Variability in Bias of Gridded Sea Surface Temperature Data Products: Implications for Seasonally Resolved Marine Proxy Reconstructions
Abstract
Seasonally resolved reconstructions of sea surface temperature (SST) are commonly produced using isotopic ratios and trace elemental ratios within the skeletal material of marine organisms such as corals, coralline algae, and mollusks. Using these geochemical proxies to produce paleoclimate reconstructions requires using regression methods to calibrate the proxy to observed SST, ideally with in situ SST records that span many years. Unfortunately, the few locations with in situ SST records rarely coincide with the time span of the marine proxy archive. Therefore, SST data products are often used for calibration and they are based on MOHSST or ICOADS SST observations as their main SST source but use different algorithms to produce globally gridded data products. These products include the Hadley Center's HADSST (5°) and interpolated HADISST (1°), NOAA's extended reconstructed SST (ERSST; 2°), optimum interpolation SST (OISST; 1°), and the Kaplan SST (5°). This study assessed the potential bias in these data products at marine archive sites throughout the tropical Atlantic using in situ SST where it was available, and a high-resolution (4 km) satellite-based SST data product from NOAA Pathfinder that has been shown to closely reflect in situ SST for our locations. Bias was assessed at each site, and then within each data product across the region for spatial homogeneity. Our results reveal seasonal biases in all data products, but not for all locations and not of a uniform magnitude or season among products. We found the largest differences in mean SST on the order of 1-3°C for single sites in the Gulf of Mexico, and differences for regional mean SST bias were 0.5-1°C when sites in the Gulf of Mexico were compared to sites in the Caribbean Sea within the same data product. No one SST data product outperformed the others and no systematic bias was found. This analysis illustrates regional strengths and weaknesses of these data products, and serves as a cautionary note against the wholesale use of a particular gridded data product for marine proxy calibration, whether for a single site or larger regional reconstruction, without considering the inherent heterogeneous bias present in each data product that we show varies among locations. Furthermore, this study has implications for comparing climate models and these SST data products.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2016
- Bibcode:
- 2016AGUFMPP53A2372O
- Keywords:
-
- 4299 General or miscellaneous;
- OCEANOGRAPHY: GENERALDE: 4999 General or miscellaneous;
- PALEOCEANOGRAPHY