AT Bourke temperatures were carefully recorded at the post office for 125 years, from 1871 to 1996. But this record is ignored by those announcing new temperature records on local Bourke radio. They rely on a record that only goes back to 1998!
From about 1952 through until 1996 the Postal Service staff carefully read and recorded the weather in the Stevenson screen at 9am, 12noon, 3pm, 6pm and then went back to the post office each evening at 9pm, 3am – yes the temperature was read manually at 3am by Postal Service staff – and then again at 6am. This information was immediately relayed to the Bureau in Melbourne.
But go to weatherzone.com.au and it suggests that temperatures at Bourke have only been recorded since 1998. What a travesty. 
http://www.weatherzone.com.au/climate/station.jsp?lt=site&lc=48245
There have been complaints to the local radio station that keeps announcing record hot days on the basis of this much-truncated record; temperatures for Bourke back to 1998. The radio station says that it relies on weatherzone.com.au for its information. Complaints to weatherzone have resulted in comment that the Bureau are unable to merge data, and can only provide Weatherzone with information on a station by station basis with the ‘Bourke Airport’ the current open weather station for Bourke.
Meanwhile, when it comes to the generation of contrived temperature data for Australia’s annual average temperature anomaly the Bureau have merged the airport and the post office, in fact the post office has been subsumed by the airport. This data has also been shortened, from 1871 to 1910, and of course ‘homogenized’.
Another travesty.
Thanks Bob FJ.
Neville @ April 18th, 9:26 pm
First of all the Phil Jones climax in the warming given as 1940 is based on earlier HadCrut3 data which has since been adjusted in a general anticlockwise direction of the time-series curves. HadCru have uniquely used within THEIR “climate science” a 21-year smoothing with weighting of linear triangulation simplified Gaussian distribution. (Don’t ask me why, and frankly, I think it is almost bizarre!). Nevertheless, if you look here at CURRENT unadjusted HadCrut3:
http://www.cru.uea.ac.uk/cru/data/temperature/HadCRUT3.pdf
The peak in the GLOBAL break-over curve is pretty well-on 1940 but also notice that the 21-year smoothing curve seems to ignore the 1942 high outlier:
If you play again with the BoM trending options on what appears to be recent Hadcrut3 unadjusted, you should be able to see that trending options affect the break-over curve in more ways than you suggest.
Keeping this brief, what you need to recognise is that the VERY MANY smoothing methods give different outcomes and may be chosen to emphasize a point of view by a researcher.
And, even linear smoothing comes in various forms. WFT use OLS (Ordinary Least Squares) and you should check this out:
http://en.wikipedia.org/wiki/Linear_regression
Gotta go
The Bolter nails the extremists on all their dud predictions. And jumping in with both feet in his gob is that dingaling Adam Bandt.
http://blogs.news.com.au/heraldsun/andrewbolt/index.php/heraldsun/comments/bandt_proves_brandis_right_yes_the_greens_want_debate_suppressed/
Thanks Bob FJ , I've had a play and yes I notice the line moves down as the moving average number goes up.
But I'll ask again why does WFTs show a higher trend for 1910 to 1945 for both HAD 3 and HAD 4? The warming trend is supposed to be from from 1910 to 1940 ( not to 1945) according to Phil Jones.
Neville @ April 17th, 8:04 am
Ok, since you ask nicely I’ll let my lunch go down before returning to work
It is a matter of “loose” statistics. First of all it is generally recognised that there is an underlying cycle in the temperature record of about 60-years, with a pronounced peak somewhere around 1940. (This was possibly made more pronounced by the WW2 El Nino, although some parts like Europe and Russia had very cold winters)
Unfortunately, the data are very noisy with a sharp break-over curve together with a high outlier in 1942, which complicates matters. In short, linear trend treatments should be used with great caution when there is an obvious underlying cycle with high noise. Instead, moving averages are arguably of better use.
Here’s an exercise for you. Go to the following link where the BoM appears to have the HadCrut3 unadjusted global average time-series, but with the advantage of being able to assess different trending methodology:
http://www.bom.gov.au/climate/change/index.shtml#tabs=Tracker&tracker=global-timeseries&tQ%5Bgraph%5D=global_t&tQ%5Bregion%5D=global&tQ%5Bseason%5D=0112&tQ%5Bave_yr%5D=3
I have set it to a 3-year moving average, but if you grab and drop the cursor on the scale you can increase it progressively up to 15-years. Notice that the centre of the break-over curve moves as you go.
Have a play with that and I’ll come back to your other points.