Site description
This work was conducted in the Gibson Jack Creek watershed in southeastern Idaho, United States, a 16.7-sq km, steep (average watershed slope ~20°), high-relief watershed managed by the US Forest Service. The climate is characterized as semiarid, and precipitation primarily falls between fall and spring. Gibson Jack spans the rain-snow transition, with an average of 0.38 m of annual precipitation at lower elevations (~1500 m), which is primarily rainfall, and an average of 0.76 m annual precipitation at higher elevations (~2100 m), which is a mix of rain and snow (Welhan 2006). Peak streamflow occurs in spring, as the seasonal snowpack melts (Hale and Godsey 2019). North-facing slopes are vegetated primarily by Douglas fir, while south-facing slopes are vegetated with sagebrush, grasses, and juniper (Kline 1978).
Our research focused on a ~200-m perennial-intermittent-perennial stream transition along the mainstem of Gibson Jack Creek, which was the focus of a previous study of subsurface controls on stream drying (Dohman et al. 2021). We instrumented three sites along the mainstem. An upstream perennial site (UP) has a drainage area of ~13.5-km2, and is characterized by more frequent hillslope-riparian-stream connectivity, with shallow lateral flowpaths contributing the streamflow, and moderate vertical flow losses (Dohman et al. 2021). An intermittent site (INT) is ~150 m downstream of UP, with a ~15.5-sq. km drainage area. Nearly all of the intervening stream reach is intermittent and a tributary contributes additional flow ~100 m upstream of INT, leading to the jump in drainage area. INT is characterized by minimal hillslope-riparian-stream connectivity and high vertical losses, leading to stream drying (Dohman et al. 2021). This reach has been observed to dry seasonally for four years between 2018 and 2023, with drying starting any time between June and September. Complete drying is usually preceded by diel drying and rewetting cycles, during which disconnected pools form along the intermittent reach (Dohman et al. 2021). The third site is a downstream perennial reach (DP) located ~30 m downstream of a large spring, with a drainage area of ~15.5-sq km (nearly identical to INT), where streamflow resumed downstream of the intermittent reach. Flow at this perennial reach is sustained by consistent longitudinal connectivity–groundwater discharge from the spring–despite minimal lateral hillslope-riparian-stream connectivity (Dohman et al. 2021).
These three mainstem sites were instrumented with piezometers and pressure transducers to measure stage (the stream or “S” sites described in Dohman et al. (2021)), as well as water quality sensors (described below). In addition to weekly sampling at these main sites, we also collected weekly grab samples at the spring outlet and the tributary. For this paper, we focus on samples and data collected from March 25, 2020 through October 30, 2020, referred to hereafter as the 2020 field season.
In situ monitoring
We used a variety of sensors to measure water quantity and quality at 15-minute intervals during the 2020 field season at the three mainstem sites.
Fluorescent dissolved organic matter (fDOM) was measured using PME cyclops with Turner Designs cyclops-7 fluorometer (excitation at 325 nm with a 120 nm bandwidth and emission at 470 nm with 60-nm bandwidth). fDOM sensors were calibrated before deployment using 400 ppb 1,3,6,8-pyrene tetrasulfonic acid tetrasodium salt (PTSA), cleaned weekly, and any fouling interference was corrected assuming linear drift. Data were corrected for temperature using sensor-specific corrections (Watras et al. 2011). We did not measure turbidity, and therefore were not able to correct fDOM for turbidity interference. However, turbidity data from 2022 suggest that turbidity is generally low at these sites, even during snowmelt (<15 NTU), only peaking during storms, which were excluded from our analysis. Since turbidity interference is minimal at low turbidity (<14 NTU) (Downing et al. 2012), our inability to correct for turbidity interference is unlikely to have significant effects on our results.
Dissolved oxygen was measured using PME miniDOT sensors. Before deployment, sensors were calibrated in 100% saturated water by bubbling water with aquarium airstones and in 0% DO water by adding yeast. Sensors were fitted with copper antifouling plates to minimize fouling and cleaned weekly. Saturation concentrations were calculated using water temperature and local barometric pressure, and saturation deficit was calculated as the difference between saturation and actual DO concentrations (in mg O2/L).
Conductivity was measured using HOBO-U24-001 conductivity sensors, and conductivity data was converted to specific conductivity using sensor-specific calibrations (U.S. Geological Survey 2019). Local light availability (in LUX) and air temperature were measured using HOBO Pendant loggers attached horizontally to rebar within the stream channel above the stream surface.
All water quality sensors were cleaned weekly, and data were corrected for fouling drift as needed. Calibrations were checked following deployment at the end of the study period and assessed for calibration drift, which was <5% for all sensors. All sensor water quality data were inspected visually for outliers, which were removed. Small gaps in sensor data (<1 hour) were filled by linear interpolation. Longer gaps were removed from the time series (including a 22-day period from July 14, 2020 to August 3, 2020 at UP due to fDOM sensor malfunction).
Water level was measured with HOBO-U20-001 unvented pressure transducers that were suspended in fully screened piezometers installed to refusal within the stream channel and corrected for barometric pressure (installation and corrections detailed in Dohman et al. (2021)). We used stage for our analyses rather than discharge due to large uncertainties in stage-discharge relationships at these sites. This limits comparisons of magnitude of change among sites because of differences in channel geometry, but because we used stage primarily to characterize seasonal and subdaily variation within sites, our reliance on stage data is still useful.
To correct for barometric pressure for each sensor, atmospheric conditions were measured every 10 minutes at a nearby high-elevation weather station and then adjusted by 5.22 kPa to account for the elevation offset. An additional pressure offset of 0.42 kPa was required for the INT sensor because the sensor consistently reported unrealistically high pressures. This additional offset was validated by observations in the field and inferred drying and rewetting of the independent fDOM sensor. Two brief, late-season storms required further attention at site INT: sedimentation of this sensor required a daily stepwise correction of 11.5 to 6.6 cm during the 7-10 Sep storm, and then a linear trend offset of up to 24.1 cm during the 11 Oct storm. We could validate these offsets because of fDOM sensor wet/dry patterns and field observations during and following the events. Furthermore, we observed dry conditions from 21 Oct through the end of the 2020 field season during multiple visits and at the fDOM sensor, so we replaced noisy measurements from the sedimented pressure transducer at INT with no-flow (zero) values. All field observations match the recorded stage, though there are three days when the fDOM sensor records slightly earlier rewetting or later drying than expected from the pressure transducer (Sept 7, 25 and 28). A more complex correction could be applied to the Sept 7 data to correct for this offset, but the surrounding days for all three dates reproduce the wet/dry patterns to within instrument accuracy, and thus, we err on the side of parsimony in our corrections.
Dissolved organic carbon sampling
Grab samples for dissolved organic carbon (DOC) concentration were collected weekly at the three mainstem sites, as well from the tributary and the spring inputs. Samples were syringe-filtered in the field through Whatman 0.45-µm nylon filters into falcon tubes and frozen until analysis on a Shimadzu TOC/TN analyzer at Brigham Young University.
References:
Dohman, J. M., S. E. Godsey, and R. L. Hale. 2021. Three-Dimensional Subsurface Flow Path Controls on Flow Permanence. Water Resources Research 57: e2020WR028270. doi:10.1029/2020WR028270
Downing, B. D., B. A. Pellerin, B. A. Bergamaschi, J. F. Saraceno, and T. E. C. Kraus. 2012. Seeing the light: The effects of particles, dissolved materials, and temperature on in situ measurements of DOM fluorescence in rivers and streams. Limnology and Oceanography: Methods 10: 767–775. doi:10.4319/lom.2012.10.767
Hale, R. L., and S. E. Godsey. 2019. Dynamic stream network intermittence explains emergent dissolved organic carbon chemostasis in headwaters. Hydrological Processes 33: 1926–1936. doi:10.1002/hyp.13455
Kline, R. 1978. Municipal watershed management plan for West Mink and Gibson Jack creeks of the Pocatello Ranger District. Caribou National Forest Open-File Report.
U.S. Geological Survey. 2019. Chapter A6.3. Specific Conductance. 9-A6.3. 9-A6.3 U.S. Geological Survey.
Welhan, J. A. 2006. Water balance and pumping capacity of the Lower Portneuf River Valley Aquifer. Idaho Geological Survey Staff Report 06–5. 06–5.
