Blog

The last few days have been very hot across eastern Australia. The mainstream media have been reporting ‘unprecedented’ conditions. This fits a popular narrative, but its inconsistent with the original historical record.

It was very hot in Australia in the late 1800s and the early years of the 1900s, particularly in outback Australia – at places like Bourke. Many of the extremely hot days were recorded using non-standard equipment, and so they are not recognised by Australia’s Bureau of Meteorology. Outrageously though, the Bureau has deleted from the raw digitised record the hottest day ever recorded in standard equipment at an official weather station. Ostensibly, because it was recorded on a Sunday – the day that the temperature-recorders normally had off in the days of manual temperature recordings.

No doubt the fellow who recorded temperatures from the thermometer in the Stevenson screen at the Bourke post office got up on the morning of Sunday 3rd January 1909, and decided he was going to work anyway. He was not going to miss what did turn out to be the hottest day of the century.

So, when I went to the trouble of getting access to the log book, now held by the National Archives of Australia, I could see that the fellow had not only written in 125 degree Fahrenheit (equivalent to 51.6 degree Celsius) – he had also underlined it!

I took a photograph of the entry, Exhibit 1.

Back then the person who recorded the temperatures each day, known as the ‘Observer’, was not the same person who compiled the ‘Abstract of Results’ at the end of the month. This supervisor (who used a red pen) sometimes made changes to temperatures as recorded by the Observer; and seems to have been initially confused as to the actual temperature on Saturday 2nd January 1909 at Bourke (originally changing it from 112 to 125).

The supervisor was clear in his summary though, writing in red that the highest temperature at Bourke for the month of January 1909 was 125 degree Fahrenheit recorded on 3rd January, as shown in Exhibit 2.

But if you go to the raw temperature data at the Australian Bureau of Meteorology’s website, there is no entry for this date, as shown in Exhibit 3.

****

If you like what I do, consider making a goodwill contribution.

After looking at hundreds of temperature series from different locations across Australia, I’ve come to understand that only cities show the type of warming reported by the IPCC, and other such government-funded institutions. Much of this warming is due to what is known as the Urban Heat Island (UHI) effect: bitumen, tall-buildings, air-conditioners, and fewer and fewer trees, means that urban areas become hotter and hotter.

For example, in a recent study of temperature variability and change for south-east Australia it is evident that maximum temperatures in the cities of Melbourne and Hobart are increasing at a rate of about 0.8 degree Celsius per century; while the rate of increase at the nearby lighthouses is half of this.

While the trend of about 0.4 degree Celsius per century at the lighthouses – as shown in Chart 1 – is arguably an accurate record of temperature change, the Australian Bureau of Meteorology changes this. To be clear, the Bureau changes a perfectly good temperature series from Cape Otway lighthouse by remodeling it so that it has Melbourne’s temperature signal – all through the process of homogenisation.

Government agencies in the USA have done exactly the same thing to temperature records for Colorado. This is all explained in detail in this new video by Monte Naylor:
 
https://vimeo.com/196878603/b9ea716a74

The video runs for about 40 minutes, and is quite technical.

The conclusions from this study have been summarized by Monte as follows:

(1) The USHCN Fort Collins station temperature record was not recognized by NOAA as having the heat bias from expanding UHI which has been easily identified by other researchers.

(2) NOAA’s homogenization program adjusted the USHCN Boulder station temperature history in a fashion that does not match any of the four other nearby rural/suburban long-term temperature histories. Nor does the NOAA-homogenized Boulder temperature history resemble the average temperature trend found by this study.

(3) NOAA’s homogenization program adjusted the Boulder temperature history to resemble the UHI-contaminated temperature history of the Fort Collins station.

(4) The best estimate of the northern Colorado Front Range temperature trend is obtained by using the TOB-adjusted Group of 5 average which shows a warming temperature trend of 1.7 °F (0.95 °C) from 1900 to 2015. The NOAA temperature trend, about 4 °F over 115 years, is more than twice the best estimate of this study.

(5) About 70% of the warming shown in the Group of 5 average temperature trend occurred before 1932. Temperatures trends of recent decades do not show anomalous warming.  Distinct warm temperature events occurred in the 1930’s and 1950’s that were much warmer than those observed since the turn of the 21st century.

(6) The Northern Front Range Group of 5 average temperature trend does not increase in a fashion consistent with increasing atmospheric carbon dioxide.

Media reports yesterday claimed that 2016 was another record hot year. However, close scrutiny of all the temperature data from the state of Victoria paints quite a different picture. When all 289 temperature series from Victoria are simply combined, the hottest years are in the early part of the record. In particular, 1914 is very evidently the hottest year on record – see blue time series in Chart 1. This finding builds on a recently published book chapter focused on south-east Australia, and is part of a larger study working towards a realistic reconstruction of Australia’s temperature history.

Yesterday the Australian Bureau of Meteorology released its Annual Climate Statement, claiming 2016 to be the fourth-warmest year on record for Australia – and also an unusually wet year. Wet years are usually much cooler years, but because the overall trend in the ACORN-SAT* time series shows significant warming, even a wet year comes out as relatively hot.

Of course there is no one place in Australia where the average temperature can be measured; so the Bureau relies on a reconstruction to determine how hot 2016 was, relative to the historical record. Their method, however, is quite subjective in terms of choice of locations to include, the method used to remodel the individual temperature series before they are combined (this is refered to as homogenisation*), and the area weighting applied – with the weighting changing on a daily and monthly basis.

Late last year, I had a book chapter, co-authored with John Abbot and published by Elsevier*, which shows historical temperature trends for south-east Australia from 1887 to 2013 based on a more transparent system – that can be easily replicated. We choose the longest continuous series, used the same series to calculate every value, and applied an area weighting based on topography and landuse – and did not remodel individual temperature series. In the chapter we conclude that temperature trends for south-east Australia are best described as showing statistically significant cooling (yes cooling) of 1.5 degree Celsius from 1887 to 1949, followed by warming of nearly 2 degrees Celsius from 1950 to 2013. The warmest year in this reconstruction is 2007, followed very closely by 1914.

A colleague at the University of Tasmania, Jaco Vlok, has compared our south-east reconstruction with a reconstruction based on all 289 temperature series for Victoria – but only from 1910. The different methodologies used to generate these reconstructions, and also the official ACORN-SAT series for Victoria, which is the series used by the Bureau to calculate the official statistics for Australia, are detailed in Table 1.

There is a very high degree of synchrony between the reconstructions, though when all the raw data is simply combined – Vlok’s approach – the hottest years are all in the earlier part of the record: 1914 (hottest) followed by 1919, 1921, 1938, 1961 and then 2014.

Postscript: I have expanded on this analysis in an article just now published by Graham Young at OLO. Jennifer, Noosa, Monday 9th January, 2017.

_____

* ACORN-SAT stands for Australian Climate Australian Climate Observations Reference Network – Surface Air Temperature and is a dataset developed by the Bureau based on a subset of available temperature series, almost all homogenised, and then combined with an area weighting and used to report climate variability and change. The annual climate statement for 2016, based on this ACORN-SAT dataset, is here: http://media.bom.gov.au/releases/333/2016-a-year-of-extreme-weather-events/

* Homogenisation involves changes to measured temperature values ostensibly to correct for non-climatic variables. These changes to the observational data are quite different from quality assurance. For example, the need for homogenisation most often results from a ‘statistical test’ detecting a break point, these breakpoints often occur after a period of missing data. In response all values preceding the breakpoint are often reduced by a specific amount back to 1910. The amount by which the measured observational values are reduced is determined through the application of algorithms and calculated relative to what are referred to as ‘neighbouring’ stations, which may be Urban Heat Island (UHI) effected, and/or located many hundreds of kilometers from the target location.

* Marohasy, J. & Abbot, J. 2016. Southeast Australian Maximum Temperature Trends, 1887–2013: An Evidence-Based Reappraisal.  In: Evidence-Based Climate Science (Second Edition), Pages 83-99. http://dx.doi.org/10.1016/B978-0-12-804588-6.00005-7 You can read more about this series here: https://jennifermarohasy.com.dev.internet-thinking.com.au/2016/12/temperatures-trends-southeast-australia-1887-part/

Charts that show year-on-year increases in temperature have become a symbol of all that is wrong with the world – or at least modernity and Western Civilization. So, why then, when such a chart was shown to a studio audience of opinion leaders, did it caused them to break-out in spontaneous applause? The Sydney Q&A audience could not conceivably have been applauding global warming – more likely this audience was applauding the affirmation that was being afforded their belief in global warming?

 

Of course, the chart that particle physicist Brian Cox held-up to the audience (as shown in the above Youtube video) is an historical reconstruction. There is no one place on Earth where the world’s temperature can be measured – and this particular chart is based on a variable number of homogenised temperatures series. By ‘homogenised’ I mean remodelled – ostensibly to ensure “non-climatic influences are minimised”; but in the process trends are changed.

I’ve just had a book chapter, which I co-authored with John Abbot, published that shows historical temperature trends for southeast Australia back to 1887 – using unhomogenised temperature series.

This reconstruction is shown by the purple line in Chart 1 (see below). It suggests a rate of warming less than half that shown in Cox’s global homogenised reconstruction; and might be better described as exhibiting considerable inter-annual variability – with cycles of cooling and warming.

The method used to develop this reconstruction perhaps represents an important first step in developing an Australian-wide reconstruction based on un-homogenised data.

Our chapter concludes that the series for southeast Australia is best described as showing statistically significant cooling (yes cooling) of 1.5 degree Celsius to 1949, followed by warming of nearly 2 degrees Celsius to the present. (If you would like a pdf copy of the chapter, email me: j.marohasy at climatelab.com.au).

The official Bureau of Meteorology reconstruction for the state of Victoria is shown in red in Chart 1; and is based on ACORN-SAT. The homogenised ACORN-SAT database only begins in 1910. It is used by the Bureau, and also CSIRO, to report climate variability and change. Eleven locations are used in the ACORN-SAT reconstruction for Victoria – with some of the series from these locations very short, and with a lot of missing data; all were remodolled.

Our southeast reconstruction is based on the same five long and continuous series with only three adjustments made to two of the series (Deniliquin and Cape Otway) to correct for equipment changes in 1908 and 1898 – as detailed in the chapter.

Despite the very different methodologies used, the reconstructions are surprisingly similar. Both reconstrucions show considerable inter-annual variability, with almost synchronous peaks and troughs: see Chart 1. In both reconstructions, 2007 is the hottest recent year, though temperatures were almost as hot back in 1914. In large part because the ACORN-SAT database only starts in 1910, which corresponds with a dip in the record, this official series for Victoria indicates an overall rate of warming of 0.9 degree Celsius per century – compared to 0.3 degree Celsius for the southeast reconstruction.

Both reconstructions are of maximum temperatures. Global warming is typically reported as an increase in the mean temperature, which is the average of the maximum and minimum temperatures. The maximum temperature is the daytime temperature; and an arguably better measure of regional temperature variation because of the higher rates of turbulent mixing of the atmosphere during the day time.

In Chart 1, I’ve also plotted temperature maxima as measured at the agricultural research station near Rutherglen in northern central Victoria – this is the series shown in yellow. This is an exceptionally high quality series because it has been measured using standard equipment at the same rural site for over 100 years. But it was not part of the southeast reconstruction because we only used series that began on, or before, 1887 for the southeast reconstruction.

Temperature minima and maxima do not always trend in the same direction. In the case of Rutherglen, temperature minima actually show cooling – consistent with other series from this region. This cooling is most obvious in spring, and probably associated with the extensive development of irrigation. Through the homogenisation process the Bureau change the cooling in the Rutherglen temperature minima to warming before including Rutherglen in the official ACORN-SAT database. I’ve written extensively about this, including in a recent research paper entitled simply ‘Temperature change at Rutherglen in south-east Australia’.

Key Reference/New book chapter

Marohasy, J. & Abbot, J. 2016. Southeast Australian Maximum Temperature Trends, 1887–2013: An Evidence-Based Reappraisal.  In: Evidence-Based Climate Science (Second Edition), Pages 83-99. http://dx.doi.org/10.1016/B978-0-12-804588-6.00005-7