The Australian Temperature Record- Part 4: South Australia

Ken Stewart, June 2010

South Australia is the driest state in the Commonwealth, proudly claims no convicts in its ancestry, and is home to some of Australia’s best wines- and worst temperature records.

Such important information as the temperature record for the last 100 years, which is rightly of great interest to the government and the public, must surely be based on high quality data, right?  The truth may surprise you.

This post could have been subtitled “The Case of the Missing Data.” 

The Australian Bureau of Meteorology (BOM) develops its climate Trend Maps and Time Series graphs form 100 sites nationwide, comprising the High Quality Australian Site Network.  10 of these are in South Australia. 

Here’s the Trend Map

And Time Series Graph:

As you can see, BOM declares a warming trend of 0.12C per decade, or 1.2 degrees C for the last 100 years.

I am engaged in a fairly lengthy study of the High Quality Australian Site Network, and as part of that project have analysed the data from the South Australian sites.  I averaged maxima and minima for all stations at each site, then compared with the High Quality means.  A spokesperson for BOM has asserted that: 

“On the issue of adjustments you find that these have a near zero impact on the all Australian temperature because these tend to be equally positive and negative across the network (as would be expected given they are adjustments for random station changes).”

Here are the results for South Australia. 

It may be comforting to see that there is no warming bias.   However, a closer look reveals some problems.

Mt Gambier and Strathalbyn are included, although both were designated as Urban stations by Torok and Nicholls (1996) and therefore not supposed to be used.

The sites are “High Quality” because they supposedly have observations over a long term: Torok and Nicholls preferred 80 years but for many sites this could not be achieved so nearby sites were combined.  Fair enough, but the South Australian data is particularly scanty.

Here is the number of years that raw data is available for each site, from least to most. 

Total station-years: 749 = 74.9%.

The official record for South Australia is based on less than 75% of the optimum data.  A quarter of the record is missing, and only 4 stations can truly be called long term with at least 80 years of data, and one of them is urban!

If the stations with less than 80 years of data are removed, the raw trend is reduced to 0.8 C per 100 years, and the adjusted trend to 0.93.  With the Urban location of Strathalbyn removed as well, the raw trend becomes 0.63 and the adjusted trend 0.9.  That would never do. 

Here is a map of the sites (compare with the Trend Map above):

As well, the combination of stations at each location has led to some questionable splices. 

Here are graphs of the raw and adjusted data from each location.

Tarcoola- in the desert, 200km inland from the Bight, about 80km west of the Stuart Highway.

Woomera- in the late 1940s the British wanted a testing range for their missile program, and the Woomera Rocket Range was started.  Records start from  1950, and we can be fairly certain the scientists would have wanted very accurate data.  BOM took over in the early 1970s.  Only 60 years.

Ceduna- on the Great Australian Bight

30 years missing.  Now the adjustments (I’ve included Streaky Bay to get data from earlier years): 1.1 cooled to 0.7.

A splice of Ceduna with Streaky Bay less 0.7C is pretty close, and gives a trend of 0.9.  Where do they get the previous 15 years?

Yongala- about 200km north of Adelaide and 60km from Port Pirie.  Yongala has many years missing, but notice the cooling trend at the start:

So why the warming from from 0.8 to 1.5/100 years?   Georgetown 49km away shows distinct cooling.

And a purely theoretical splice of Yongala and Georgetown less 1.5 gives a trend of 0.35!

Rayville Park- (Snowtown site until 2000)- about 100km north of Adelaide. +0.45?

Here’s the HQ plot:I couldn’t work out how they created this series with a trend of 0.8 until I realised they had adjusted 90 years on the basis of 2 years overlap! They reduced the whole series by about 0.6, with some extra cooling at the start.  I did a similar splice, but only reduced the Snowtown  data to 1956 by 0.2 and increased Rayville Park to match and got the following:

0.55C, which is exactly the same as nearby Clare. 

Strathalbyn Racecourse, in the south east, not far from the mouth of the Murray.  This is listed as urban by Torok and Nicholls, therefore should not have been used:

The two sites spliced gives a trend of 1.3 which BOM reduces to 1.

Robe- on the coast in the south east

Here’s HQ and a splice identical to it, except for adjustments:

+0.65 to +0.4.

Mt Gambier- in the far south east, near the Victorian border, in rich farm country, and also Urban.

Notice two records, but 30 years missing data.  The trend of the raw data spliced gives 1.5, the same as HQ:

But Robe is less!! 

Actually, only the minima are missing- the maxima are available and show something interesting. Here’s the maximum data less 4.9 (to match the mean):

I spliced the two records and plotted with HQ:

0.1! 

Cape Northumberland- on the coast 35km south of Mt Gambier, has only48 years of data!  But BOM still manage to make a trend:

At least it has been cooled from 1.2 to 0.9.   But where did they get the rest of the data?  Robe?  What about Hamilton in Victoria?

Why not stick with Hamilton?

Cape Borda- on the western end of Kangaroo Island. 

Progress report on the Australian High Quality Site Network:

Sites checked:             65 out of 100

Raw trend:                  +0.8 degrees C/ 100 years

High Quality trend:     +1.00

Average difference:    +0.19

Warming bias:           23.95%