"Right now, incomplete or infrequent water quality data can give
people an inaccurate picture of what's happening — and making
decisions based on inaccurate data can be risky," says Dr.
François Birgand, an assistant professor of biological and
agricultural engineering at NC State and co-author of a paper
describing the work. "Our approach will help people get more
detailed data more often, giving them the whole story and
allowing them to make informed decisions."
In addition to its utility for natural resource managers, the
technique will allow researchers to develop more sophisticated
models that address water quality questions. For example, the
researchers are using data they collected using the new
technique to determine the extent to which fertilizer runoff
contributes to water pollution in specific water bodies and the
role of wetlands in mitigating the effect of the runoff.
To collect water quality data, the researchers used existing
technology called "UV-Vis" spectrometers, which are devices that
measure the wavelengths of light absorbed by water. The upside
to these devices is that they can collect data as often as every
15 seconds and over long periods of time. This is far more
frequent than is possible with conventional water sampling and
lab analysis techniques. The downside is that they are designed
to monitor only a handful of key water quality parameters:
nitrates, dissolved organic carbon and turbidity, or how clear
the water is.
But the NC State research team developed a technique that
uses a suite of algorithms to significantly expand the amount of
information that can be retrieved from the spectroscopy data
collected by UV-Vis devices. Specifically, the new technique
allows researchers to get information on the levels of organic
nitrogen, phosphates, total phosphorus and salinity of the
water. This water quality data can offer key insights to a host
of questions, including questions about nutrient pollution.
The researchers tested the new technique in a restored
brackish marsh that experiences approximately 70 centimeters of
tidal variation – and a salinity that can vary from freshwater
to saltwater within minutes when the tide turns.
"We found that the automated results using our technique were
comparable to the results we obtained by testing water samples
in the lab," Birgand says. "So we gain a lot in terms of
monitoring frequency, without sacrificing accuracy."
The paper, "Using in situ ultraviolet-visual spectroscopy to
measure nitrogen, carbon, phosphorus, and suspended solids
concentrations at a high frequency in a brackish tidal marsh,"
is published online in Limnology and Oceanography: Methods. Lead
author is former NC State Ph.D. student Randall Etheridge.
Co-authors are Birgand; Dr. Jason Osborne, an associate
professor of statistics at NC State; Dr. Christopher Osburn, an
assistant professor of marine, earth and atmospheric sciences at
NC State; Dr. Michael Burchell, an associate professor of
biological and agricultural engineering at NC State; and Justin
Irving of s::can Measuring Systems.
[to top of second column]
The work was supported by National Science Foundation grant
DGE-0750733, U.S. Environmental Protection Agency grant EPA 2871 and
the North Carolina Water Resources Research Institute.
A copy of the study abstract follows:
"Using in situ ultraviolet-visual spectroscopy to measure
nitrogen, carbon, phosphorus, and suspended solids concentrations at
a high frequency in a brackish tidal marsh"
Authors: Randall Etheridge, François Birgand, Jason A. Osborne,
Christopher L. Osburn, and Michael R. Burchell II, North Carolina
State University; Justin Irving, s::can Measuring Systems
Published: online March 2014, Limnology and Oceanography: Methods
Abstract: The collection of high frequency water quality data are
key to making the next leap in hydrological and biogeochemical
sciences. Commercially available in situ ultraviolet-visual (UV-Vis)
spectrometers make possible the long-term collection of absorption
spectra multiple times per hour. This technology has proven useful
for measuring nitrate, dissolved organic carbon, and total suspended
solids in many environments, but has not been tested in tidal marsh
conditions where upstream freshwater mixes with estuarine waters,
resulting in rapid changes in concentrations and salinity. These
three parameters encompass only a portion of the nutrients that are
of interest in these systems. To test the potential of spectroscopy
to measure these and other nutrient concentrations, spectrometers
were installed in a constructed brackish tidal marsh and absorbance
spectra were compared to lab analyses for coinciding discrete
samples. Variable selection techniques, including partial least
squares regression, lasso regression, and stepwise regression, were
used to develop models with which nitrate, total kjeldahl nitrogen,
dissolved organic carbon, phosphate, total phosphorus, total
suspended solids, and salinity in brackish marsh waters can be
predicted from UV-Vis spectrometer measurements. Significant
relationships between the absorption spectra and the laboratory
measured concentrations were observed for all of the parameters.
Phosphate and total phosphorus were the only nutrients which had R²
values less than 0.86 for their best calibrations. This study shows
the potential to collect multiple water quality parameters at a high
frequency in brackish waters using in situ spectrometers and gives
the tools to replicate this analysis in all environments.
news release received from
North Carolina State]