Statistical modeling approaches for PM_(10) forecasting at industrial areas of Malaysia

摘要

Major industrial areas in Malaysia experience number of unhealthy days because of extreme impermanent PM_(10) incidents which are detrimental to human and the environment. In order to lessen the threat of acute air pollutant levels, a short-term forecasting algorithms is needed to advise people in general of unsafe air pollution happenings, and additionally to formulate air pollution management strategies. In this approach, statistical models consisting of MLR and PCR are employed for PM predictions at 4 major industrial areas of Seberang Prai, Pasir Gudang, Kemaman and Nilai in Peninsular Malaysia. Gaseous pollutants, meteorological factors and 8 years data of daily PM_(10) form 2007 till 2014 was applied to predict PM_(10) concentration levels. Results showed that MLR performed better than PCR and the major primary sources are road traffic and industrial emissions whilst wind speed display inversely proportional relationship with the PM_(10) concentrations.