A
Proposed Surface Water Quality Monitoring
Network for the Lake Michigan Watershed
Introduction
Water-quality
monitoring in the Lake Michigan Watershed is conducted by many entities,
including federal, state, local, and tribal governments, academia, the private
sector, non-profit organizations, and volunteers. There currently is little coordination of monitoring efforts or
structured information exchange among these entities. The recognition of this situation, and a desire to improve
individual and collective monitoring efforts, led to the formation of the Lake
Michigan Monitoring Coordination Committee (LMMCC).
One of the top
priorities of the LMMCC is to provide guidance and recommendations for
coordinated and comprehensive monitoring networks for the Lake Michigan
Watershed. As a pilot or test case for
the network approach, this proposal suggests guidelines for a surface water
monitoring network. This example
focuses only on water sampling and analysis from Lake Michigan
tributaries. It does not address other
important monitoring activities such as contaminants in fish, sediments,
biological integrity, habitat, and air deposition. The LMMCC envisions that these components will be added to this
framework in the future as time and resources permit.
The primary goal
of the coordinated water-quality network is to monitor tributaries to Lake
Michigan to determine spatial and temporal trends of selected parameters. In addition to the primary goal, there are
three secondary goals of the network:
1. Determine whether individual rivers are
meeting water-quality standards;
2. Detect new, emerging chemicals/problems;
and
3. Determine environmental and anthropogenic
sources of chemicals.
It is understood
that individual agencies may have additional goals and objectives for water
quality monitoring. The network
proposed here is meant to provide a basic framework. It can be tailored to meet the individual needs and requirements
of monitoring organizations.
The tributaries
(integrators) selected for regular assessment should be those that are
currently known to contribute the greatest flow and/or contaminant levels into
Lake Michigan. The network also should
include some larger tributaries that represent “reference” conditions (e.g.
Pere Marquette and Escanaba).
Specifically, these would include the following rivers:
·
Grand
Calumet
·
St. Joseph
·
Kalamazoo
·
Grand
·
Muskegon
·
Pere
Marquette
·
Manistee
·
Manistique
·
Escanaba
·
Milwaukee
·
Sheboygan
·
Fox
·
Menominee
Smaller
tributaries (integrator and indicator) should be sampled at a lower frequency
(perhaps monthly?) as “spot-checks”, as resources allow. The individual monitoring entities should
identify these smaller tributaries based on their program needs and knowledge
of the waters.
Additional
upstream indicator basin tributaries should be selected to provide a reasonable
assessment of the effects of environmental and anthropogenic features that
influence water quality. This probably
would be done by the individual entities, but could be done in a coordinated
fashion.
Scheduled
monthly samples, collected regardless of flow conditions at the time of
sampling, should be collected from each site to measure chemical loadings. Additional samples may be collected during
high-flow events to provide information on maximum concentrations. High-flow events are defined as stream flow
at or above the 20% exceedance flow or an increase in stream flow of 100% above
the preceding base flow.
The sites should
be assessed annually, although the level of effort will vary. The LMMCC recommends that sites be sampled
once every five years for the purpose of calculating contaminant loadings to
Lake Michigan. In all other years, the
sites will be sampled to measure average concentrations of selected
parameters. This will require the
collection and analysis of six samples from each location. Dates will be pre-scheduled and distributed
throughout the open-water season, with samples to be collected regardless of
flow conditions at the time of sampling.
At a minimum,
the water samples should be analyzed for the following parameters:
Total/Ortho
phosphorus Temperature Total suspended sediment
Chloride Mercury Copper
Nitrogen
((N)2+NO3, KJ, NH4)
Field parameters
– pH, dissolved oxygen, alkalinity, flow (continuous), chlorophyll
Other parameters
should also be considered, particularly total PCBs and herbicides. These analyses tend to be expensive, and
therefore they likely will be performed less frequently based on the
availability of adequate funding. To
detect emerging problems, a full scan of organic and inorganic parameters (e.g.
metals, conventionals, herbicides/pesticides, base neutrals, BTEX, other
organics) should be conducted on one sample from each location. Other parameters can be added as necessary,
including pharmaceuticals, E. coli,
and caffeine.
For each
parameter or class of parameters, all participating organizations should agree
on a collection method, particularly whether to use continuous, grab, or
temporally and/or spatially composited samples for analysis.
Methods used for
the analysis of metals and the selected organic contaminants should be able to
detect these parameters at the low levels in which they typically occur in
surface waters. Sample collection for mercury and trace metals should use clean
sampling techniques to prevent any potential contamination. For mercury analysis, USEPA Method 1631,
with a detection level of 0.5 ng/l, should be used. For the trace metals, ICP-MS is the most widely used analytical
method. Other methods that may be used
for metals analysis should have demonstrated comparability to USEPA Method 1631
and ICP-MS.
If it is decided
that other contaminants should be added to the network, such as PCBs,
pesticides/herbicides, pharmaceuticals, etc., then the LMMCC should identify
preferred analytical methods with which other proposed methods must be
comparable.
Methods
comparability tests should be developed and implemented to verify comparability
where differing field or laboratory methods are being used.
A Quality
Assurance Project Plan (QAPP) should be prepared for the coordinated monitoring
effort as part of the network design process.
As a general guideline, a minimum of 15% of the samples should be
devoted to quality assurance. These
samples should include trip blanks, field blanks and replicates, lab blanks and
duplicates, and analyte spikes. The
QAPP also must address issues such as data quality objectives, chain-of-custody,
data verification, and data validation. In addition, the laboratories that
analyze the samples must have approved procedures in place to ensure the
quality of their analyses.
A coordinated
database should be developed. The
following metadata should be collected and made available for each sample
included in the database:
·
Location
info (ID#, station type, sample media, lat/long, waterbody)
·
Date and
time
·
Reason for
sample collection
·
Sample
description (collection method, sample type)
·
Sample
analyses (preservation, method, detection level, regulatory reporting level,
precision, and accuracy)
·
Data source
(data owner, sampling entity, laboratory)
There are
several tests available to measure temporal trends. Linear regression is one method.
It requires several data assumptions to be met, including constant
variance, data independence, and normal distribution. If any of these assumptions are not met, then a nonparametric
method should be used. The Mann-Kendall
Tau test is one nonparametric procedure for analyzing trends. This method determines whether more
increases or decreases in concentration occur over a specified time period than
would be expected by chance. If
concentrations are not changing over time, the levels would increase and
decrease at about the same frequency.
Another nonparametric method is the Seasonal Kendall Test. This test examines seasonal differences in
water quality, as data in one season are compared only to data in the same
season in later years. Thus, variations
in concentrations throughout the year do not add to data variability, which
must be overcome before a trend can be discerned.
It is well known
that flow variability often has an effect on contaminant levels found in rivers
and streams. Load and concentration
data should be adjusted for flow by plotting the residual concentrations versus
time, where the residual concentration is the difference between the observed
concentration and the expected concentration based on the flow at the time of
sampling. We can then evaluate the
flow-adjusted data for trends.
Organizations
likely will use other data analysis methods to meet specific needs, including
source identification, attainment of water quality standards, and evaluating program
effectiveness.
The use of a
Geographic Information System (GIS) would be very valuable for the spatial
display and analysis of data.
Prior to sample
collection, two reports should be prepared.
The first is a study design report, which describes the goals, network
design, and sampling and analytical methods.
The second is the QAPP. Both of
these are one-time reports, although minor updates may be necessary as
modifications and refinements are made to the network.
The data
generated from this network should be formally summarized in a report once
every five years. This corresponds to
the recommended 5-year interval for calculating tributary loads. The report would be distributed
independently and included in future updates of the Lake Michigan Lakewide
Management Plan. A 5-year cycle makes
sense for several reasons. First,
report preparation at a higher frequency could become very resource intensive. Second, the primary goal of the network is
to assess trends, which requires several years of data to be meaningful. It does not make sense to report on trends
at a greater frequency. Of course,
individual federal and state agencies, non-profits, local governments, and
other stakeholders can use the data for their own needs and reporting
requirements, at whatever frequency they deem necessary (e.g. some states may
want to prepare an annual report on data from their tributaries). Therefore, the data will be used as it is
generated, but the formal report only will be prepared every 5 years. Responsibility for this report will have to
be determined in future discussions.
This example is
meant to serve as a starting point for discussion. We encourage LMMCC members to raise questions and identify key
issues as this proposal is developed in more detail. To stimulate discussion, some initial questions are provided for
consideration.
1. Should additional tributaries be a part
of this network? When selecting
tributaries, are there any criteria besides flow and expected contaminant
levels that should be considered? What
about sampling reference sites?
2. Is the proposed sampling frequency
sufficient? Are 12 samples sufficient
to calculate contaminant loads? Are 6
samples enough to reasonably estimate mean concentrations?
3. Are there additional parameters that
should be added to the core list identified above? How do we design a sampling program to identify new contaminants?
4. A key premise of the LMMCC is that
identical sampling and analytical methods are not necessary for data generated
by different entities to be comparable.
What level of effort is necessary to establish comparability, and how
should such studies be designed?
5. Are there additional meta data items that
should be added to the list presented above?
Can STORET be used as the common database, or should a new one be
developed? If we decide a new one is
needed, who will develop it, input data, and maintain it?
6. We proposed that a formal report be
prepared every five years. Should a
report be prepared more frequently? Who
will be responsible for producing the report?
Who will be responsible for developing and maintaining the GIS
component?
7. What is the role of volunteers in this network? What level of training and quality assurance is required to ensure that volunteer data (and agency data for that matter) are defensible?