Akademik Veri Yönetim Sistemi
XVI International Scientific Agriculture Symposium "Agrosym 2025", Sarajevo, Bosna-Hersek, 2 - 05 Ekim 2025, cilt.1, sa.1, ss.466, (Özet Bildiri)
This study evaluates the spatial variability of water quality and identifies pollutant sources in the Riva (Çayağzı) Stream watershed, located in Istanbul, Türkiye. Monthly water samples were collected from five strategically selected locations throughout 2024. In situ measurements included physicochemical parameters (pH, electrical conductivity, dissolved oxygen, water temperature, and stream flow velocity). At the same time, laboratory analyses included major cations, nutrients, and heavy metals (Pb, Cd, Cr, Cu, Ni, Zn). A multivariate statistical approach, comprising Principal Component Analysis (PCA), Hierarchical Cluster Analysis (HCA), and Pearson correlation, was employed to identify pollutant patterns and site similarities. These analyses were further integrated with land use classification derived from Landsat-8 imagery, using the Random Forest algorithm. PCA results revealed distinct pollution signatures: Öyümce was characterized by pronounced heavy metal contamination (Pb, Cd, Cr, Cu, Ni, Zn); Ömerli exhibited moderate metal levels; Riva was dominated by high ionic content (EC, Na, Mg) and suspended solids; Pasamandıra displayed intermediate pollution; and Ömerli Dam Outlet represented the least-impacted, reference site with high dissolved oxygen and minimal contaminants. HCA supported these spatial groupings, clustering sites by their dominant pollution profile. Land use assessments indicated significant urban and industrial pressure in the downstream Riva area, agricultural influence in Öyümce and Ömerli, and forest dominance in the dam’s upstream watershed. Strong correlations among pollutants suggested that urban runoff and agricultural activities were the primary drivers of contamination. These findings highlight the value of integrating statistical and remote sensing-based spatial analyses for source apportionment.