Multivariate Statistical Techniques to Identify the Source of Pollution and Assessment of Surface Water Quality
Principal component analysis (CPA) and multiple liner regression (MLR) analysis was applied on the data for 14 physico-chemical parameters of surface waters from Tunggak River adjacent to the Gebeng Industrial Estate, Pahang, Malaysia during February 2012 – January 2013 with the objective of identif...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
National Institute of Ecology
2013
|
| Subjects: | |
| Online Access: | http://umpir.ump.edu.my/id/eprint/11998/ http://umpir.ump.edu.my/id/eprint/11998/ http://umpir.ump.edu.my/id/eprint/11998/1/Multivariate%20Statistical%20Techniques%20to%20Identify%20the%20Source%20of%20Pollution%20and%20Assessment%20of%20Surface%20Water%20Quality.PDF |
| Summary: | Principal component analysis (CPA) and multiple liner regression (MLR) analysis was applied on the data for 14 physico-chemical parameters of surface waters from Tunggak River adjacent to the Gebeng Industrial Estate, Pahang, Malaysia during February 2012 – January 2013 with the objective of identifying sources of pollution and their contribution to the variation in water quality. Physico-chemical parameters were determined for a period of 12 months by following standard methods of analysis. Results reveled that most of the parameters including BOD, COD, conductivity, NH4-N and phosphorus were in concentrations greater than the national standard of Malaysia. PCA was applied to identify the source and MLR analysis was done to determine their contribution. PCA yielded five VFs; which extracted 74.72% of total variance that established its validation. Results showed that, surface water quality was strongly influenced by ionic groups of salts, soil erosion and agricultural runoff, organic and nutrient pollutions from domestic wastewater, industrial sewage and wastewater treatment plants. Vicinity of industrial parks resulted in low DO concentration all over the basin. MLR showed the contribution of every variable to be highly significant (p<0.01). |
|---|