UPDATE from on the road: Eli has been off planet for a week. Ms Rabett has taken control, but she is fluffing in the shower and we are in a place with wireless. Anyhow, brief looking about shows that Steve McIntyre has found a mistake in some of the GISS adjustments (actually Eli does not even know if the mistake was with the GISS or the USHCN bunches, as Climate Audit appears to be off the air because of a DRDOS attack, or at least that is what they say).
However, junior management wishes to point out that as we pointed out below, the issue was not the physical layout of the stations, Tony Watts' and Roger Pielke Sr. thing. After going through the metadata about Detroit Lakes, Eli concluded there was a problem, but the problem had to be in the adjustments. Eli, being the lazy Rabett, and ear deep in work at the time left it at that without looking into the matter. McIntyre found one, which from the comment slipstream appears to have been fairly trivial (sort of like Spencer and Christy's minus sign), but, as Tony pointed out and is quoted below)
But hey, they can "fix" the problem with math and adjustments to the temperature record.They did.
Tony Watt's surface station watchers have landed a whopper, Detroit Lakes, Mn, a rural station in western Minnesota (ID: 212142)
As he says
This picture, taken by www.surfacestations.org volunteer Don Kostuch is the Detroit Lakes, MN USHCN climate station of record. The Stevenson Screen is sinking into the swamp and the MMTS sensor is kept at a comfortable temperature thanks to the nearby A/C units.According to Don Kostuch, the AC unit was moved from the roof of the building 5/99 and the chart from GISS (uncorrected USHCN data) shows ~ 4C jump about 1999
This has everyone (well isn't Climate Audit everyone) jumping up and down and has even made it over to Germany
Knashing of Hansen
We codda done it better
Ah, but little bunnies, if we actually go and get the data from GISS or from CDIAC (sadly in F) we see that the jump
really comes in 1997/1998. The lab bunnies, who do RTFR also noted that there were a whole bunch of months missing in 1997 and earlier in the decade, (USHCN fills in even long stretches, GISS 999.99s them out) so they went to the historical record for Detroit Lakes and they saw that the station moved from one side of the lake to the other August 30, 2002. The satellite picture shows no building near the old site. Thus the jump happened two years before the station moved to its current location near the air conditioner.
UPDATE: Eli and the bunnies want to drive a stake into that air conditioner theory. Detroit Lakes is in rural western Minnesota. They do not run A/C during the winter. It is too cold for a heat pump. If there were any effect from the A/C it would raise the temperature during the summer. Let us look at the difference between the summer (J-J-A) and the winter (D-J-F). Listening to the ranting anonymice there should be a jump when the A/C was installed (1999)
Well what about that scuzzy Stevenson screen. August 11, 2006 the max/min thermometer was replaced by the MMTS sensor. Nothing under the screen, why pay any attention to it. One might speculate that when they put in the automatic MMTS they pushed it to the side and left it. The shelter has since gone through a Minnesota winter. Eli, a plush white Rabett, would look like that after one of those suckers.
So what DID happen in 97/98 according to the historical record?
Two things. On January 1, 1997 the station was moved to another management section and it looks like the time of observation changed from 17:00 to 7:00 and on July 16 1998 the observer changed.
Looks like a lot of the new anonymice were wrong when they thought a photograph alone was important. Of course, we are depending on the station historical record available to everyone through the majic of the INTERTUBES. Still, there does appear to be a problem here, IEHO associated with an imperfect homogeneity adjustment
Currently all data adjustments in the USHCN are based on the use of metadata. However station histories are often incomplete or changes that can cause a time series discontinuity, such as replacing a broken thermometer with one that is calibrated differently, are not routinely entered into station history files. Because of this we are developing another step in the processing that will apply a time series discontinuity adjustment scheme described in Peterson and Easterling (1994) and Easterling and Peterson (1995). This methodology does not use station histories and identifies discontinuities in a station's time series using a homogeneous reference series developed from surrounding stations.Eli will let Tony Watts have the last word, because at least he got it right
But hey, they can "fix" the problem with math and adjustments to the temperature record.