Skip to Main Content
Analysis of the spatio-temporal patterns and source apportionment of water pollution is important for proper management and protection of water resources. In this paper, different multivariate statistical methods were used to explore the spatio-temporal patterns of water pollution, and quantitative relationships between the important pollution parameters and environmental variables. Eleven significant parameters measured in 22 monitoring sites were preprocessed spanning between 2001 and 2007. Results of the hierarchical clustering analysis (HCA) demonstrated that this method had high flexibility for efficient classification of the monitoring sites. Results of discriminant analysis (DA) revealed that a high number of parameters contributed in discrimination of classes in the spring and summer seasons, especially in the April and September months. Recorded data of river water temperature (RWT), runoff, and two products of the MODIS sensor including the monthly Normalized Difference Vegetation Index (NDVI) and land surface temperature (LST) for the period of 2002-2007 were used as the explanatory variables. Test of NDVI and LST was based on extraction of their average values in different buffers of 250 up to 1500 m around the streams. Monthly data of group 1, a group with the highest number of monitoring sites resulting from the clustering procedure, was used for the analysis. Rotated principal component analysis (rotated PCA) was used for exploration of the quantitative relationships between the pollution parameters and environmental variables. Absolute principal component score- multivariate linear regression (APCS-MLR) was applied to quantify the source contributions for each pollution parameter. Results showed that NDVI and runoff can be considered as the efficient indicators of the non-point pollution sources such as the agricultural activities and surface weathering. NDVI showed an important role in reduction of TDS and No3-. Multiple buffe- s of NDVI showed temporally variable relations with different pollution sources. LST showed high discrimination potentials for distinguishing pollutions related to biochemical activities. Although the tested environmental variables revealed some relationships with those of the water pollution sources, nevertheless for more detailed analysis of the water pollution problem, the role of other latent environmental variables should be taken into consideration.