At the gaging station site, a record of water surface height, or stage, was compiled to compute a record of discharge from stage-discharge relations. The gage datum was arbitrary, and was not the same each year. The methodology for stage recording was constrained by the presence of frozen ground, sub-freezing air temperatures, the need to protect stream recorders from unanticipated (and unknown) high discharge and lack of a site free of current at all stages.
The frequency with which stage was recorded was also determined by the field site, equipment and logistics. Water-stage recorders provided a continuous record during periods of 1991 and 1992 when they could be used; manual measurements were made at other times, rarely more frequently than twice-hourly.
In 1990, streamflow began rather suddenly two days after field personnel arrived, marked by the pulsed advance of a slushflow through cold, dry snow filling the channel. An initial gaging station site was immediately selected at a constriction in the channel, where the stream flows from the canyon onto the delta, and a staff gage was emplaced (1030 h June 9). Virtually the entire stream channel perimeter at this point was comprised of snow, resulting in a highly unstable cross section. (IMAGE: early streamflow) The staff gage was temporarily displaced several times by subsequent slushflows.
As the delta stream channel morphology became more apparent, a decision was made to relocate the gaging station to a more suitable site. The staff gage was reinstalled at 0930 h on June 11, at the section of stream channel utilized for all subsequent stage measurements. (IMAGE: hydrological station)
At the new site, the staff gage was initially situated close to the center of the channel, due primarily to the existence of slush levees on either side. This staff gage was used for stage measurement until 1700 h on June 17, when an additional staff was installed closer to the right edge of water. The second gage was considerably less influenced by standing waves, and immediately became the reference gage. A six day period of overlapping observations was used to convert heights recorded on the midstream gage to the datum of the gage at the right side of the stream (using third order regression).
In 1991, a securely braced staff gage was installed by 1900 h on June 4 at the same site used in 1990, less than four hours after a slushflow again initiated streamflow for the season. The staff gage was used throughout the field period, with observations again dependent upon how rapidly stage varied. During waking hours, readings were typically hourly or bi-hourly, and whenever discharge was measured.
During segments of the field period, a water-stage recorder was also operated, in conjunction with a stilling well and float sensor. The recorder was a Stevens type F model 68, with a Quartz multispeed timer, 1:1 gearing between the float pulley and the chart drum, and an F4 chart. A 4 inch float was used inside an approximately 45 cm diameter steel barrel, which served as the stilling well, and upon which the recorder was housed.
The water-stage recorder and barrel (hereafter called the barrel gage) were situated close to the staff gage on the right side of the stream. The recorder location was chosen to balance a desire to maximize the range of heights able to be recorded, with a need to protect the recorder from influence by current and wave action. Additionally, the location was constrained by an inability to alter the frozen stream bed. As a consequence of the recorder location, the stilling well inlet hole was sometimes above the stream water surface, which prevented the recording of stage height when discharge was less than ca. 0.3 m^3 s^-1. The recording gage datum height was measured by autolevel, relative to a large boulder, on June 18 and July 2. The two values were within 1 mm, indicating no net movement of the gage during the period.
Streamflow began in 1992 as a small volume of snowmelt began to flow on June 22, both through and at the surface of snow filling the stream channel. Discharge was determined for the first day by direct measurements, made in a snow-free portion of the channel approximately 200 m above the lake. On June 25 a recording gage was started, in an attempt to record stage height over a greater range than had been possible the previous year, while also reducing the risk of recorder loss. The device used a 3 inch float sensor inside a 3.25 inch inside diameter tube (hereafter termed the tube gage), which was linked to the recorder by cable and pulleys. The outside of the tube was calibrated to serve as a staff gage. A second water-stage recorder, the barrel gage, was installed on July 2, in approximately the same location used in 1991. Although this gage was again unable to record low stages (i.e. in 1992, when discharge was less than 0.75 m^3 s^-1), the recorder proved more sensitive than the tube gage recording system. Both gages were operated until August 8, and complemented by manual observations.
The measurement of discharge in the High Arctic is problematical for a number of reasons, as described by Woo and Heron (1986). The conventional current-meter method was the technique selected for use at Taconite Inlet, largely because of the frequency with which discharge had to be measured. Although 15 to 30 minutes were required for each measurement, the method worked well. At the time of each measurement, other observations were made, including slush concentration, degree of bottomfast ice in the channel, channel wall material, and the rate at which stage was changing.
The 204 discharge measurements made in the inlet stream during the three years were all done by measuring water depth and velocity at a series of verticals across the stream. The locations of these verticals were defined by tags along a polypropylene tag line, set up each year above the rated section. The principal difference between measurements over the three years was in the number of verticals used.
Except for the first 11 measurements of 1990, the velocity measurements were all made by a Swoffer Instruments, Inc. model 2100 current meter. This is a horizontal-axis, propeller-type electronic instrument. The resolution of the meter was 1 cm/s, and manufacturers stated accuracy is within 1 percent, given periodic calibration tests and adjustments by the user. These were in fact carried out, both in the field and subsequent to the field seasons. Velocity measurements recorded at any one vertical were the average of at least three, 6-second averages displayed by the meter, following at least 6 seconds at the depth being measured. An attempt was made to always measure velocity at six-tenths of the water depth, measured down from the surface. Depth was measured using a calibrated rod, with a 3 cm diameter foot to prevent penetration into the channel bed.
The first three discharge measurements of 1990 were made at the initial, upstream gaging station site (all on June 10). The first 11 measurements were made with a General Oceanics current meter (June 10 and 11) which, in a field comparison of three full discharge measurements, resulted in values within three percent of those by the Swoffer meter used subsequently.
Through the 1990 season, the Swoffer meter sensor was mistakenly used on the wrong side of the wading rod. The propeller, however, was properly oriented. Several tests were conducted in 1991 to ascertain the effect of this configuration: in two tests, the average correctly measured velocity was 10 percent slower than when the meter was backward, and in another, the correctly measured velocity was 20 percent faster. Based on these equivocal results, and the fact that the difference is not likely to be linear with velocity, the data were not adjusted.
Water temperature measurements made in 1990 were primarily made approximately 175 m downstream from the gaging station site, close to camp. This downstream site was also used in 1991 and 1992 to measure water temperature, whenever conductivity measurements were made. The primary water temperature measurement site in 1991 and 1992, however, was at the gage site upstream.
Water temperature measurements at the downstream site were made with a Yellow Springs Instrument Co. (YSI) model 33 S-C-T meter. The 1991 and 1992 measurements with this instrument were used only for comparative and back-up purposes. At the gaging station site, a continuous series of water temperature was acquired through the period of streamflow during 1991 and 1992. Measurements were made every minute by a Campbell Scientific Instruments (CSI) model CR21 micrologger and a model 101B thermistor, which were averaged on the hour (n = 60), and recorded. Although the thermistor used for these measurements was not intended for continuous submersion, the sensor was modified by coating the probe tip with white Plastic Rubber. In 1991 the sensor was then wrapped in aluminum foil, while in 1992 an additional coating of silicon was applied, both in an effort to reduce solar heating.
The gaging station site sensor was held at a fixed position in the streamwater, close to the right edge of the stream. Generally, the sensor was 5 to 10 cm above the channel bed. Water depth over the sensor varied according to stage.
Suspended sediment samples were obtained almost exclusively with a depth-integrating suspended-sediment wading-type hand sampler, model US DH-48, developed by the Federal Inter-Agency Sedimentation Project. (IMAGE: US DH-48 in use) This sampler collects streamwater, and sediment suspended in the water, in a round 500 mL (16 oz.) bottle. The bottle fills isokinetically while submerged in flows of velocity less than 2.7 m s^-1. The sampler used at Taconite Inlet was made by Product Manufacturing Co., and was fitted with a 6.4 mm (0.25 in.) nozzle and a standard wading rod. Glass collection bottles were used in 1990 and 1991, while plastic bottles from the sampler manufacturer were used in 1992.
The DH-48 can sample to within 90 mm of the stream bed, resulting in an unsampled zone just above the bed. However, across the sampled cross section at the field site, bed roughness probably induced sufficient turbulence to disperse suspended sediment relatively uniformly through both the unsampled and sampled zones. For example, several samples from the sampled zone contained material up to 4 mm in diameter.
The electrical conductivity (EC) of streamwater was measured to provide a proxy estimate of total dissolved solids (TDS) being transported to the lake. EC measurements were not correlated with analytically determined TDS values. Therefore, only an estimate of TDS magnitude and temporal variability was obtained.
All EC measurements were made manually with a Yellow Springs Instrument Co. (YSI) model 33 S-C-T meter. This instrument provides an internal temperature correction to conductivity measurements. Meter and probe accuracy are stated by the manufacturer to be +/- 3 and +/- 2 percent of reading, respectively. In the worst case additive error situation, this translates to 5 µ mhos for a 100 µ mho reading. Excepting the first few weeks of 1990, all measurements used the same cable and sensor.
Most conductivity measurements were made at a main channel monitoring site approximately 175 m downstream from the gaging station site. The exceptions to this were 1991 readings through early July, which were made in the lower volume distributary, located closer to camp. Several comparisons indicated that differences between the two sites were minor, unless flow in the camp channel was greatly reduced. In two tests at such times, the EC in the camp channel was 30-35 percent greater.