Deciphering catchment filtering effects on export regimes and temporal memory of river water quantity and quality

Water from precipitation and constituents from multiple sources are integrated along river networks and exported at catchment outlets after undergoing hydrological and biogeochemical processes. Accordingly, stream water quantity and quality exhibit emergent behaviors rather than reflecting a simple sum of inputs, typically expressed as power-law relationships between discharge (Q) and concentration (C) which distinguish water quality parameter’s export behavior (C–Q pattern), and as temporal persistence (memory) in riverine output. How catchments generate these characteristics has long been of interest. Previous studies have shown that source availability governs which C–Q pattern emerges, and that catchment filtering of stochastic inputs can generate legacy sources that produce memory. These findings naturally raise core follow-up questions: (1) Even when the same C–Q pattern emerges, what controls variations in its intensity?; (2) Does catchment filtering invariably lead to the formation of legacy sources?; and (3) given that both C–Q patterns and memory arise from catchment filtering, do they share common underlying physical drivers? To address these questions, we aim to examine how catchment characteristics affect C–Q patterns and temporal memory formation, and whether systematic linkages exist between them. Across 24 catchments in South Korea, we analyze daily precipitation and discharge data along with 8 water quality parameters (pH, EC, DO, TOC, TN, TP, Turbidity, and Chlorophyll-a). To investigate catchment spatial characteristics that account for source availability and existence of legacy sources, we employ 14 explanatory variables representing catchment geography, land use, social indicators, and water-use characteristics. The C–Q pattern analyses report that, for some nutrients, the pattern in which concentrations increase with discharge due to unlimited non-point sources is consistent across all catchments, but the increasing magnitude gets sharper in natural catchments. This indicates that the relative contributions of point and non-point sources act as a key factor that shapes C–Q patterns. Temporal memory is examined using power spectrum analysis. Our results show that under natural cover conditions - where legacy sources are more likely to form – river discharge tends to exhibit long-term memory; however, for some ionic parameters, short-term memory gets strengthened. This suggests that persistent input signals from anthropogenic sources are disrupted during the catchment filtering process. Finally, for discharge-flushed water quality parameters, we identify a decreasing power-law relationship between C–Q patterns and memory, suggesting that surface-runoff-dominated export of water quality parameters represents a stochastic input signal that is not buffered through soils, and therefore its intrinsically memoryless signal is preserved. These findings corroborate a process-based understanding of catchment filtering mechanisms and provide a scientific basis for integrated monitoring and management of water quantity and water quality at the catchment scale.

Acknowledgements

This work was supported by the Creative-Pioneering Researchers Program through Seoul National University and by the National Research Foundation of Korea (NRF) grant funded by the Korean government (Ministry of Science and ICT) (No. RS-2025-00523350).