Vous êtes sur la page 1sur 5

Continuous Monitoring

Optical Sensors
for Water Quality
Brian A. Pellerin and Brian A. Bergamaschi

hifts in land use, population, and monitoring gaps, and provide science- matter (DOM) convert absorbed light into
climate have altered hydrologic based decision support across a range of other forms of energy, and include the
systems in the United States in ways issues related to water quality, freshwater re-release of energy at longer wavelengths
that affect water quality and ecosystem ecosystems, and human health. (e.g., fluorescence) by humic substances.
function. Water diversions, detention The wavelength and amount of light
in reservoirs, increased channelization, State of the Technology absorbed and emitted provides important
and changes in rainfall and snowmelt One of the most promising advances information on the type, size, and
are major causes, but there are also more in recent years is the increasing use of concentration of constituents in water.
subtle causes such as changes in soil optical sensors for water quality studies. Field optical measurements related to the
temperature, atmospheric deposition, Optical sensors rely on the absorbance, concentration and types of suspended
and shifting vegetation patterns. The fluorescence, or scattering properties of particles in water have been around for
effects on water quality are complex and materials that are dissolved or suspended more than 40 years, with turbidity – a
interconnected, and occur at timeframes in water (Figure 1). Recent interest has measure of the relative clarity of water –
of minutes (e.g., flash floods) to decades focused on the ability to measure the perhaps the most common example. While
(e.g., evolving management practices). concentration or type of some dissolved the ability to make relatively simple and
However, water-quality monitoring constituents through absorbance and inexpensive optical measurements of
has historically focused on discrete fluorescence. Certain types of dissolved DOM and nitrate in the laboratory has
samples collected weekly or monthly, and constituents such as nitrate and organic been known for even longer, advances in
laboratory analyses that can take days or
weeks to complete. Low-frequency data
and delayed access hampers a timely
response during events, limits the ability
to identify specific causes or actions, and
may result in poorly quantified effects on
ecosystems and human health at local to
regional scales.
Recent advancements in
commercially available in situ sensors,
data platforms, and new techniques for
data analysis provide an opportunity
to monitor water quality in rivers,
lakes, and estuaries on the time scales
in which changes occur. For example,
measurements that capture the variability
in freshwater systems over time help
to assess how shifts in seasonal runoff,
changes in precipitation intensity, and
increased frequencies of disturbances
(such as fire and insect outbreaks) affect
the storage, production, and transport
of carbon and nitrogen in watersheds. Figure 1. Optical sensors make measurements based on the interactions of light from a sensor
Transmitting these data in real-time also with particles or dissolved constituents in water. Certain types of dissolved constituents, such as
provides information that can be used nitrate and organic matter (DOM), convert absorbed light into other forms of energy, including
for early trend detection, help identify the re-release of energy at longer wavelengths (e.g., fluorescence) by certain humic substances.

Spring 2014 / LAKELINE 13

electronics and sensor technology over the “dead zone” during summer
past 20 years has led to the development months. The recent deployment
of field-rugged, compact and low power of UV nitrate sensors at key
optical sensors for direct measurements locations such as in the lower
of these constituents in water. Nitrate Mississippi River at Baton Rouge
and DOM have been the focus of much (Figure 3) allows for monthly
recent interest for optical sensor use and loading estimates to be refined
development and are further discussed while reducing the uncertainty in
here. those estimates, leading to a better
understanding of the timing and
Optical Sensors for Nitrate magnitude of nitrate transport
On the forefront of new sensor within the basin.
technologies for water-quality monitoring
in freshwater systems are ultraviolet Optical Sensors
(UV) photometers for continuous nitrate for Organic Matter
measurements (Figure 2). UV nitrate Fluorescence-based optical
sensors have been used during the past sensors also present an emerging
few decades for wastewater monitoring opportunity to better understand
as well as for coastal and oceanographic organic matter in rivers, streams
studies, but have gained broader use in and lakes. Organic matter includes
freshwater systems only in the last few a broad range of organic molecules
years. The current generation of UV of various sizes and composition
nitrate sensors is now being designed that are released by all plants and
specifically for freshwater applications animals (living and dead) and
with rugged housings, internal data have important implications for
loggers, built-in wipers, and data drinking water quality, contaminant
processing tools that better account for transport, and ecosystem health.
particles and other interferences common Measuring the fraction of dissolved
in rivers, streams, and lakes. Optical organic matter (DOM) that absorbs Figure 2. Continuous UV nitrate sensors and other
water-quality instruments deployed in the Mississippi
nitrate sensors operate on the principle light at specific wavelengths and
River at Baton Rouge (USGS gage 07374000) allows
that nitrate ions absorb UV light at subsequently releases it at longer
for a better understanding of nitrogen dynamics at the
wavelengths around 220 nanometers. wavelengths (e.g., fluorescence) rates in which changes occur.
Commercially available sensors utilize is diagnostic of DOM type and
this property of nitrate to convert spectral amount. Studies have often used the
absorption measured by a photometer to excitation and emission at 370 and 460 variables such as mercury concentrations
a nitrate concentration, using laboratory nanometer (nm), respectively, to quantify are also strongly correlated with in situ
calibrations and on-board algorithms. the fluorescent fraction of colored DOM FDOM measurements. For example, in
This allows for calculating real-time (referred to as FDOM). Sensors for situ sensors were deployed seasonally
nitrate concentrations without the need for FDOM have a long history of use in on Browns Island, a tidal wetland in the
chemical reagents that degrade over time oceanography as an indicator of terrestrial San Francisco Bay-Delta, to measure
and present a source of waste (Pellerin et organic matter entering the coastal ocean, optical properties related to DOC and
al. 2013). but have only recently been adopted dissolved methylmercury (MeHg) across
Nitrate is the largest component of for use as water-quality monitors in tidal cycles and seasons (Bergamaschi et
total nitrogen in most freshwater systems freshwater systems. al., 2011). In situ FDOM measurements
and, in many locations, represents the In situ FDOM sensors have been explained almost 90% of the variability
most significant concern for algal blooms used in many different environments in dissolved MeHg concentrations across
and human health. One such example is to provide a relatively inexpensive, two channels and three seasons, allowing
in the Mississippi River Basin, where high-resolution proxy for dissolved researchers to develop accurate and cost-
the addition of optical nitrate sensors organic carbon (DOC) concentrations. effective flux estimates of MeHg in a
at key U.S. Geological Survey (USGS) This has been useful to understand the highly dyamic tidal system (Figure 4).
discharge gaging stations is providing transport of DOC from watersheds, but
new information about the sources and has also been used to better understand State of the Art
processes that deliver nitrogen to the the internal sources of DOC in drinking The advantages offered by in situ
coast. For example, USGS discrete water reservoirs (Downing et al. 2008) optical sensors over discrete sampling
and model data on nutrient loads from and the ability to predict the formation or other in situ approaches (ion
the Mississippi River basin to the of disinfection by products such as selective electrodes and wet chemical
Gulf of Mexico have been critical for haloacetic acids following drinking water sensors) are many – rapid sampling
understanding the role of nutrients in treatment (Carpenter et al. 2013). In rates, low detection limits, low power
the formation of a low dissolved oxygen some cases, other related biogeochemical consumption, no chemicals, easy field

14 Spring 2014 / LAKELINE

Figure 3. Sensors capture data during all hydrologic events, which results in higher accuracy and lower uncertainty than modeled loads that are
based on discrete samples. The data from the USGS site on the Mississippi River at Baton Rouge (USGS gage 07374000) demonstrates the complex
relationship between nitrate concentrations and discharge, which reflects both the sources of nitrate within the basin and the accumulation of nitrate
in soils prior to flushing.


Figure 4. Data from an instrument deploying at a tidal wetland (Browns Island) in the San Francisco-Bay Delta. (A) the correlation between in
situ FDOM sensor measurements and dissolved methylmercury concentrations from discrete samples. (B) Picture of the in situ sensor deployment
package that includes FDOM and a variety of other optical sensors. (C) Time series data of methylmercury (MeHg) fluxes from Browns Island,
indicating a net off-island flux across the spring-neap tide. High-resolution data also illustrate the influence of tidal cycles and wind events on MeHg
flux in the San Francisco Bay-Delta system. See Bergamaschi et al. (2011) for more detail.

Spring 2014 / LAKELINE 15

servicing, and long-term deployment possible across a wide range of turbidities rate of change to observe the true time-
capability. Optical sensor technology is and DOC concentrations. dependence. Sampling bias can occur
sufficiently developed to warrant their Interferences, matrix effects, and when constituent concentrations change
broader application, but generating data other challenges for collecting high significantly between samples, which can
that meet high standards requires future quality water-quality data in situ will lead to overestimates or underestimates
investments into common methodologies best be solved by continuing to work of watershed loads, inaccurate pollution
and protocols for sensor characterization with manufacturers and the broader user assessments, and potentially obscured
and data management. For example, community to fully characterize sensors seasonal or long-term trends. Traditional
optical nitrate sensors were originally and develop mechanical solutions and/ monthly discrete sampling approaches
developed for very different environments or correction schemes that will work may be particularly susceptible to bias in
– coastal oceans with very low turbidity across the typical range of conditions more dynamic freshwater systems such as
and color, versus wastewater treatment encountered in rivers, streams and lakes. streams and small rivers.
facilities with very high turbidity and Similarly, continued development There are many instances where high
color. Therefore, use in rivers, streams, of common methodologies and protocols temporal resolution data are critical for
and lakes requires careful consideration of are critical to ensuring comparable understanding drivers of water quality
instrument design, such as the appropriate measurements across sites and over time and effects on human health, ecosystem
optical path length and wipers or other (Pellerin et al. 2012). Such investments function, or water management. For
anti-fouling techniques. will continue to increase the number of example, continuous measurements may
Similarly, optical sensor sites at which these technologies are used improve upon nutrient load estimation
measurements may be influenced by as well as increase the types of parameters techniques if discrete sampling does not
a variety of matrix effects including that can be measured by sensors in real- fully capture the concentration-discharge
water temperature, inner filter effects time. range or where the concentration-
from highly colored water, turbidity, discharge relationship is poor. However,
and the presence of bromide (for UV Evaluating the Need for Continuous Data not all freshwater systems are subject to
photometers). Some in situ optical Given the current costs to purchase rapid changes in water quality, and some
sensors require further characterization optical sensors alone (approximately monitoring or research goals may be
for interferences and correction schemes $2,000-5,000 for FDOM and $15,000- sufficiently addressed with less frequent,
prior to widespread use in river, stream, 25,000 for UV nitrate) and the ongoing discrete data collection. Potential users
and lake monitoring. For example, expenses related to instrument service and should assess existing discrete water
laboratory tests with standard reference maintenance, potential users may want to quality data and continuous sensor data
materials demonstrated that measured carefully consider whether “continuous” for other parameters (such as specific
FDOM values were strongly influenced data (for example, multiple samples per conductance and dissolved oxygen) to
by temperature, turbidity, and DOC hour or day) are really needed. Although determine if a site would benefit from
concentrations in matrix conditions explicit guidance is not available, basic continuous measurements from an optical
similar to those observed in many rivers time-series analysis requires that the sensor. Temporary sensor deployments
and streams (Downing et al. 2012, Figure rate of sampling be greater than the could also provide short-term data on the
5). However, these interferences appear
to be predictable and corrections may be B

Figure 5. (A) Attenuation of the FDOM signal in an optically clear standard solution (left), as well as a solution with DOM (center) and suspeneded
particles (right). (B) The percentage of FDOM attenuated with several types of in situ FDOM sensors as a function of turbidity. See Downing et al.
(2012) for more detail.

16 Spring 2014 / LAKELINE

degree of variability before investing in References
a longer-term continuous measurement Bergamaschi B.A, J.A. Fleck,
effort. B.D. Downing, E. Boss,
B.A. Pellerin, N.K. Ganju,
Looking to the Future D.H. Schoellhamer, A.A.
The current generation of Byington, W.A. Heim, M.
commercially available optical sensors Stephenson and R. Fujii.
will significantly improve the temporal 2011. Methyl mercury
resolution for measurements of nutrients, dynamics in a tidal wetland
organic matter, and other parameters. quantified using in situ
However, a variety of new and improved optical measurements.
sensors for applications in rivers, streams, Limnol Oceanogr, 56(4):
and lakes are on the horizon. For example, 1355-1371.
recent improvements in LED technology Carpenter K.D., T.E.C.
in the lower UV range (e.g., 250-290 Kraus, J.H. Goldman,
nanometers) will open up new avenues J.F. Saraceno, B.D.
for the use of custom fluorometers for Downing, G. McGhee and
the detection of wastewater and other T.Triplett. 2013. Sources
contaminants from urban and agricultural and Characteristics of
landscapes (Figure 6). Several wet Organic Matter in the
chemical sensors also show significant Clackamas River, Oregon,
promise for long-term, in situ monitoring Related to the Formation of
of soluble reactive phosphorus and Disinfection By-products
ammonium in freshwater systems. In in Treated Drinking Water:
addition, advancements in real-time U.S. Geological Survey
data transmission and communication Scientific Investigations
with sensors will provide numerous Report 2013–5001, 78 p.
benefits including monitoring sensor Downing B.D., B.A. Figure 6. Field testing a custom in situ fluorometer for
performance, providing an early warning Bergamaschi, D.G. detecting wastewater effluent near Madison, Wisconsin. Photo:
of water-quality issues, allowing for Steve Corsi, USGS.
Evans and E. Boss. 2008.
adaptive sampling, and increasing public Assessing contribution of
awareness. The user community should DOC from sediments to a drinking- Center in Sacramento,
also continue to work with software water reservoir using optical profiling. California. Despite his
developers to continue development of Lake Reserv Manage, 24: 381-391. title, most of his work
tools for automating quality-assurance Downing B.D., B.A. Pellerin, B.A. actually takes place in
and quality-control (QAQC), storage and Bergamaschi, J. Saraceno and T.E.C. water. He uses a variety of
retrieval, and visualization of real-time in Kraus. 2012. Seeing the light: The tools to better understand
situ optical sensor data and statistics. effects of particles, temperature and watershed biogeochemistry,
Perhaps the greatest scientific “bang inner filtering on in situ CDOM but recent efforts focus on
for the buck” lies in the development of fluorescence in rivers and streams. the application of in situ
inter-calibrated networks of water-quality Limnol Oceanogr: Methods, 10: 767- optical sensors for carbon and nutrient studies in
sensors that provide information about 775. rivers and streams. He can be reached at bpeller@
water quality across the continuum from Pellerin B.A., B.A. Bergamaschi, B.D. usgs.gov.  
headwater streams to lakes, reservoirs, Downing, J. Saraceno, J.D. Garrett and
and ultimately coastal rivers and L.D. Olsen. 2013. Optical Techniques Brian A. Bergamaschi
estuaries across the United States. The for the Determination of Nitrate in is a research chemist at
information provided by such a network Environmental Waters: Guidance the U.S. Geological Survey
would assist environmental and water- for Instrument Selection, Operation, California Water Science
quality managers as an early warning of Deployment, Quality-Assurance, and Center in Sacramento,
problems, help assess long-term trends, Data Reporting. USGS Techniques and California. His current
and provide data to evaluate the effects of Methods Report 1-D5, 37 pp. projects generally
management and mitigation actions across Pellerin B.A., B.A. Bergamaschi and J.S. coalesce around the idea
multiple scales. However, standardized Horsburgh. 2012. In situ optical water- of using intrinsic organic
sensor measurement protocols, data- quality sensor networks – Workshop geochemical properties – molecular, isotopic,
collection strategies, and common QAQC summary report. USGS Open-File and spectroscopic – as tracers of environmental
approaches will be necessary to develop Report 2012-1044, 13 p. processes occurring in both terrestrial and aquatic
an inter-calibrated network of in situ environments. He can be reached at bbergama@
optical sensors with different agencies and Brian A. Pellerin is a research soil scientist at the usgs.gov. c
users. U.S. Geological Survey California Water Science

Spring 2014 / LAKELINE 17