Validation of GOME ozone profiles and evaluation of stratospheric transport in a global chemistry-transport model.

This paper presents a validation of GOME O3 profiles which are used to evaluate stratospheric transport in the chemistry-transport model (CTM) Tracer Model version 5 (TM5) using a linearized stratospheric ozone (O3) chemistry scheme.
A comparison of GOME O3 profile measurements with independent O3 sonde measurements at mid-latitudes shows an excellent agreement. Differences are smaller than 5 %, well within the uncertainty of the O3 sonde measurements. Within the tropics the GOME O3 profiles differences are larger with a clear lower stratospheric negative O3 bias with compensating positive biases in the troposphere and higher stratosphere.
The TM5 model with linearized O3 chemistry simulates realistic lower and middle stratospheric spatial and temporal O3 variations on both short (daily) and long (seasonal) timescales. Model stratospheric O3 is significantly overestimated in the extratropics and slightly underestimated in the tropics, as also shown in a comparison with TOMS total O3 column measurements. This model bias predominantly occurs in the lower stratosphere and is present throughout the year albeit with seasonal variations: the bias is larger during local winter compared to local summer. The particular spatial and seasonal variations of the model bias suggests a too fast meridional stratospheric transport in TM5, which agrees with earlier found shortcomings of using winds from data assimilation systems (DAS winds).
The model results are very sensitive to the data assimilation method in the numerical weather prediction (NWP) that provides the model wind fields. A large reduction (up to 50 % of the bias) in modeled lower stratospheric mid-latitude O3 was found when winds from four-dimensional (4DVAR) instead of three-dimensional (3DVAR) data assimilation were used. Previous work has shown that using different forecast periods was important for improving the age-of-air. Model results differed with different forecast periods (up to three days), although the effect was mainly confined to high latitude lower stratospheric O3. Apparently using different forecast periods is more important for age-of-air calculations than for stratospheric O3 calculations.
A positive bias in the extra-tropical lower stratosphere of about 20% remained, possibly related to the lack of heterogeneous polar stratospheric O3 destruction in TM5.