G'day John,
A really good question but it doesn't match the thread. But since the Climate threads are not staying on topic for more than a post or so anyway, why should this make any difference.
I was quite happily having a email conversation with an expert at a Climate centre. He was happily providing some detail of some of the extraordinary problems with indidivual weather bureaus, within states, within country and across states and countries. The conversation came to an abrupt stop when I asked a question relating to a study that he apparently didn't like to be challenged in any way.
But here is the problem. I can tell you for instance that in Western Australia they record a record high or low as the day that it is read. In New South Wales, where I live, the record the record high or low as the day it most likely occurred on (the previous day). That makes little difference to averages but a very big difference to some other studies related global warming.
I can tell you how the average is currently calculated by the Australian Met Bureau but not how they did it 15 years ago. But a great deal of bureaus provide their data to say the UK Met Bureau on a proprietary basis and no one else is able to access the raw data. Worse, the averaging system also has the same anonimity and might have changed more than once even within a year, let alone over several years.
So if all you have is monthly averages, there is next to no chance at all of determining how the average was calculated for an individual station or state, region or country. This is one of the reasons why raw data is something I'm currently so focused on. At least with raw data, all I have to do is look at the monthly average and from the raw data can determine how the average was created and what the difference is due to that particular method of calculation. Indeed, I would like to produce a data set that includes an average always calculated the same way; plus the method adopted in the main data sets currently used, and the difference this produces, plus modifications to the data, undertaken by the group that produced the data set, and how this differs from the raw data, etc.
It should show pretty clearly just what different methods achieve but actually getting the raw data is the most difficult part. Finding out what happened to a weather station can only be achieved if a sample group is made. It is just way too difficult to make the various enquiries needed to find that level of detail for more than a very few stations.
Regards
Richard
