Impact of meteorological analyses and chemical data assimilation on modelled long-term changes in stratospheric NO2
1Institute for Climate and Atmospheric Science, School of Earth and Environment, University of Leeds, Leeds, UK
2Belgian Institute for Space Aeronomy, Brussels, Belgium
3Instituto Nacional de Técnica Aeroespacial, Spain
4NIWA, Lauder, New Zealand
*now at: Earth and Planetary Sciences, UC Berkeley, California, USA
**now at: 1981 Omakau-Chatto Creek Road, Alexandra, New Zealand
Abstract. We have used a three-dimensional (3-D) off-line chemical transport model (CTM) to investigate long-term changes in stratospheric NO2. The basic model was integrated from 1977 to 2001 using ECMWF (European Centre for Medium-Range Weather Forcasts) ERA-40 reanalyses. Additional model runs were performed which assimilated HALOE observations of long-lived tracers to constrain the model trace gas distributions. Assimilation of a single long-lived species (CH4) improves not only the distribution of all other long-lived species, via tracer-tracer correlations, but also shorter lived radical and reservoir species. Assimilation of the long-lived species corrects for errors in the model, due to horizontal transport from the ERA-40 reanalyses, and allows a more direct test of the model's chemistry.
The basic model significantly underestimates the observed column NO2 from mid-latitude ground-based sites in the mid-late 1990s. The mean underestimate is ~ 26% for summertime values between 1992 and 1998. Moreover, as the model agreement is better in the early 1990s, it underestimates the increasing trend throughout the decade. However, when the model assimilates HALOE CH4 data both comparisons are greatly improved. The mean model-observation difference reduces to 8% for summertime values and the trend improves. This indicates that given realistic wind fields to constrain the tracer transport, the model chemistry and aerosol schemes are able to reproduce the observed trends in NO2. Implications of this for using analysed wind fields to determine dynamical ozone trends are discussed. Ozone trends derived directly from transport models forced by analysed winds are likely subject to similar errors.