A comparison of the lower stratospheric age spectra derived from a general circulation model and two data assimilation systems

We use kinematic and diabatic back trajectory calculations, driven by winds from a general circulation model (GCM) and two different data assimilation systems (DAS), to compute the age spectrum at three latitudes in the lower stratosphere. The age spectra are compared to chemical transport model (CTM) calculations, and the mean ages from all of these studies are compared to observations. The age spectra computed using the GCM winds show a reasonably isolated tropics, in good agreement with observations; however, the age spectra determined from the DAS differ from the GCM spectra. For the DAS diabatic trajectory calculations there is too much exchange between the tropics and midlatitudes. The age spectrum is thus too broad, and the tropical mean age is too old as a result of mixing older midlatitude air with tropical air. Likewise, the midlatitude mean age is too young because of the in‐mixing of tropical air. The DAS kinematic trajectory calculations show excessive vertical dispersion of parcels in addition to excessive exchange between the tropics and midlatitudes. Because air is moved rapidly to the troposphere from the vertical dispersion, the age spectrum is shifted toward the young side. The excessive vertical and meridional dispersion compensate in the kinematic case, giving a reasonable tropical mean age. The CTM calculation of the age spectrum using the DAS winds shows the same vertical and meridional dispersive characteristics of the kinematic trajectory calculation. These results suggest that the current DAS products will not give realistic trace gas distributions for long integrations; they also help explain why the extratropical mean ages determined in a number of previous DAS‐driven CTMs are too young compared with observations. Finally, we note that trajectory‐generated age spectra show significant age anomalies correlated with the seasonal cycles. These anomalies can be linked to year‐to‐year variations in the tropical heating rate. The anomalies are suppressed in the CTM spectra, suggesting that the CTM transport scheme is too diffusive.