ERRORS IN CSIRO TEMPERATURE CHANGES

IMPORTANT NOTE (APRIL 2005): ERRORS IN CSIRO TEMPERATURE CHANGES

In March-April 2005 it was discovered that the CSIRO SRES temperature data employed in the Ruosteenoja et al. (2003) report were erroneous. Due to an error in the model specifications provided to the IPCC DDC for all CSIRO SRES runs, the labels for model skin temperatures and model air temperatures were reversed and therefore skin temperatures were inadvertently analysed rather than air temperatures. These specifications have since been corrected at the DDC (in late 2002).

The error only concerns the CSIRO temperature projections (see below). Temperature responses of other models are correct, as well as all precipitation projections. Unless users specifically wish to analyse CSIRO temperature change results, the main messages of the report and associated scatter diagrams concerning the range of uncertainties in modelled projections of regional climate remain unaffected.

The median, 90^th and 99^th percentile of the absolute errors are 0.05K, 0.19K and 0.52K respectively. The largest errors are +0.83K (region 2 – Arctic ocean, 2070-2099, December-Feb, A1FI) and -0.84K (region 5 – Alaska, NW Canada, 2070-2099, A1FI).

A revised text file has been prepared with the corrected values:
REVISED PLOTPOINTS3.TXT

A full list of the differences between the two versions is available here: plotpoints_diffs.txt

.