(Manuscript recieved 25 September 1991, in final form 29 May 1992)
Any discussion of intraseasonal and interannual variability in the atmosphere must presume a reliable assessment of the observed variability. In spite of continued improvements in observing systems, quality control techniques, and data analysis schemes, however, and also because of them, this assessment remains difficult in the tropics.
In this paper the authors examine the mean tropical circulation during two Januarys, 1988 and 1989, as described by the circulation analyses produced at two weather prediction centers, the National Meteorological Center (NMC) in Washington, D.C., and the European Center for Medium-Range Weather Forecasts (ECMWF) in Reading, England. In particular, the authors' focus is on the change in the circulation between 1988 and 1989 as estimated by these two sets of analyses, especially the change in the 200-mb wind divergence associated with organized deep convection. The authors find that in many regions the discrepancy between these estimates is of the order of the change itself. A comparison with maps of the outgoing longwave radiation (OLR) is not quantitatively useful in this regard.
One way out of this dilemma is to derive divergence fields that are consistent with the 200-mb vorticity balance. The authors do so by solving the "chi problem" of Sardeshmukh ans Hoskins. Because the large-scale voricity fields generated by NMC and ECMWF are highly correlated (~98%), the divergence fields derived in this manner are also better correlated than the analyzed fields and enable a more reliable assessment of the observed change between these two periods.