German-wide analysis of high-frequency, continuous concentration-discharge relationships

High-frequency observations of solute and particulate concentrations (C) and associated discharge (Q) measurements allow the analysis of C-Q relationships on the event-scale. Together with catchment attributes and event properties, C-Q relationships can be used to disentangle the fine-grained dependencies between catchment properties, hydrologic processes and water quality.

We collected 72 high-frequency (sub-hourly), continuous C and Q time series across Germany in catchments ranging from 8 to 122.000km² (median of 6.051km²). Besides discharge at all 72 locations, the database contains water temperature (60 locations), turbidity (34), conductivity (59), oxygen (57), pH (53), ammonia (16), nitrate (22), phosphate (10) and chlorophyll (12) in varying lengths over a maximum period of 20 years. Event filters were used to extract single discharge events. Hysteresis indices and classes of each event were used to describe the C-Q relationships. Event properties such as season, magnitude, variability, length, antecedent conditions, rise- and fall characteristics and physical catchment characteristics were assigned to each C-Q relationship.

We used a Random Forest model to explain the hysteresis properties based on the event- and catchment characteristics. Particularly the variables turbidity and conductivity revealed spatio-temporal dependencies, which we relate to interplays of land use and soil characteristics. We further found that the C-Q patterns are significantly impacted by the identifiability and definition of the extracted discharge events across the heterogeneous catchments as well as the resolution and quality of the measurements of the different C variables.