3.1 Hydrometric Monitoring and Tracer Experiment:
Stream levels in the North St Vrain River followed a seasonal pattern consistent with snow driven hydrographs of the Southern Rockies with rising streamflow starting in late April, peak flows in late May to early June and falling streamflow throughout the late-summer/early-fall months (June – September) (Figure 1a). Several summer convective storm events occurred in July and August but did not strongly influence the seasonal hydrograph (Figure 1a). Using stage and geochemical patterns at Inflow (Figure 1a), we categorized four distinct hydro-periods: (I) rising limb (May 01, 2018 – May 15, 2018); (II) peak flow (May 16, 2018 – June 18, 2018); (III) falling limb (June 19, 2018 – July 10, 2018); and (IV) recession (July 11, 2018 – Sept 30, 2018) (Figure 1).
Analysis of patterns between Inflow stage and relative stage (i.e., stage z-scores) at target sites indicated that target sites generally followed the broad seasonal pattern of streamflow at Inflow but also demonstrated distinct site-specific behavior (Figures 2 & 3). We used inflection points in source-target stage relationships to infer changes or thresholds in hydrologic connectivity between Inflow and floodplain sites. The stage at which inflection points (Istage) in source-target stage relationships occurred varied between sites and spanned a range of Inflow stages from 366 to 642 mm (Figures 2 & 3, Table S1). At the major channel sites, relative stage was strongly coherent with Inflow stage throughout the study period and inflection points in the source-target stage relationship represented only small changes in slope (0.001- 0.004, see Figure 2). At Outflow, the inflection point occurred at very high flow (Istage: 642 mm) and the slope change was very low (0.001) suggesting the potential for a false positive, possibly driven by a shift at the site to overbank flooding.
At side channel sites, relative stage generally followed patterns similar to Inflow stage with the exception of Side-01 that exhibited hysteretic behavior with higher stage relative to the Inflow during the rising limb compared to the falling limb and recession (Figure 3: panels Side-01, -02, -03 and -04). Inflection points were identified across a wide range of streamflows ranging from the lowest at Side-02 (Istage: 366 mm) to the highest at Side-03 (Istage: 597 mm).
Both of the surface connected ponds (Pond-Con-01 & Pond-Con-02) and the isolated pond (Pond-Iso) had high water levels starting in the rising limb that did not fluctuate strongly as a function of Inflow stage (slopes 0.001 to 0.003). At all three ponds, water levels declined rapidly relative to Inflow stage below the inflection point (see Figure 3). The inflection point occurred during the falling limb at Pond-Con-02 (Istage: 621 mm), and at much lower flows during the recession at Pond-Con-01 (Istage: 368 mm) and Pond-Iso (Istage: 398 mm) (Figure 2). Like Side-01, Pond-Con-02 exhibited hysteretic behavior with higher stage relative to the Inflow on the rising limbs than on the falling limb. The high water levels in pond sites during the rising limb suggest sampling began after ponds had mostly filled with groundwater, local snowmelt, and streamwater. Pond-Con-01 and Pond-Con-02 went dry in mid-September while at Pond-Iso, levels dropped below our water level logger in early September and the pond went completely dry in late September (Figure 3).
The hysteretic behavior observed at Side-01 and Pond-Con-02, which are connected to each other by a surface channel (Figure 1b), may be related to a failure of a beaver dam during peak flows. While we did not identify the specific failed dam, such failures are common in beaver impacted systems and change the thresholds in river stage at which floodplain features have surface connections (Westbrook et al., 2011).
The tracer injection experiments conducted during high (June 13, 2018, Inflow stage: 635 mm) and low flows (July 30, 2018, Inflow stage: 384 mm) demonstrated the presence or absence of surface water connectivity between Inflow and a subset of target sites (Table 1). While tracers can also move through sub-surface flowpaths, the instantaneous tracer injection cannot detect flowpaths with very long residence times, and as such tracer arrival primarily reflects surface connectivity within our system. We did not observe arrival of injected tracer at Pond-Iso during either experiment, providing evidence of a lack of strong surface connectivity between Inflow and this site (Table 1). Tracer arrivals at other sites were variable, and we only observed tracer arrival during both the high and low flow injections at the Major Channel sites (Table 1). During the high flow tracer injection, tracer arrival was first observed at Main-Mid with a time to peak (TTP) of 22.5 minutes, followed by Side-01 (TTP: 35 min), Outflow (TTP: 46 min) and a more delayed arrival at Pond-Con-02 (TTP: 101 min) and Pond-Con-01 (TTP: 196 min) (Table 1). Modal velocity, which is defined as the most common velocity along a flowpath, was highly variable at connected sites (range: 0.09 - 0.87 m s-1, Table 1) indicating variable residence times along connected flow pathways. During the low flow tracer, the tracer arrival was only observed at the Main-Mid site (TTP: 40.8 min, Velm: 0.48 m s-1) and Outflow site (TTP: 85 min, Velm: 0.43 m s-1). Given the limits of detecting tracers at high residence times noted above, the lack of response at Side-01, Pond-Con-01 and Pon-Con-02 during the low flow injection cannot confirm a complete absence of surface connectivity. However, these results demonstrate that during the low flow experiment, Side-01, Pond-Con-01 and Pon-Con-02 were not strongly connected with the Inflow site.