Have we been underestimating midlatitude air-sea interaction?

Some traditional, climate-oriented sea surface temperature (SST) observational datasets do not generally include satellite data and are typically based on in-situ observations with a coarser spatial resolution (1 to 2 degrees), prominent examples being the Extended Reconstructed SST from NOAA (ERSST) and the Hadley Centre SST, version 3 (HadSST3). Other datasets combine both, in-situ and satellite observations, such as the Hadley Centre Sea Ice and Sea Surface Temperature dataset (HadISST).

The main objective of this work is twofold. First, we globally characterize and compare SST climatology and variability at grid-point level, considering seasonal averages (DJF, MAM, JJA, SON), between two standard, climate-oriented datasets, HadISST (1° resolution) and ERSST v5 (2° resolution), with the GHRSST product developed by the European Space Agency Climate Change Initiative (CCI) (0.05° resolution). Secondly, we assess the impact of temporal and spatial resolution in such SST characterization as well as on air-sea interaction, estimated by correlating SST with turbulent heat flux (THF; latent plus sensible). The study spans over 1982-2016 (35 years) that corresponds to the record of the satellite product (CCI).

Our results show that the coarser datasets (ERSST-HadISST) overall have a warmer mean-state, except in the more dynamically-active oceanic regions such as the western boundary currents where they yield a colder SST climatology. More interestingly, the high-resolution dataset (CCI) markedly displays larger SST variability in these dynamically-active oceanic regions, which is consistent along the seasonal cycle. Likewise, we also find higher correlations between SST and THF over the western boundary currents in CCI as compared to ERSST-HadISST, indicating a stronger ocean-atmosphere coupling. Our results suggest that the high temporal and spatial resolution provided by remote sensing is key to better resolve air-sea interaction.