Tuesday, October 16, 2007

Statistical Incompetence at the Australian Bureau of Meteorology

The Australian Bureau of Meteorology, might well be good at collecting data, in fact they are world recognised as amongst the best. All that is however completely useless if you cannot analyse the data well. It would seem that either they don't employ statisticians to analyse their data or that the statisticians that they employ are well, incompetent.

Let me tell you why.

The Australian Bureau of Meteorology, on their website, love to look at climate extremes. Like how many very hot days have we had etc. All point to the obvious; that we are having more hot days and less cold ones, more hot nights and less cold ones - An obvious conclusion when considering the theory of Global Warming.

But lets concentrate on the nights. They claim that Australia has seen an increase in very hot nights, a veyr large increase in hot nights, whilst also having a decrease in cold nights, a decrease in frost nights (thank goodness - the farmers will be happy) and a decrease in cool nights.

But lets look at how they came to these graphs. This website shows their analysis, where you will notice that the above information is solely put together using the minimum temperature only.

We pointed out before that minimum temperatures in Australia have been increasing since the second world war, but we also showed that temperatures at Midnight, 3am and 6am have shown no statistically significant increase in temperature.

This is because the minimum temperature actually occurs usually around 15 to 60 minutes after sunrise. Thats right, the minimum temperature does not occur at night. Hence the minimum would be influenced by the sun, and is not a reliable statistic to measure overnight temperatures. It doesn't even occur at night.

But that doesn't stop the Australian Bureau of Meteorology from announcing increasing warm nights and decreasing cold nights based on the minimum temperature. Complete Incompetence indeed.

It is a shame that the ABOM cannot analyse their own data statistically well, because their collection methods are one of the best. Generally they get the CSIRO to do the analysis, and, well, we all know how good they are at doing statistical analysis too.

Incompetence or perhaps intolerance.

No comments: