Previous- - 3.3 Pervasive Systematic Errors and Biases Index Next - 3.3.2 Bias Adjustments 1941 to Present

3.3.1 Bias Adjustments 1850 to 1941

The need for adjustments to minimize the cold bias associated with bucket measurements in the period from 1850 to 1941 is well established. Folland and Parker [1995] calculated adjustments using a simplified physical model of the buckets used to make SST measurements combined with fields of climatological air-temperature, SST, humidity, wind and solar radiation. Some parameters in their model were taken from literature and others were estimated from the data. The length of time between the water sample leaving the sea surface and the measurement was estimated by integrating their model until a seasonal cycle in the SST was minimized. The fractional contributions of canvas and wooden buckets were estimated by assuming a linear change over time from a mix of wooden and canvas buckets to predominantly canvas buckets by 1920. The rate of this change was estimated by minimizing the air-sea temperature difference in the tropics. The same method was also used in Rayner et al. [2006] and Kennedy et al. [2011c].

Smith and Reynolds [2002] took an alternative approach. They adjusted SSTs based on statistical relationships between Night Marine Air Temperature (NMAT) and SST. The resulting adjustments were different to those produced by Folland and Parker [1995] although the magnitude of the global average adjustment was similar. Both Folland and Parker [1995] and Smith and Reynolds [2002] found a long term increase in the magnitude of the adjustments that is, an increasing cold bias from the 1850s to 1941. Huang et al. [2015, 2016] and Liu et al. [2015] extended this approach to the full SST record and also created an ensemble data set.

The methods employed by Folland and Parker [1995] and Smith and Reynolds [2002] are not independent as they both rely on NMAT, which have their own particular pervasive systematic errors [Bottomley et al., 1990; Rayner et al., 2003; Kent et al., 2013]. The use of NMAT to adjust SST data is, to an extent, unavoidable as the heat loss from a bucket does depend on the air-sea temperature difference.

In data sets based on a ICOADS release 2.0 and later, the earlier bucket adjustments were found to over-adjust SST in the period 1939-1941. Rayner et al. [2006] and Smith et al. [2008] ramped the adjustments down to zero over this period. Kennedy et al. [2011c] showed that the ramp-down corresponded to new data in that release of ICOADS that included a large fraction of ERI measurements.

Previous- - 3.3 Pervasive Systematic Errors and Biases Index Next - 3.3.2 Bias Adjustments 1941 to Present