Perth temperature history
Also see Busselton relocation
On May 2, 2011, The West Australian newspaper published the following story, claiming that "Perth has experienced its hottest start to the year on record, with a mean maximum temperature of 31.5C for the first four months, breaking the previous record of 31.3C set in 1978".
But can we trust the BoM calculations? Was the mean maximum in the first four months of 2011 really 31.5C?
Well, we can do it the BoM way and calculate the average of the rounded monthly figures for January (32.5C), February (34.1C), March (31.9C) and April (27.3C). Yep, that's an average 31.45C, which further rounds out to 31.5C. Too easy.
But what if we don't average the rounded monthly figures and instead average the actual maximum temperatures recorded on each day during the first four months? It's pretty easy to do. Just check the monthly figures for Perth Metro (station 9225), which is the official "Perth" site for the BoM, for January, February, March and April 2011.
The actual maximum temperatures in all 120 days of January, February, March and April 2011 can be seen here. Get your spreadsheets busy and figure out the average maximum. You'll find it's 31.42, which rounds to 31.4C.
So the actual mean maximum for Perth in the first four months of 2011 was 31.4C, not 31.5C as claimed by the BoM.
On March 3, 2011, the BoM claimed that Perth's 2010/11 summer was the equal hottest on record with a mean maximum of 32C. But if you use the daily temperatures instead of the rounded monthly figures, the average summer maximum was actually 31.89C, or 31.9C when rounded. Again, their calculations were based on an average of the rounded monthly maxima rather than the days themselves (see Was Perth's summer of 2010/2011 a record?).
2010/11 was the third hottest summer recorded in Perth, but not the equal record.
Ah, but this is being a bit pedantic, is it not, since the first four months of 2011 were still the hottest on record?
That depends upon whether you also want to research a bit deeper into the record itself. The BoM references its official "Perth" temperatures from station 9225 in Mt Lawley, which opened and recorded its first annual temperature means in 1994 - picture below in May 2011 courtesy Stephen Harper.
Before 2004, the official Perth temperatures were recorded at Perth Regional Office (station 9034). Check the first four months of 1978 and the mean maximum was 31.3C, the previous record as stated by the BoM.
Stations 9034 and 9225 are about four kilometres apart and the evidence suggests the Mt Lawley site is slightly hotter during the day than the Perth Regional Office site. Site 9034, which opened in 1876 and closed in 1992, has a 19 metre elevation.
Site 9225, which has had a stable mean temperature since opening in 1993, has an elevation of 25 metres. Proximity to broad river waters and the UHI influence of inner city and university development are probably relevant.
The average annual maximum at Perth Metro over 17 years is about .4C warmer than it was earlier at Perth Regional Office. If you suspect the increase is because of global warming, not the four kilometre change in location, consider the annual minima in "Perth" from 1975 to 2010:
The average annual minimum at 9225 over the past 17 years was about 1.6C cooler than it had been at 9034, the drop happening immediately the station change was made. Maybe a rapid onset of global cooling, but a more robust explanation is that the four kilometre change in station locations significantly affected temperature recordings and made historic pre-1994 comparisons dubious at best and irrelevant at worst.
Perth Regional Office (9034) started recording from 1897 and finished in 1992. This location was originally known as the Observatory and started recording temperatures in 1897 at the Mt Eliza site now mostly occupied by Dumas House, adjacent to Kings Park. Although the BoM gives no clues that Perth Regional Office has a history of more than one location, station 9034 is the benchmark for historic temperature comparisons in the capital of Perth.
View of Observatory grounds soon after completion with Fraser Avenue and Kings Park in the background. The meteorological instruments including a Stevenson Screen can be seen in the right of the photograph.
The Bureau of Meteorology Perth Regional Office at 127 Wellington St, East Perth, facing north with the instrument enclosure in the foreground
And for Perth's temperature record from 1897 to 2010 ...
To further illustrate the distortion caused by moving the Stevenson screen temperature recording instrument a few kilometres from its original location, the Mt Lawley 9225 temperatures four kilometres inland show higher maxima and significantly lower minima.
Based on the temperatures charted above, the last two decades in Perth have had the hottest days and coldest nights since records began in 1897. Why are records claimed for average maxima but minima and mean are ignored, and why are any of them relevant for comparison with other locations before 1994?
It's worth looking a little more closely at Perth Metro ...
Annual temperature variations at Perth Metro are clearly linked to cloud cover, as evidenced by a comparison of the chart above with Perth's annual rainfall at Perth Metro: 1994 - 740mm, 1995 - 905mm, 1996 - 889mm, 1997 - 699mm, 1998 - 743mm, 1999 - 822mm, 2000 - 796mm, 2001 - 689mm, 2002 - 738mm, 2003 - 854mm, 2004 - 637mm, 2005 - 875mm, 2006 - 467mm, 2007 - 703mm, 2008 - 808mm, 2009 - 607mm, 2010 - 504mm.
It's worth looking at Results of Meteorological Observations Made in Western Australia During 1908 (PDF 29mb) published in 1912 by Commonwealth Meteorologist Henry Hunt:
Perth Gardens was Perth's first temperature recording location as of 1876, situated about 1.5 kilometres from the Observatory site on Mt Eliza in an area since renamed as the Supreme Court Gardens (see map above).
The 1912 Hunt document also contains a table with monthly and annual mean temperature readings at Perth Observatory from 1897 to 1907, and at Perth Gardens from 1876 to 1907, which provides a useful 11 year overlap.
The difference between station locations and the influence of inner city heat is further illustrated (Excel download 25kb) by comparing Perth Regional Office 9034 with Perth Airport 9021, about nine kilometres to the east.
Perth Airport opened in 1944 while Perth Regional Office relocated from Mt Eliza to inner Wellington St in 1967 and closed in 1992.
From 1944 to 1967, the average monthly minimum at Perth Airport was 12.1C and from 1967 to 1992 it was 12.1C - no difference. From 1944 to 1967, the average monthly maximum was 24C and from 1967 to 1992 it was 24.1C - a .1C warming.
From 1944 to 1967, the average monthly minimum at Perth Regional Office was 13.3C and from 1967 to 1992 it was 13.8C - a .5C warming. From 1944 to 1967, the average monthly maximum was 23.3C and from 1967 to 1992 it was 23.8C - a .5C warming.
On May 31 2011, the BoM stated in its seasonal update for Perth that autumn 2011 had the hottest mean maximum on record:
The BoM also references the average number of days above 30C at Perth Metro "from observations beginning in 1897". Perth Metro did not open until 1993. Perth Observatory atop Mt Eliza opened in 1897, again four kilometres away.
When informing the public that record high temperatures have been experienced, the Bureau of Meteorology should ensure the accuracy of its figures by averaging daily temperatures instead of rounded monthly calculations.
Rounded averages of rounded averages are not accurate and a .1C error is significant if claiming record temperatures, particularly when Australia's economic and political future is likely to be determined by a public understanding of the need or otherwise for a carbon dioxide tax.
The BoM should also acknowledge that temperatures recorded at its Mt Lawley 9225 station are being compared with stations in different locations where local environmental factors have a significant influence on Stevenson screen thermometers.
Perth's temperature history is too disjointed to make meaningful comparisons and present-day recordings at Mt Lawley should only be compared to recordings from the same site beginning in 1994.
It might mean Perth has a very short history of temperature recordings, but at least the data will be relevant.
Top of page