Three-dimensional variational data assimilation of ozone and fine particulate matter observations: some results using the Weather Research and Forecasting - Chemistry model and Grid-point Statistical Interpolation

In operational air-quality forecasting, initial concentrations of chemical species are often obtained using previous-day forecasts with limited or no account for the observations. In this article we assess the role that assimilation of surface measurements of ozone and fine aerosols can play in improving the skill of air-quality forecasts. An assimilation experiment is performed using the Weather Research and Forecasting - Chemistry model and Grid-point Statistical Interpolation, a three-dimensional variational assimilation tool. The modelling domain covers the northeastern region of North America. The measurements come from the United States Environmental Protection Agency AIRNow network and are available hourly. Background error covariance statistics are derived from forecasts in July 2004. Comparison of forecasts issued in August and September 2006 and initialized with and without the assimilation follows. Results show that forecasts of ozone and fine aerosol concentrations benefit from the assimilation in terms of standard verification scores for a period of at least 24 hours. However, significant reduction of errors as a consequence of the assimilation is accompanied by fast model error growth in the early forecast hours. Published in 2010 by John Wiley & Sons, Ltd.