Haze episodes: identification of air pollutants and meteorological factors in Borneo, Central, Eastern, Northern, and Southern regions of Malaysia

• Published in: Desalination and water treatment Vol. 219; pp. 31 - 39
• Main Authors: Yusof, K.M.K.K., Jalal, K.C.A., Abdullah, S., Azid, A., Sani, M.S.A., Yunus, K., Isa, N.M., Jamalani, M.A., Zulkifli, Z.R., Rani, N.L.A.
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01-04-2021
• Subjects: Artificial neural network; Generalized linear model; Haze episode; Principal component regression; Sensitivity analysis; Principal component regression; Generalized linear model; Haze episode; Sensitivity analysis; Artificial neural network
• ISSN: 1944-3986
• DOI: 10.5004/dwt.2021.26780

This research was based on selected daily haze occurrences between the year 2006 until 2015 in five regions of Malaysia (Borneo, Central, Eastern, Northern, and Southern regions). The generalized linear model (GLM), principal component regression (PCR), incorporation of artificial neural network and sensitivity analysis (ANN-SA) techniques were applied in this study to generate respective models namely as MLP-HM-GLM, MLP-HM-PCR, and MLP-HM-LO and identify the relationship of air pollutants and meteorological factors to particulate matter (PM10) variability. The performances of these models were compared based on coefficient of determination (R2), root-mean-square error (RMSE), and squared-sum error (SSE). From the findings, ANN-SA that generated the MLP-HM-LO model was the most suitable technique to identify the main contributors to PM10 variability. Ultraviolet-b (UVb) and sulphur dioxide (SO2) were found as the most significant pollutants that affected the PM10 variation. UVb also had consistently influenced PM10 variability over five regions. MLP-HM-LO model had rendered the highest R2, with the lowest RMSE and SSE values compared with MLP-HM-GLM and MLP-HM-PCR models. Thus, the ANN-SA technique was highly practicable in determining future haze circumstances in Malaysia.