Impact of different ozone sounding networks on a 4D-Var stratospheric data assimilation system

Observing system simulation experiments (OSSEs) are performed to examine the impact of ozonesonde observations on a data assimilation system during a simulated satellite data gap during February 2003. Using the four-dimensional variational chemical data assimilation system SACADA, the relative influence of launch rates and station coverage is investigated. Starting with the VINTERSOL/MATCH campaign, different network and sounding configurations are evaluated. To quantify the performance of assimilation experiments in the lower stratosphere, analysis skill and linear pattern correlation with respect to ERA-Interim reference data are assessed for the 20 km altitude level. Using first-guess and analysis minus observation error statistics, a priori error settings are tuned to optimise the assimilation of simulated and real-world ozone soundings. In summary, it is found that, during satellite data gaps, ozonesonde data can have a significant positive impact on the mean analysis skill depending both on the number of observations and the network layout. A better distributed network based on the GAW system, with 28 stations and three soundings bi-weekly, proves clearly superior to VINTERSOL/MATCH, showing a positive gain in skill of 0.26 compared to a free-running model.