Stratospheric transport using 6-h-averaged winds from a data assimilation system

[1] Stratospheric transport calculated using assimilated winds has been shown to be too fast in many cases, because of excessive mixing and an overstrong residual circulation. It is shown that the use of 6-h-averaged wind fields instead of instantaneous analyses can substantially reduce this problem for NASA's Goddard Earth Observing System version 4 (GEOS- 4) sequential data assimilation system. Two examples are used to illustrate impacts in an off-line chemistry transport model. An age-of-air computation shows that the stratosphere becomes substantially older when time-averaged winds are used, yet still not as old as that determined from observations. An ozone assimilation experiment reveals improvements in the spatial structure of assimilated ozone, better agreement with independent observations, and a 40-60% reduction in observation-minus-forecast residuals. The averaged meteorological fields have also been incorporated in the on-line transport computations: This is equivalent to using a 6-h square-wave filter at the analysis times. Sub-6-h noise in the system is substantially reduced with this approach. Two additional examples of transport are shown. A simulation of 2004-2005 shows that the descent of N2O in the Arctic polar vortex is represented more accurately than in previous studies. A tape recorder signal is evident in the tropical moisture; it ascends about 30% faster than that in the real atmosphere. In summary, the use of 6-h-averaged winds substantially improves the transport characteristics of the assimilated data, although the circulation remains too fast.