Raising Arizona

Anthony, sorry I should have included this reference as well:
Easterling, D. R., T. R. Karl, E. H. Mason, P. Y. Hughes, D. P. Bowman, and R. C. Daniels, T. A. Boden (eds.). 1996. United States Historical Climatology Network (U.S. HCN) Monthly Temperature and Precipitation Data. ORNL/CDIAC-87, NDP-019/R3. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, Oak Ridge, Tennessee.
I suppose it’s strange that this reference is from 1996 if a BAMS paper with version 2 was published last year. On the same site they have Version 2 as a 1990 reference:
Karl, T. R., C. N. Williams, Jr., F. T. Quinlan, and T. A. Boden (ed). 1990. United States Historical Climatology Network (HCN) serial temperature and precipitation data. ORNL/CDIAC-30, NDP019/R1. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, Oak Ridge, Tennessee.
see http://cdiac.ornl.gov/r3d/ushcn/ushcn.html#source for more info…

One last thing…being a scientist, I don’t naturally look at a figure and assume it is correct. I tend to want to investigate its accuracy. Usually I only look at the Arctic sea ice stuff since that is what I’ve been working on the last few years, but today I had some time to take a look at this posting on the temperature record at a station in AZ. I found it strange that the adjusted temperatures would be higher than the raw data since I would expect it to be the other way around.
So first I did a google search for the data from the station and the first link was to a Revision 3 data set which showed different temperatures. When Steve gave the link to the data he showed in this posting, I then downloaded that data. What I found is that the way Steve put the two plots together was misleading. In reality, the adjusted data has lower temperatures than the raw data by 1.31-1.32 F until 1973 when suddenly the temperatures match to within a tenth of a degree.
I think the real thing that should be looked at is why is there a constant offset applied until 1972 and then the raw and the adjusted temperatures are about the same. Does this constant offset have to do with a bias in the heating/cooling of the instrument?
Also, I would imagine the Urban heat island affect would become more important post 1972 (though I don’t know anything about the town around the Ajo site). So why does it appear that no adjustment was made (at least during the 1973-1984 period of overlap)? And why is there no raw data after 1984? I know the temperature record is of great interest to WUWT, so hopefully some of you can answer these questions.
thanks, Julienne

Julienne
Great questions. I would guess that there was a time of observation adjustment made around 1972. If the station readings were previously made in the afternoon, and later made in the morning, USHCN would (rightly or wrongly) add on to the post 1972 measured temperatures and subtract from the pre-1972 measured temperatures..
What concerns me about that is that the switch in polarity occurs coincident to an adjustment. It looks very suspect, particularly when the data set became sparse around the same time. The adjusted data has all many symptoms of being flawed.

On your overlay graphs of Ajo Arizona, raw and corrected, you state that “they (the graphs) are identical until 1970, when the magical adjustments kicked in.” Actually they are identical AFTER the magic adjustments kick in. The post 1970 adjustments amount to zero. If there is any magic it occurs prior to 1970, when the temperatures for the corrected data were reduced. It is a minor point and this was mentioned earlier by Regg. I believe this to be an honest error, perhaps a result of an overzealous need to prove a point. It happens to all of us.

Julienne,
GISS normally quotes trends from the 1970s onward, and the adjusted trend is actually fairly steep during that time period. So I disagree abiut the significance.

Steve, unfortunately for this station the raw data ends in 1984, so it’s a bit hard to say much about a difference in slope between raw and adjusted data from 1970-1984 (I would have to do a significance testing to see if you even have enough data records for any slope to be significant for only 15 years of observations). But even during this time-period, the adjusted data only give a slope of 0.013F/year. Would be good to see all the raw data since obviously there must be raw data from 1984 onwards if they have adjusted data during that time…

Anthony, I would disagree with “always”. At NSIDC we have all the raw satellite data we archive available, plus the algorithms that are applied to that raw data, and we do Climate Science…;)
REPLY: Which is why I didn’t list NSIDC. Also see above on USHCN VERSION2, I think you missed it -A

Julienne
The reason I wrote this article was a response to the Stanford paper suggesting a major increase in heat waves in the four corners states. I think we both know that idea is laughable in Colorado, and looking at the USHCN data it became clear that maximum temperatures have not increased in Arizona either.
Low temperatures in Arizona have increased in urban areas, because of grass, trees, irrigation (anything which increases humidity and/or causes increased condensation,) asphalt, buildings, air conditioners, etc.
USHCN claims to adjust out UHI effects, but looking at the adjusted data for Arizona it does not seem credible that they are doing an adequate job.

What the world needs now is…
(sarc on)
a NASA super-launch that installs 1.21 gigawatts, I mean 1.21 mini giga-weather-robots all around the world, on land and sea, that transmit their readings back to the mothership, which in turn transmits the data to Houston and Chyenne Mountain, the one in Colorado Springs, which processes it through a super-dupper super-computer, and transmits it it via the web to everyone who wants to know.
(sarc off)
when the citizens of the empire lost faith in Rome, the world grew dark, and cold…

TOBS adjustment is quite interesting. Evidently there’s been a number of changes in the overall habits of people who record and reset the min/max temperatures.
I understand that resetting near the time of daily lows can cause a daily minimum to be recorded for two consecutive days when it should have been recorded for just one day and resetting near the afternoon high can cause two consecutive days of daily maximum to be recorded when it should have been recorded for just one day. I question the method used to establish the amount of bias this introduces. It appears to hinge on just a single study which is in itself questionable practice and there is also no verfication that the software which makes the corrections is faithfully following the procedures used in the study.
Also curious is how the TOBS changes from net positive to net negative biases. One might reasonably think that shifts from morning-to-afternoon or afternoon-to- morning would occur in equal numbers. By the shape of the TOBS corrections it would appear that there have been a number of large simultaneous shifts in the population from “morning people” to “night people” and vice versa. Interesting sociological phenomenon there. I’d love to see a graph made from station metadata that shows the average time of observation. It should pretty much exactly match the shape of the TOBS adjustment curve. If it doesn’t then something is rotten in the state of Denmark.
SHAP is also interesting. This adjustment is made for station changes other than time of observation. I would assume largely for changes in position. I suspect this is the most computationally complex adjustment which also means it has the greatest room for error. It adds a remarkably large and growing warming bias to the record beginning in about 1920. The rate of growth in adjustment magnitude doesn’t vary by huge amounts but there are some notable changes in it in 1950 where it increases faster than before and in 1990 where it flattens out.
Again, this is curious behavior. One would expect these kind of station changes to be a wash – i.e. moves from a colder to warmer locations and moves from warmer to colder would occur in equal numbers. But that isn’t the case. Nowhere near the case. Somehow all the station changes show a large bias towards moving from warmer to colder locations thus requiring either an upward adjustment to data prior to the move or downward adjustment to data before the move. I can’t think of anything that would drive a more or less constant trend to relocate stations to cooler locations as time goes on. I understand how things like changing the paint to something more reflective could result in a cooling step change, or moving from less windy to more windy location could add evaporative cooling, and going from a lower to higher altitude, or moving into a seabreeze, and so forth. What I question is why there is such a pronounced bias towards moving to cooler situations. The kind of paint being used if more a more reflective kind reflective type gradually replaced a less reflective but there doesn’t seem to be any other category of change that would do it.
And once again this adjustment relies on a methodology described in just one paper by the same person (Karl) who wrote the paper on TOBS bias. By all appearances the entire warming trend displayed in the fully homogenized data for the United States rests on findings in just two papers written by the same researcher. That in and of itself is alarming.