City Warmth Island Research – Do Pristine Climate Stations Even Exist? – Watts Up With That?
By: Geoff Sherrington.
Scientist, Australia.
This article tests the hypothesis:
Measured historical factors can be used to distinguish between urban and pristine stations.
Australian weather stations and data were tested. Australia’s low population density plus its many weather stations allow many “pristine” station candidates to be examined for studies of Urban Heat Island (UHI) effects.
From an initial list of 1,000+ stations, those with adequate data were narrowed down subjectively to 45 pristine station candidates. Some of their general properties are tabled below.
https://www.geoffstuff.com/45 pristine candidate comparison stations.xlsx
If several stations in a set are truly pristine, they should have similar temperature trends over time. If they do not have similar time series trends, then some factor more than natural variation must be affecting them; they cannot all be pristine.
Here is a graph of the trends of the 45 candidate stations.
https://www.geoffstuff.com/trens 45 pristine candisates.jpg
This is for maximum temperatures, Tmax, years 1910 to 2020. In the following table, the linear least squares fit is calculated and expressed in the equivalent of ⁰C per century for maximum (Tmax), minimum (Tmin) and average (Tav) temperatures.
| PRISTINE | Tmax trend 1910 to 2020 | Tmin trend 1910 to 2020 | Tav trend 1910 to 2020 | RMS error Tmax | RMS error Tmin |
| BALLADONIA | 0.2 | 0.6 | 0.4 | 6.6 | 4.99 |
| BENCUBBIN | 1.1 | 1.2 | 1.1 | 7.63 | 5.62 |
| BIDYADANGA | 2.0 | 1.9 | 1.9 | 3.52 | 5.08 |
| BRUNETTE DNS | 0.9 | 1.3 | 1.1 | 5.29 | 5.98 |
| CAPE BORDA | 2.4 | 1.1 | 1.7 | 4.64 | 3.21 |
| CAPE BRUNY | 1.6 | 1.4 | 1.5 | 4.05 | 2.96 |
| CAPE CLEVELAND | 0.1 | 2.9 | 1.5 | 3.37 | 3.15 |
| CAPE LEEUWIN | 0.9 | 0.7 | 0.8 | 3.37 | 2.89 |
| CAPE MORETON | 0.8 | 1.1 | 1.0 | 3.28 | 3.67 |
| CAPE OTWAY | 0.9 | 0.7 | 0.8 | 4.66 | 3.09 |
| COCOS ISLAND | 0.4 | 2.0 | 1.2 | 1.09 | 1.17 |
| EDDYSTONE PT | 1.4 | 1.1 | 1.2 | 3.59 | 3.49 |
| ELLISTON | 2.9 | 1.2 | 2.1 | 5.45 | 4.29 |
| EMU CREEK | 3.2 | 2.3 | 2.8 | 6.42 | 5.59 |
| EUCLA | 1.8 | 0.2 | 1.0 | 5.78 | 4.61 |
| FLINDERS IS | 1.7 | 1.6 | 1.6 | 4.35 | 4.39 |
| GABO IS | 1.4 | 0.3 | 0.9 | 3.48 | 3.49 |
| HUME RESERVOIR | 3.3 | 2.3 | 2.8 | 7.57 | 5.67 |
| JERRYS PLAINS | 3.1 | 1.0 | 2.1 | 6.21 | 5.91 |
| KYANCUTTA | 0.9 | 1.0 | 1.0 | 7.41 | 5.38 |
| LADY ELLIOT IS | 2.1 | 0.9 | 1.5 | 3.24 | 3.17 |
| LARRIMAH | -0.3 | 1.5 | 0.6 | 3.69 | 5.33 |
| LEARMONTH | 3.1 | 0.8 | 2.0 | 5.72 | 5.04 |
| LOCKHART R | 1.5 | 2.4 | 1.9 | 2.24 | 2.79 |
| LOW ISLES | 1.0 | 0.8 | 0.9 | 2.98 | 2.17 |
| MAATSUYKER IS | -0.1 | 1.3 | 0.6 | 3.79 | 2.68 |
| MACQUARIE IS | 1.0 | 0.9 | 1.0 | 1.99 | 2.53 |
| MANDORAH | 2.1 | 1.8 | 2.0 | 4.35 | 5.49 |
| MANGALORE | 2.6 | 0.6 | 1.6 | 7.28 | 5.33 |
| MARDIE | 0.8 | 3.1 | 1.9 | 5.01 | 5.47 |
| MARRAWAH | 1.5 | 1.5 | 1.5 | 3.72 | 3.07 |
| MARREE | 3.1 | 2.8 | 3.0 | 7.87 | 4.31 |
| MONTAGUE IS | 1.4 | 2.5 | 2.0 | 3.51 | 3.31 |
| NEPTUNE IS | 1.7 | 1.7 | 1.7 | 3.34 | 2.54 |
| OENPELLI NT | 0.4 | 1.7 | 1.0 | 2.52 | 3.16 |
| OMEO VIC | 0.5 | 1.6 | 1.1 | 6.78 | 5.02 |
| PALMERVILLE | 0.6 | 0.3 | 0.4 | 2.82 | 4.04 |
| POINT HICKS | 3.1 | 1.4 | 2.3 | 4.95 | 3.59 |
| RABBIT FLAT | 2.8 | 0.8 | 1.8 | 5.92 | 7.31 |
| TABULAM | 3.3 | 1.7 | 2.5 | 5.14 | 4.14 |
| TARALGA | 3.1 | 1.2 | 2.1 | 6.92 | 5.36 |
| VICTORIA RIVER | 0.8 | 0.4 | 0.6 | 4.16 | 6.21 |
| WARMUN | 0.4 | 0.5 | 0.5 | 4.39 | 5.52 |
| WARRUWI | 1.4 | 1.3 | 1.3 | 2.03 | 2.27 |
| WILLIS ISLAND | 0.4 | 1.0 | 0.7 | 1.99 | 1.77 |
| Simple average | 1.5 | 1.3 | 1.4 | 4.54 | 4.14 |
The linear trends for Tmax range from a high of 3.3 ⁰C per century at Hume Reservoir to a low of -0.3 ⁰C per century at Larrimah. These trends are far larger than the uniformity expected for true pristine sites with similar natural variation.
Therefore, the hypothesis that Measured historical factors can be used to distinguish between urban and pristine stations is shown to fail, because measured historical factors cannot even distinguish between one plausibly pristine station and another.
https://www.geoffstuff.com/trend table pristine.xlsx
For wriggle matchers, here is the same data without the linear line of best fit, but with some lightly smoothed character retained.
The wriggles do not easily fall into a recognisable, simple, or systematic pattern. One might infer that there is “noise” from sources such as different start and end dates for the stations, plus station shifts at various dates, instrument changes, changes after quality testing and so on. Known system changes such as from ⁰F to ⁰C in November 1972 and from Liquid-In-Glass to Pt Resistance thermometry mostly in 1991 to 2001 have been studied for diverse stations. Their step changes if any are likely to be below 0.5 ⁰C, but the “noise” in these wriggles is an order of magnitude greater as shown by the RMS error figures tabled above in ⁰C units.
Willis Eschenbach has calculated a figure that can be compared with the wriggle graph above. It shows the expected earth surface temperatures derived from incoming radiation measured by the Ceres satellite system.
The trends from the Eschenbach graph using CERES data are about 1.5 ⁰C per century equivalent.
To the extent that comparison is valid – and I know of no reason to doubt – these satellite-based temperature estimates for random grid cells on the earth show time trends similar to each other and with matching wriggles over the last 20+ years. Something happens to the relation between the temperatures from satellite measurements and the land surface thermometer estimates we are discussing. Part of that “something” could be UHI, but my stations were chosen to have minimal UHI. How small is minimal? At Palmerville station,
“The same observer performed observations for 63 years from 1936 to 1999, and was the area’s sole remaining resident during the later part of that period. “
http://www.bom.gov.au/climate/data/acorn-sat/stations/#/23000
https://www.www.geoffstuff/palmerville.jpg
It is possible to create “adjusted temperatures” for these land surface stations. Most of the lighthouse stations in this candidate set have been separately adjusted by the ACORN-SAT procedures published by the Bureau of Meteorology. (A future article looks at the future for future adjustments).
Most past studies of Urban Heat Island effects start with this reasonable, logical, simple equation:
Tuhi = Turban – Tpristine
That framework works only if Turban and Tpristine can be measured. This article starts to show that there are impediments to the useful definition of Tpristine by measurement rather than by assignment. The next article in this series compares these 45 pristine candidate stations with 37 urban stations, to seek systematic differences between the two groups.
Through these articles, you are invited to consider if these temperature sets from Australia are much different to those from other countries, with the point in common that they are not fit for the purpose of influencing very expensive government policies.
(END)
Like this:
Loading…
Comments are closed.