JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 101, NO. D6, PAGES 10,311-10,334, APRIL 30, 1996 Comparison of U.K. Meteorological Office and U.S. National Meteorological Center stratospheric analyses during northern and southern winter G. L. Manney, • R. Swinbank, 2 S. T. Massie, 3 M. E. Gelman, 4 A. J. Miller, • R. Nagatani, • A. O'Neill, • R. W. Zurek • Abstract. Meteorological data from the United KingdomMeteorological Office (UKMO), produced using a data assimilation system, and the U.S. National Meteorological Center (NMC), produced using an objective analysis procedure, are compared for dynamically activeperiods during the Arctic and Antarctic wintersof 1992. The differences seenduring these periodsare generallysimilar to thoseseen during otherwinterperiods. Both UKMO and NMC analyses capture the large-scale evolution of the stratospheric circulation during northernhemisphere (NH) and southern hemisphere (SH) winters. Stronger verticaland horizontal temperature gradients develop in the UKMO than in the NMC data during stratospheric warmings; comparison with satellite measurements with better vertical resolution suggests that the stronger vertical temperature gradients are more realistic. The NH polar vortex is slightly strongerin the UKMO analyses than in the NMC in the middle and upper stratosphere, and midstratospheric temperaturesare slightly lower. The SH polar vortex as represented in the UKMO analyses is stronger and colderin the midstratosphere than its representation in the NMC analyses. The UKMO analyses on occasion exhibit some difficultiesin representing cross-polar flow or changes in curvature of the wind field at very high latitudes. In addition to the above study of two wintertime periods, a more detailed comparison of lower-stratospheric temperatures is donefor all Arctic and Antarctic winter periods since the launch of the Upper Atmosphere Research Satellite. In the NH lower stratosphere during winter, NMC temperatures are consistently lowerthan UKMO temperatures and closer to radiosonde temperatures than are UKMO temperatures. Conversely, in the SH lower stratosphere during winter, UKMO temperaturesare typically lower than NMC and are closerto radiosonde temperature observations. 1. Introduction Instruments onboard the Upper Atmosphere Research Satellite (UARS) have been measuring stratospheric temperatures, winds, and constituent speciessince its launch in September 1991. Among the correlative data that aid in interpretation of UARS data are two sets of global meteorological analyses. One is provided by the U.K. Meteorological Office (UKMO), the other by • Jet Propulsion Laboratory/California Institute of Technol- ogy, Pasadena. 2Meteorological Office, Bracknell, England, aNational Center for Atmospheric Research, Boulder, Colorado. 4ClimateAnalysis Center,NMC, NWS, NOAA, Washington, D.C. 5Centre for Global Atmospheric Modelling, Reading, England. Copyright 1996 by the American Geophysical Union. Paper number 95JD03350. 0148-0227 / 96/95 J D- 03350 $05.00 the U.S. NationalMeteorological Center(NMC). Both analysesincorporate measurements made by radioson- des and by the NOAA series of polar-orbiting satellites. though the analysistechniques are quite different. The UKMO uses a data assimilation technique to produce daily analyses of geopotential height, temperature, and winds[$winbank and O'Neill, 1994a]. The NMC uses an objective analysisscheme to produce daily analyse•,. of temperatures and geopotential heights (from which winds may be computed) [Gelman et al., 1994].These two data sets are both being widely used to help in- terpret UARS measurements; it is thus important to determine how consistently they represent the strato- spheric circulation. The degree to which the two data sets agree gives an indication of the reliability of both data sets. Where they disagree,it is sometimes possi- ble to relate differences to known shortcomings in the analysisprocedures.Understandingthe differences be- tween the two data sets will help those who wish to use meteorological data in scientificanalyses to determine to what level of detail they may rely on such data. We begin to address these questions by comparing 10,311