variavel0=<b>Reginaldo Rosa Cotto De Paula</b> - rcotto@bol.com.br CEFET-ES
<b>Jane Méri Santos</b> - janemeri@npd.ufes.br UFES-ES
<b>Valdério Anselmo Reisen</b> - valderio@cce.ufes.br UFES-ES
<b>Abstract.</b> This paper studies the application of time series modelling to the description of turbulent concentration fluctuation of atmospheric pollutants around an isolated building under neutral and unstable atmospheric conditions. The "instantaneous" concentration data was obtained from field experiments carried out by Santos (2000). A rapid response photo-ionisation detector was used to obtain high frequency data (50Hz). The time series of concentration on the windward wall showed episodes of zero concentration, which is probably related to the larger turbulent eddies than the plume width reaching the obstacle. The thermal stratification has influenced the concentration fluctuation intensity on the leeward wall. The concentration fluctuation intensity is about 50% higher under unstable conditions. The stochastic quality of the models indicates that the results obtained using the Box- Jenkins methodology satisfy the main criteria of the verification process. The high variability of variance related to the original time series measured on the windward wall is probably the reason for the failing of the ARMA (p,q) model on the Ljung-Box test, however, the model forecast of concentration mean and concentration fluctuation intensity tend to the same values of these statistical moments of the original series. Thus, the results indicate that the time series analysis may be an adequate tool to represent the atmospheric dispersion of pollutant around buildings.
<b>Keywords.</b> concentration fluctuations, atmospheric turbulence, time series.