Saturday, 18 July 2015

Australian Bureau of Meteorology Station Data Review


In a previous post I described my adventures downloading and converting the Bureau of Meteorology’s (BOM’s) Climate Data Online (CDO) and its Australian Climate Observation Reference Network – Surface Air Temperature (ACORN-SAT) datasets.

The BOM claims “The Australian Climate Observations Reference Network – Surface Air Temperature (ACORN-SAT) dataset has been developed for monitoring climate variability and change in Australia. The dataset employs the latest analysis techniques and takes advantage of newly digitised observational data to provide a daily temperature record over the last 100 years. This data will enable climate researchers to better understand long-term changes in monthly and seasonal climate, as well as changes in day-to-day weather. ”

Another BOM statement is more worrying: “ACORN-SAT is a complete re-analysis of the Australian homogenised temperature database.” The emphasis on ‘re-analysis’ is mine.

As Gary Smith says in his wonderful book, Standard Deviations: Flawed Assumptions, Tortured data and Other Ways to Lie with Statistics, self-selection bias leads researchers to find what they are looking for and to ignore anything that contradicts their pet theories.

In the case of climate change research and temperature records, this may lead researchers to find the past temperatures colder ‘than first thought’ and more recent temperatures warmer. I’m keen to see if there is any evidence of this taking place with Australia’s BOM.

I sincerely hope not.

You may recall from the last post on this topic that the CDO database contains temperature data from 1,871 weather stations from around Australia, dating as far back as1855. There are separate records for maximum and minimum temperatures and to be useful, both maximums and minimums must be available for the periods of time in question.  On this basis, CDO contains data for 1,718 ‘useful’ stations, as I’ve chosen to call them.

The BOM has selected 112 of these stations to form the ACORN-SAT dataset.  Why 112? the BOM explains “The locations are chosen to maximise the length of record and network coverage across the country.” Sounds fair enough, but leaves the question what’s wrong with the others?

BOM Question 1 Is  descriptive data consistent

Note: The black dots are ACORN-SAT stations. The red dots are CDO stations.

While reviewing the documentation associated with the data, I discovered another interesting fact: The 112 ACORN-SAT stations are actually composed of records from 202 CDO weather stations.

Confused?  I was.

Only 38 of the 112 station records are based on a single weather station.  The remaining 74 are composites of the data from 2, 3 or in one case 4 separate weather stations. While there are, no doubt, excellent reasons for doing this it makes before and after adjustment comparisons more difficult.

Even the exact makeup of the composite stations was difficult to clarify.  BOM documents describing the final station makeup gave different results. This crucial information was not available electronically.

Never mind.  I got it in the end.

I’ve compiled an extensive analysis of the station data that’s available as a PDF from the link below.

It’s based on queries of my downloaded database and was analysed with Mathematica. It’s got maps and is pretty self-explanatory, but probably a bit too detailed for some.

Click here to download the PDF file.

There’s one more bit of analysis I intend to do prior to looking at the actual temperature data. In my next post I’ll address the timespan of both the CDO and ACORN-SAT stations. The BOM claims that it picked CDO stations with the longest periods of observation for ACORN-SAT.

We’ll see.

No comments:

Post a Comment

Got a comment for me?