Abstract
It is widely agreed that the exposure in air pollution causes serious respiratory illnesses and that the weather may also contribute to the seriousness. However, it is difficult to quantify the effects of pollutants and weather conditions due to the high unknown nonlinearities in the impact of these environmental and weather factors on the illnesses and possible interactions amongst these factors. In this paper, the multiple index model is employed to explore such relationship in Hong Kong. Time lags are incorporated to address the incubative period from the impact to the symptoms. We found that the weather factors are important in triggering the illnesses. However, the similarity in the number of illnesses and weather pattern causes the seriousness of ambient air pollutants being distorted. We restrict the time periods in winter and summer in which the weather is relatively stable. We then clearly identified the importance of air pollutants to respiratory illnesses. The results also revealed the difference between sub‐regions. By comparing with linear and several nonlinear models, we understood that the nonlinear relationship is supportive. We also found that our model is better than the single index model, indicating that further fine‐tuning is necessary to achieve better forecasting results. Copyright © 2009 John Wiley & Sons, Ltd.