Derived from river monitoring data, concentration-discharge (C-Q) relationships are useful indicators of riverine export dynamics. A top-down synthesis of C-Q patterns was conducted for suspended sediment (SS), total phosphorus (TP), and total nitrogen (TN) for nine major tributaries (15 monitoring sites) to Chesapeake Bay, which represent diverse characteristics in terms of land use, physiography, and hydrological settings. Model coefficients from the recently-developed Weighted Regressions on Time, Discharge, and Season (WRTDS) method were used to make informative interpretation of C-Q relationships. Unlike many previous C-Q studies that focused on stormflow conditions, this approach allows simultaneous examination of various discharge conditions within an uncertainty framework. This synthesis on WRTDS coefficients (i.e., the sensitivity of concentration to discharge) has offered new insights on the complexity of watershed function. Results show that watershed export has been dominated by mobilization patterns for SS and TP (particulate-dominated species) and chemostasis patterns for TN (dissolved-dominated species) under many river discharge conditions. Among nine possible modalities of low-flow vs. high-flow patterns, the three most frequent modalities are mobilization vs. mobilization (17 cases), chemostasis vs. mobilization (13 cases), and chemostasis vs. chemostasis (7 cases), representing 82% of all 45 watershed-constituent pairs. The general lack of dilution patterns may suggest that none of these constituents has been supply-limited in these watersheds. For many watershed-constituent combinations, results show clear temporal non-stationarity in C-Q relationships under selected time-invariant discharges, reflecting major changes in dominant watershed sources due to anthropogenic actions. These results highlight the potential pitfalls of assuming fixed C-Q relationships in the record. Overall, this work demonstrates the utility of WRTDS model coefficients for interpretation of river water-quality data and for generation of sensible hypotheses on dominant processes in different watersheds. The approach is readily adaptable to other river systems, where long-term discretely-sampled data are available, to decipher complex interactions between hydrological and biogeochemical processes.
Read full abstract