Seasonal Shifts in Correlation Structure of Stream Water Quality Following Artificial Cloud Seeding

This study investigates the influence of wintertime artificial cloud seeding on the correlation structure among water quality parameters in a mountainous watershed. Two representative periods with contrasting hydrological and physicochemical conditions―March (immediately after seeding) and October (non-seeding season)―were compared. Using long-term monitoring data from the Odaecheon-1 site collected between 2013 and 2023, Pearson correlation coefficients were calculated for 18 water quality variables, and the inter-month differences (Δr) were used to assess seasonal structural changes. In March, strong positive correlations were observed among organic matter indicators, such as COD-TOC (r = 0.84) and BOD-TOC (r = 0.83), suggesting that snowmelt during the post-seeding period contributed to elevated organic loads. In contrast, October showed marked negative correlations between organic matter and nitrogen-related parameters, including COD-TN (r = -0.85) and COD-DTN (r = -0.86), which likely reflect the influence of nonpoint source pollution from decaying vegetation or agricultural runoff. The TP-NH₄⁺-N pair showed higher correlation in March (Δr = 1.17), implying the simultaneous influx of ionic nutrients following seeding-induced snowfall and meltwater transport. On the other hand, DTP-Fecal Coliform (Δr = -0.73) and PO₄^3--P-Fecal Coliform (Δr = -0.92) correlations were more pronounced in October, indicating the co-transport of pathogens and phosphate in non-seeding periods. This study quantitatively reveals that specific parameter pairs can reflect seasonal hydrological inputs and the potential influence of artificial precipitation experiments. Multivariate correlation analysis is shown to be a useful approach for evaluating the environmental impacts of cloud seeding operations.