htm
There is a plethora of data to play with by poking around the various US & state gov’t sites.
Temperature – average gets tabular data like this:
TUTTLE CREEK DAM, KS
Monthly Average Temperature (Degrees Fahrenheit)
(148259)
File last updated on Oct 18,
*** Note *** Provisional Data *** After Year/Month 200707
a = 1 day missing, b = 2 days missing, c = 3 days, ..etc..,
z = 26 or more days missing, A = Accumulations present
Long-term means based on columns; thus, the monthly row may not
sum (or average) to the long-term annual value.
MAXIMUM ALLOWABLE NUMBER OF MISSING DAYS : 5
Individual Months not used for annual or monthly statistics if more than 5 days are missing.
Individual Years not used for annual statistics if any month in that year has more than 5 days missing.
YEAR(S) JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC ANN
1959 —– z —– z —– z —– z —– z —– z —– z 82.13 a 68.45 a 51.58 35.48 h 35.95 l 67.39
1960 25.53 n 27.32 l 24.52 c 58.06 l 62.86 f 71.18 74.90 77.74 70.65 57.61 42.52 28.40 55.94
1961 25.37 a 33.12 42.06 48.85 59.92 a 71.82 77.79 75.31 63.57 56.21 39.48 23.55 51.42
1962 20.50 31.96 36.58 51.53 71.40 b 72.00 76.21 77.37 64.75 58.94 42.55 31.44 52.94
1963 16.68 31.43 43.57 c 55.61 b 63.73 76.17 81.13 78.77 70.82 66.18 45.31 a 22.76 54.35
1964 30.42 31.26 36.68 54.50 67.08 71.47 81.18 74.34 67.43 53.77 44.75 27.21 53.34
1965 28.03 26.86 31.34 54.35 67.82 72.67 77.79 75.45 64.88 57.18 45.58 37.63 53.30
1966 24.16 28.57 44.81 50.13 d 62.40 73.10 82.87 72.69 64.97 55.23 42.65 28.87 52.54
1967 27.94 30.12 43.71 55.83 59.61 70.58 73.26 71.77 62.77 54.21 40.53 32.39 51.89
REPLY: Thanks Eric, I was not aware of this website having tabular data, when I get a free moment I’ll plot this to see if jumps can be seen when the station was changed from Screen to MMTS.
And thanks for your continued pilgrimage for station surveys!
To: RBerteig:
The proximity of the anemometer to the pan of water suggests to me some sort of evaporation experiment relating to humidity. Would be interesting to learn what the actual purpose is.
Or maybe this one deserves the subtitle: “How Not To Measure Wind Speed”. What do you think, Anthony?
REPLY: Its an old piece of standard equipment called an evapotranspiration pan.
http://www.photolib.noaa.gov/htmls/wea00915.htm
Ahh, our government at work.
I’ve wondered whether there was a Hurst effect in temp data; if the total of all temp records from the numerous temp stations in an area or region are averaged, does that bulk averaging eradicate the temp tainting that exists at each or most of the seperate stations? I came across a paper which suggests that the microclimate inhomogeneities that you are discovering at almost every data collection point can be overcome by some sort of Hurst rescaling;
http://ams.allenpress.com/perlserv/?request=get-abstract&doi=10.1175%2FJCLI3663.1
This Runnalls and Oke effort assumes the thermal micro inhomogeneities are enhanced at night and can be overcome by averaging between 2 stations so that flaws in the combined averaging of temps over the stations are nullified. But this smacks of the average temp concept, which is at the heart of AGW, being established by some sort of regression; the issue of regionalism is sacrificed on the pretext of eradicating the localised micro inhomogeneities; andt the Hurst rescaling is tautological in that it assumes a uniformity which the process then establishes by removing the locational factor along with the micro inhomogenities. The baby is thrown out with the bathwater. Far better to have a uniform standard of temp collection so that localised variation can be preserved; but if there are too many local deviations from the mean, then the average temp over an area, which is used to base a trend on, must be flawed because of a Hurst effect?
Cohenite, I read the abstract. To my reading, it does not suggest averaging. Instead, it appears to suggest a means for detecting and quantifying microclimate biases. It appears that the significant element is that microclimate affects high and low temperatures differently. If I were to do that, I would want to be very sure that the specific site had the elements the technique was calibrated for, and not other unknown reasons for bias.
It seems to me that it might be possible to use the technique to clean the historical data using this technique. It might also be possible to use the technique to extract the UHI bias as well.
That would be science, though, not something I’d expect from an organization that can say “Oh, its fallen off before and it wasn’t a problem then, so it won’t be a problem now”.
Doesn’t look like NH ice is melting the way AGW people would like it to…LOL
http://arctic.atmos.uiuc.edu/cryosphere/IMAGES/current.365.jpg
In fact its at 1975 levels for summer (still 2 months to go though) wait and see….
http://arctic.atmos.uiuc.edu/cryosphere/IMAGES/seasonal.extent.1900-2007.jpg
looks like 9.5 millions square km = 1975 summer levels
Dishman; wouldn’t an averaging have to occur between the nominated 2 stations?
[…] USHCN Stations From My Desk 18 06 2008 In How Not to Measure Temperature Part 65, I showed a COOP-A station near Manhattan, KS at Tuttle Creek Lake that had been surveyed. In […]
Cohenite,
It’s not the same as averaging.
Here’s an example in a different domain. I’ve got this nifty CCD imager I’m taking hundreds of long (30 minute) exposures with. With images that long, there’s a problem with cosmic ray hits filling the wells on one or more pixels. Cosmic ray hits have a distinctive signature, in that the pixel value is typically >50 standard deviations out. This is the critical common step – identifying the signature of the noise pattern.
In my case, because the effects of the noise burst can’t be further characterized, I just throw data point away.
In the case of the paper you cited, the authors argue that the noise can, in fact, be accurately characterized and quantized. Once the bias has been identified, it can be subtracted out as a step function over the duration of each corresponding site change.
I would say that this method is probably significantly superior to the GISS method.
As I see it, the GISS method applies knees to approximate step functions. If your sample set includes an increasing number of step functions (as CRN appears to), the over all trend will be up.