Quantifying Haze Effect using Air Pollution Index Data

Abstract

Malaysia has been misfortunate with intermittent haze episodes since 1997 which affect the airquality tremendously.In Malaysia, an instrument named as air pollution index (API) is utilizedin determining the quality of air, which is influenced by the presence of haze. API values arecalculated by considering the concentration of harmful particles in haze. So, any haze episodeheavily affects the API values and can be considered as a determining factor. Since Malaysiais prone to haze, it is crucial to identify and quantify the haze effect on the API values.Therefore, four models– an autoregressive integrated moving average (ARIMA), regressionmodel with ARIMA errors (ARIMAX), time series regression and Prophet models areemployed. It is found that ARIMAX (4,0,1) with non-zero mean is the best model in describingthe API data with presence of haze as external regressor based on the smallest adequacy anderror measures for training and test datasets. In conclusion, the effect of haze is significant indescribing the API values and thus, proper health management is required during haze episodes.