Volume 22, Issue 137 (11-2015)                   RJMS 2015, 22(137): 31-43 | Back to browse issues page

XML Persian Abstract Print

PhD, Professor of Environmental Health Engineering, Department of Environmental Health Engineering, Faculty of Health, Tehran University of Medical Sciences, Tehran, Iran , mesdaghinia@sina.tums.ac.ir
Abstract:   (5579 Views)

Background: Air Quality Index (AQI) quantifies the relationship between air quality and the level of health. The value of AQI may be predicted using neural network model for a day in advance, based on the meteorological variables and autocorrelation behavior of the index in Kermanshah, a city in western Iran.

Methods: Data for air pollution and meteorological variables, collected during three years, were lagged   for two proceeding days. The AQI for a next day was considered as dependent variable and other were used as predictors. The performance of model was assessed with correlation coefficient (r). The most important variables to predict the AQI were identified using sensitivity analysis.

Results: The r coefficient for the training, validation and testing the model was 0.75. Among the meteorological variables, the horizontal view and precipitation had a greater impact on the AQI. One day proceeding precipitation can significantly reduce the amount of AQI for the next day. An inverse relationship was found between AQI and horizontal view.

Conclusion: The proposed model can be used to predict the Kermanshah’s AQI index. Regarding to the issue of air pollution in this city, especially fine particulate pollutions if such a model is used dynamically to predict the AQI, it can be useful tools for the declaration of an air pollution alert. The preparation of an online model-based system for the prediction of AQI index for Kermanshah city is suggested to be conducted in future studies.

Full-Text [PDF 1427 kb]   (2845 Downloads)    
Type of Study: Research | Subject: Neonatology

Rights and permissions
Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.