Previous - 3.3.5 Assessing the Efficacy of Bias Adjustments Index Next - 3.4 Sampling Uncertainty
The need to adjust SST data prior to 1941 to account for a cold bias associated with the use of canvas and wooden buckets is well established. There is also good evidence for the need to adjust data after 1941. Adjustments for these pervasive systematic errors have been developed. There are, at all times, three different estimates of the bias adjustments, which are in general agreement, but locally differ, and give a first indication of the structural uncertainty. Evidence for the efficacy of the adjustments comes from laboratory tests, wind tunnel tests, comparisons with coastal sites and consistency with subsurface ocean temperatures, marine air temperatures and land air temperatures. Contrary evidence comes from a recent field experiment in the Pacific. Uncertainty could be better understood by: improvements in metadata; carefully designed fields tests of buckets and other measurements methods; the creation of new independent evaluations of the biases; and continued comparison between SST and related variables.
Previous - 3.3.5 Assessing the Efficacy of Bias Adjustments Index Next - 3.4 Sampling Uncertainty