What is Up With This
Posted by Jeff Id on October 18, 2009
Ok, say you’ve got two sets of temperature data from the same island in the Antarctic. Suppose further that you have a team of qualified scientists recording and archiving the data. Imagine that there are two highly qualified multi-million dollar programs for maintaining qc checking and dispensing the data for the good of science. What would you expect to see if you examined the data recorded from a single thermometer and two different highly qualified government organizations?
I’ve been discussing data from the Australian Bureau of Meteorology with a friend by email who does not wish to be credited. He has provided me daily temperatures for four stations from the BOM archive which match four stations used in Steig et al. The BOM requires payment to receive their data which he has done for quality evaluation purposes. We’ve been investigating whether there are significant differences between the datasets.
This first plot is from Macquarie Island. It was one of the stations used in Steig et al which employed the British Antarctic Survey version of the data (red line). The black line is the Australian BOM. I have processed the data by grouping months together so that both series are by month.
The data from both groups is clearly the same information by the excellent visual match. However my next step was to anomalize the data and look at trend significance. Steig et al used a 1957 – 2007 window of the data but we’re looking at QC issues so this is the whole series.
Both show significant trends over the length of the dataset in relation to AR1 corrected weather. However, the BOM version is almost double the trend of BAS. The next graph is a plot of the difference between the two datasets.
The difference in the datasets is a whopping 8 sigma greater than trends created by short term variance. Unfortunately I can find little discussion of what is obviously a correction either by BAS or BOM or both. This is the worst example of the four stations I’m looking at currently. Perhaps the correction is justified though due to the rampant industrialization of this global exporting powerhouse.