Beschreibung:
<jats:title>Abstract</jats:title>
<jats:p>Reanalysis data provide a good estimate of global atmospheric temperature and wind fields. However, the available reanalysis datasets reveal nonnegligible discrepancies in their mean state and temporal variability. In this study, the quality of eight reanalysis datasets is evaluated by examining their dynamical consistency in the extratropical stratosphere. The dynamical consistency is quantified by computing the residual of the zonal-mean momentum equation. The residual is generally small in the lower stratosphere, especially at and below 30 hPa, but increases significantly aloft in both hemispheres poleward of 45°, where the effect of parameterized gravity wave drag becomes important. However, at most levels, a large difference in the residual is found among the datasets. This interdata difference is mainly caused by an uncertainty in the Coriolis torque. The non-quasigeostrophic terms, such as those associated with the vertical motion, also play a nonnegligible role when the polar vortex accelerates or decelerates.</jats:p>
<jats:p>The latest reanalysis datasets exhibit smaller residuals than their earlier counterparts. For example, ERA-Interim is dynamically more consistent than ERA-40. This improvement over the generations is largely attributed to a better representation of the Coriolis torque. This is not likely achieved by the increase in satellite data observations over the past few decades. In fact, the dynamical consistency is only weakly sensitive to the analysis period. Instead, model-specific factors, such as data assimilation technique, model resolution, and physics, likely play a crucial role in improving the dynamical consistency.</jats:p>