Verification of a prognostic meteorological and air pollution model for year-long predictions in the Kwinana industrial region of Western Australia

A prognostic air pollution model (TAPM) has been used to predict meteorolog y and sulphur dioxide concentration in the Kwinana industrial region of Wes tern Australia for 1997, with a view to verifying TAPM for use in environme ntal impact assessments and associated air pollution studies. The regulator y plume model, DISPMOD, developed for the Kwinana region has also been run using both an observationally based meteorological file (denoted DISPMOD-O) and using a TAPM-based meteorological file (denoted DISPMOD-T). TAPM predi ctions of the meteorology for 1997 compare well with the observed values at each of the five monitoring sites. Root mean square error and index of agr eement values for temperature and winds indicate that TAPM performs well at predicting the meteorology, compared to the performance of similar models from other studies. The yearly average, 99.9 percentile, maximum and mean o f the top 10 ground-level sulphur dioxide concentrations for 1997 were pred icted well by all of the model runs, although DISPMOD-O and DISPMOD-T tende d to overpredict extreme statistics at sites furthest from the sources. Ove rall, TAPM performed better than DISPMOD-O, which in turn performed better than DISPMOD-T, for all statistics considered, but we consider that all thr ee sets of results are sufficiently accurate for regulatory applications. T he mean of the top ten concentrations is generally considered to be a robus t performance statistic for air pollution applications, and we show that co mpared to the site-averaged observed value of 95 mug m(-3), TAPM predicted 94 mug m(-3), DISPMOD-O predicted 111 mug m(-3) and DISPMOD-T predicted 125 mug m(-3). The results indicate that the prognostic meteorological and air pollution approach to regulatory modelling used by TAPM, gives comparable or better results than the current regulatory approach used in the Kwinana region (DISPMOD), and also indicates that the approach of using a currently accepted regulatory model with a prognostically determined meteorology fil e will be an acceptable approach for regions without meteorological observa tions. There is nothing in TAPM that was developed specifically for the Kwi nana region, and so we feel that TAPM could be applied equally well to othe r regions, with similar performance. The disadvantage of TAPM is the approx imately 1000 times longer CPU time required to run the model compared to mo dels such as DISPMOD, and so we anticipate that the two approaches will com plement each. other for regulatory applications for the near future. (C) 20 01 Elsevier Science Ltd. All rights reserved.