The anomalies are computed relative to only 2003-2006 because those years were relatively free of El Nino and La Nina activity, which if included would cause temperature anomaly artifacts in other years. Thus, these anomalies cannot be directly compared to, say, the Reynolds anomalies which extend back to the early 1980s. Nevertheless, they should be useful for monitoring signs of recent ocean surface warming, which appears to have stalled since at least the early 2000′s. (For those who also track our lower tropospheric temperature ["LT"] anomalies, these SST anomalies average about 0.20 deg. C cooler than LT since mid-2002, but there is considerable variability in that number).
The SST retrievals come from Remote Sensing Systems (RSS), and are based upon passive microwave observations of the ocean surface from AMSR-E on NASA’s Aqua satellite, the TRMM satellite Microwave Imager (TMI), and WindSat. While TMI has operated continuously through the time period (but only over the tropics and subtropics), AMSR-E stopped nominal operation in October 2011, after which Remote Sensing Systems patched in SST data from WindSat. These various satellite SST datasets have been carefully intercalibrated by RSS.
Despite the relatively short period of record, I consider this dataset to be the most accurate depiction of SST variability over the last 10+ years due to these instruments’ relative insensitivity to contamination by clouds and aerosols at 6.9 GHz and 10.7 GHz.