Realism of Global Reanalyses over Antarctica and the Southern Ocean across the 20th and Early 21st Centuries

Abstract:

Thanks to their ability to synthesize a wide variety of meteorological data and “produce data where there are none”, nowhere are global atmospheric reanalyses more needed than over Antarctica and the Southern Ocean, where long-term records are few and far between. At the same time, because of the reduced observational constraint, nowhere else do reanalyses face more challenges, which can significantly reduce the reliability of their products. Most notably, the transition into the modern satellite era in 1979 greatly affected the skill of the NCEP-NCAR and ERA-40 reanalyses and explains why most reanalysis-based investigations of Antarctic climate change have since focused on the post-1979 period. The most recent generation of reanalyses has benefited from advances in numerical weather prediction, data rescue/quality-control and lessons learned from previous projects, all of which have improved their overall quality in the southern polar region. Alongside comprehensive reanalyses, recent data assimilation experiments using only surface or conventional observations have tried to overcome the 1979 “barrier”, but other problems have become apparent, including the necessity to specify realistic ocean boundary conditions (sea ice, SST) around Antarctica prior to 1979. The presentation will provide an overview of the skill of recent global reanalyses in high southern latitudes from the 1900s onward. It will discuss how far back in time we can reasonably go and in what areas caution is needed. The datasets covered will include ERA-Interim, CFSR, MERRA, and JRA-55, as well as recent century-long efforts, 20CR and ERA-20C.