Water quality pattern recognition of the Muda River Basin, Malaysia

Muda River Basin (MRB) in the state of Kedah, Malaysia with the range values of Water Quality Index (WQI) were between 55.8 and 91.0 during the period 1998-2013. The wide variations in water quality (WQ) indicate that the MRB is affected by various sources. Agricultural-related cover approximately 55% of the total area of the MRB. Meanwhile, about 35% of the catchment area is still covered by forests. Therefore, agriculture, logging activities, and agro-based industries are the main human activities in the area which contributes to water pollution. While, the presence of point sources such as factories may cause river water quality at the same level degraded differently in the same catchment. Therefore, there is a growing interest to determine the sources of pollution and land use classes responsible for this deteriorated WQ and determining how the desired WQ can be secured on a sustainable basis. The overall aim of this study is to recognise the pattern of water quality across different land classes within the Muda River system The objectives of this study were to (i) elucidate the water quality pattern that best represent the water quality variation of the Muda River Basin; (ii) determine the impacts of various land use on the water quality of Muda River Basin; (iii) develop the artificial neural network model for the prediction of spatial clusters and water quality status among the nine monitoring stations of the Muda River Basin; and (iv) forecast the status of water quality for the year 2020 for each monitoring station in Muda River Basin. The study employed secondary water quality and land use data. The land use data were put into four land use categories; agricultural, forestry, urban areas, and others. Investigation of the impacts of various land use on the WQ were carried out at four buffer zones(500 m, 1000 m, 1500 m, and 2000 m) as well as the whole river basin. The data was processed using multivariate analysis, artificial neural network (ANN) technique, and geographic information system (GIS). The study results elucidated that four sources of contamination in the MRB were organic pollution (contaminants that can be biodegraded by microorganisms), turbidity factor (include high flow rates, soil erosion, urban runoff), agricultural runoff (the portion of rainfall that runs over agricultural land) and natural factor (include geology, soil types, topography, precipitation intensity). Station MD02, which was adjacent to the rubber factory, should be the focus of remediation efforts. The monitoring stations that best represent the water quality variation of the MRB were MD02, MD03, MD05, and MD09.Hence, there is potential in improving the efficiency of the monitoring network in MRB. Subsequently, the study found that the land use types only had a minor effect on WQ in MRB. The ANN models developed to predict the spatial clusters, the WQI, and the water quality class (WQC). The results suggest that the ANN models can be used by environmental planners and decision makers for WQ management purposes. Moreover, ANN models were utilised to forecast the status of WQ for the year 2020. The predicted WQ remained in class II except for stations on the Jerung River (MD02 and MD03), which was in class III. The predicted value of WQI for the year 2020 assumed that no other land use types will take place in the river basin from 2007 to 2020. The findings of this study provide information to authorities responsible for river basin management to address the issue of WQ deterioration. These efforts can help to make plans in advance to guarantee the continuity of good water quality for generations to come