Research papersConcentration-discharge patterns in a small urban headwater stream in a seasonally dry water-limited tropical environment

•Concentration (C)-discharge (Q) patterns were studied in an urban headwater stream.•'Broken stick' or threshold behaviours suggest three biogeochemical hypotheses:(1)Initial dilution then discharge-invariant C (negative to 0 slope linear curves).(2)Initial dilution then enrichment/flushing (negative to positive linear curves).(3)Initial solute flushing then dilution (positive to negative linear curves).

The simplicity of the hydrochemical stationarity concept renders it attractive for partitioning solutes between geogenic and anthropogenic sources. The current study used a small urban headwater stream in a seasonally dry environment to address two research questions: (1) What concentration (C)-discharge (Q) patterns exist in small urban headwater streams?; and (2) Do the C-Q patterns persist across C-Q metrics and temporal scales? Four C-Q metrics were tested: concentration-discharge (C-Q), concentration-cumulative discharge (C-ΣQ), load (L)-discharge (L-Q) and normalized concentration-normalized discharge (NC-NQ). C-Q and NC-NQ revealed discharge-invariant behaviour for Ca, two linear relationships with threshold-like transitions from negative to positive slopes for Mg, K and Na, and positive linear relationships for Fe, Pb and PO43−. The threshold-like transitions with distinct breakpoints were more apparent in C-ΣQ patterns for all solutes. These patterns are consistent with three hypotheses: (1) negative linear to zero slope relationships indicate dilution followed by discharge-invariant behaviour (Ca); (2) negative to positive linear relationships (Mg, K and Na) point to dilution followed by solute enrichment or flushing; and (3) positive to negative linear relationships (Pb, Fe and PO43−) suggest initial solute mobilization followed by dilution. The three dominant behaviours were robust across weekly, fortnightly and monthly timescales. Significant linear L-Q relationships were observed for all solutes, suggesting that loads can be predicted from discharge. Our findings suggest that C-Q relationships are highly dynamic, and multiple processes control streamflow hydrochemistry at different times depending on antecedent discharge. The application of multiple C-Q metrics provided additional insights not apparent by using a single metric. The insights are critical to understanding of catchment hydrology and conceptual representation of hydrochemical processes in models.

Graphical abstractDownload high-res image (182KB)Download full-size image