Clustering Pollutants Concentration from Sumatra Peat Fire using Fuzzy C-Means Algorithm

Abstract

Indonesia has large peatland areas which can significantly impact both regional and global environments. The forest and land fire in 2015 produced high level of air pollutants concentration in Sumatra, especially in South Sumatra, Riau, and Jambi provinces. This study aims to cluster pollutants concentration from peat fires in Sumatra during July until November 2015 using fuzzy c-means algorithm. The dataset used in this study is concentration of pollutants including CH4, CO, CO2, NH3 and PM2.5. Based on the experimental result, pollutants concentration data are grouped into 8 clusters. The result of this study indicates that the highest pollutants concentration is found in the cluster which has only one member with concentration 14,500 μg/m3 of CH4, 145,980 μg/m3 of CO, 1,183,831 μg/m3 of CO2, 13,800 μg/m3 of NH3 and 6,290 μg/m3 of PM2.5. Another cluster containing high concentration has 6 objects representing haze locations with the average concentration 8,265 μg/m3 of CH4, 83,427 μg/m3 of CO, 676,558 μg/m3 of CO2, 7,913 μg/m3 of NH3, and 3,597 μg/m3 of PM2.5. The largest cluster with 590,816 haze locations contains the lowest concentration of pollutants. Although its average concentration levels are low, yet the maximum concentration are still in moderate or high enough levels. In general, high concentration of pollutants was distributed mostly in South Sumatra, Jambi, and Riau.