Choosing the Right Sea Surface Temperature Datasets: Benchmarking and Guidance for Climate Applications

Abstract Sea surface temperature (SST) datasets underpin many climate applications, including climate monitoring, the study of historical climate variability and change, attribution, model evaluation, ecosystem assessment, paleo-proxy calibration, and use as boundary conditions for atmospheric model simulations. Many different SST products are available. This paper addresses why SST products differ, what these differences mean for climate analyses, and which products are best suited for various purposes. Differences among SST products are first reviewed with respect to improvements in bias adjustments, gridding and infilling techniques, and uncertainty quantification. The implications of these advances are then assessed through historical case studies, evaluation of spatial patterns, and comparison of global means and key regional indices. Substantial discrepancies in trends are found during the satellite era using older SST products, but recently released datasets are much more consistent. Recent datasets also show a more-consistent SST evolution during World War II and in trends associated with Tropical Pacific zonal gradients. Disagreements persist, however, with respect to early-20 th -century warming and in data-sparse regions such as the Southern Ocean and Arctic. To assist users across disciplines, we articulate principles for dataset selection based on application needs and highlight the NCAR Climate Data Guide and an accompanying web-based data-selector tool that provides updated benchmarking and access to SST products.