Artificial surface pressure trends in the NCEP-NCAR reanalysis over the southern ocean and Antarctica

An examination of 50 years of the National Centers for Environmental Predic tion (NCEP)-National Center for Atmospheric Research (NCAR) reanalysis from 1949 to 1998 reveals that significant spurious trends occur in the surface pressure held. Long-term surface pressure reductions are apparent south of 45 degreesS. The largest trend in surface pressure is near 65 degreesS whe re an approximately steady long-term pressure reduction of about 0.20 hPa y r(-1) (10 hPa in 50 yr) is located. The negative pressure trend represents a gradual reduction in a positive bias for the reanalysis Observations at A ntarctic stations do not support this long-term trend, although short-term interannual variations are reasonably well captured after about 1970. The n egative pressure tendency near 65 degreesS continues well into the 1990s al though a reasonable number of stations between 65 degrees and 70 degreesS b egan taking observations along the coast of east Antarctica during the 1950 s and 1960s. Few Antarctic observations, however, are used by the reanalysi s until about 1968, and the quality of the pressure field for the reanalysi s appears poor in high southern latitudes prior to then. The trend in high southern latitudes appears to be a component of global temporal variations in the reanalysis;some of which are supported by observations but others ar e not. In the Southern Hemisphere, the sea level pressure difference between 40 de grees and 60 degreesS, an indicator of westerly wind intensity, increases a pproximately from 20 hPa in the early 1950s to 25 hPa in the early 1970s an d 28 hPa in recent years. The relatively high density of observing stations along the Antarctic Peninsula, however, results in an approximately steady local surface pressure after the pressure fell about 4 hPa during the late 1950s. Based upon these findings, researchers should account for jumps and long-term trends when making use of the NCEP-NCAR reanalysis.