| Summary: | Sungai Beranang is one of the main reaches of the Langat River, and one of the main sources
of water supply in Selangor. It is subjected to pressure due to rapid urbanization and other
anthropogenic activities within its proximity areas such as agricultural, residential, recreational
and industrialization. All of these activities have altered the land use pattern and deteriorated
the water quality of the Sungai Beranang. This study was conducted to assess the spatial
temporal variation of water quality in Sungai Beranang, identity the possible water pollution
source and develop Multivariate Water Quality Index (MWQI). Water quality of Sungai
Beranang (physical and biological parameters) was mea ured based on the National Water
Quality Standards (NWQS). Multivariate statistical techniques, such as Hierarchical
Agglomerative Clustering Analysis (HACA), Principal Component Analysis (PCA),
Discriminant Analysis (DA), Multiple Linear Regre sion (MLR), and Artificial Neural
Networks (ANN) were used to assess the spatial/temporal differences in the water quality of
Sungai Beranang covering a one-year period of September (2017) to November (2018) for the
wet and the dry seasons at nine different locations. Multivariate water quality index (MWQI)
using the varimax rotation fP A was developed to categorize the water pollution levels, which
are low, moderate and high. The analy i output clearly indicated that the Sungai Beranang
Basin was lassified under Class III of Malaysia river water quality standard as resulted from
multi-point sources of pollution. The Water Quality lndex WQI trend analysis revealed that the
upstream water quality of ungai Beranang was relatively good. HACA yielded three different
clusters according to the level of pollution; which are High Pollution Area (HPA), Moderate
P llution Area (MPA) and Low Pollution Area (LPA). The output of DA backward mode
revealed that the source apportionment discrimination was achieved with the classificatio
matrix accuracy of87.74 % with pH, Salinity, COD, TSS, NH3N, N02, N03, S04, EC, Cd, g"
Zn and Cr. Those significant parameters ha a high variation of spatial distribution. While,
variation of temporal distribution was achieved with the classification matrix accuracy tlf94.44
% with Temperature, Salinity TSS, NH3-N, Turbidity, EC, O&G, E. coli, e, Zn, Pb, Cu. PCA
varimax factors (VFs) are responsible for 87.15%, 80.05% and 77.78% of the total variance for
LPA, MPA and HPA, respectively. Furthermore, five pollutants (organic, nutrient, chemical,
mineral and natural) were the factors aggravating the quality of Sungai Beranang. ANN
prediction model was to predict the most significant variables that contaminated Sungai
Beranang. MLP-FF-ANN Model B was the best prediction model with the R2 and RMSE values
of 0.969 and 1.33, respectively. MWQI yielded three categories of water quality pollution level
which are low, moderate and high. DA-MLR model gave a good accuracy model performance
for forecasting the WQ1. The R2 value was 0.90 and the model exhibited 90% variability of
WQI. The multivariate techniques in this study was the most effective and reliable in solving
the source apportionment of pollution in Sungai Beranang with time and cost savings. The
policy maker and relevant government agencies should adopt this approach for monitoring
program of the Sungai Beranang as part of the water resources management.
|
|---|