June 2025 | Oceanography
69
et al., 2020) equipped with a Cyclops-7F rhodamine fluorome
ter (Turner Designs). Discrete water samples were later analyzed
in the laboratory using a benchtop fluorometer. The Niskin bot
tle was lowered into the water column and triggered at depths
ranging from 0.5 m to 45 m. The sampling was conducted on a
release day, a pre-release day (1 day prior) and a post-release day
(1–3 days later). The PIXIE captured the vertical RWT/Chl-a con
centration profiles to demonstrate its multichannel functionality,
while the Niskin bottle provided a ground-truth measurement
of the RWT concentration at the surface and above the seafloor.
Data collected by each method were used to semi-quantitatively
validate the performance of the PIXIE as an in situ fluorometer.
RESULTS AND DISCUSSION
Calibration
A total of 37 data points was collected during the calibration of
the PIXIE’s RWT channel. These consisted of six standard con
centrations across six temperature setpoints. The 60 ppb mea
surement at 5°C saturated the device. A replacement 6.9°C tem
perature set point was also collected. A total of 37 data points
were collected during the calibration of the Chl-a channel. The
80 ppb measurements at 5°C and 8°C saturated the device. A
replacement 9.6°C temperature set point was also collected.
The raw fluorescence data for each data point were collected
at the PIXIE’s maximum sample rate of 16 samples per second.
In line with previous work (Park et al., 2023), the raw data were
downsampled through a moving average of 16 samples, for an
effective sample rate of 1 sample per second. Fifteen minutes
of raw data were collected for each data point. During the last
five minutes, 300 samples were collected, and the mean of these
300 samples was taken as the calibration data point. The prior
10 minutes of data were inspected to ensure an apparent equilib
rium fluorescence had been reached.
The six standard concentrations for each fluorophore
were used to generate a best-fit line for each temperature (see
Figure 3). In the 60 ppb RWT case, a 5°C data point was first
extrapolated from the six unsaturated temperature set points
(6.9°C as well as the original five). In the 80 ppb Chl-a case,
5°C and 8°C were first extrapolated from the five unsaturated
temperature set points (9.6 °C as well as the original four). The
parameters of the resulting curves do not change significantly
between the inclusion or exclusion of the extrapolated points.
The coefficient of determination (R-squared) for each calibra
tion curve exceeded 0.99 for RWT and 0.98 for Chl-a.
FIGURE 3. (a) Rhodamine water tracer (RWT) calibration curves, with a dashed line indicating saturation. The extrapolated point is encircled. Inset: Plot
with concentration-equivalent 10-sigma error bars. (b) RWT temperature compensation “slope of slopes” curve. (c) Chl-a calibration curves, with a
dashed line indicating saturation. Extrapolated points are encircled. Inset: Plot with concentration-equivalent 1-sigma error bars, illustrating sensitivity to
bubbles. (d) Chl-a temperature compensation “slope-of-slopes” curve.