Sites and Data Used
The algorithm we developed was designed around a single streamwater SC
sensor that was deployed in Clay Brook, a lotic system in New Hampshire,
U.S.A. (Figure 2), which repeatedly exhibited all three SC patterns. We
used data from May to November for the 2013 and 2014 calendar years. The
SC sensor recorded both electrical conductivity (EC) and water
temperature. Using these the two measures, the EC was corrected for
temperature and reported as if recorded at 25 °C, giving the SC. SC was
calculated using
where T was water temperature in °C and both EC and the resulting SC
were in µS cm-1.
Clay Brook is located in the hills at the southern end of the White
Mountains. Its headwaters begin on Plymouth Mountain and, flowing
generally northward, ends with its confluence with the Baker River, a
tributary to the Pemigewasset and Merrimack Rivers, which drain to the
Atlantic Ocean. The location of the sensor and pour point of the studied
catchment area is roughly halfway through the stream’s full watercourse,
encompassing a catchment area of 7.49 km2 (Figure 2).
The median elevation of the basin is 359 m, with an average slope of 9
degrees. The catchment is 89.3% forested with a mix of hardwood and
coniferous trees. Common species include American beech, red maple,
sugar maple, red oak, yellow birch, eastern hemlock, and red spruce. The
remaining 10% of the catchment’s land cover is split between
agriculture and wetlands, including 1.3% development. The soils of the
region are generally coarse textured Spodosols of a range of drainage
classes.
We used meteorologic data from the Plymouth Regional Airport, Plymouth,
NH (NOAA ID K1P1) located 4.6 km from the sensor site at 43.78°
latitude, -71.75° longitude, at an elevation of 154 m. We used data from
the airport’s weather station, which measured air temperature, relative
humidity, and precipitation in 20-minute intervals.
The SC sensor data from Clay Brook were measured with an Onset Hobo Data
Logger U24, which recorded water temperature and EC at a 4-minute
interval. The sensor has an EC accuracy of 3% of the reading value, or
5 µS cm-1, whichever is greater, and a temperature
accuracy of 0.1°C. The sensor’s operating range is -2°C to 36°C in
non-freezing conditions and has a resolution of 1µS
cm-1 and 0.01°C. The data were screened to remove any
records that were consistent with the sensor being out of the water at
low flows, and random sensor malfunctions that resulted in one data
point being excessively high compared to the preceding and following
data points. Independent lab analysis of field grab samples at this site
suggest that the sensor accuracy is within 5%. We further screened data
to only include periods of time from May 1 to October 31 when regional
rainfall-runoff responses are generally not complicated by snow and ice
(Figure 3).