The media reports today that 2005 is among the hottest years on record. This claim is based on the global average surface temperature record, which as discussed several times on this weblog is fraught with serious data quality issues. Our recent paper has even shown that a warm bias exists in the data.
The media supports this claim of the hottest year (or nearly so) by stating
âFour separate temperature analyses released Thursday varied by a few hundredths of a degree but agreed it was either the hottest or second-hottest year since the start of record-keeping in the late 1880s.â?
This is a misleading statement. The âfour separate temperature analysesâ? are mostly from all or a subset of the same raw data!
While the statement is clarified later in the article,
âThe groups use the same temperature data but differ in how they analyze them, particularly in remote areas such as the Arctic, where there are few thermometersâ?,
this important caveat is missing in the earlier statement in the article (moreover, as we will show in a soon to be submitted paper, other areas also have a sparcity of data; for 20N to 20S, for example, 70% of the grid areas over land have 1 or less observation sites).
The raw surface temperature data from which the four analyses are derived are, therefore, essentially the same. That the four analyses produce similar trends should come as no surprise!
A question to the different groups which has been posed to several of them, but they have not answered, is what is the degree of overlap in the data sets? While some of the analyses use subsets of the raw data, the raw data is almost identical. To frame this question another way, what raw surface temperature data is used in each analysis that is not used in the other analyses? The best estimate we have seen is that 90-95% of the raw data is the same.
Not to highlight this important issue is an example of cherrypicking; this time by the analyses groups that are releasing the surface temperature data.