Deconstructing the flow duration curve with stream geochemistry in a forested headwater catchment (Luxembourg)

Abstract:

Flow duration curves relate streamflow magnitude to frequency and have been used to relate physical basin characteristics to hydrographic response, regionalize basin behavior, classify catchment response and assess inter-annual variability. The flow duration curve (FDC) has been applied extensively for water planning in hydropower generation, irrigation and stream pollution management. While much has been done relating hydrologic process and physical basin characteristics to FDC shape, little research has focused on how the FDC relates to stream geochemical response - the integrative catchment signal seen as a key measure of catchment function. Here we apply stream geochemical data to deconstruct the FDC and assess our current perceptual model of catchment behavior.

The flow duration curve is composed of three distinct regions, a high-flow tail, mid-flow slope, and low-flow tail, see figure 1. Rainfall runoff models have been used to relate physical basin and climactic factors to the corresponding catchment FDC - largely dividing the FDC into these component parts representing the interplay of precipitation (high flow tail) and evapotranspiration (low flow tail), which are separated by subsurface drainage (middle FDC region). The shape of the upper tail of the FDC essentially reflects the most intense rainfall events in the data record, when discharge is highest. The mid-slope region of the FDC reflects subsurface flow processes. Decreasing soil depth, and hence storage, or increasing soil conductivity, within the rainfall-runoff model, resulted in shorter mid-slope regions that gave way more quickly to the lower evapotranspiration tail.

While physical catchment characteristics - like geology, soil depth, slope and vegetation cover - have been linked to FDC shape, little research has focused on relating stream geochemical response to catchment FDC. One study used the slope of FDC to compare hydrographic response of lake catchments with stream outlet chemistry. An overall slope term, the ratio of discharge at 0.50 to 0.95 exceedance probability, essentially a ratio of baseflow to high flow and a rough index of hydrograph response variability, was positively correlated with the sum of base cation export. To our knowledge though, no studies have specifically looked at stream geochemical response, a key indicator of catchment function, throughout the FDC.

Here we seek to directly relate stream geochemical response to the catchment flow duration curve and test our perceptual model of catchment function. The results presented here are an extension of the comparisons amongst catchment base cation export and FDC shape, and augment the results from others who employed models to test how hydrologic processes correlated with regions of the FDC. One major limitation of much of previous research relating modeled basin characteristics to the FDC has been the lack of observational data available for testing hypotheses. Here we leverage existing stream discharge and chemistry data to ask if patterns in the FDC indicate threshold-like shifts among contributions from streamflow sources? We hypothesize that shifts in the FDC slope correlate with activation/deactivation of contribution from distinct streamflow sources. We test this hypothesis by looking for geochemical signatures during certain regions of the FDC, before and after any apparent changes in FDC slope.

We assessed the stream geochemical response within the framework of the catchment FDC by constructing a FDC based on multiple years of data to incorporate annual variability of the precipitation and subsequent catchment response. The period-of-record flow duration curve for the 47 ha study catchment (Weierbach, Luxembourg) is shown in figure 1. The FDC shows a distinct shift between the low-flow and high-flow tails, presumably where the dominant FDC-shaping process shifts between evapotranspiration and rainfall intensity. This lends a somewhat convex shape to the mid-slope region of the FDC.

Stream water samples were taken from the Weierbach bi-weekly between 2007 and 2012 and analyzed for conductivity as well as the concentration of major anions/cations (Ca, Mg, Na, K, Cl, NO₃, SO₄and Si). Most chemical species show a general decrease in concentration with increasing flow, though anions like nitrate and sulfate increase in concentration with increasing flow. Additional samples from groundwater wells and suction-cup lysimeters throughout the Weierbach suggest that stream chemistry reflects mixing from groundwater and soil water sources. Though both exhibit high variability, groundwater base cation concentrations tend to be higher than the stream and soil water base cation concentrations tend to be lower, figure 1.

The relationship between cation concentration and FDC probability appears nonlinear, with slope breakpoints appearing to coincide with the convex-shaped mid-slope of the FDC. Moreover, cation concentrations appear to stabilize at flows greater than the 0.40 probability of the FDC. We applied regression analysis across the whole of the FDC as well as across the two regions divided by the mid-point (~0.40) of the FDC. Regression analysis supported our visual interpretation that indeed, the slope of concentration of cations is different between the two regions of the FDC.

This stabilization in stream geochemical response with within the probability domain of the FDC potentially indicates a threshold when a maximum number of flowpaths and sources are actively contributing to the stream to culminate in a geochemically averaged signal. High concentration sources (groundwater) continuously mix with low concentration sources (soil water) to contribute to the observed geochemical signal at the catchment outlet. A likely area where this mixing occurs is within the near-stream saturated zone. Infrared imagery taken throughout a small headwater reach of the catchment was analyzed to determine if the expansion and contraction within the saturation extent of the near-stream zone correlated with the FDC or changes in the slope of geochemical response. Saturation extent increased non-linearly with increasing flow and appears strongly mitigated by antecedent wetness conditions.