Global 1 degree monthly sea surface temperature comparisons

Abstract

Comprehensive global comparisons have been made between two satellite-based sea surface temperature (SST) products and related blended satellite/in situ SST products. The satellite SST data sets were those derived from advanced very high resolution radiometer (AVHRR) data using the modified Pathfinder SST (MPSST) and operational NOAA SST (ONSST) algorithms (collectively referred to as satellite SST data). The blended satellite/in situ data included the NOAA Optimal Interpolation (OI) and 2-dimensional variational analysis (2DVAR) SST data sets, and the UK Meteorological Hadley Center Global Sea-Ice and Sea Surface Temperature (HADISST) data set (collectively referred to as "analysis SST data"). Global monthly comparisons for co-located 1 degree squares using standard statistics were computed. Only data from each SST data set from 1985-1997 were used in these comparisons and the satellite data were separated into day/night time periods. Statistics for each comparison set (e.g. MPSST/OI, ONSST/OI etc.) were then generated to determine the global mean bias, RMS differences, standard deviation of bias, and correlation values. These statistics were examined to determine how well the MPSST product performs with respect to the ONSST on a monthly and seasonal basis. Preliminary results indicate that with respect to the mean bias, the ONSST comparisons generally have a lower bias in the low variability central ocean basin region while the MPSST has the significantly lower bias in high variability areas (e.g. western boundary currents). The RMS difference comparisons indicate that the ONSST is on average lower during the nighttime while the MPSST performs better during the daytime.