Several recent posts and hundreds of associated comments have focused on the subject of temperature adjustments. TOBS, homogeneity, scribal records, rural versus urban … it is enough to make one’s head spin. I understand the desire to adjust data, but I often wonder if the problem is simply intractable and the adjustments we do have in place are a bit naive. I often ask myself: can the rich historical record of a station overwhelm the well-meaning attempts to adjust for observation times, UHI, and other factors? How often were stations relocated? What were the surroundings like? How accurate were the instruments? Were the instruments properly calibrated? What happened when the curator became ill, or had pressing business matters to attend to, or overslept? Multiply this by a hundred, or five thousand … do the variations simply become noise we can filter out? I think when we are comparing years to an accuracy of hundreths of a degree, these questions become important.
Last summer I stumbled across a paper published January 21, 2005 by Stephen R. Doty and Dr. Lesley-Ann Dupigny-Giroux titled “THE HISTORY OF SURFACE WEATHER OBSERVING IN BURLINGTON, VERMONT, 1832-1973″. I posted a copy of it on surfacestations.org. One of the stated goals of this paper was to produce a document that future studies can use to evaluate the validity of the data that were collected here, judge the trustworthiness of the observers who collected them, and determine the climatological significance of the whatever variability may be discerned. I found the paper an interesting and revealing study of one of the many USHCN sites across the country, and it is the most complete historical description of any USHCN site I have run across to date. It describes in great detail how the Burlington station moved its way eastward from its original location in downtown Burlington to its present location at the airport. Along the way it was moved to rooftops, windows, close to ground level, close to porches, near parking lots and runways. I was particularly drawn to this history because I lived for nearly 20 years in the Burlington area beginning in 1982, and I am very familiar with the city and locations described throughout the paper.
I wrote this post not to prove a point for or against AGW. In fact, one should see that no conclusion can be made with respect to AGW by examining this station’s record (well, at least I don’t think a conclusion can be drawn). I wrote it really to point out that the surface record is a poor measure of climate change, at least when we are attempting to measure with sub-degree accuracy. Unless the complete station history for each station in the record is fully understood and compensated for, we are doing nothing more than guessing when comparing the present to the past.
The following image is taken from the paper, showing the various station locations between 1832 and 1973:
I have tried to summarize the changes this station has seen over its lifetime. Below are paraphrased descriptions of the station locations since 1880. I selected the years after 1880 since they are covered by the GISS record, although FILNET does go back to 1837. I also used an alphabetic label next to my descriptions for use on several plots. In a few cases I note additional thermometer information taken from the USHCN station history file, such as change in instrumentation and/or height.
(A) December 1878 – June 1883: Howard Opera House at the SW corner of Bank Street and Church Street (44° 29′ N 73° 12′ W, Elevation 213 feet). The type of thermometer is not indicated, but it’s height is recorded as 56 feet above the ground, placing it on the roof.
(B) The location of the station from July through September, 1883 is unclear. The address listed for October 1883 is City Hall Park, which is about a block from the Howard Opera House.
(C) November 1883 – November 1891: 55 Elmwood Avenue (44° 29′ N 73° 12′ W, Elevation unknown). The type of thermometer and its height above the gound is not indicated. Note that the USHCN station history file lists the height as five feet from 1883 to 1906.
(D) December 1891 – January 1906: 57 Elmwood Avenue (44° 29′ N 73° 12′ W, Elevation unknown). The thermometer is listed as a “Green’s max and min, and ordinary” mounted in a double blind (Weather Bureau pattern) shelter located on the west side of the house, 12 feet from the ground. This obviously differs from the USHCN station history file.
(E) February 1906 – March 1906: 301 South Willard Street (44° 29′ N 73° 12′ W, Elevation unknown). The shelter (with presumably the same thermometer) was listed as attached to north window of house about 1 foot above sod (basement window of house.) As reported in the paper, the relocation to South Willard Street represented a move to a higher elevation and a residential setting that was further removed from the shores of Lake Champlain.
(F) Late March 1906 – June 1943: Weather Bureau Building at 601 Main Street (44° 29′ N 73° 11′ W, Elevation 398 feet). The unspecified thermometer is described as being 12 feet from the ground. Figure 3 in the paper is of a postcard of the site showing the location of the weather station off the front porch. The location was 1 mile east southeast from the previous location, further from Lake Champlain and at the top of the hill occupied by the University of Vermont campus. It was recorded by Weather Bureau personnel that “The location is excellent for meteorological purposes but rather far out from the business portion of the city.”
According to the USHCN station history file, this move also resulted in a lowering of the thermometer height from 12ft to 6ft above the ground. However, note that for two months in 1906, the station was 1 ft above the ground.
The following image is taken from a 1906 USGS survey map of Burlington. I added a green dot to indicate the location of the station at the beginning of 1906 and a red dot to indicate the location at the end of 1906. To me, it looks like the station moved from n urban to almost a rural setting. But was urban in 1906 like urban today? What, for example was used to pave the streets?
(G) 1934 – February 1950: Administrative Building, Burlington Municipal Airport (44° 29′ N 73° 09′ W, Elevation 331 feet). Until March 1939 the station was operated by National Airways, Inc, and then by the U.S. Civil Aeronautics Authority until June 1943. At that time the Weather Bureau consolidated the two observing sites in Burlington at the airport location, closing the City Office. The Weather Bureau occupied the same office as the CAA had been using. (In my opinion, it is likely the temperature record reflects the readings from the 601 Main Street location until June, 1943.) A diagram of the Administrative Building indicates the thermometer was located in a CRS about 25 feet north of the building. It was mounted on a Townsend support. Maximum and minimum thermometers and a fan were installed.
I found the following passage from the entry quite interesting: In September 1943, the station inspector recorded the following, “Exposure only fair. Too close to runways, building, etc., and over sand soil. Maxima several degrees higher than City Office readings on sunny days; mimia lower on clear nights, due to radiation.”
The following image is taken from a 1948 USGS survey map of Burlington. I added a red dot to indicate the location of the station in 1943 and a blue dot to indicate the location in 1944.
(Gi) 1959 – according to the USHCN station history file, an H06x series Hygrothermometer was installed at the site.
(Gii) 1964 – according to the USHCN station history file, the thermometer was lowered from 6ft to 4ft above the ground.
(H) February 1950 – July 1973: New Administrative Building, Burlington Municipal Airport (44° 29′ N 73° 09′ W, Elevation 331 feet). This represented a move of 0.3 miles north from the previous location. A passage in the report says: A note on the Exposure and Installation of Instruments form dated May 16, 1955, reveals that the “Shelter is located between parking lot and ramp, both of which are blacktop.” The Weather Bureau maximum and minimum thermometers were housed in a “large shelter” 5 feet above the ground. The Figure 11 caption indicates the shelter was 30 feet north of the building. It appears to me that the shelter would be in the shade at noon for most of the year except the summer months. It would not be in the shade earlier in the morning and later in the afternoon.
(Hi) The final update in the USHCN station history file occurs in 1985, when the H06x is replaced with an H08x. No change in height is recorded.
(I) From July 1973 on (which is not covered by the paper) it is unclear looking at MMS as to when the station was relocated and when further equipment changes were made. None of the location entries indicate that a relocation ocurred, although a slight GPS location change occurs in February, 1996. This may correspond to the change to an ASOS which we know is located today between runways, away from the terminals and control tower. I do know that when I lived in Burlington the 1973 location was completely paved over and used as a rental car parking lot. The number of actual location changes between 1973 and 1996 remains unclear.
If one looks at the USHCN station history file, however, there is an indication that an MMTS was put into service in 1973, at a height of 4ft above the ground. The H06x series Hygrothermometer is still listed as being in use.
The following is a recent satellite image from Google Earth which I originally placed on surfacestations.org. I added two additional labels showing the location of the station in 1943 and 1973. Because the site has undergone significant changes in the past 15+ years, what you see is not representative of what the area looked like 35 and 65 years ago.
I plotted the GISS raw and adjusted temperatures for the Burlington station, and added markers to indicate the points in time when the station was relocated. For example, the line marked “A” indicated the last year the station recorded data from “location A” as described above. Given we do not have a control to compare against, it is impossible to know how the station moves affected the record.
I also tried plotting the trends for each period of time the station was at a different location. While there are marked differences in the trend lines, I am not convinced they are meaningful because most of the time periods are short.
Finally, I plotted GISS adjusted minus GISS raw to see if there was any correlation to the station moves or instrument changes. Not surprising, there does not appear to be any such correlation, because GISS does not adjust for such changes. Instead, GISS made an adjustment based on nearby rural stations. Nearby means stations within 500km, or up to 8 hrs driving from Burlington.
What can one conclude from all of this? Looking at the raw Burlington data, nothing really stands out. The adjustments made to bring the station in-line with rural stations does not seem to make a lot of sense in that it does not correspond to any station move or instrument change. Yet common sense tells us that the location, height, and instrument changes all affected the record, and probably more than 0.6C. It is unlikely, however, that we will ever know the extent with which those parameters affected the record. Given that, is the Burlington record important? Can it give us the accuracy required to estimate global temperatures to 0.01 degrees?