Geoscience Reference
In-Depth Information
plastics into sensor production to minimize adhesion of biofouling. This is a particularly
promising invention, as it removes common side effects of bio-wipers such as impedance
of the sensor signal due to mechanical positioning on the sensor emitter/detector surface,
interference due to wiper shading the signal, and wiper failure.
Biofouling can impact data by decreasing or increasing fluorescence signal (Delauney
et al., 2010), the former due to physical blocking of optical path and the latter if the foul-
ing material fluoresces at same wavelengths measured by the sensor (
Figure 6.13c
). In the
event that biofouling impacts data quality, portions of a data stream can either be excluded,
or attempts to correct the impact can be conducted. This is largely dependent on the degree,
type, and duration of fouling. In a study in 2009 by Cetinic et al., measurements from a
biofouled sensor were corrected by assuming a linear drift in instrument response as a func-
tion of increased biofouling. In the paper, analysts are warned that complications with this
approach can arise because biofilm growth is exponential and undergoes a five-stage series
of development, meaning that the response may not in fact be linear. However, given that in
most deployments, calculating true biofilm growth rate is not possible, linear corrections may
be the only reasonable option. In addition, the manner of sensor deployment is also a factor,
where correcting data from a profiling mooring or autonomous underwater vehicles (AUVs)
may be more challenging (due to sensor sampling multiple water types) than from a moored
stationary sensor where biofouling rate may be calculated. In any case, analysts should note
that if biofouling on any optical window is sufficiently developed to the point of swamping
the ambient signal, even the best calibration procedure will not yield high data quality.
6.5.1.6 Understanding NOM Sources Within Environments
Commercially available sensors are often configured to measure at excitations and emis-
sions which are representative of humic fluorescence peaks at longer wavelengths (e.g.,
peak C; Coble,
1996
). Many studies have found peak C-based NOM fluorescence useful
as surrogates for dissolved organic carbon (DOC) concentration in a myriad of pristine
and anthropogenically influenced watersheds. Analysts should be forewarned that changes
in fluorescence signal from single-channel sensors may not stem from a change in fluor-
escence intensity alone, but rather from a shift in the wavelengths of the fluorescence
peak. These shifts result from source and compositional differences of organic material.
Consider, for example, a case where a freshwater system has three dominant sources of
CDOM: wastewater, plant leachates, and autochthonous production. Variations in sensor
signal could arise from changes in concentrations of the material, but could also result from
the sensor detecting one source over another. Sensors configured at a single fixed set of
wavelengths cannot measure the shift and can compromise signal intensity. Before deploy-
ment of a peak C sensor, it is recommended that a preview of excitation-emission matrix
spectroscopy (EEMS) from the study site to preestablish FDOM/DOC relationships that
are to be considered. Review of EEMs and DOC will usually yield a sense of how reliable
the sensor proxy will be for that system. In addition, ongoing collection of discrete (cali-
bration) samples for EEMs analysis may also prove to be necessary to evaluate continued
instrument performance as well.
