This study evaluates the temporal variability of the Antarctic surface mass balance, approximated as precipitation minus evaporation (P − E), and Southern Ocean precipitation in five global reanalyses during 1989–2009. The datasets consist of the NCEP/U.S. Department of Energy (DOE) Atmospheric Model Intercomparison Project 2 reanalysis (NCEP-2), the Japan Meteorological Agency (JMA) 25-year Reanalysis (JRA-25), ECMWF Interim Re-Analysis (ERA-Interim), NASA Modern Era Retrospective-Analysis for Research and Application (MERRA), and the Climate Forecast System Reanalysis (CFSR). Reanalyses are known to be prone to spurious trends and inhomogeneities caused by changes in the observing system, especially in the data-sparse high southern latitudes. The period of study has seen a dramatic increase in the amount of satellite observations used for data assimilation.The large positive and statistically significant trends in mean Antarctic P − E and mean Southern Ocean precipitation in NCEP-2, JRA-25, and MERRA are found to be largely spurious. The origin of these artifacts varies between reanalyses. Notably, a precipitation jump in MERRA in the late 1990s coincides with the start of the assimilation of radiances from the Advanced Microwave Sounding Unit (AMSU). ERA-Interim and CFSR do not exhibit any significant trends. However, the potential impact of the assimilation of rain-affected radiances in ERA-Interim and inhomogeneities in CFSR pressure fields over Antarctica cast some doubt on the reliability of these two datasets.The authors conclude that ERA-Interim likely offers the most realistic depiction of precipitation changes in high southern latitudes during 1989–2009. The range of the trends in Antarctic P − E among the reanalyses is equivalent to 1 mm of sea level over 21 years, which highlights the improvements still needed in reanalysis simulations to better assess the contribution of Antarctica to sea level rise. Finally, this work argues for continuing cautious use of reanalysis datasets for climate change assessment.

Snowfall was measured at 11 sites in the McMurdo Dry Valleys to determine its magnitude, its temporal changes, and spatial patterns. Annual values ranged from 3 to 50 mm water equivalent with the highest values nearest the coast and decreasing inland. A particularly strong spatial gradient exists in Taylor Valley, probably resulting from local uplift conditions at the coastal margin and valley topography that limits migration inland. More snow occurs in winter near the coast, whereas inland no seasonal pattern is discernable. This may be due, again, to local uplift conditions, which are common in winter. We find no influence of the distance to the sea ice edge. Katabatic winds play an important role in transporting snow to the valley bottoms and essentially double the precipitation. That much of the snow accumulation sublimates prior to making a hydrologic contribution underscores the notion that the McMurdo Dry Valleys are indeed an extreme polar desert. Copyright © 2009 Royal Meteorological Society