The update from v2 to v3 processing is described in
Smith et al., 2008.
The v3 improvements are justified by testing with simulated data. ERSST v3 has improved
low-frequency (LF) tuning that reduces the SST anomaly damping before 1930 using the
optimized parameters. Both in situ and satellite (AVHRR) SST data were used as inputs
in the Smith et al., 2008. However, the addition of satellite SSTs introduced
residual cold bias (in the order of .01 deg C).
AVHRR is an infrared (IR)-based instrument. IR measurements can only be
obtained in clear-sky conditions, and cloud contaminated data are often
difficult to identify. This contamination leads to a
cold SST bias in the retrievals.
An attempt was made to correct for these biases as mentioned in
Reynolds et al., 2007, but the
adjustment did not fully compensate for the cold bias. While this small difference did not
strongly impact the long term trend, it was sufficient to change rankings of warmest month
in the time series, etc. Therefore, the use of satellite SST data was discontinued.
Except for the removal of the satellite aspect, ERSST v3b processing is identical to version 3.
More details can be found in the papers listed in the