Spatial and temporal patterns of SST uncertainty

This study investigates the spatial and temporal structures in the uncertainties provided in five sea‐surface temperature (SST) Level 4 gridded products: Operational Sea Surface Temperature and Sea Ice Analysis (OSTIA) near‐real‐time and reprocessed datasets, European Space Agency Sea Surface Temperature Climate Change Initiative (ESA SST CCI) and Copernicus Climate Change Service reprocessed SST analyses, v2.1, ESA SST CCI v3, and Daily Optimum Interpolation Sea Surface Temperature v2. These product uncertainties are compared to the standard deviation of SST across these products, and also the standard deviation of the Group for High Resolution Sea Surface Temperature Multi‐Product Ensemble and ERA5. Across all domains examined ( to , United Kingdom, Gulf Stream and tropical Pacific), there are some clear expected structures: a decreasing trend in SST uncertainty with time (attributed to the addition of new sensors, improved sampling, and reduced observation and retrieval uncertainty), as well as increased uncertainties in regions of higher SST variability, such as in western and eastern boundary currents. Unexpected structural patterns (and steps in uncertainty value) were found to be associated with changes in processing scheme as well as increasing uncertainty with proximity to the coast (identified via cluster analysis). Both effects are attributed to the methods employed within the OSTIA system. Although individually well evidenced by physical or processing drivers, the cumulative effect at regional scales may be misleading and interpreted incorrectly if assumed to be purely observationally based. Such resultant structure in the uncertainty fields may impact applications where the SST products are used as boundary conditions, such as in numerical weather prediction and within coupled models.