Evaluation of water quality using pattern recognition technique in Melaka River Basin, Malaysia

The unorganised expansion and uncontrolled urbanisation development had led to environmental issues involved with river water pollution that occurred within the Melaka River basin. There are three objectives study to solve the water quality of the Melaka River basin through (1) to identify the potential sources of Melaka River pollution in correlation with the temporal land use classification, (2) to determine the land use land cover changes in 2015 and predict the future land use land cover for 2022, and (3) to verify pattern recognition based on the land use classification between 2015 and 2022 in the determination of pollutant sources within the Melaka River. The results indicated HCA have two cluster areas (C1 and C2). The DA indicated 12 variables were found to be the most significant parameters with a high variation in the spatial distribution. The PCA found that the C1 was attributed to contamination from the agricultural and residential activities; while the C2 included the pollution sources from the agricultural, residential, industrial, animal husbandry, and sewage treatment activities. In the Pearson correlation analysis and ANOVA analysis shown Vegetation Area (VA), Non-Industrial Area (NIA), Industrial Area (IA), Open Space Area (OSA), and Farming Area (FA) were correlated with majority of the physico-chemical and biological variables. LULC incorporated with CA-Markov chain model analysis showed NIA (67.50%), IA (29.15%) and FA (3.35%) were increased from 2001 to 2015 and 2022; while the VA (89.59%), OSA (3.32%) and WB (7.08%) continued to decrease. In GIS, the LISA analysis indicates 2015 showed the clustered area of C1 with S1 as the main contributor; while C2 of S5 as the main contributor. The Moran I analysis from high-to-low pollutant sources are NIA (0.80) > VA (0.59) > IA (0.42) > OSA (0.40). In 2022, the contamination from high-to-low are VA (0.59) > NIA (0.58) > IA (0.50) > OSA (0.41) in the Melaka River. The hotspot analysis for year 2015 indicated C1 of S1 as the main contributor; while C2 of S5 as main contributor. General G-statistic analysis from high-to-low are IA (1.099 x 10-3) > NIA (5.39 x 10-4) > VA (6.9 x 10-5). In 2022, the hotspot analysis in C1 indicated S7 as the main contributor; while C2 remains the same as S5 which is the main contributor to have pollutant sources from high-to-low are IA (3.012 x 10-3) > NIA (5.95 x 10-4) > VA (7.6 x 10-5). In conclusion, Moran I and general G-statistic had gained support especially in benefiting from the PCA in recognising the pattern of pollutant sources (e.g. land use classes) with precise details. Simultaneously, LISA and hotspot analysis have an advantage over HCA by recognising the clustered area more specifically, which was based on the pattern of pollutant sources.

Preview Text FPAS 2018 12 - IR.pdf Download (778kB) | Preview

Actions (login required)

View Item