I BRAF1
6258
I 001D86100
I
I
I
I
I
I
JOINT EVALUATION OF
UPSTREAM/DOWNSTREAM NIAGARA RIVER MONITORING DATA,
1984-1985
Prepared by
Data Interpretation Committee
May 15, 1986
-------
I
I
t
I
I
I
1
I
I
I
I
I
I
1
I
I
I
t
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
OFFICE OF
WATER
Dr. Abdel H. El-Shaarawi
Chairman, Data Interpretation Subcommittee
for the Niagara River Monitoring Plan
Canadian Centre for Inland Waters
867 Lakeshore Road
Burlington, Ontario L7R4A6
Canada
Dear Abdel:
On behalf of Rollie Hemmett of the U.S. Environmental
Protection Agency, Region II, Simon Litten of the New York
Department of Environmental Conservation, and other subcommittee
members, I am submitting to you the enclosed draft report for
consideration by the subcommittee.
The draft report, titled, "Joint Evaluation of Upstream/
Downstream Niagara River Monitoring Data", focuses on background
information, assumptions inherent in the monitoring network
design, data reliability, descriptive statistics for the monitor-
ing data, statistical analysis methods, and results. On the
issue of data reliability, we are in dire need of feed-back
from the analytical subcommittee.
Each subcommittee member south of the border has seen a
version of this draft report. Although there seems to be
agreement on the general thrust of the draft document, none of
our members has fully endorsed the draft; each of them reserves
the right to comment. Our mutual goal is to produce a product
that can be accepted by the subcommittee as a group for submission
to the management committee for its consideration.
-------
I
i
I
I
f
I
I
I
I
i
i
I
f
i
i
i
I
I
I
-2-
The U.S. Members of the Niagara River Monitoring Plan
management committee, accordingly, have not commented
on the draft. They await a joint, consolidated submission by
the subcommittee.
I recommend you convene a subcommittee meeting in the
near term to resolve any differences. I look forward to seeing
you then.
Sincerely,
Maurice E. B. Owens, Ph.D
Chief, Program Evaluation Branch
Office of Water Regulations and
Standards (WH-586)
Enclosure
cc: Rollie Hemmett, USEPA, Region II - 2 copies
Simon Litten, NY, DEC - 2 copies
Dave Dolan, USEPA, ORD - 1 copy
Peter Wise USEPA, GLNPO, Region V - 1 copy
-------
i
i
i
i
t
i
i
i
i
i
i
i
I
i
i
i
I
f
i
TABLE OF CONTENTS
Page
CHAPTER
1 EXECUTIVE SUMMARY 1-1
2 INTRODUCTION . 2-1
2.1 THE PURPOSE 2-1
2.2 THE RIVER 2-3
2.3 THE SAMPLING SYSTEM 2-7
2.4 THE MEDIA 2-14
3 GENERAL ANALYTIC FRAMEWORK 3-1
3.1 ENVIRONMENTAL MODELING 3-1
3.2 CONCENTRATION/FLOW REQUIREMENTS 3-4
3.3 DATA QUALITY 3-7
4 DATA AND STATISTICAL ANALYSIS 4-1
4.1 SAMPLING PERIOD 4-1
4.2 DATA AND DETECTION LIMITS 4-3
4.3 SEASONALITY 4-11
4.4 ANALYSIS 4-25
4.4.1 Calculating Mean Concentrations 4-25
4.4.2 The Screening Test 4-36
4.4.3 Application of Screening Test ..... 4-40
4.4.4 Loadings Estimates 4-42
REFERENCES
CHAPTER 2 2-20
CHAPTER 4 4-53
APPENDICES
A DETECTION LIMITS A.-1
B CALCULATION OF THE CONFIDENCE INTERVALS B-l
C SCREENING TEST RESULTS C-l
D DIFFERENTIAL LOADINGS D-l
-------
I
1
i
1
t
I
I
1
i
I
i
i
i
I
i
i
i
t
i
LIST OF FIGURES
Page
2-1 Primary Features of Niagara River 2-5
2-2 2-6
2-3 2-8
2-4 Scale Distorted Map of Potential Sources of Toxic
Substances to the Niagara River ...... ... 2-9
2-5 Locations of IWD Niagara River Sampling Stations 2-11
2-6 Niagara-on-the-Lake Intake Assembly ..... 2-12
2-7 2-16
2-8 2-17
4-1 Sampling Period 4-2
4-2 Plot of Percentage of Nondetects and Below Detection
Limits for All Pollutant Classes for the Water Phase .... 4-9
4-3 Plot of Percentage of Nondetects and Below Detection
Limits for All Pollutant Classes for the Suspended
Sediment Phase 4-10
4-4 Plot of Flow (L/Day in Billions) at Fort Erie 4-15
4-5 Plot of Flow (L/Day in Billions) at Niagara-on-the-Lake . . . 4-16
4-6 Plot of Pentachlorodibenzo Furan in the Water Phase
Versus Time 4-17
4-7 Plot of Dichloro in the Water Phase Versus Time 4-18
4-8 Plot of DDT in the Water Phase Versus Time 4-19
4-9 Sediment Concentration (mg/1) Versus Time 4-20
4-10 Plot of Octachlorodibenzo-P-Dioxin in the Sediment
Phase Versus Time 4-21
4-11 Plot of Phenathrene in the Sediment Phase Versus Time .... 4-22
4-12 Plot of Lead in the Sediment Phase Versus Time 4-23
ii
-------
I
*
1
i
I
1
1
I
1
I
I
4
i
f
I
1
i
f
I
I
LIST OF FIGURES (Continued)
Page
4-13 Scatter Plots of Flow Versus Concentration for Four
Pollutants 4-24
4-14 Schematic Diagram of Confidence Bounds for Mean Daily
Concentrations of Some Parameters Sampled at Fort Erie . . . 4-35
LIST OF TABLES
2-1 Metal Concentrations in Niagara Gorge Native
Geological Units 2-19
4-1 Pollutant (Classes) Sampled 4-4
4-2 Total Number of Samples by Class and Phase 4-6
4-3 Percent of Nondetects by Station and Phase 4-7
4-4 Dependency of Nondetects at FE and NOTL 4-8
4-5 Mean Daily Concentrations 4-27
4-6 Differential Concentrations by Phase 4-41
4-7 Summary of Estimated Differential Loadings 4-43
4-8 Summary of Estimated Loadings by Pollutant Class (FE) . . . 4-47
iii
-------
I
I
i
i
t
t
I
t
i
i
f
i
f
i
i
I
i
t
i
1. EXECUTIVE SUMMARY
The Data Interpretation Subcommittee examined historical upstream/
downstream Niagara River monitoring data collected by Environment Canada's
Water Quality Branch-Ontario Region for the period December 12, 1984, to
October 2, 1985. The Subcommittee offers the following conclusions:
1. The adequacy of the FE and NOTL sampling stations is open to
question. The FE station, which is intended to capture the
influent contaminant load from Lake Erie, is suspect of missing
that portion of the load originating in Lackawana and Buffalo.
To some extent, these losses can be reconciled by redrawing
the beginning of the river from a point immediately down-
stream of the FE intake and immediately upstream (south)
of the mouth of Smokes Creek. This solution, however,
ignores the possibility of a much larger scale plume that
may come from far beyond the Buffalo area.
2. Because of the unusual nature of the sample collection
procedure and the very large sample volumes, some standard
quality control procedures could not be performed. Other
aspects of the quality control program were found to be poorly
documented, and conclusions about the adequacy or inadequacy
of the laboratory work cannot be made. This is especially
serious because of the extremely low concentrations reported.
3. Despite grave uncertainties about data quality, the Subcom-
mittee proceeded with a two stage statistical analysis. The
first stage used a nonparametric test, the Wilcoxon signed
rank procedure, to identify substances whose concentrations
increased between FE and NOTL. This statistical analysis found
33 chemicals in the sediment phases, water phase or both. For
those substances that increased, the Student's t was used to
estimate confidence intervals in loadings. This step showed
another seven chemicals having confidence intervals including
zero; these chemicals were accordingly dropped from further
consideration. The 26 substances shown to increase are:
• Octachlorodibenzo-furan (suspended sediment only)
• Arsenic (both media)
1-1
-------
1
I
i
t
i
i
i
1
i
i
i
i
i
i
i
I
i
t
i
l'
i
• Chromium (both media)
• Mercury (suspended sediment only)
• Zinc (both media)
• Aluminum, extractable (water only)
• Iron (water only)
• Manganese (water only)
• Nickel (water only)
• Lindane (suspended sediment only)
• 1,2,3-trichlorobenzene (suspended sediment only)
• 1,2,3,4-tetrachlorobenzene (both media)
• Mirex (suspended sediment only)
• PCBs (suspended sediment only)
• Pentachlorobenzene (both media)
• Hexachlorobenzene (suspended sediment only)
• Acenaphthene (suspended sediment only)
• Benzanthracene (suspended sediment only)
• Benzo(a)pyrene (suspended sediment only)
• Benzo(b and k)fluoranthene (suspended sediment only)
• Chrysene (suspended sediment only)
• Fluoranthene (suspended sediment only)
• Fluorene (suspended sediment only)
• Ideno(l,2,3-cd)pyrene (water only)
• Phenanthrene (both media)
• Pyrene (suspended sediment only).
4. The Subcommittee has selected the paired Wilcoxon(or signed rank)
statistical test as the appropriate procedure to determine whether
individual pollutant concentrations or loads increase over the
course of the Niagara River. Although there is general agreement
that the basic test should be a nonparametric one, the issue of
which nonparametric test should be used has proved to be quite
controversial. Other candidate nonparametric statistical test
include the paired sign test and a 2 by 2 contingency table
1-2
-------
1
I
I
I
I
1
I
1
I
I
i
I
1
1
1
I
1
f
I
DRAFT
analysis to test relationships between the number of detections at
the two stations.
Technically, each of these three procedures is used to test a
different hypothesis, but each such hypothesis is related to
detecting change between the measurements at the two stations.
The null or no change hypothesis tested by the Wilcoxon statistic
is that the distribution of the differential concentrations(or
loads) for an individual pollutant is symmetric about zero.
This hypothesis is equivalent to the statement that the measure-
ments for a given pollutant at the two stations behave in the
same statistical manner. The Wilcoxon statistic, accordingly,
incorporates the sign as well as the magnitude of each differen-
tial measurement. By contrast, the sign statistic is used to
test the null hypothesis that the median of the differential
measurements is zero, and the sign statistic depends only on
the sign of the differential. The Wilcoxon test hypothesis,
therefore, was judged to be the appropriate for the Niagara
River application.
A technical difficultly arises in using the Wilcoxon test with
the Niagara River data. The difficulty arises when within a
measurement pair the concentration for one station is a non-
detect and that for the other station is a quantified detection.
In this case the differential is not well defined. This
difficulty can be resolved operationally by setting nondetects
equal to a specific value, say zero. The statistical level of
the test is not changed by this procedure, but the sensitivity
(or power) of the test can be affected together with the
results of a particular outcome. The effect of using this
procedure on the results is judged to be minor.
The actual test procedure has been implemented as follows:
First the Wilcoxon test is applied to the concentration by
media by pollutant. For a given pollutant, denote the differen-
tial concentration (NOTL minus FE) by Di with nondetects set
to zero. We test the null hypothesis
HQ: no change or distribution of D
symmetric about zero
against the one-sided alternative hypothesis
H^: increased concentration or
distribution shifted or skewed
to positive values
1-3
-------
1
I
I
I
I
1
I
1
I
I
I
1
1
I
1
i
i
I
i
at the 95% level. Discard all Ds which are equal to zero, and
denote by n the number of remaining Di. Rank from 1 to n the
absolute value of those Di in ascending order. If a set of Di
have the same absolute value, replace the ranks of the Di in the
set by the average of ranks for the set. Then let T+ be the sum
of ranks corresponding to positive Di. Then compute
T+ - n(n+l)/4 - 0.5
1/2
[n(n+l)(2n+l)/24]
and accept H^:(increase) at the 95% level if
z > 1.645.
For those pollutants within media determined to show a statistically
significant increase in concentration by the Wilcoxon test, a point
estimate of mean daily differential mass loading was computed. A
90% confidence interval based on the student's t-distribution is
also provided. Whether on not the confidence interval includes zero
is equivalent to testing the no increase hypothesis at the 95% level
based on the t-distribution. If a confidence interval for a pol-
lutant includes zero, the finding of increased loading, therefore,
is in question.
1-4
-------
I
I
I
I
I
I
I
1
1
I
I
I
1
I
1
I
1
f
I
2. INTRODUCTION
2.1 THE PURPOSE
The United States and Canada have been concerned about toxic pollutants
in the Niagara River for more than a decade. These pollutants, which origi-
nate from industrial and municipal discharges as well as hazardous waste
disposal sites, have the potential to compromise the quality of biological
life in the Niagara River and the Lake Ontario basins. The mutual concern of
both Canadian and United States environmental agencies resulted in a decision
to study the contaminants entering the Niagara River. That decision led to
the formation in February 1981 of the Niagara River Toxics Committee (NRTC),
a binational group of representatives from Environment Canada, Ontario's
Ministry of the Environment (MOE), the United States Environmental Protection
Agency (EPA), and New York's Department of Environmental Conservation (DEC).
The NRTC assumed the tasks of identifying sources of toxic pollutants entering
the Niagara River and recommending long-term monitoring programs for the
Niagara that would allow evaluation of the effectiveness of control programs.
In 1984, NRTC proposed an ambient large volume water and suspended
sediment monitoring program. The NRTC specified three objectives(l):
1. To determine the trends in the concentrations of specific
chemicals in the water and suspended sediments of the
Niagara River.
2. To compare the measured concentrations of chemicals of
concern to given ambient criteria, standards, guidelines,
and objectives, and to report on significant exceedences
that require agency action.
3. To estimate the relative differences in the loadings of
chemicals of concern between the river's source and mouth.
2-1
-------
I
I
I
I
I
I
1
1
I
I
I
I
1
I
1
I
I
I
I
The NRTC suggested that two sites be sampled, a downstream station at
Niagara-on-the-Lake (NOTL) and an upstream station at Fort Erie (FE). In a
discussion of the anticipated outcome of such a project, the NRTC wrote(l):
The results of the large volume water and suspended sediment
sampling can be assessed by both parametric and non-parametric
statistics. Work on this project suggests that for most
substances, the detection limits may not be sufficient to
allow the use of parametric statistics. Therefore, procedures
will have to be worked out to analyze trends in terms of non-
parametric statistics.
As yet there is no final agreement on the methods for determining trends
in upstream/downstream differences (objective #1), and statistically explicit
definitions of exceedence of criteria, standards, etc., are also unavailable
(objective #2). Agreement on statistical methods for detecting upstream/
downstream differences was reached (objective #3).
The jurisdictions have accepted a test which looks at the sign (+ if
greater at NOTL, - if greater at FE) of the upstream/downstream difference
in loadings. This statistical sign test rests on the fewest assumptions
about the quality of the data. All participants in the monitoring process
would agree to make the fullest possible use of the available ambient moni-
toring information, but the sign test may be overly cautious for optimal
evaluation of some substances.
More recently a new binational group, the Four Party U.S.-Canada Niagara
River Toxics Management Plan has taken up the monitoring issue. The objec-
tives of the new Plan call for the parties to:
• Determine toxic chemical loadings to the Niagara River
from Lake Erie (input)
2-2
-------
I
I
1
1
1
I
I
1
I
I
I
I
1
I
1
I
1
I
I
• Determine total toxic chemical loading from the Niagara
River to Lake Ontario (output)
• Determine toxic chemical loadings from sources along the
Niagara River by comparing the differences between the output
from the river and input to the river from upstream sources
(input-output differential monitoring identified by the NRTC).
As can be seen from the preceding discussion, the NRTC was skeptical
that such arithmetic loading operations could be carried out with the
quality of data then available. The Four Party group established three
technical groups that are charged with developing sampling, analytical, and
interpretive procedures sufficient to meet these new objectives by modifying
the water and suspended sediment monitoring program at FE and NOTL. The
Data Interpretation Group has the further charge of examining existing FE/
NOTL ambient data and accompanying quality control information and reaching
a joint interpretation.
This report by the Data Interpretation Group will extract as much
information as possible. Considerations of sampling design, chemical
analysis, laboratory quality control, and distribution of resultant data
(e.g., proportion of samples with contaminant concentrations below the
detection limit) will be made in selecting appropriate statistical methods.
To the extent that future monitoring will produce data similar to those
discussed here, this report can also give insights into the interpretation
of new observations. Decisions on interpretive methods are best made prior
to sampling so that the optimal sampling strategies can be used and so that
all parties have realistic expectations about the outcome.
2.2 THE RIVER
The Niagara River is, by virtue of the falls, a nonnavigable 37-mile
strait flowing from Lake Erie northward to Lake Ontario. The Niagara provides
84 percent of the input water supply to Lake Ontario, which is the source of
2-3
-------
I
I
I
I
I
I
I
I
I
I
I
I
1
I
1
I
1
I
I
drinking water for the majority of the Canadian population. As a barrier
to navigation, the falls have served historically to concentrate industry,
particularly iron, steel, and coke manufacture, flour milling, and oil
refining, at a geographical position where transshipment becomes necessary.
Hydropower from the falls also encouraged the American birth of electro-
chemical production. Both the industrial and attendant population concen-
trations were not achieved without environmental costs.
Figure 2-1 shows the primary features of the Niagara River area.
Illustrated is a large central island (Grand Island) that splits the river
into the Chippawa Channel (57 percent of the flow) and the Tonawanda Channel
(43 percent of the flow). The mean influent flow of the Niagara over the
past 124 years is 204,000 cubic feet per second (cfs). Unlike smaller
streams, seasonal flow differences for the Niagara, which has a drainage
area of 264,000 square miles, are small. Using the mean daily flows by
month for the last 13 years, we see that the ratio of the difference between
the highest flow months (209,000 cfs for May and June) and the lowest
(189,000 cfs in January) to the mean for the 13 years was only 10 percent.
This may be compared with a more typical river of that latitude, the Genesee.
The Genesee has a drainage area of 2,467 square miles and a mean flow of
2,813 cfs. The ratio of the difference between its highest flow month
(6,580 cfs in April) and its lowest (836 cfs in August) to the mean is
192 percent(2). Figure 2-2 gives a 5-year hydrograph (1976 to 1982) for the
Niagara. To the extent that flow alone affects contaminant concentration and
loading, seasonal effects due to flow are expected to be small.
Power intakes for the Robert Moses reservoir, on the U.S. side of the
river, and the Sir Adam Beck reservoir, on the Canadian side, have maximum
capacities of 110,000 and 65,000 cfs, respectively. These intakes divert
water around the falls in accordance with the 1950 Niagara River Treaty
specifications of at least 100,000 cfs over the falls during daylight hours
of the tourist season and at least 50,000 cfs otherwise. The power diversions
2-4
-------
LAKE ONTARIO
Occasional shape
of p/ume under
southeasterly \
Usual shape of plume under
predominant northwest winds
of time)
Toronto-£4K£ ONTARIO
Figure 2-1. Primary Features of Niagara River
2-5
-------
1 DRAFT
I
I
I
I
I
I
1
•
I
t
I
1
I
1
I
^ Figure 2-2.
l
-------
I
I
I
I
I
I
I
I
I
I
f
I
1
I
1
I
1
I
I
are significant in that time of travel, nominally 18 hours, is greatly
complicated. Figure 2-3 shows the effect of hydroelectric discharges on
lower river flows. The significance of the effect is that water sampled at
NOTL is a mixture of waters that passed FE at different times so that
comparisons of a single parcel of water before and after it passed through
the river are not possible. Furthermore, the Robert Moses intake removes
water from the eastern bank. This portion of the river probably contains a
large proportion of the contaminants added to the river.
Over 100 significant industrial and municipal discharges occur along
the river or its immediate tributaries. Other discharges enter the river
from storm sewers. Sixty-six hazardous waste landfills have a significant
potential for contaminant migration to the Niagara River. Figure 2-4 is a
scale-distorted map indicating potential sources of toxic substances to the
river. Because of the nature and practices of the manufacturing activities
in the Niagara River area, the discharges and landfills contain toxic sub-
stances. Toxic substances of concern in the Niagara include polycyclic
aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs), chlorinated
benzenes, chlorinated dibenzodioxins and dibenzofurans, metals, and phenols.
Some of these substances cause toxic effects in the local biota and some
bioaccumulate to the point where they may threaten the well-being of wild-
life and humans consuming fish from the Niagara and Lake Ontario.
2.3 THE SAMPLING SYSTEM
The Water Quality Branch-Ontario Region (WQB) of Environment Canada
established an ambient monitoring system in 1984. This system consists of
two trailers located at FE, the upstream station, and NOTL, the downstream
station. Submersible pumps bring water from the intakes placed out in the
river and off the bed up into the trailers where a coarse filter (1-mm pore
size) removes macroalgae and other large debris. The coarsely filtered
water is then led into a model KA-02-06-075 Westfalia continuous centrifuge,
where positive density particles (suspended sediments) are removed at
2-7
-------
Figure 2-3.
2-8
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
1
I
1
I
I
NIAGARA FALLS WPCP
CVANAMIO
(NIAGARA FALLS!
CYANAMIO
(WtLLANOl
(TIIL
OONNCW'HANNA COKI
Figure 2-4. Scale Distorted Map of Potential Sources of Toxic
Substances to the Niagara River
2-9
-------
I
I
I
I
I
I
I
\
I
I
I
I
I
I
I
I
I
I
1
IMF
9,500 G. Previous studies from other water bodies (Allen, 1979) determined
that the separator is 90-95 percent efficient when operated at a flow
rate of 6 L/min(3). K. Kuntz reported mean recoveries of 81 percent and
83 percent for particulate nitrogen and particulate carbon, respectively
(n = 11), from Niagara River water(4). Research supported by DEC is cur-
rently examining the size distributions of particles in raw and centrifuged
water samples. Effluent water is collected every 2 weeks as a grab sample,
and suspended sediments are removed from the continuous centrifuge at the
same time. The centrifuge is run for 24 hours to collect enough mass for
analysis, 5 to 50 grams from almost 9,000 L. Water and suspended sediments
are treated separately to maximize extraction efficiencies in the different
media except that volatile organics were analyzed only from centrifuged water
and aqueous metal analyses were done on uncentrifuged water. Figure 2-5
shows the locations of the two sampling stations, and Figure 2-6 illustrates
the sampling train at NOTL.
The choice of sampling stations was dictated by appropriateness and con-
venience. There is no single best site. Sampling carried out by Environment
Canada and Ontario MOE (reported in NRTC, 1984) indicated apparently uniform
chemical concentrations from samples taken along a transect across the source
of the Niagara. Concentrations were highest on the eastern (New York) side,
but high mercury levels were also found on the western (Ontario) side. FE
samples miss a turbidity plume that hugs the river's east bank which, from
inspection of aerial photographs, seems to come from the Buffalo River and,
further into Lake Erie, Smokes Creek. The sampling site itself is slightly
downstream from the City of Buffalo's principal sewage treatment plant.
Because of the high velocity of the upper river, it is thought that crossover
of east bank water is unlikely. Some of the difficulties in accepting FE
observations as representing influent contaminant loads could be resolved
by redefining the river mouth as a line running from immediately below, or
north of, the FE intake to immediately above, or south of, the mouth of Smokes
Creek. This redefinition, however, ignores the possibility of a much larger
scale shore hugging plume extending as far as the mouth of the Detroit
River. The existence of such a plume is suggested by a hydraulic model of
2-10
-------
I
I
I
t
I
I
I
I
I
I
I
I
1
I
I
I
1
I
I
DRAFT
Niagara-On-The-Lake (NOTL)
Distance from Shore
Distance from Bottom
Distance from Surface
Flow Rate
200 ft
18 ft
30 ft
6 L/min
Fort Erie (FE)
Distance from Shore 100 ft
Distance from Bottom 6 ft
Distance from Surface 6 ft
Flow Rate 6 L/min
Figure 2-5. Locations of IWD Niagara River Sampling Stations
2-11
-------
I
I
I
I
I
1
I
I
2-12
fi-
U Is
V.8
,1
*i
I j; z§ £i> »
1 111
1} ii i S < *—p-
!!!i!l i
lilllM
ill
e
OJ
en
to
0)
^
(0
J-l
c
0)
4-J
I
o
«
u
(8
I
CSJ
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
Lake Erie(5). Unfortunately, the scale of the physical model is inadequate
to define the continuation of the plume at the FE station. For the time
being, we recognize the possibility that FE data underestimate loading
from Lake Erie.
The horizontal location of the NOTL station is less likely to be
biased than that of FE because of mixing by the falls and the distance in
a relatively quiescent stretch of the lower river from the outfalls of the
power dams. Chan (1977) addressed horizontal stratification by sampling
16 sites on three transects on 3 days during five sampling periods(6).
Major ions, conductivity, pH, and nutrients were analyzed. The data were
subjected to a three-factor analysis of variance test for temporal, cross-
sectional, and spatial effects where each sampling period was examined
separately. Chan concluded, "There is no lateral or crosssectional varia-
tion in the water chemistry at the mouth of the Niagara River." However,
the results of 64 cross-sectional tests indicate 19 significant (p < .05)
effects. Some substances (N02 and Si02) consistently display significant
cross-sectional effects and other substances do not. A reanalysis of Chan's
data where all sampling periods are lumped may eliminate the apparent cross-
sectional effects, but as of now there is no strong evidence for cross-
sectional homogeneity.
Considerations in the vertical placement of the sampling intakes are a
desire to avoid sampling the layer of sedimentary material carried along the
river's bed and the necessity to stay clear of boats. The success in avoid-
ing bed loads is unknown, but it should be pointed out that the suspended
sediment loading calculated from NOTL is usually greater than that from FE
even though the FE intake is nearer the river's bottom. The size and composi-
tion distribution of suspended sediments in the water column is not well known
and could be greatly influenced by seiches in both Lake Erie and Lake Ontario,
as well as operations of the Robert Moses and Sir Adam Beck power stations.
2-13
-------
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
The extent of these problems in sampling suspended sediments is mitigated by
the relative unimportance of suspended sediment loading for most contaminants.
The sampling intakes are 100 and 200 feet long at FE and NOTL, respec-
tively. The 25 feet of intake line nearest the trailers are heated to pre-
vent winter-time freezing. It can be assumed that, as with most surfaces
exposed to natural waters, a biofilm will grow within the intakes. This
postulated biofilm would be a likely sink for hydrophobic substances. It
will also occasionally slough off. This sloughing off may result in apparent
contaminant spikes.
Fifty-liter water samples were brought to Burlington, Ontario, for
processing in a specially designed extractor, the Aqueous Phase Liquid
Extractor (APLE). The APLE device sprays methylene chloride through a large
volume of water (50 L). Details on the operation of the APLE system are
given by McCrea and Fischer (1984)(7). After this mixing, the solvent was
evaporated down to five mL for Florisil clean-up and GC injection. Sediments
were extracted for organic contaminants using Soxhlet apparati, again with
methylene chloride. Trace metal analyses were performed using atomic absorp-
tion spectrometry except for arsenic and selenium, which were analyzed by
inductively coupled argon plasma spectrometry.
2.4 THE MEDIA
The IWD river monitoring program assumes that contaminants are car-
ried in two phases, as an aqueous solution and sorbed onto suspended sedi-
ments. Separate extraction procedures maximize recovery from each phase.
C. Yin (1985) and Hassett and Milicic (1985) suggested that naturally occur-
ring macromolecules derived from decomposition of plant material can impede
recoveries of certain hydrophobic substances like mirex and PCBs(8; 9).
Preliminary results of studies currently supported by DEC indicate that
oxidizable dissolved substances do affect the liquid/liquid recoveries of
2-14
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
1
I
I
I
I
PCBs, chlorinated benzenes, DDTs, and mirex in Niagara River waters. The
abundance of interfering macromolecules probably varies with season and
location in the river. Sewage treatment plants are believed to be important
sources. The two-phase model of contaminant distribution may not be comple-
tely sufficient and could result in underestimating the contaminant loading
to the river. It may also explain some of the very high variability observed
in contaminant concentrations. DEC's macromolecular interference study is
expected to be complete by March 1987.
Suspended sediments can originate from autochthanous production of
calcite and microalgae in Lake Erie, from wind-induced resuspension of bottom
sediments in eastern Lake Erie, from tributary loading to the river, and from
erosion at the face of Niagara Falls. Figure 2-7 relates wind vectors and
speeds at Buffalo International Airport to suspended sediment loading at FE.
Figure 2-8 relates precipitation events at Buffalo and Niagara Falls to
upstream/downstream suspended sediment differences and also presents the
relation between estimated flow from the Buffalo River (given as the sum of
flows from Cayuga Creek, Cazenovia Creek, and Buffalo Creek) and suspended
sediment differential.
The mean differential loading of suspended sediments was 2,481 tonnes
per day (3,966 and 6,447 tonnes per day at FE and NOTL, respectively).
If the falls are assumed to be receding at the rate of 1.5 feet per year
across a width of 3,200 feet and a 550-foot incline, then the suspended
sediment contributed by their erosion would account for 537 tonnes per day
or 8 percent of the NOTL material. The bedrock of the falls, measured
vertically, is dolostone and limestone, 172 feet (density of 2.7); shale,
145 feet (density of 2.6); and sandstone, 73 feet (density of 2.48). The
mean density of the face is about 2.62. This applies to 2.64 x 106 cuft
or to 200,000 L/day of eroded rock. Using the calculated mean density,
this would yield 537 tonnes/day. Three western New York State streams
(Cattaraugus, Tonawanda, and Genesee rivers) have a mean 5-year suspended
2-15
-------
6? /^
Figure 2-7.
-------
/ -f -
c «?
I
I
m Figure 2-8.
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
sediment concentration of 57 mg/L. If the mean flow addition to the Niagara
(9.84 x 10^ L/day) contained 57 mg/L suspended sediments, it would account
for another 561 or 9 percent of the NOTL total. This leaves 21 percent still
not accounted for and increases the suspicion of an east shore hugging plume
that misses the FE station.
Work supported by DEC is seeking to find the sources of suspended sedi-
ments to the Niagara using scanning electron microscopy and x-ray energy
spectroscopy to characterize particles by morphological and elemental
criteria. DEC workers are examining native rock units in the Niagara River
gorge for trace metals. Preliminary results, shown in table 2-1, indicate
that some of the easily erodable shales have concentrations of arsenic,
chromium, copper, lead, and zinc similar to those seen in industrially af-
fected sediments. These shales were deposited 420 million years ago. The
results of this study could demonstrate that the natural background load
to the Niagara of these metals is a substantial portion of the total differ-
ential increase.
2-18
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
DRAFT
Table 2-1. Metal Concentrations in Niagara Gorge
Native Geological Units
Depth Below
Robert Moses
Dam (ft) Unit**
Lockport As Cd
0 Dolostone
10
20
30
40
50
60
70
80
90
100
110
<2.0 <0.2
<2.0 <0.2
<2.0 <0.2
17.0 0.2
120 Rochester Shale
130
140
150
160
25.8 <0.2
15.0 <0.2
170 Irondequoit Limestone
180 Reynales Limestone
190 Neahga Shale <2.o <0.2
200 Thorold Sandstone
Metals
(mg/k«)
Cr Cu Pb Hg Ni Zn
<0.2 5.0 <2.0 2.8 <0.08 0.66 28.0
<0.2 5.2 <2.0 3.2 <0.08 0.98 40.0
<0.2 12.6 7.2 4.0 <0.08 1.7 18.2
0.22 28.0 7.4 10.4 <0.08 12.0 28.0
28.0 12.8 340.0 <0.08 10.8
52.8 340.0 10. 0 <0.08 7.0
17.0
4.8 <2.0 <0.08 3.2
34.0
72.0
12.6
**
***
Source: S. Litten Per Comm, 1986
Field specimens were identified by J. Garry and E. Zuk using curring logs
and the stintgiaphy published in: Johnson, R.H. 1964. Ground Water in
the Niagara Falls Area, New York, With Emphasis on the Water-Bearing
Characteristics of the Bedrock. USGS, State of New York Conservation
Department. Bulletin GW-52.
Rock specimens were crushed to a min particle size of 2 prior to acid
extraction.
2-19
-------
CHAPTER 2 REFERENCES
1. Niagara River Toxics Committee. October 1984. Report^ A Joint
Publication of New York State Department of Environmental
Conservation, Environment Canada, U.S. Environmental Protection
Agency, and Ontario Ministry of the Environment.
2. USGS Water Resources Reports, Western New York, 1966 to 1984.
3. Allen, R. 1979. Sediment Related Fluvial Transport of Contaminants:
Advances by 1979. Report No. 107. Inland Water Directorate; Western
and Northern Region.
4. Kuntz, K. Per Comm.
5. Rumer, R.R., Jr., K.M. Riser, A. Wake, and K.H. Yu. 1976. Hydraulic
Model Study of Lake Erie. Water Resources and Environmental
Engineering Research Report No. 76-1. Department of Civil
Engineering. State University New York/Buffalo.
6. Chan, C.H. 1977. Water Quality Surveys on the Niagara
River - 1974. Report No. 48. Inland Waters Directorate;
Ontario Region. Water Quality Branch. Burlington,
Ontario.
7. McCrea, R.C., and J.D. Fischer. 1984. Construction and Operation
of a Continuous Sampling System for Monitoring Organochlorine
Contaminants in Natural Waters. Water Quality Branch-Ontario
Region.
8. Yin, C. 1985. Effect of Aqueous Organic Matter on the Fugacity of a
Toxic Hydrophobia Solute: Laboratory and Field Studies. Ph.D. diss.,
College of Environmental Science and Forestry, State University of New
York, Syracuse.
9. Hassett, J.P., and E. Milicic. 1985. Determination of Equilibrium and
Rate Constants for Binding of a Polychlorinated Biphenyl Congener by
Dissolved Humic Substances. Environmental Science and Technology
19: 638-643.
2-20
-------
I I)
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
3. GENERAL ANALYTIC FRAMEWORK
The purpose of this chapter is to review the basic assumptions underlying
the sample collection and analysis efforts. The following sections accomplish
this review:
• Section 3.1, Environmental Modeling. This section reviews
general considerations of pollutant transport, fate, and
transformation.
• Section 3.2, Concentration/Flow Relationships. This section
describes the relationship between net loadings, differential
concentrations, and differential flows.
• Section 3.3, Data Quality. This section reviews the quality
control procedures used in developing the Niagara River study
data.
3.1 ENVIRONMENTAL MODELING
To obtain preliminary pollutant loadings estimates in the absence of a
full-scale modeling effort, the Niagara River Toxics Management Plan has
adopted an approach that uses input/output differential monitoring to esti-
mate pollutant loadings to the Niagara River. With this approach, flow and
concentration values for the Fort Erie (FE) sampling station are used to
estimate mass input to the river from Lake Erie sources, while flow and
concentration values for the Niagara-on-the-Lake (NOTL) sampling station
are used to estimate mass output from the Niagara River to Lake Ontario.
Finally, differences between mass input to the Fort Erie station and mass
output from the Niagara-on-the-Lake station are compared to determine toxic
chemical loadings to the Niagara River from sources along the river. In an
effort to increase the correlation between upstream and downstream loadings
over the 10-month sampling period, researchers collected simultaneous
samples at the upstream (FE) and downstream (NOTL) stations. Efforts
3-1
-------
I
i
f *>
•r
•
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
to correlate upstream and downstream loadings may be enhanced in the future
when sampling at the two stations will be lagged to reflect the approximate
travel time (18-36 hours) for water moving from the upstream to the down-
stream station.
An input/output modeling approach incorporates several simplifying
assumptions that should be carefully considered in the interpretation of
pollutant loading estimates. These assumptions include:
• Negligible hourly/daily variation in pollutant loadings
to FE and NOTL
• Complete mixing at FE and NOTL
• Negligible pollutant sources and losses, besides point/non-
point source discharges, between FE and NOTL.
Each of these assumptions is discussed below.
Negligible Hourly/Daily Variation in Pollutant Loadings to FE and NOTL
Because previous sampling efforts have employed simultaneous sampling
at the FE and NOTL stations, upstream and downstream sampling, when compared,
do not reflect differential pollutant concentrations for the same parcel of
water, but rather for different water parcels separated by the 18-36 hour
travel time required for water to move from the FE station to the NOTL
station. Accordingly, sampling results may be skewed by significant hourly
or daily variation in pollutant loadings, particularly where pollutant load-
ings may be characterized by regular daily or weekly cycles for pollutant
loading. Concerns about short-term variation in pollutant loadings are
accentuated by the use of grab samples, rather than time-composited samples,
to characterize pollutant concentrations in aqueous fraction samples at
the two stations. For these reasons, it should be assumed that temporal
3-2
-------
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
variation, especially cyclical variation, in pollutant loadings is negligible
for both the FE and NOTL transects.
Complete Mixing at FE and NOTL
Because pollutant concentrations in samples collected at the FE and
NOTL sampling stations are used to estimate mass input and output for the
entire river, these samples must be assumed to be representative of pollutant
concentrations for the entire water mass crossing the FE and NOTL transects.
In other words, it must be assumed that there are no lateral or vertical pol-
lutant concentration gradients across these transects. For all practical
purposes this assumption seems reasonable for the NOTL (downstream) transect
due to the strong mixing action provided by Niagara Falls and the lesser
significance of potential toxics sources located in the relatively quiescent
stretch between the falls and the NOTL transect. Use of the mixing assump-
tion for the FE (upstream) transect, however, would appear to be more proble-
matic. As discussed previously, technical studies of hydraulics for eastern
Lake Erie suggest the possible presence of a horizontal concentration gradient
across the FE transect not resolvable by simple redefinition of the river's
source. Consequently, it is possible that FE sampling data may not accurately
represent pollutant concentrations at the FE transect.
Negligible Pollutant Sources and Losses, Besides Point/Nonpoint Source
Discharges, Between FE and NOTL
Because pollutant transport in the Niagara River system is dominated by
rapid, downstream flow from Lake Erie to Lake Ontario, a simple input/output
loadings model based on a comparison of flow and concentration for the FE and
NOTL transects should provide a reasonable approximation of toxic chemical
loadings from sources along the Niagara River. This approach may result,
3-3
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
\ A
II
however, in overestimation or underestimation of point/nonpoint source load-
ings due to the possible effects of additional pollutant sources and losses
within the Niagara River system.
Most significantly, volatilization effects associated with the turbulent
mixing action of the Niagara Falls probably result in substantial pollutant
losses of volatile compounds entering the river above the falls. Possible
effects stemming from other fate processes are more difficult to assess.
Although strong, rapid flow of the Niagara River will minimize the effects
of sediment deposition and resuspension within the river channel, storage
of water in power reservoirs may increase the importance of these processes.
Also, little is currently known about the magnitude of biochemical processes,
such as biodegradation and oxidation, within the river channel and power
reservoirs. Efforts to refine pollutant loadings estimates for point/
nonpoint source discharges along the river would require consideration of
other possible pollutant sources/losses within the Niagara River system.
3.2 CONCENTRATION/FLOW RELATIONSHIPS
This section explores the concept of differential loading estimates and
the role that concentration and flow measurements play in developing the
overall differential estimates.
The following expression is the differential or net loading of pollutant
mass in water as it travels down the Niagara River:
AL = % QN - Cp Qp
where:
3-4
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
AL = differential mass load
CN is the total concentration (sediment plus water phase) of
the pollutant at NOTL
QN is the flow at NOTL
Cp is the concentration at FE
QF is the flow at FE.
By adding and subtracting the product C^ OF* ths equation changes to
AL - CN(QN - QF) - QF(CF - CN)
= CN (AQ) + Op (AC)
where AQ and AC denote differential or net flow and concentration.
Substantively, the differential mass load is expressed above as a sum of
two terms: (1) the product of the (final) concentration at NOTL and the
increase (or decrease) in flow through the Niagara plus (2) the product of
(initial) flow and the increase (or decrease) of the concentration of the
pollutant as water passes through the Niagara. If there is no net change in
concentration (AC = 0), the differential load is not zero, but rather Cfj(AO).
In absolute terms, C^(AQ) is substantial because, from the study's data base,
the average value of AQ is 9.22E+09 liters per day. That is, for a pollutant
concentration of 1 microgram/liter with zero differential, the increased dif-
ferential load due to the increased differential flow would be 9.22 kg/day,
accepting prima facie this estimate of average differential flow.
To examine the relative contribution of the differential flow term to
the total differential load, the following equation is used:
3-5
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
a(AQ) .
AL aQN - QF
where a is the relative net change in concentration or
a = CN/CF.
From the data base, Qp averages 583E+9 liters per day while QN averages
592E+9 liters per day, a net increase of 1.58 percent. Given these values,
the differential flow term contribution is not the dominant of the two.
For instance, if there is a differential increase concentration of only
25 percent, then the differential flow term accounts for only 7.3 percent of
the total increase in net load of the pollutant. That is, the 25 percent
increase in concentration accounts for 92.7 percent of the increase in dif-
ferential load. Thus, given the contribution of differential concentration
relative to that of differential flow, one in essence can neglect the dif-
ferential flow term contribution to differential loads. This assertion is
based on joint consideration of the magnitude of the differential flow term
contribution and the precision and accuracy of the concentration measure-
ment as follows.
It is well recognized that measurements of ambient pollutant concentra-
tions, especially such measurements at small concentrations, are subject to
substantial relative errors in precision and accuracy. Although the rela-
tive error in flow estimates is small compared to those of ambient concentra-
tion measurements, no basis exists to extend our confidence to precision and
accuracy of differential flow estimates in this application. The differential
flow term Cjj(AQ), therefore, compounds the imprecisions and inaccuracies
of concentration and differential flow measurements.
As for inaccuracies or bias in concentrations, any common to concentra-
tion measurements at both stations cancel out one another in estimates of
differential concentration. Any imprecision of the concentration measure-
ments, however, is compounded in the estimate of differential concentration.
3-6
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
DRAFT
Because of direct multiplication of the FE flow term Op, the relative preci-
sion of the differential concentration estimate is translated directly to the
estimate of differential load.
Precision of differential load estimates can be, in good part, assessed
through statistical analysis of these data and reduced through replicated
and repeated concentration measurements. Within this monitoring data base,
little can be done to assess the level of accuracy of measurements and dif-
ferential loads. Accuracy of the measurements is best assessed with reliable
quality control information and confirmed with results of independent studies.
This conclusion indicates the essential role of reliable quality control
information in estimating differential loads.
Thus, precise and accurate measurement of differential flows and differ-
ential concentrations are critical to precise and accurate estimation of dif-
ferential load. Within these two measurements, the reliability of the rela-
tive differential concentration measurement is more critical than that of the
flow.
3.3 DATA QUALITY
When the NRTC recommended the two station monitoring program for the
Niagara River, the committee recognized that the level of pollutant concen-
trations would strain the detection limits of analytical methods employed.
This recognition was based on the assessment that the order of magnitude of
the pollutant concentrations being measured was at or below the detection
limits of state-of-the-art analytical methods. Generally, circumstances such
as these lead to data that are controversial, at least to some degree. These
data are no exception.
The Data Interpretation Subcommittee has attempted to assess the quality
of the Niagara River monitoring data from the standpoint of an end user, not
3-7
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
a group of experts in analytical methods. The subcommittee, accordingly, has
examined the quality control documentation and results of the undertaking.
It is understood that quality control in chemical analysis has come to embrace
a broad range of meanings. Most would agree, however, that the end result of
an adequate quality control program is the production of data of known quality.
Three items fundamental to the Niagara River monitoring program are:
• Specific definition of detection limits
• Rate of false positives
• Rate of false negatives.
Precise information on these items has not been developed. Specifically, a
clear definition of detection limits does not exist. Some pollutants have no
detection limits, and a substantial number of pollutant measurements are
quantified below their respective limits of detection.
Precision and accuracy data are not reported for all pollutants. When
such information is given, there are a very small number of replications, on
the order of five.
Taken together, these observations indicate that the concentration data
generated to date are not of known quality. Given the small level of pol-
lutant concentrations and the number of nondetects, the quality of these
data are questionable. The Data Interpretation Subcommittee, nonetheless,
has proceeded with the analysis of the data generated to date with reserva-
tion.
The Data Interpretation Subcommittee refers the issue of generating data
of known quality to the Analytical Subcommittee for resolution of the techni-
cal underpinnings and for recommendations on reporting results. The Data
3-8
-------
•
I
DRAFT
I Interpretation Subcommittee urges the Analytical Subcommittee to ensure
that the generated data are of known quality and to suggest any required
I change in emphasis on quality control versus data production.
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
3-9
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
tlKAFT
4. DATA AND STATISTICAL ANALYSIS
The purpose of this chapter is twofold: (1) to review general charac-
teristics of the 1984/1985 Niagara River upstream/downstream data and to
determine appropriate statistical methods for analysis; and (2) to conduct
a statistical analysis of these data to determine increases in loadings to
the Niagara River. To accomplish these ends, this chapter is organized
into the following sections:
• Section 4.1—Sampling Period
• Section 4.2—Data and Detection Limits
• Section 4.3—Seasonality
• Section 4.4—Analysis.
4.1 SAMPLING PERIOD
As can be seen from figure 4-1, this analysis comprises three data
sets collected over three different periods:
Flow calculations, suspended sediment and, to some extent,
trace metals in suspended sediment were collected from
November 1983 to September 1985.
Trace metals in the aqueous phase were analyzed from
October 1984 to August 1985.
Organic pollutants in both phases were analyzed from
December 1984 to September 1985.
4-1
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
Flow Calculations
Suspended Sediment in Water
X
tf>\
Metals in Ambient Water
(n = 39)
^*
Trace Metals in Suspended Sediment
(n = 33)
Organo-Chlorines in Water
* (n = 22)
Organo-Chlorines in Sediment
ln = 211
Volatiles in Water
(n = 23)
Phenols, Etc. in Water
(n = 19)
Phenols, Etc. in Sediment
(n = 20)
Dioxins in Water
(n = 19)
n indicates the number of sampling events
for the parameter in the medium
Dioxins in Sediment
(n = 21)
Furans in Water
(n = 19)
Furans in Sediment
(n = 21)
Figure 4-1. Sampling Period
4-2
-------
I
I
I
uHAFT
Figure 4-1 also shows that the number of observations available for
_ analyzing each pollutant parameter according to phase ranges from 19 to 39.
| Measurements were collected for each component on the same day at each
station in order to pair upstream and downstream data. Since different
I pollutants were analyzed over different periods, the number of upstream-
downstream pairs will be smaller for some pollutants than for others at the
I two stations.
1
I
I
I
I
I
I
I
I
I
I
I
Table 4-1 specifies pollutant classes sampled. As indicated, the number
of pollutants analyzed in each class ranged from 5 (for dibenzofurans and
r
dioxins) to 31 for organo-chlorines. Table 4-2 also shows that not all pol-
lutants were sampled in both phases. For example, none of the volatiles was
sampled in the suspended sediment phase.
Most of the data discussed in this report were from samples collected
every 2 weeks from December 1984 to September 1985.
4.2 DATA AND DETECTION LIMITS
As shown in table 4-2, 6,608 samples, consisting of 3,570 in the aqueous
phase and 3,038 in the suspended sediment phase, were analyzed for concentra-
tion levels:
4-3
-------
I
I
1
I
I
I
I
I
I
I
1
I
I
I
I
I
I
I
i
en
in
O < 0
UJ
UJ
UJ
o
o
uj a
Z 0
UJ -1
N X
tu z u
Z
X
o
a.
uj uj uj m a
NINJNIOh-
ZZZKuj
UJUJUJUJUJUJUJ
MNNMCQcncQ
azzzooo
UJUJIUUJ
ZZZZ
Kujiu
OK
ZUJ
lluZ
uiNuj
§g
z5
X H
U Q
t- O
o a
uj x x a. o a.
3
CJ
UJ X
a o
a. £ '
. u
a. £
x
UJ
m
30
ma:
O
O^^J^QZ *CQUJ
DcxxxH-cxjaa^
__ _O UJ UJ
> a. x
ua uj uj
Z Z Z
UJ UJ IJLt UJ
-i z u a
>• UJ UJ < >•
H t— UJ X ~
X X U H <
a. a.
-------
I
I
I
t
I
I
I
I
I
1
I
I
I
I
f
i
I
i
I
a)
3
d
o
CJ
a.
to
C/3
CO
CO
d>
c
to
CM
OJ
O
in
§
in
H- 3
uj lu
X Z
at
>-
a.
—
a
u
t
in m
h-
—I UJ
^5
a. z
u. w Z 1*1
x eg
uj H uj -
ZZZ-«
1U < UJ _
(na:o:o
>-OOZ
uj Z
Z UJ
UJ K
< in
x in -a
H-JOiuuj
X O Z X X
a. z uj a. a.
o o
x >•
t- t— a cr
x Q. a; CE
a o a o
UJ 1
zc\j
- -
UtKOXX
O O — 1 (_> U — 1
i _i x < < o
ZXUQil-Z
i u >-" »— z uj
3 in
-i m
o -i
a. u
a uj
M Z UJ
g 35
i Sxs
u < t- ui
< X UJ O
a: i- o a
i- E u er o
uj a £ o —i
t- o o -J x
uj u. a- x u
z z o o u n
ui o a -J < a
r>4 ta o x a:
z a: -i o i- -
X
UJ
>-• U g 3
r 1-1 5 « K ir
M Z M £ -J UJ
5 uj £ o < a. z
3 v> a a: -
_- i
Z 0£ —' I-*
< 3 ui z
a o u x: uj u
_-. — _ — < ^ a: u -i z
—iK4xoao:ui
-------
1
1
1
*
i
i
i
l
i
t
i
i
i
i
i
i
i
i
i
Table 4-2. Total Number of Samples by Class and Phase
# of Aqueous # of Suspended
Pollutant
Classes
Furans
Dioxins
Metals
Organo-Chlorines
Phenols, PAHs
Phthalates
Volatiles
TOTAL
Samples
(FE + NOTL)
180
180
935
1,322
679
274
3,570
Sediment Samples
FE + NOTL
210
210
645
1,258
715
0
3,038
Total Samples
390
390
1,580
2,580
1,394
274
6,608
As indicated in table 4-3, a large percentage of the samples itemized
in table 4-2 were nondetects (NDs), meaning the reported value for these
samples was zero. Four of the six classes of pollutant samples at FE were
characterized by 50 percent or greater NDs in the aqueous phase. Despite
point and nonpoint loadings from sources along the river's course, a high
percentage of aqueous samples taken at NOTL were analyzed as NDs. The non-
detects actually increased, although insignificantly, for dioxins and organo-
chlorines in the aqueous phase at NOTL. The percentage of NDs was also high
in the suspended sediment phase at these stations.
4-6
-------
I
I
1
1
i
l
i
I
i
i
i
i
i
i
i
i
i
i
i
Table 4-3. Percent of Nondetects by Station and Phase
Class
% NDs at FE
Aqueous Susp. Sed.
% NDs at NOTL
Aqueous Susp. Sed.
Furans
Dioxins
Metals
Organo-Chlorines
Phenols, PAHs,
Phthalates
Volatiles
50.6
84.7
50.5
61.0
69.0
77.2
48.6
61.9
0.0
72.1
50.6
NA*
32.7
89.4
44.5
60.2
67.2
69.9
22.9
56.2
0.0
62.9
32.7
NA*
*No suspended sediment samples were analyzed for volatile organics.
Table 4-4 identifies the dependency of detects and nondetects at FE
and NOTL. As noted, if a pollutant is detected at FE, then it is more likely
to be detected than nondetected at NOTL. This is 'true for all classes of pol-
lutants in both the sediment and the aqueous phase. If a pollutant is a non-
detect at FE, then it is more likely to be a nondetect at NOTL. Except for
furans in the sediment fraction, this is true for all classes of pollutants
in both the sediment and the water phase.
The data also include observations with recorded values below the detec-
tion limit (BDL), which is a positive value recorded between zero and the DL.
The DLs for these observations appesr in the table in Appendix A.
NDs and BDLs together, as illustrated in figures 4-2 and 4-3, comprise
a large percentage of samples. Since it is difficult to interpret values
reported below the detection limit as well as nondetected concentrations, the
4-7
-------
I
i
I
i
i
I
I
f
i
I
t
i
I
i
i
i
t
t
i
DRAFT
Table 4-4. Dependency of Nondetects at FE and NOTL
FURANS IN
SEDIMENT
N D
0
T ND
L
Fl
D
40
14
54
L. ER]
ND
41
10
51
IE
81
24
105
DIOXINS IN
SEDIMENT
N D
0
T ND
L
FT.
D
29
11
40
ERIE
ND
17
48
65
j
46
59
105
FT. ERIE
ORGANO-
CHLORINES j
IN SEDIMENT! D I ND
N D
0
T ND
L
132
43
175
101
352
453
233
395
628
PHENOL ' s
PAH's &
PHTHALATES
IN SEDIMENT
N D
0
T ND
L
FT. ERIE
D
165
6
171
ND
68
107
175
233
113
346
METALS
IN SEDIMENT
N
0
T
L
D
ND
F1
D
315
3
318
r. ER]
ND
0
0
LE
315
3
318
FT. ERIE
FURANS
IN WATER
N D
0
T ND
L
D
165
6
171
ND
68
107
175
333
113
346
DIOXINS
IN WATER
N D
o n
T ND
L
FT.
D
5
8
13
ERIE
ND |
4
68
72
9
76
85
ORGANO-
CHLORINES
IN WATER
N D
0
T ND
L
FT
D
208
50
258
. ERI!
ND
55
347
402
^
263
397
660
PHENOL ' s
PAH's &
PHTHALATES
IN WATER
N D
0
T ND
L
FT. ERIE
D
82
20
102
ND
26
201
227
108
221
329
METALS
IN WATER
N D
0
T ND
L
D
214
17
231
FT.
ND
45
191
236
ERIE
259
208
467
VOLATILES
IN WATER
N D
0
T ND
L
}
D
27
4
31
"T. El
ND
14
91
105
HE
41
95
136
4-8
-------
I
t
1
I
i
i
I
i
i
I
i
i
i
i
i
i
i
i
i
M UJ
H (JO
U) U
_i
8
UJ
Q.
4-9
T
a
UJ
"• 1
T
UJ
0.
UJ
"• 1
T
tfl
1
4 I
UJ
UJ
"• 1
4 T
0
M
a
w
fc 1
T
o
M
Q
U
"• 1
UJ
a.
o
u-4
CO
4J
1-1
e
c
O 0)
i-1 03
4-1 «
a ^
10 c to T- i—
c T- •— co
m x ITS
-------
I
I
f
f
I
t
I
I
1
I
I
i
1
i
t
i
I
I
i
UJ o
CO O
UJ
u
DC
UJ
Q.
o
_i
HI
u.
LU
u.
ID
IJL
UJ
O
00
o
CO
O
CM
4-10
en
en
i-
<
i-
en cj
T
UJ
a
1
CO
cr
o
1
T
i-
OJ
I
i T
o
1
CD
M
a
UJ
"• 1
-z.
o
M
t-
CJ
UJ
I-
LU
a
UJ
a_
>
\-
DRAFT
u
o
i-l 0)
e to
0.
c
O 4J
•H C
4_l (U
o e
01 0)
o on
3 T!
O CD O)
3-i- QJ S- J=
4- -O S O Q.
QQ O I— OJ UJ
i—i p—i uj oi o:
Q Q Z O Q-
C3
UJ
-------
I
t
i
i
i
t
I
l
i
I
f
i
t
i
i
i
i
t
i
statistical analysis and interpretation of the data are complex and require
considerable care. A quick review of the figures indicates that, in most
cases, a higher percentage of FE samples are NDs and BDLs as compared to
samples taken at NOTL. Further, when comparing FE to NOTL, one sees a
tendency of NDs to decrease from FE to NOTL. At the same time, the number
of BDLs is greater at NOTL. This suggests that concentration measurements
increase from FE in NOTL. Section 4.3 uses statistics to analyze the ac-
curacy of this statement.
4.3 SEASONALITY
The fact that the data set consists of NDs, BDLs, and pollutants measured
at low concentrations complicates data analysis because the accuracy of the
data is brought into question. The analysis is futher complicated by other
factors, including seasonality. This section addresses three objectives.
First, it provides a graphical representation of the flows at FE and NOTL.
Second, plots of flow versus concentration are produced to assist in determin-
ing the'relationship of flow and concentration. No patterns are detected.
Third, a rationale for pairing observations for controlling potential seasonal
effects is presented. The analysis portion of this chapter uses this method.
The accompanying figures, 4-4 through 4-13, appear at the end of section 4.3.
From November 1983 to September 1985, the flow of the Niagara River
ranged from a low of about 425 x 109 I/day at FE to a high of 665 x 109
I/day at NOTL. As shown in figures 4-4 and 4-5, the flow at NOTL is nearly
always greater than at FE. The mean flow recorded was 5.82 x 109 I/day at
NOTL and 5.73 x 109 I/day at FE. The river experienced higher than mean
flow in the spring and lower than mean flow in the winter, as plotted in
figures 4-4 and 4-5.
4-11
-------
I
i
i
i
i
t
i
t
i
i
1
t
t
i
i
i
t
Figures 4-6, 4-7, and 4-8 provide some typical plots of pollutant
concentration values versus time for the aqueous phase. These plots reveal
typically large numbers of nondetect values with spikes of pollutant con-
centration values intermittent throughout the sampling period. A large
spiked value for all pollutants appeared in summer 1985.
Figure 4-9 is a plot of the sediment concentration in water. These
data are used with figures 4-10, 4-11, and 4-12 to convert concentration of
pollutants in sediment to water concentration of sediment-borne pollutants.
Figures 4-10 through 4-12 plot pollutant concentration values versus
time for the sediment phase.
The plots for the sediment phase, as compared to those for the water
phase, exhibit fewer values at or below the detection limit and fewer non-
detects. There is more variability, and a clear spike for all pollutants
is visible for summer 1985.
The plot of sediment concentration coefficient versus time, figure 4-9,
clearly shows the large variability between values over time. In general,
the overall picture indicates that the Niagara-on-the-Lake values for con-
verting sediment fraction values to water are often greater than the values
for Fort Erie. Larger values occur in the winter of 1984-85 and in the
summers of 1984 and 1985. Summer/winter differences seem to be a random
phenomenon.
Holding loadings constant and assuming conservation of mass, one would
expect that the concentrations at FE would be higher than at NOTL because
the greater NOTL flow would dilute the concentration. (See section 3.2
for a further discussion.) Figure 4-13 plots flow versus concentration for
four pollutants. As illustrated, no correlation exists between flow and
concentration.
i
4-12
------- An error occurred while trying to OCR this image.
-------
I
I
1
f
1
1
I
1
I
f
1
I
1
I
1
I
1
1
I
It also should be noted that the period of collection of most .of these
data was 10 months. If there is a 1-year seasonality cycle, these data will
then be biased due to the relatively abbreviated period of sampling.
If seasonal effects result in changes over a day or two, brief lag times
between the two paired readings would not affect the control of seasonal
variations facilitated by the pairing. However, if changes in input at the
river's head occurred slowly, then the lag time built into the pairing process
would be insensitive.
Upstream to downstream correlation coefficients on the daily loadings
values were computed. In general, these values on a statistical basis were
significantly greater than zero, as is to be expected. This positive correla-
tion, and covariance, is used later when analyses are presented concerning
mean daily differential loadings. That is, the standard error of the differ-
ences is reduced by the covariance terms. This makes confidence statements
about the mean difference more precise.
4-14
-------
I
1
1
I
I
1
I
1
I
I
1
I
1
I
t
I
f
f
I
O X
4-15
-------
1
i
i
t
i
i
i
i
i
t
i
i
i
i
i
i
i
i
i
U. _1 O X
I
0)
4->
o
I
to
CO
•i-l
IZ
03
C
O
PC
C
s
o
u
O
m
I
S-l
3
4-16
-------
t
I
1
I
I
I
I
I
I
I
1
I
I
I
I
I
t
I
I
DRAFT
o
o
o
in
o
o
o
o
o
o
m
o
o
o
o
o
i
O
in
o
(9
in
o
1
in
CD
CD
U.
*4
o
>
o
z
ID
(0
T
O
<
(VI
o
CD
CO
- UJ
u.
tn
CO
h O
1-1
H
03
3
UJ
oc
o
£.
p.
Cfl
z
U)
UJ
H
t£
UJ
tr
o
u.
c
CO
i-i
3
O
to
c
•H
•a
o
C
01
4J
O
o
o
a
3
(JO
•H
4-17
-------
I
I
1
I
I
1
I
1
I
1
1
I
1
I
1
I
1
I
I
A-
n
o
in
o
CD
in
. o
" I
in
o
o
K
z
o
d
<
CD
z
A
O
0)
CD
S
cc
UJ
-T Q
(D
> CD
O Z
Z H
U)
o
o
tn
o
o
cu
o
o
O
CO
T
OD
OJ
O
. 00
ID
- UJ
U,
tn
o
J- o
UJ
H
(C
UJ
cc
o
U.
o
p
o
o
U
o
3
be
4-18
-------
I
I
t
1
I
1
I
1
I
1
t
I
1
I
I
I
i
i
i
m
inoomin
uozuujzt-a:
-------
t
I
1
t
I
1
I
1
I
1
f
i
I
I
i
i
i
I
i
tn
CO
o
s
(0
ID
CD
<
CVJ
o
in
GO
en
uj
LL
v4
o
_
O
cn
3
to
UJ
T D
CD
> CO
O Z
Z M
•H _J
O D.
<
CD
(0
in
m
o
tn
in
ru
o
cu
tn
in
• o
CO
• o
CO
CD
- UJ
U.
«•«
• o
. m
CO
1- o
o
M
cn
c
o
c
0)
o
c
o
CJ
c
0)
e
•fl
"O
OJ
CO
ON
I
3
bC
CJOZUtIIZh-tZ
-------
I
I
1
I
I
1
I
t
I
1
1
I
1
I
1
I
1
I
I
10
o
o
S
,
I o
s
-
o
UJ
Q
> CD
O Z
Z H
. O
2
s
v<
I- o
1
, w
I o
I. o
O
o
o
o
o
o
(M
O
O
o
o
o
o
o
o
o
o
o
CO
o
o
o
o
o
o
cu
4-21
CRAFT
IU
H
O
cr
CD
cr
UJ
cr
o
u.
o
M
t-
o
o
s
•H
H
3
tn
Ol
CO
fl
x:
c
0)
e
•H
•o
O)
Q
I
O<
O
N
C
•H
•a
o
O
o
O
rH
0-1
•
O
rH
I
-------
I
I
1
I
I
1
I
t
I
I
1
I
1
I
i
I
i
I
i
in
a
in
ID
9 2
z M
03
00
m
K
. oo
en cu
uozuujzi-
-------
I
I
1
I
I
1
I
I
I
1
I
I
1
I
I
I
I
I
I
(U O
_J
UJ
o
tr
o
UJ
M
IT
UJ
i
u.
M
I-
CJ
o
en
3
en
cu
03
tB
P
0)
•H
T3
0)
Cfi
T3
rt
0)
o
I—I
p-
0)
(-J
3
(1C
4-23
-------
1
1
1
1
1
t
1
1
1
1
1
1
1
1
1
1
1
1
4
W •
-J W
Is.
u:
E- ac
2 C 44
2 3 a
•4? C/4 O3
0! '
£g«
£f g
S u3 *"*
> u
— z z
[&. M 2
O — -
ae
v rj3
3 H *
*" 2£
!H!
< u.
s °
5 e->
u o
< a.
44 • 4 4 4 * •« *
4 U 4 4 _
• *«««.*«<*a**
4 4
'X
-J
a.
Is *
W V]
5W
5 3 M
I S*S
~i Cc. Z U
5 c^ " u
3 C M J
6- 5
C u
a. H
^ CJ
S £ c a
•< < 3
U C 4
W tiJ _J
Z Cb 4
u
CO U.
> c
Is- -
u o
a.
^
4
4
4
4
4
£ 4 4 "
SASSSSS^SAS
4
i
u
•oj
o7 en
y:
Z fN
C H
< Z OS
Z tri «"
u cn a
q M — •
- . ? " •*
i « 0. <
1 > =
5 3 W " «
300
U -J Ed Z
C*. X W
a. HS
C Z -J
H
32
o. a:
< £ 3
5s sj3
3S -
U
a. H
3 - -
4 44
• ^•MOJ«O«*4I
»3**-n*J**-» *
4-24
-------
I
I
1
I
I
I
I
I
I
I
I
I
1
I
I
I
I
I
I
DRAFT
4.4 ANALYSIS
This section encompasses analyses that lead to the presentation of the
following:
(1) The mean daily difference in loadings from upstream
to downstream on the river
(2) The mean daily loading upstream at Fort Erie
(3) The mean daily loading downstream at Niagara-on-the-
Lake.
When the loadings analysis is presented, the differential loadings will
be discussed first. The reason for this approach is that the purpose of the
FE/NOTL comparative station study is to generate a loading estimate to Lake
Ontario from sources, both point and nonpoint, along the 34-mile stretch of
the Niagara River.
Several preliminary analyses are required before proceeding to the
loadings calculations. First, concentration measurements must be analyzed,
as is done in section 4.4.1. Then, there must be a determination of which
parameters to use in the differential loadings calculation. This will be
accomplished through the nonparametric screening test described in section
4.4.2 and applied in section 4.4.3. The loadings analysis will be presented
in section 4.4.4.
4.4.1 Calculating Mean Concentrations
The large number of NDs and BDLs in the data set raises questions con-
cerning their manner of treatment. Standard laboratory practice would dictate
that both ND and BDL would be recorded as zero. For the purposes of this
report, however, alternative formulations for values of ND and BDL were
considered as follows:
4-25
-------
I
I
I
I
I
I
I
I
I
I
I
I
t
I
I
I
I
I
I
(1) Do not use NDs in any calculations.
(2) Calculate NDs as they are recorded, that is, ND = 0.
(3) Calculate NDs as (DL + 0)/2.
(4) Calculate NDs = LMV (least measured value) or DL,
whichever is less.
(5) Calculate NDs = DL.
Similar formulations were considered for BDL measurements. Method (1)
was dropped from further consideration because ND should be given equal
weight with other measured values. Methods (2) and (5) were accepted as
reasonable alternatives for determining NDs since their alternative con-
sideration as 0 and DL would work to allow an estimate of the ambiguity
associated with the existence of these values in the average daily concen-
tration computations. An explanation of the method for computation is
discussed below.
The average daily concentration figure for each class of pollutants was
determined by using the data for each pollutant as presented in the original
data, including the nondetected values and the positive values below the
detection limit. The total daily concentration for a class of pollutants was
computed by summing the daily concentrations for each constituent member of
the class.
Table 4-5 presents the mean daily concentrations, and the associated
standard errors for those means, for each pollutant at each station in each
medium. The column labeled "Below Detection Limit Ambiguity" is the amount
of increase in the mean daily concentration that would occur if all nondetect
values were replaced by the detection limit. The mean daily concentration,
then, with the nondetected measurements replaced by the detection limit would
be the mean as reported, plus this ambiguity value in the table. An approxi-
mate 95-percent confidence interval can be computed as the mean concentration,
plus or minus twice the reported standard error.
4-26
-------
I
I
1
I
I
I
I
I
I
I
I
I
t
I
I
I
I
I
I
CO
c
o
,-J
4-1
tfl
£
c
CD
O
q
o
o
^
r— I
•H
nj
£j
C
0)
S
m
1
0)
•H
43
H
*
1 t
5 i
a i
H* 1
z
UJ K
O I
Z u
o :
5 ^
O II
in
u
M
1ARY STATIST
ID-CrlOT CDTC
in i
M
£
M
-I
M M
2 (J O
0 UJ M
-J *- CD
UJ UJ £
sa a •<
Is
§0:
2
< UJ
i-
10
z
o
Q;
Z
UJ
^
Q
U
z
UJ
3
•4
i 5-uj
1! £ N
• •< M
w cn
i
7
*
j
•)
j
^
if
j
£
3
.
I
\
1
»—
3
— J LU
-1 £
Sz
H
Z
^_
3 in
-j in
£2
a. u
CM
*
01 R! CJ M
[BENZO-FURAN
[BENZO-FURAN
JENZO-FURAN
[BENZO-FURAN
o o a o
See O K
OKU
-J -/ O -J
X X -J Z
u u x u
< < u «c
UJ UJ UJ IU
H Dp X X
£
UI
N
U)
Q.
in
Z
i
u.
OJ
•*
0
rO
CO
••
OJ
1ENZO-FURAN
a
o
|
X
u
g
•* i« •*
0 O M
^ ^ ft|
O* «^ "N
T
•^
s
o
•o
OJ
«
—
-r
e>j
Z
M
X
M
I
I
O
M
Z
LORODIE
X
u
g
OOOQOOOOOOOOOOOOOOOOCOOOOOOOOOO-4
ooooooooooooooooaoooooooooooooo
OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
ooooooooooooooooooooooooooooooo
oooeoooooooooooooooeooooooooooo
cuc\JCNJCsiegNCJcgcjNNCuc\jcac\JCJ«\icjc>JNCJCJC\jctjcjc\Jcycgcvj- — oa:z
3 Q Q o z -i-i >• o: oxiiaaaixjoi-
a zooxzafZuuaaaaa r x >-• t-* •-. i ( c **uxa.
I IMI IX— J*-tZU4UJ-*"«tU»-«ZLJUJ --""**» "IUUJLJ
\
tn
UJ
H-
_J
<
£ X
^ 1
§ §
in <
z in
i 5
-i a.
x
u
1 01
0 -1
z o
< z
U) UJ
S £
f» m -T
r*. -« -N
OJ >6 O
O 0 0
Kl ftj O
o e o
00-T
0 0 O
M »N OJ
UJ
z
UJ UJ <
f- UJ X
a. < z
U Z UI
m
o
0
o
LA
o
CM
o
OJ
UJ
z
UJ
>~
a
UJ
CO
<6 •*
0 0
o o
o o
9* OJ
o o
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
OJ
3
C
C
O
CJ
0)
c
O
O
c
O
CO
c
I
0)
CO
H
§(
H
M t-
5 i
a c
z u
y =
Z i
O <
U h
z £
O II
n
u
in H
M C
H- U
^
H t.
• u
cr
< i
| |
in i
M
M
_J
z >
a t-
H- 3
x o 3
O UJ M
-J H- ffi
UI UI £
CO O <
s<*
Q£ Q£
< o
§i
< UI
in
g
M
5 cr
H H
Z
J UJ
< y
i i
u
5 3
J r
3
J
r»
UI
1 r N
i in in
•
5
L
.
!
1LLUTANT
WE
a. z
i-
3 in
-j in
£2
a. u
ui o o • X
UJ
UJ Z
Z UI
3 ?
:NZO(6,H,IIPER1
1RYSENE
.UORANTHENE
LUORENE
>ENO(I,2»3-CD)I
VPHTHALENE
ffl U U. U. M Z
o o o o
oooo
o o •» o
slss
o o KI o
-~~~
UI
*-
lu,
15
1ENANTHRENE
rRENE
IS(2-ETHYLHEXYI
[-N-OCTYL-PHTHJ
0.100
IA ^ CM IA
— 0 CO 0
O O O O
e o o e
- CM -1 CM
in in
—i _j
(A ^^ ^^
[CHLOROPHENOLS
JICHLOROPHENOL!
ETRACHLOROPHENt
iNTACHLOROPHENl
o i- >- a.
.00000000000
ooooooooooo
ooo'-o'-^oo'o'm
<0 KI
ssisIssfsKs 1
• • • • I
eu >r *•* r* i
u
(
h
•
u
1
i
u
1
g
(
1
_,85§ >".,§
O Z M E -J uj D UJ Z
zui£O «
>)• Kl
7 *9*
33
§s
oi
KI Nt
Z Z
UI UI
:HLORODI
:HLOROO]
TETRAl
PENTAt
j
0
i
V)
3
§
u.
~ss
*"*«
352
ss^
-0 0> 10
- S S
lA CM CO
KI -* U>
* KI
<«3
a 3 at
U. 1 U.
i o i
ilui
tLORODIE
:HLORODI
4LORODIE
HEXACl
HEPTAt
OCTACt
IT
ji
i-
S5
CM«
0 U>
0 CM
O O
O KI
Z Z
M IH
X X
n n
u o
a a
1 1
1 1
0 O
Z Z
UI UJ
:HLOROOI
:HLORODJ
TETRAI
PENT At
^
£
X
0
a
\ '•"' ', '--
i "; •- '*' v-
-' , 1 p
..'**-
SS2SS
j-Kir-oo
S2S25
^r^K^oo
CM * -l 0 ,0
S - cJ 3, o
KI in CQ e <6
*J- ^ ^ o CM
•4
z
Z IH Z
M X M
11 O H
o i a
a i a.
l O 1
111
§w 8 3 3
cr O u.
cr O cr -J
-i x >J in
u < u a u
x n. £ ui S
UJ UJ Cj 1 1
x x o < <
i
in
z
g
X
u
<
10
cr
o
»^ -
°0
00
s"
00
0- CO
0- S
>o o
0 0
UI
z
<
a
zo
< -j
§s
lj <
4-28
-------
I
I
I
I
I
I
I
I
3
C
C
O
CJ
en
c
o
•H
4J
CO
01
o
o
d
CO
01
u-1
I
•a-
co
H
Kl
§ :
M C
H- h
2 C
h~ *
Z C
UJ U
o (
u h
«
i 3
in
u
M U
1— f
in a
l-l U
H-
•< |-
£ cc
i
1 *
£ o
Z) (
t_
M
r
(H
-J
it
M M
H" 3
3 O O
O UJ M
— i H aj
uj uj £
oj a <
S*
Q£ Cc
2S
§a
4 UJ
»-
in
§
M
H
«
CC
H
Z
UJ
U
z
UJ
r
i UJ
H ^
3Q» UJ
r IM
c < M
c in in
e
:
c
j
H
e
j
§
j
i
But
£l
*^ «b
h-
Z
<
h-
=> in
-i in
-j <
S_i
u
o o o
S«* o
eg 03
o o o
000
1A O O
-• eg O
0 O O
0 O 0
•O M 0
•o m o
Kl sO 0
O 0 O
O O 0
eg eg eg
eg tg CM
IU
z
UJ <
z a
2§
S^z
-J U H
X 1 QC
u > a
I X "
in o ^
tn
eg
0
o
o
i
o
o
K»
in
.r
eg
o
eg
eg
5
EX
O
UJ
M
O
01
^
•a
o
0
O
03
0
o
•o
•f
CM
0
o
CJ
OJ
z
M
Q£
1
o •»•
o o
in KI
o o
o *^
O CJ
O 0
o o
0 CJ
o >o
o
0 0 O O O
O O O 0 O
-i O in Kl LD
o r« in eg o
eg -T in •» in
0 0 0 O O
0 O 0 O O
eg eg eg eg eg
eg eg eg eg eg
§
_l
X
u
>»
K K O UJ X
So a a a
o a o >-
a. a. a. a. ?
• o » > UJ
r% O t (L 1C
^o^Kt^^utr^-ooj^cO^ooajNiritfcjoo
r*a3O'-^iflo^N^)*oKimKi^o^oooooo
oooooooooooooooooooooo
u*ioc\ji.ncMfOor*>«*>«ooo«)»>o<-4OU«OOOOOOOOOOOOOOO
ojoin^iAKio-HtAQoirtinNoeMo-o^oocj
L^ooajiflcMor^^-cocg^Kt^o — tficsjocgoin
Ktor**Ui-«O9t0>4-4QO-4OOCvJOO
oo>^tnKtojsr^<-Hr^ocjo-^oooooooo
oo^ocj-«>>.o-^eoooooooooooo
« eg
cvj cu AI cvJcsjcjesjcvjcsjcvjfyNftjcJWHiooKiOfO o
NCUCSJCSJCg>
OZZZOOOZMNUJ o: o; cc
-uj<>» — ^- *
cr oxxxaoeo:'\-'OH-ujx^ - * - LU
M -iouut-i-i-^-wx-ja:ot-'<^ »t CSJ ^ ^
O O 0 -4 0
O 0 O 0 O
o o o o o
e\j .-,—,« —
UJ
z
UJ
ce
>-
a.
i
UJ 1 UJ
Z KI UJ Z
uj * Z uj
£ CXJ UJ QC
h* UJ --IX
Z Z -« < H
< uj - X Z
DC CE O »— <
SO Z X Z
3 UJ CL UJ
-J -I Q < X
u. ^ w Z ^
I
eg
o
o
o
CO
o
o
o
<±
in
o
o
o
eg
IU
Z
U
CL
>-
0»
• •
eg in
0 0
0 0
0 0
CO O
in -i-
9* •£
eg o
O 0
,0 r-
V .J
5&
"TS
eg i
OTZ
i-t w
ffl a
•* w
' , • r !
^r* ,
a) r^ Z
U M K Z UJ
M a: uj iu x
Q h* h~ Q. CL
4-29
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
T3
0)
3
e
•H
4-1
C
o
^x
en
C
0
•H
cfl
i-i
4-1
C
cu
o
c
o
u
^
1— 1
ft
cfl
n
c
CO
CU
•
1/1
1
OJ
i— 1
CO
H
*f
» I
H> h
W Li
H. <
Z C
UJ U
§ '
U 1
§ 3
W
cn
u
H U
1- h
tn a
H U
SUMMARY STAT
f*"
M
HI
«J
X U 19
O IU W
-j H eg
uj iu £
oi a <
a
§ac
Q£
< IU
in
§
H
<
z
UJ
' O
i O
H
• z
3 <
( IU
: r
:
1 3
< M
n in
j
H
e
j
I
'
.
' H
L
0 <
H
5 u H- <« 3
IU < X HI UJ - ^
CO U 0 Q H. _ <
-J
Z
V)
§
HI
H
<
a:
H1
Z
UJ
O 3
U N
£
-J »-
M tO
H- ^
< <
-i H
O uj
> r
0*40000*00000
ooooooooooo
ooooSoooooo
ooooooooooo
ooooooooooo
oooS^oooSSo i
ooooooooooo u
t
,
u
:
1-
c
u
u
•
u
e
«
(
(.
4
H
!
UJ 1
or§ £>. i E
ZM££ <3UJZ 1
0£ S
3
J
1
J
J
J
:
I
t
>
(
4
•
H>
3 z
3 <
: H.
-I \U
a. z
h-
5
H
3 in
-j in
a. u
-e\jinK>r»«r«r^
cCSKSfi?SS
•*OOOO^*f
^^
caegcurueucacgca
z z
HI HI Z
X X HI
Z Z Z O 0 X
rfrfyrf^MMfl
22<2<3oS
D 5 IE 3 CK I I Q
u. u. 3 u. 3 a. a. i
1 1 U. 1 U. 1 ID.
O O 1 O 1 O O 1
NMOMOKJNJO
ZZNZNZZN
(OOujaiuJOioiuj
HiMCDMajHiHtai
QOHIQMQOM
Q£D£OaOQ:Q:o
uuxuxuux
Q:H<»- IHII IX-JMZOIUJ
xo<<-i
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
l"*V?
UH
AIT a
^^
,
r-J
T-)
Q
C
CO
0)
•
m
i
0)
^H
CO
H
in
*
(.
<
i £
w
^ H
^ •
K i
H 2
Z '
Ul C
«J 1
Z «
U
§ u
i
VI «
u
M
h- U
in ?
M f
t-
s !
o:
i ;
g t
1
t
H
M
C
M
.J
§^
S
H M
fc» ^
3UU
O IU M
-J H «
UJ UJ £
a o <
a
S3
try
at
Ul
H-
in
§
M
<
Q£
UJ
U
u
3 z
H <
1 £
t
1!
IU
^
1 n in
i £ N
: < «
•i m in
i
i
j
e
c
j
j
5
c
1
-4
•
"
I 1
i §u
£<
z
h-
3 VI
-J VI
0. U
ooooooooo
ooooooooo
— OOOO — 0 O~ •f
ooooooooo
« o r. r^ - vn o « »
ooooooo-^o
ooooooooo
^
IU
t-t
a
<
a: H
2 S
X X O
O UJ QC
>- ae o
H- i- a iu x M _j
a a a a o £ x
a a a a >- x a u
a.o.d.d.Haoffjx
a.oaa.££Q.Q.x
o*
0
o
In
o
0
o
.OROBENZENE
X
u
M
o
N
0
o
0
o
0
0
.OROBENZENE
X
u
M
a
to
r*
0
o
0
o
o
,
o
o
.OROBENZENE
5
o
•f
000
000
o o o
2*2
O 0 0
-00
O 0 O
000
zzz
[CHLOROBENZE
[CHLOROBENZE
[CHLOROBENZE
t-se
1 1 1
to * in
ftj OJ tO
o
0
o
0
o
in
o
0
IU
z
IU
2
UJ
fETRACHLOROB
r~
1
-»
CJ
O O
0 0
o o
S !n
o o
osyirtin
** _i c\j m
— « *•< CXJ CJ ••« ft* OJ, • ~ -. -J-J
< < >• a. «j-xtn-Jzz
UJ UJ X O UJ Q XK-JOUJUJ
ZZKDCL U UJXOZXX
Z o o: < u. M z mujz -JIUJXOQ
uj uj < >- * ' — • uj *zuj -"••JXtt.ttQ:
h-UJX— --*-*UJI— UJ--JX H ^ O CC — * -J E
xuf-<xzuj i OO-JUU— 'M5w>— o:
^lS55iw^§§^<^?2My"£5^^°^SSuj
<
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
IT* TI
T3
0)
3
c
c
o
en
c
o
o
c
o
M
c
I
• o o o e
OH- O O O O
MM o o e o
I U IS
2 " 3
ui ui £
to o <
o o a a
§ ig
u
z £
O u.
P 8?
I
in
u
<
h-
in
5 §
a M
" 5
£
-i u
£ 5
5 u,
IS -I
« a. ui
M EN
Z < M
ii in in
a
§
g H-
-J UJ
-* r
Sz
03 J- M <>
4 <\J O . 5
§^ M
UJ Z
U * ul U
a u -J z
UJ l-t UJ 1^
E Z V) Nl
j
H
1
:
:
x
u
:
i
u
$
J
U
:
E
<
£
«
A
<
H
S
i
L
|
9
H
M
)_ 3
3 U 13
O Ul M
ui ui £
a o <
o
||
H
0
O*0*O*O*O»O*Ok9'9tO*CtfCJ-4M-h-i i i ixui
1-rxo.i-i-rxa.i-ujiDzouvQiuQaQ.an.a.n.i-a:
UJUIIUUIUUIUILUUIIJ i i M i i X-^MZUIUI * * * ~LU»-I
M&XXOH'&XXO^^^J^IiSO^OuiXX&O^CLiCC
_j
(9
Z
in
Z
— M
— .j a
-ISO
\ UJ -J
u> a. x
& w tj
~ in i
S 5 i
< X <
(V o u
3 M a
u. a o
*Si?'» ';>• :
^&V
ooo — ^- — oo
oooooooo
isslslll
?s?i?ss°
°S-S;<>JS-"V
CXJCSJCMCMNNCMCVI
«JOJMOJN(MPJCg
UJ Ul
z z
Ul Ul
IU Ul Ul Ul Nl Nl
Z Z Z Z Z Z
M Nl Nl Nl ID m
Q Z Z Z O O
< ui ui ui o: ex
x o a a o u
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
,>
*C
3
5
4-1
S I
O t
v^ Z (.
o «
M C
M 1- u
<— <
X CC C
O (- u
i-l Z t-
U 0 3
J-l O
4J U
C Z U
-
•H in «
S s S
s i !
0) mo
s J
1
•
1
— _j _i j
<<>• a. -J < ui o O
cc cc cc — >- x in _i z z
IUIUXQUJ a XH-JOlUUJ
z z H 5 a. u luxozxx
uj uj z -J — iu i iu xa.ziua.a
U CC < u. w Z KIIUZ J I UJ X O O
«*>• — —•• IU -Zul >- -1 X 0- CC CC
aax^x x niujce z >- a. o O o
X*"- *>^ -IU^-UJ *-JX *- H- O CC -J — 1
i-«zzz — -MMtHaciuuj
09 a o i-
r- m -»
o o a
o
UJ
)L
ENE
3N TETRACHLORID
UJ Z CC
m u
i
o
M
ce
H
Z
O
U
_J
M
1—
<
5
djAjoineooooooooooo
OOMOOOOQOOOOOOQO
o — -roKioooor~oooooo
^«-4OkKtr^oaoo>oo<-^oooo
or^oooooooooooooo
UJ
Z iu
IU Z 1-
-j •< a
IU > X <
z x i- cc
•< t- IU l-
x uj a x
K O CC UJ
C Ul CC O
Sr o -j z iu
o -i x S r - O
OOUMZM3MCC ZK-JM
QXOS HUI£OQ»ZQC9CJ3£ UJ (J
-JU)-iMZ!inQceCLO
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
fcf*
The data in the table suggest that some pollutant parameters are measured
so that the ambiguity factor is very small, as is the case for copper and
zinc in the suspended sediment phase. As noted earlier, no NDs were recorded
for metals in the suspended sediment phase at either station. On the other
hand, a large number of nondetects causes the ambiguity factor to be very
large, as is the case with dioxins measured at FE in the aqueous phase. The
sample variance of concentration values is the sum of squared deviations
about the mean divided by (n - 1). The sample standard deviation of concen-
tration values is the square root of the variance, and the standard error of
the means is the standard deviation divided by the square root of n. The
standard errors associated with the average daily concentrations can be used
in computing a 95-percent confidence interval about the mean. For some para-
meters, this interval includes zero. For example, while the mean concentra-
tion for hexachlorodibenzofuran in the suspended sediment stage at FE is
computed at 83.1, the standard error 57.1 results in a 95-percent confidence
interval ranging from -31.1 to 197.3. On the other hand, a number of para-
meters exist for which the standard error is a small fraction of the mean
and, as a consequence, tight bounds about the mean result.
Figure 4-14 displays a few representative constituent parameters with
these upper and lower bound concentrations and a 95-percent confidence
interval about the mean. See Appendix B for a description of the method for
calculating confidence intervals. Some of these parameters, while yielding a
positive average mean daily concentration, are associated with 95-percent
confidence intervals that include zero. Consequently, for these pollutants
and any similar ones in the table, we cannot conclude from the data that the
mean daily concentration is different from zero on the basis of statistical
hypothesis tests. A somewhat easier way to make this judgment is to recognize
whether the standard error is larger than one-half the mean value.
4-34
-------
1
1
rC
1
1
1
1 «•
1
1
1
1
1
1
1
1
1
1
H^^^H
H
^
^
5 Value
• Figure 4-14.
1
1
DRAFT
Heptachlorodibenzo-Furan
Upper Bound, 46.8 = Mean, 38.4 = S.E. I
Lower Bound, 43.5 = Mean, 38.6 = S.E. I
P9/I
Octachlorodibenzo-P-Dioxin
Upper Bound, 14.0 = Mean, 3.0 = S.E.
Lower Bound, 6.9 = Mean, 3.7 = S.E. •^^••^ pg/i
pg/i
Chromium
Upper Bound, 8.6 = Mean, 1.6 = S.E.
Lower Bound, 8.6 = Mean, 1.6 = S.E.
mg/kg
mg/kg
Benzo(a) Pyrene
Upper Bound, .2530 = Mean, .1014 = S.E.
Lower Bound, .2528 = Mean, .1015 = S.E.
Hexachlorobenzene
Upper Bound, .5749 = Mean, .1830 = S.E.
Lower Bound, .5586 = Mean, .1852 = S.E.
Figure 4-14. Schematic Diagram of Confidence Bounds for Mean
Daily Concentrations of Some Parameters Sampled
at Fort Erie.
4-35
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
DRAFT
4.4.2 The Screening Test
The next step in the analysis is to determine whether the upstream to
downstream concentration changes exhibit an increase that can be considered
statistically significant. If a statistically significant difference in
concentrations is found, then one can conclude that the loadings contributed
by sources along the river are affecting recorded downstream concentrations.
If no difference is found, that is, if the null hypothesis is accepted, then
the contribution was not found to have an effect on downstream concentrations
and computation of the change in loading over the course of the river is
meaningless.
The following discussion explains two nonparametric statistical tests,
the sign test and the signed rank test, that may be used to determine which
of the parameters have statistically significant increase in concentration.
The purpose of this assessment is to select the most appropriate test for
determining whether the data provide evidence of an increase in differential
concentration or loading of pollutants as water travels down the Niagara
River. The next paragraphs present the rationale for use of the signed rank
test.
As discussed in chapter 2, the sign test was initially recommended as
the preferred approach. The sign test is a statistical method to identify
those pollutants for which the data suggest an increased differential concen-
tration or loading. It is not a powerful test, however. Differences between
the two monitoring stations—based on the data at hand—are not obvious, and
this fact casts some doubt upon the appropriateness of using the sign test.
In comparison, the nonparametric signed rank test is commonly used.
The null hypothesis (no increase) tested by the signed rank test is more
appropriate than that for the sign test for this analysis. Furthermore, the
signed rank test is the more sensitive of the two tests to change. Because
4-36
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
DRAFI
the sign test is less likely to detect change, it is also less likely to give
a false positive.
In an early analysis in this application, the use of the sign test was
prompted by chemical analytical data being reported as "nondetect," trace, and
a quantified detected value (reference 2). In this data base, no values are
reported as trace. All values are reported as nondetects, quantified detected
values, or missing values. That rationale, accordingly, does not apply.
A number of data reliability issues (e.g., false positives) exist that
cannot be resolved within the confines of the current monitoring network and
quality control design. The best that can be done is to isolate apparent
findings and treat them as tentative unless and until they are confirmed by
further work. To this end, the use of a sensitive statistical tool to
identify candidate findings is more beneficial, and the signed rank test for
initial testing for differential loadings is thus recommended.
The setting for this assessment is that data are drawn from two popula-
tions—the sampling stations at FE and NOTL. Furthermore, the data are
analyzed in pairs so that there is some relationship to observations taken
at the same time. The analysis indicates that the concentration and flow
observations exhibit seasonal patterns and that pairing observations in time
removes the seasonal variation component. Accordingly, an analysis approach
not based on pairings would be much less sensitive to detecting change across
the two stations because the seasonal variation would introduce a large
component of noise into the analysis.
Both tests are based on the differences of paired observations across
the two populations. Specifically, let d^ denote the concentration or load-
ing at NOTL minus that at FE. Basic assumptions are common to both tests.
The first is that d^ is well defined; that is, the differences have meaning.
For example, if in an experiment, individual subjects are classified into one
of three groups, differences between two numerical classifications may not
have any substantive meaning. In this application, differences relate to net
4-37
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
DRAFT
concentrations or mass loadings that have definite meaning. An operational
difficulty, however, is introduced due to the presence of a nondetect at
either or both stations. This difficulty must be overcome in applying either
of the two tests.
The second assumption common to both tests is that the d^ are statisti-
cally independent. The analysis suggests that there is no evidence that the
dj_, whether based on concentrations or loadings, is correlated.
THE SIGN TEST
The sign test is used to test the null hypothesis:
H0: Md = 0.
The median of the distribution of d± is zero. With this hypothesis, the
differences d^ are, in expectation, equally likely to take on positive
or negative values. The sign test statistic used to test HQ is merely the
sum of g^, where gj_ is +1 (or -1) if d^ is positive (or negative), thus
giving rise to the name sign test. If any d^ is zero, then that difference
is excluded from the sign test statistic computation.
An operational difficulty can arise whenever the g^ is computed based on
one or two nondetections in the difference pair. One approach to this diffi-
culty is to ignore all pairs d^ that do not have at least one member of the
pair derived from a concentration at least as large as the detection limit.
By definition, the sign test does not take into account the magnitude of
the differential and, consequently, all magnitudes of the same sign are
treated equally. The test, accordingly, can detect only a shift away from
zero in the median of the distribution of d^. Strictly speaking, a positive
median is neither a necessary nor sufficient condition to infer that the
average differential concentration or loading has increased.
4-38
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
1
DRAFF
THE SIGNED RANK TEST
The signed rank test is used to test the null hypothesis:
HQ: the distribution of d^ is symmetric about zero.
This hypothesis is equivalent to the statement that the distribution of
differentials from FE to NOTL is the same as the distribution of differentials
from NOTL to FE. This hypothesis is dependent on the entire distribution
being symmetric about zero, which is a precise statement that the measurements
at both stations behave in exactly the same manner, that is, with no change.
The signed rank test is designed to detect any shift of the distribution of
d± toward positive values.
To compute the test statistic rank test, which is also called the
Wilcoxon test, one proceeds as in the sign test to compute g^. The next
step is to rank d-[ in ascending order according to its absolute value and
denote the rank of d^ by R-^. If Rj_ = 1, then d^ is the smallest differential
in absolute value, and if R-s is equal to the total number of paired differen-
ces, then dj is the largest. The sign rank test statistic is computed by
adding the product of g^ and R^ over all i, the null hypothesis being rejected
for large values of this sum. Thus, the sign rank statistic considers the sign
as well as the comparative magnitude of the differential through its rank.
The discussion of the sign test noted a potential operational difficulty
in determining whether the d^ is positive or negative whenever observations
in the pair are nondetections. One encounters this same difficulty in deter-
mining the signs of the rank test, and it is further compounded in determining
the ranks. To rank d^ one must compare the absolute value of d^ with that of
all the djs. This determination can be somewhat ambiguous in that either or
both d^ or dj are based on observations that include nondetections. Based on
an assessment of the data and computations of these test statistics, different
4-39
-------
1 1/fttfT
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
approaches to treatment of this ambiguity have little effect on the values of
the computed signed rank statistic.
4.4.3 Application of Screening Test
The details of the application of the Wilcoxon test and a table of the
test results appears in Appendix C.
Thirty-three pollutants met the Wilcoxon test screening criteria, as
shown in table 4-6. Of these, 27 met the screening criteria in the suspend-
suspended sediment phase; 8 of the 27 that met the screening criteria in the
suspended sediment phase also met the criteria for the aqueous phase; and an
additional 6 pollutants met the screening criteria in the aqueous phase. Of
the eight pollutants significant in both phases, all were more significant,
as judged by the significance levels, in the suspended sediment phase.
The statistical test is of the null hypothesis that the median of the
distribution of differences (upstream-downstream concentration) is zero
versus the alternative hypothesis that the median of the distribution is
positive. That is, the alternative hypothesis in this test is that there is
a positive shift in the concentration value distribution in the course of the
river so that the difference value distribution tends to be positive. The
test then, is a one-sided test where positive difference tend to confirm the
alternative hypothesis. The detailed table of tests results is presented in
Appendix B.
On the basis of the signed rank test results, parameters were selected
to conduct a mean daily loadings analysis for the pollutants determined to be
statistically significant. Only the parameters passing the screening test,
having a statistically significant increase in concentration downstream, were
used to calculate mean daily differential loading. This procedure was adopted
because the analysis of effect of statistically insignificant concentration
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
Table 4-6. Differential Concentrations by Phase
Significance Parameters by Phase
Pollutant Suspended Sediment Aqueous
Dibenzofurans
Hexachlorodibenzo-furan X
Heptochlorodibenzo-furan X
Octachlorodibenzo-furan X
Metals
Arsenic X X
Chromium X X
Mercury X
Zinc X X
Aluminum (extract) X
Iron X
Manganese X
Nickel X
Organic-Chlorine
ABHC (?) X
1,2,3 Trlchlorobenzene X X
1,2,3,4 Tetrachlorobenzene X X
Hexachlorobutadiene X
Mirex X
PCBs X
Pentachlorobenzene X X
1,2,4,5/1,2,3,5-Tetra X
1,2,4-Trichlorobenzene X
Hexachlorobenzene X X
Ideno (1,2,3-CD) Pyrene X
Phenols, PAHs, Phthalates
Acenthe (?) X
Benzanthracene X
Benzo (A) pyrene X
Benzo (BrK) Fluoranthene X
Benzo (G,H,I) Perylene X
Benzo (A,H) Anthracene X
Chrysene X
Flouranthene X
Fluorene X
Phenanthrene X X
Pyrene X
4-41
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
DRAFT
changes on loadings provided no additional information as to loadings to Lake
Ontario. Table 4-7 summarizes the results, explained below.
4.4.4 Loadings Estimates
The mean daily loadings at Fort Erie and Niagara-on-the-Lake were paired
to estimate the mean daily difference in loadings upstream to downstream
indicated in table 4-7. Flow data, concentration data in water and in sedi-
ment, and concentration of sediment in water data must be available on the same
day to estimate the mean daily difference, which is the most significant informa-
tion given in table 4-7.
In the case of benzo(a,h)anthracene, the mean daily difference of 3.3385
has an associated standard error of 1.8304. A 90-percent confidence interval
for the mean daily difference of (-0.0660, 6.7429) covers the zero value.
This means that if a paired t-test were run on the mean daily difference,
the conclusion would be that one could not reject the hypothesis that the
true difference is zero.
For analyzing benzo(a,h)anthracene, the signed rank test used nine
nonzero differences, eight of which were positive or higher downstream than
up. The single observation that showed a higher level at Fort Erie than
Niagara-on-the-Lake was the fourth largest observation. For this pollutant
parameter, then, application of the signed rank test results in the conclusion
that there is a statistically significant increase from upstream to downstream.
However, the standard deviation of the mean daily difference in loading,
which is computed as the difference in the product of the flows and the same
concentrations, is sufficiently large so that the associated t-test cannot
reject the hypothesis that the true difference is zero.
The confidence interval column of table 4-7 indicates that only a small
number of parameters yielded confidence intervals including the value zero.
4-42
-------
s
M
UJ
S
Z u.
u -J
il
OOCMOOOOOOOOOO-KMOCVJOMKIOMCMOOOl
11 II II
•H
T3
JJ
c
(U
(4
0)
to
s
•H
4J
CO
w
l-i
(8
CO
r-
I
a:
UJ
u
UJ
a
o i
u -J
h-
X O
S5
55
UJ Z
a. iu
E N
< M
in in
|COCMr***^>OCM'^iAM«n
OKjcM-pef.-.inS-ISKlSin
o KI in -f r*. o» •« -* o -* -r m HI .^ «?
-HincMr.Kt.HCMcocooinoaSS
-S2SS3SSSg?iS?
o-: o o o « - « o o g -; o o
u1tMMin
rx«rr«.o.coin-^jSo$SiXS
o KI o o
CM O
CM CM CM -4 «
' O O O O»
325
§2S
U. I U.
i O i
O M O
S53
IU 01 UJ
co M m
MOM
§g§
gsg
xSx
U < U U X
< t- < X UJ I
x a i- a3 ac i
iu uj u i MI
OUI
a:Mt
oo:
->i--
xi
UJ UJ IU
z x z
UJ H UJ
uj U Z -J
Z < < >•
iu a: a a:
N UJ UJ X O UJ
z z z i- 3 a
UJIUUJUJZ-^-- UJ UJ
SZUK-4U.M Z Z
IU < >- — " * IU LU
gxa:a.xa£x x ac
HX— « * -UJHUJX
-IXt-<-
<
a
to
HI
UJ
| §
H- in
3 to Z
-IV) <
-J < a:
a -i z>
a. u u.
ac
o
£
X
a.
i
x
<
a.
I
1
O
(0
2
H
w
j
t
g
u
e
u
t
c
1
.
3
L-
1-
D
-J UJ
^§
a. z
K
<
^S
o'S
a. u
IU
z
UJ
NJ
IU UJ Z
Z Z iu UJ
LU iu CO Z
N N O uj
z z a: uj ac
iu uj o z uj >- f-
03 CO — 1 IU Z 0- CJ
O O X M Ul •" <
ac ac u z N o a:
o o < iu z u t-
_i _i a co uj i iu x
xxt-oaJKiziu
u u uj a: o « uj
MMKOOfCMCEC
a:a:i-io«x3 C
^(— -fXJ--
<
a
N
10
X
35
UJ
t-
<
_1
-• <
>- X
< 1-
a x
\ 0.
C9
- Z
in 5 —
LU ^
z in -t
M -a
g 5 ^
-i a. o
5 - £
1 U) —
o -i in
Z O -1
< z •<
U) LU H
CE X UJ
a a. E
UJ
in
UJ
5uj
ilS^
M = ZK
4-43
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
:a
J
Thus, the two test results, the signed rank test on concentrations and the
paired t-test loadings, would produce consistent conclusions for most para-
meters.
One can observe from the confidence interval column of table 4-7 that
the lower bound values are often very small. For example, the mean daily
loading differential for octachlorodibenzo-furan is 2.6376 grams per day.
For chromium, the mean daily differential loading is 0.0120 metric tons per
day or 12 kilograms per day. The upper bound on the mean differential daily
loading for chromium, on the other hand, is 0.0497 metric tons per day or
49.7 kilograms per day.
Table 4-7 also shows that the furan loadings in sediment are all very
small. The furans did not pass the screening test, and so they are not includ-
ed in the loadings analysis in water fraction. The loadings of furan in the
water fraction are, then, insignificant. The metals loadings in the sediment
fraction for arsenic and chromium have a lower bound of 9 and 12 kilograms
per day, respectively, and an upper bound of 40 and 50 kilograms per day,
respectively. Mercury is less than 1 kilogram per day, and zinc has a lower
bound of 70 and an upper bound of 260 kilograms per day. In the water frac-
tion, arsenic varies between 28 and 94 kilograms per day, while chromium
varies between 50 and 650 kilograms per day. All the differential daily
loadings for metals in the water fraction are large.
None of the dioxins, which were analyzed in both the water phase and
sediment phase, had a significant increase in concentration downstream and,
consequently, dioxins do not appear in table 4-7. Also, none of the vola-
tiles, which were only measured in the water fraction, had significant
increases downstream. Thus, none appears in table 4-7.
In summary, daily differential loadings were calculated for 33 para-
meters. Of these parameters, eight had a mean daily loading confidence
bound including zero and, consequently, those loadings are not significantly
4-44
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
different from zero on a statistical basis. The remaining parameters have
confidence bounds on the mean daily loading that do not include zero and,
hence, can be considered statistically significant in _their contribution
to the pollutant loading in Lake Ontario. Additional detail regarding these
loadings differential calculations are presented in Appendix D.
Table 4-8, given at the end of section 4.4, presents the mean daily
loading at Fort Erie and at Niagara-on-the-Lake. At Fort Erie, the mean
daily pollutant loading in the water phase and a 90-percent confidence
interval about that mean daily loading are presented. These figures are
followed by the mean daily loading of the pollutant in sediment and a 90
percent confidence interval for that mean. The sample sizes for both of
these means are included. In a few instances, the sample size and the mean
value presented in table 4-8 do not agree with the figures presented in table
4-7. This happens because the figures in table 4-7 are calculated from
observations paired upstream and downstream. Table 4-7 presents the means to
determine the differential loading source. Table 4-8 presents the means to
identify the loading at that station; an increased sample size is sometimes
available for the calculation.
Mean loadings at a station in a medium for a particular parameter should
not be further analyzed if the confidence interval for that mean loading
includes zero. This is so because, as discussed with respect to a similar
matter in table 4-7, a key test of the hypothesis that the mean daily loadings
are greater than zero would not be acceptable in this case.
In addition, table 4-8 indicates that at Fort Erie, only two furans have
significant average daily loadings in sediment and none in the water phase.
At Niagara-on-the-Lake, two furans had significant loadings in water and two
also had significant daily loadings in sediment.
Of the dioxins, two parameters are significant in the water phase and
three are significant in the sediment phase at Fort Erie. However, at
4-45
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
Niagara-on-the-Lake, only one dioxin had an average daily loading signifi-
cantly bigger than zero, and that parameter occurred in the sediment phase.
Nearly all the metals had average daily loadings significantly bigger than
zero in both phases. The exceptions to this statement at Fort Erie are
aluminum, iron, and manganese in the sediment phase, as well as cobalt and
cadmium in the water phase. At Niagara-on-the-Lake, aluminum, iron, and
manganese in sediment are insignificant parameters, as is cobalt in water.
The organo-chlorines at the Fort Erie station present a widely varying
picture. Most of the parameters are significant in both phases. However,
the average daily loadings are very small except for two parameters in the
water phase: 1,2 dichlorobenzene and 1,4 dichlorobenzene. At Niagara-on-
the-Lake, about half of the organo-chlorines are insignificant in the water
fraction. Only the two that were previously mentioned have large values
in the water at Fort Erie. In the sediment phase, nearly all the organo-
chlorines are significant at Niagara-on-the-Lake; however, all of the average
daily loading figures are very small.
For the phenols at Fort Erie, only a few parameters are significant in
both the water and sediment phases; they have means significantly greater
than zero and fairly large sizes. Tetrachlorophenols has a large daily mean
in water; however, it has a very large standard error. As a consequence, the
confidence interval covers zero and the parameter should not be analyzed
further. Similarly, in the sediment phase total phenols has a sizable mean
daily load. The confidence bound includes zero, however, and thus total
phenols also should be not considered further. At Niagara-on-the-Lake, most
of the phenols are significant in the water phase, and many of them have
large average daily loading values. In the sediment phase at Niagara-on-the-
Lake, five phenols are insignificant, and several have relatively large mean
daily loadings values.
4-46
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
/-s
^•^
en
en
r- 1
U
4-1
C
•U
3
r-H
.— 1
0
—
en
bC
£
•H
•a
o
cu
€
T-l
4J
en
0
^s
«
S
CO
•
oo
CD
rH
(0
H
•4
>
4
C
N
(.
b
h
c
H
C
1
4
C
•
H
I
u
H
a
u
h
C
U
!
C
H
H
_J
>
UJ
z
M
IU
u
ut Z
O C
o> —
H-
z
IU
C
1-4
iu in
c -
C H-
3 Z
<. UJ
> UJ £
-• ^ M
i a. iu a
E C hd IU
5 < >H in
•i
j
_j
' <
; z
> 5
•
i uj
; z
J UJ
j a
3 M
. U.
J OK
4 U U
5 x5
O 3
. °. ±
5
•
5 K
•* UJ
i i ?
J **
^^
uj a;
-J UJ
Q. UJ t—
< « 3
in tn -
POLLUTANT
NAME
•f CM m KI
— 0 KI -O
o in 4 in
o « x KI
es o o •»
•» o >r o
a cj o co
0 — — CM
o o o •><
o o o o
1 1
in -4 -* CM
O (7- — CJ
O -O O -T
o o o r»
CO CO CO <0
O CO " CO
o cj -r -o
^i ^«
O CM O *
O f^ KI KI
0 9- -* V
O O -* *o
0 0 0 M
1 1
0 ^5 J5 —
0 CO 0 N.
O CJ U- CJ
O -4 O KI
TETRACHLOROOIBENZO-P-DIOXIN
PENTACHLORODIBENZO-P-OIOXIN
HEXACHLORODIBENZO-P-DIOXIN
HEPTACHLORODIBENZO-P-OIOXIN
t*
o
CM
CM
CM
CM
0
CM
CM
»
>
4
« 5
i
E
i
i
•4 «
CO
>r i.
*
t
o
KI {
O U
1 e
c
u
H
J
u
in :
KI C
in »
03 h
*
KI C
OCTACHLOROOIBENZO-P-DIOXIN
_j
<
IU
h-
M
IU
UJ
a —
M H-
U. Z
Z uj
O C
U M
a
x iu
o
^ *r o KI
o o o •*
O 4) O O
1 1
cj N «4 CO
CM * CJ CO
CJ CJ -4 IT-
-* **• CJ O
0 CO 0 ->
CO CO co CO
KI co in r*
CM * in -o
+
c
. i
u
4
h
u
:
1 :
«
o :
in
in S
•< U
1 >•
B
h
C
r^ :
^ c
•M h
Kl -
t.
OCTACHLORODIBENZO-FURAN
_j
>
at
LU
z
M
UJ
UJ
Q -*
M ^-
r S
Q C
U M
UJ
zS
•» •< uj
c r ~
«s
•
•) uj
C -J M
o. iu a
J E M uj
: < M in
i in m —
n
j -i
E £
I £
j
J IU
2(J
IU
J O
: u. ^_
U UJ
: f H-
o» _
•
5
^
j CK
j w
E -
iu K
-J IU
POLLUTANT SAM
NAME SIZ
(MA
o>
KI
in
o
o
in
CM
0
0
KI
O
0
KI
-> in
KI KI
CM O
O
si
•O CM
a o
*
CO CM
— O
ALUMINIUM EXTRACT 3
ARSENIC 3
CM KI O O
o in CM o
0 0 0 ~
o o o o
U1 — O —
CM CO 0 CO
o — — in
0000
o o o e
KI *f JJ O
O KI — CO
o o o o
o o o o
KI KI m KI
•o * r*
00 M
KI O O»
O> CO 0
0 KI -.
o KI r*.
o o o
1
» 9> r-
O -T CO
O O O
< x o a
•
KI in -• -o o
•* o in KI o>
•O O f» 0 -T
-i O O O KI
O 0000
in cj o r> co
r-. o -o -4 in
C*> 0 KI 0 CO
O 0 0 0 -»
O O 0 0 0
o o in CM r*
m o at o -o
M O O O CJ
O 0000
m -4 KI o in o (Q
CM O CM O O O O
•4
£rj2o2-S
-o
V =
iu >• 3
< 3 iu Z
a iu < iu HI uj M
uiAFT
4-47
-------
I
I
I
I
I
I
I
I
1
I
i
i
I
i
i
i
i
i
i
0)
3
c
c
o
03
en
CO
c
(0
en
br
CO
O
•a
01
4-1
CO
W
M-J
O
S-l
CO
00
CO
H
a
v.
19
n
ui
U
in
in
z
2
UI
M
Q!
UI
u
ui
Q —
o in
IT- —
z
UJ
M
z o
< UJ
ui in
a. LU a
c M ui
< M in
«n tn —
2
a
z
ui
z1
UI
a
u.
U ui
ooooooooooooo
oaooooo
ce
ui
u. Z
Z ui
o r.
U M
> o o e o o o a
ooooooooooooo
ooooooooooooo
OOCMO-*OCMOOOO<
ooooooooooo
z o
< UI
ui in
— -.OCV —
OOOOOOOOOOOOOOKIO-«-
a*
UI
t
HI
UI
U
Z
UI
a
M
U,
Z "^
O Q!
U UI
O
o
o
o
0
o
o
0
o
0
•^ CJ CO f** f*1 O O
in o r-. — ^ o -
m o *o in co o -3
^ 3* •* f** Is* o ^
tsj 9> C\J O CU O *4
-* CJ
o CM o o o o r-
CNJ co ut -T a* o ir
^ ro r» — • %y o -d
<0 o m N •-« o ~
o — o o o o -
1 1 1
oooooocMor-o-oooooooo
£
M Z
o r«i
03 -3-
»> o -3
-o o *j
O.Z <<-l0
CMCJCMCMCMMCMCMCMCJMCMCUtM
£
a ui
M >H
x o
a <
a. oc. \~
a a x x o
O O U U Qf
z -J -J >• a o
zaZuuQoaao i: x
Q£ — J Of H" H* iiii 3T u ^ tn ^
ouiDa.a.o.Q.Q.ai-Q:ocQx
- UI UI <
t- x z Z a:
K CX Q. <'Z
3 •< < ac. <
-J ui Z Z X N
O < U (J Z ui
Q» Z < < < ffl
O* CM CJ*
UI
z
UI UI
Z X
UI I—
u z
o: or
UI X O
Z 1- 3
UJ Z -1
CE < u.
>- — — •
< <«
CD O 5
Kl NJ N
UI UI UI
CD CD aj
CM 9
UI
UJ
_J
o:
IU
a,
M
r
"• IL
— UM
O W1
KJ >
UJ Z
OJ (J
4-48
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
13
0)
S
(2
c
o
03
CD
r-l
U
4J
S
o
O-i
03
M
•H
T3
to
O
T3
0)
CO
e
CO
3
00
I
(0
H
KAFT
at
uj
Si
°s
N: ui
O V)
>•
<
o
a ••
<-> t
u. Z
Sr
°3
X UJ
o in
a.
1
I
in
u
H
I
S
<
u
u
z a
< UJ
uj in
z
UJ
uj r
-J M
a. uj a
£ M uj
< HI in
in in —
UJ
y
i =
U uj
a
55
U Z
< HI z
in in —
r»or»r*icoor»o>oo>oeaeo
coo-ooocococoocoococa
z a
< u
uj in
< z
at u
h- 111 C
Z -i HI
uj a. u Q
U C N UJ
Z < HI in
5 in in —
u
-<-
S<\joKimr»^-4 — -. — c
cxjf*i^ojr^ * in r» a]
4) M — i Kl
in
in uj
3 z
U uj
a
5 S.
50 at
u uj
•o
o
U1
•o
0
tn
CM
*
1*1
CM
1*1
0
CO
•O
CO
o
CM
O
CM
0
„
o-
o>
(•4
CO
9*
t>»
•0
S
1*1
o
•T
0
e>
1*1
^4
••
o
•o
o
1*1
0
in •«
•T
in
£
UJ
HI
a
ae
o
X <
O 3
17- —
— 03 in o KI -«
i« •« m —i > CM CM CM CM
1PYRENE
a
u
UJ 1 UJ
Z 1*1 uj z
UJ - Z UJ
x CM uj a:
H- UJ • -J X
Z Z - < H.
< UJ — X Z
§g z x z
3 uj a. uj
J -i a < x
o*0»e>»CMOcM^o *
c
at
*-
3
^
X
H> UJ
a. < mm
— -J _i -j
-j •< in o o
>• x tn -i z z
x >- -i a uj m
uj z g z z x
x o. z uj a. a.
-J I uj X O O
>• -i x a. o: a
x v a. o o o
H. H- O ot _J _l
UJ U K Q X X
UJ 1 O O -J U U -J
z CM i _i x < < a
uj-~zxureH>z
C£ 10 1 U i-l H- Z UJ
^MMMQJUJUJX
Q.OJOOI— I-CLQ.
( —
J UJ Q£
3 -J UJ
j a. uj H-
< M 3
tn in —
H-
Z
H"
—1 UJ
^5
a z
OO>C»
ACHLORIDE
HANE
ETHYLENE
OETHANE
at H. o ct
H- E UJ CK O
uj a: £ a -J
H- O O -I X
uj u. a: x u
Z Z o a u HI
uj o at -J < Q
N 03 O X Ot
Z Q! J U H. CM
UJ < X HI UJ «•
en o u Q i— -.
4-49
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
DRAFT
«•
/— V
•o
3
p
•H
4-1
c:
o
/^•^
w
x_/ «
e
*">
CO u
CO -
3 I
4J H
e c
§
4-> «
.3 5
0 C
>> 2
•
cn -
M C
c •
•H y
O
|J U
J
4_) j
tO
5 i
CO .
M !
O i
1
^ !
1
•
i
i
» ^
; >
C IU
: t
J S
j
2IU
u
J IU
£ O
J U.
u 5 ce
" N-<
5 P* i
c
( CK
f 55
5 !c —
H
<
-1 IU OC
-J IU
O. IU K
«n S —
H
-i IU
KI in « r» 43
e «j 49 co o
o o «J tg co
e — — »» 4)
»«•««-
O 4} —• t* CO
o -• 4) -i * <0
•M eg 9* m KI
o KI eg in >r
0 •»• m CO »»•
o o o eg KI
*
>
e
\
i
co CO « CO CO S
G
u
i
u
u
4
,j> M r*. 4) o> *•
o 4) **^ (^ eg .
0 CO — 43 -• =
o eg x KI co t
e
rv r- to -y c- c
«« e •« r«" 4) .
eg ^ tv o> 0 J
o o o e —
i i i i i u
a
_>_<_>>. >•
4
KI -O 9- — 0 J
eg eg in o * '
e i« -T in * ;
o — o — co
t
f
«
t
c
TETRACHLOROOIBENZO-P-DIOXIN
MNTACHLOHOOIBEHZO-P-OIOXIN
HCXACHLOROOIBENZO-P-OIOXIN
HEPTACHLOROOIBENZO-P-OIOXIN
OCTACHLOROOIBENZO-P-DIOXIN
-i
<
IU
M
IU
IU
O ~
M H-
lr
U -i
v 5
o in
a- —
H
UJ
x:
M
< U
. iu m
C —
I
i __
i H-
P "««•
t UJ
3 J £
O. IU O
E M UJ
< 1-1 in
i cn in —
i
j
' ^
. ^
; 1
- IU
2 z
J M
L IU
y
i UJ
( 0
J l-l
•^
E O EC
U u
0 3
C ~
e
9
C **
4 0!
E iu
i Z >-
2 5
(
j —
-1 IU
O. UJ h-
in in <_
POLLUTANT
NAME
4) O C^ O> 4>
OD eg — 43 r»
in h* c> o 4}
o o in « 4)
o co eg eg o
r* in eg o *
*y «"^ ••• *y ••)
O <6 ^ •& f^t
o M o o m
i i i
o KI .y o tn
-o -o co r^ —
in eg KI rv ^
O KI — — 4> >
4
5
2
:
u
CO co CO «0 00 *
1
0
u
4
(.
in KI m co o S
in o KI KI KI *
KI u^ v^ M •
«
o o in eg 4) u
r» eg eg «0 o «
o eg «r -T eg. •
eg eo «• — o u
i ii p
f
j
«
r» eg o o co I
co o -« r* in •
eg O o eg r^ ¥
0 in KI CO 43 :
43 o> r« in •* ;
— eg c
!•
«
\
00 « 43 CO CO
TETRACHLORODIBENZO-FURAN
PENTACHLOROOIBENZO-FURAN
HEXACHLORODIBENZO-FURAN
HEPTACHLORODIBENZO-FURAN
OCTACHLOROOIBENZO-FURAN
_i
>
5
iM
^
IU
7 Z
A P
U M
*"
H-
: c
5 -rS
: < iu
5 iu in
: c"
i
j
K —
iu C
-J M
1 O. IU O
t C N Uct
c y cu ftj »M
« o o ey o m
o o o o •* -•
o o o o o o
* CSJ C\J «M O — K» O3 O*
4) * o -• 4) •> r^>
43 0
eg
«r KI o eg co in o
KI co o >r -f •a •»
oooo — in o
o r*
— • eg
1—
et
X
IU
5ur5
Z M D M t- Q;
5mot-ccoQ.o<
e o o in
oooo
in — o —
o o> eg >r
O o o eg
o ^ eg r»
o 43 o eg
O 0 0 -f
oooo
• » eg eg eg
~
O O — KI — 1
co o in in —
eg o 40 o *a*
Kl O Oj — O
r- o o o -
IU
in r
iu >- 5
(d CJ ^ IU U
Z K U -J Z
4-50
-------
3
s
Q-
1 B
• UJ
z1
UJ
Isz
II
^ oS
• -c §2
1 1
•^ t
4-1 Z
__ C »
• o-5
• U >- 20
•, v_x < < UI
•1 a ui in
-7 i =-
pTV| ^
15 z
UI f-
CO M W
co a u E
J5 -1 Q. UI O
7; x r M ui
•1 ^ u «( M in
• i in «n —
I c 1
TO IB -j
3 g 2
1.3 " Z
I-H m 5-
o S z
fc < M
* ^ 5
f 5 E
| 5 S
C J z -
•H j 5 a
-a o u u
IS x <
C 3 *"
T3
01 iu —
U X K
ICO ^ iu
e z 5 5
3 s ii
CO <
w 5
• «w S u. S
• o 2 " 5
• z a. u fr-
iable 4-8 . Summary
LOCATION?
SAM
SIZ
IMA
• I
• g
1 £i
1
1
^
?
y
eeooeeeoooeoo ziu
•• (jw
X UI
a- —
i
IU
— oooooooooo-*o >. z Q
oooeeoeeeoeoo s C —
in —
^ 5
M IU
gui c
^ M
u < M m
i in «n —
13 -1
g >
M^^«o««jr» ui
« .J Ott
gU UJ
II II X <
-J
<0-^KI^KOC!^0 — ^0- S *
s
< iu a
M -J IU
» r M <
M
5
-i
IU
a
M
X
g
Z Z iu
£ zza ** z
-t < < a 55 <
3 a a 5 z -i -J i-
inuiKB-li-ixXH>»- 3
o zooxZKZuooo -i ui
ou^aj^o^o^^^^ao ^c
ZXBXXIQC^DE^^II a<
4-51
»f^O>->0-.KOO-ONNI/<« — — *J
1
sslsllslislls5gs£
KI^BJO«VOtf(l«* — K^l^^f^rv^
SSS5KSi?SS5SSS;S!2RS
•» •« —
Kt(U«l«ir«««l^««m(MI>.«- a oxria:o:ai\-'G
etux M -Juuut-t-^-^trtr-J
o Q o r XMMMI i i » - u x
&&^KOoDXM^*TtMC\jf^^i(wXX
* *iu*4XUtoJ • • • • * * * * tu uj
DRAFT
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
ID
3
C
C
O
03
W
(0
1—I
u
4-1"
CO
w
3
C/3
OO
I
sr
to
H
>
IU
iO*iOcg«.*cg«.?(g-«oo — meo» — —
i i
X UI
e in
I i
^ UI
3! M
IU Z O
h* < UJ
2 £2
i
I
»
m
tu a
N ui
*-t in
in in —
O
eu
.0
(0
be
C
•1-1
•a
o
iJ
tu
z
UI
£
in
in
u
1
|
s
>
ac
UI
1
UI
UI
a
*•<
O K
SJ Wl
^•v
§!
evi eg w -• o o -r
KI P< eg
e ocuoo — o->-<
•« M i in
ac e
ui e
i§
1
M
i
ui CK
-I IU
Ot U! h- M
IB
^ ^ ^
UI UI X Q IU
z z t 5 a
•too
O X H J Q ui ui
U iu X O Z X X
i ui x a. z ui E S
UI Z -J i iu X O O
w >- -J z £ a g
§,
_l ;
Si
4-52
ac
2
a
DRAFT
2 5
o ac
N: «t
03
r»
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
CHAPTER 4 REFERENCES
1. Dixon, W.J., and F.J. Massey. 1969. Introduction to Statistical
Analysis. New York: McGraw-Hill Co.
2. El-Shaarawi, A.H., S.R. Easterly, N.D. Warry, and K.W. Kuntz. 1985.
Evidence of Contaminant Loading to Lake Ontario from the Niagara
River. Canadian Journal of Fish Aquatic Sciences. 42, p. 1278.
3. Gibbons, J.D. 1976. Nonparametric Methods for Quantitative Analysis.
New York: Holt, Rinehart and Winston.
4. Hollander, M., and D. Wolfe. 1973. Nonparametric Statistical
Methods. New York: John Wiley & Sons, Inc.
5. Mood, A., F. Graybill, and D. Boes. 1974. Introduction to the Theory
of Statistics. New York: McGraw-Hill Co.
4-53
-------
1
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
POLLUTANT
CLASS
FURANS
DIOXINS
ORGANO-CHLORINES
PHENOLS, PAH'S AND PHTHALATES
VOLATILE CONCENTRATIONS
APPENDIX A
Detection Limits
POLLUTANT
NAME
TETRACHLORODIBENZO-FURAN
PENTACHLORODIBENZO-FURAN
HEXACHLORODIBENZO-FURAN
HEPTACHLORODIBENZO-FURAN
OCTACHLORODIBENZO-FURAN
TETRACHLORODIBENZO-P-DIOXIN
PENTACHLORODIBENZO-F-DIOXIN
HEXACHLORODIBENZO-P-DIOXIN
HEPTACHLORODIBENZO-P-OIOXIN
OCTACHLORODIBENZO-P-DIOXIN
A-ENDOSULFAN
A-BHC
A-CHLORDANE
G-CHLORDANE
ALDRIN
DIELDRIN
ENDRIN
HEPTACHLOR
HEPTACHLOR EPOXIDE
P,P-DDT
O.P-ODT
METHYOXYCHLOR
MI REX
PHOTOMIREX
PCBS
HEXACHLOROBUTADIENE
1,2-DICHLOROBENZENE
1,3-DICHLOROBENZENE
1,4-DICHLOROBENZENE
1,2,3-TRICHLOROBENZENE
1,2,4-TRICHLOROBENZENE
1,3,5-TRICHLOROBENZENE
1,2,3,4-TETRACHLOROBENZENE
1,2,4,5/1,2,3,5-TETRACHLOROBENZENE
PENTACHLOROBENZENE
HEXACHLOROBENZENE
BENZO(A)PYRENE
BENZO(A,H)ANTHRACENE
BENZO(B+K)FLUORANTHENE
FLUORANTHENE
PYRENE
BISC2-ETHYLHEXYDPHTHALATE
DI-N-OCTYL-PHTHALATE
TRICHLOROPHENOLS
PENTACHLOROPHENOLS
PHENOL
BENZENE
CARBON TETRACHLORIOE
DRAFT
SEDIMENT
UNIT
PG/GM
PG/GM
PG/GM
PG/GM
PG/GM
PG/GM
PG/GM
PG/GM
PG/GM
PG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
UG/GM
MG/KG
MG/KG
MG/KG
MG/KG
MG/KG
MG/KG
MG/KG
MG/KG
MG/KG
MG/KG
DETECTION LIMIT
(SEDIMENT)
2.000
5.000
5.000
5.000
10.000
2.000
5.000
5.000
5.000
10.000
0.002
0.001
0.001
0.001
0.002
0.002
0.002
0.001
0.002
0.004
0.004
0.005
0.002
0.002
0.010
0.001
0.100
0.100
0.100
0.010
0.010
0.010
0.002
0.002
0.001
0.001
0.001
0.001
0.001
0.002
0.002
0.002
0.002
0.002
0.005
0.010
f
WATER
UNIT
PG/L
PG/L
PG/L
PG/L
PG/L
PG/L
PG/L
PG/L
PG/L
PG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
NG/L
UG/L
UG/L
UG/L
UG/L
UG/L
UG/L
UG/L
UG/L
UG/L
UG/L
NG/ML
NG/ML
DETECTION LIMIT
(WATER)
2.000
5.000
5.000
5.000
10.000
2.000
5.000
5.000
5.000
10.000
0.080
0.050
0.050
0.050
0.080
0.080
0.080
0.050
0.050
0.080
0.080
0.080
0.080
0.080
0.500
0.050
0.500
0.500
0.500
0.100
0.100
0.100
0.050
0.050
0.050
0.050
0.002
0.002
0.002
0.004
0.004
0.005
0.005
0.050
0.050
0.100
0.500
0.500
A-l
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
DRAFT
POLLUTANT
CLASS
POLLUTANT
NAME
SEDIMENT
UNIT
DETECTION LIMIT
(SEDIMENT)
WATER DETECTION LIMIT
UNIT (WATER)
METALS
CHLOROFORM
TETRACHLOROETHYLENE
1,2 DICHLOROETHANE
ALUMINIUM EXTRACT
ARSENIC
CADMIUM
CHROMIUM
COPPER
IRON
LEAD
MANGANESE
MERCURY
NICKEL
SELENIUM
ZINC
MG/KG
MG/KG
MG/KG
MG/KG
MG/KG
MG/KG
MG/KG
MG/KG
MG/KG
0.05
0.50
0.50
0.10
0.50
0.01
0.10
0.05
0.10
Note 1; To convert pollutant concentrations in suspended sediment to con-
centration equivalence in water multiply the pollutant concentration in
sediment by the appropriate mean concentration of sediment in water from
the following table.
NG/ML
NG/ML
NG/ML
MG/L
MG/L
MG/L
MG/L
MG/L
MG/L
MG/L
MG/L
MG/L
MG/L
MG/L
MG/L
0.50000
2.00000
0.50000
0.00100
0.00010
0.00100
0.00100
0.00100
0.00100
0.00100
0.01000
0.00002
0.00100
0.00010
0.00100
FE
NOTL
OVERALL
Sediment Concen-
tration Mean
6.4412
8.8576
7.6314
Standard Error
of Mean
1.1079
1.7783
1.0328
n
34
33
67
max
26.4
39.6
39.6
min
2.0
1.3
1.3
The standard error, etc. is presented above to give an overall view of
the measurements.
Note 2; The following table gives a global view of the equivalence of
sediment concentration units and water concentration units:
sediment unit
1
1 yg/g
1 Pg/g
average water unit equivalent
7.6 x 10-6 mg/1
7.6 x 10-6 mg/!
7.6 x 10-6 mg/i_
A-2
-------
I DRAFT
I
I
I
I
I
I
APPENDIX B
Calculation of the Confidence Intervals
The confidence intervals presented in tables 4-7 and 4-8 are calculated
by reference to the Student's-t statistics. Specifically, the formula is:
• — / a \
I lower endpoint = x - t — , n-1 and upper endpoint where
• I 2
.
x = X/n and
(x. - x)2(n-l)
In the application in table 4-7, the x-j/s are the upstream/downstream daily
differences in calculated loadings.
/
2/(n
I
I
I
I
I
I
I
I
I
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
UfiAM
APPENDIX C
Screening Test Results
The table in this appendix presents the details of the screening test.
The daily concentration values were used as the basis for computing this
table. It lists, by pollutant, the sample size used to compute the signed
tests, the asymptotic significance level, and the details needed to determine
the exact significance level. The last column is the alpha level of the test
as computed through the z score, which is a large sample approximation. The
stars beside the alpha level, which is .05 or smaller, indicate the statisti-
cally significant parameters.
This screening test is of the null hypothesis that the distribution of
the differences and concentration of the parameters upstream to downstream
is symmetric about zero. This means that there is no difference between the
upstream and downstream concentration values. The alternative hypothesis
considered is that there is a positive shift in concentration values in the
course of the river. Consequently, large values of the Wilcoxon test statis-
tic indicate that the null hypothesis is not true and the alternative hypo-
thesis is more plausible. That is, large values of the test statistic give
evidence that there is a significant increase in concentration down the
river.
The steps used in formulating the test statistic are as follows:
(1) Compute the d^ = downstream loading - upstream loading. Ties
in upstream and downstream concentration, di=0, are omitted.
(2) Rank the absolute values of the di from a smallest of 1 to a
largest of n, where n is the sample size; omit zeros in the
di (tied loading observations); denote the ith d's rank by R^.
Where tied observations exist, d^=dj, the average of the tied
ranks are assigned to each.
C-l
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
(3) Construct gj_ = 1 if the associated d^ is positive. Use
gj_ = 0, otherwise.
(4) The Wilcoxon test statistic is
n
T+= 2 §i Ri> which is the sum of the ranks associated
with positive difference values
(5) The test statistic distribution has been tabulated, f9r example,
in Hollander
sample sizes:
in Hollander and Wolfe (1973) for small sample sizes.* For large
4 -0.5
Z =
has an approximate N(0,l) distribution. The value of the Z
statistic appears in the next to last column of the table, and
its associated (upper tail) probability level is given in the
last column. Values of a level of .05 or smaller indicate
statistically significant scores.
Extra columns are presented in the table to provide the information
needed to utilize a table of the tail probabilities of the null distribution
of the Wilcoxon rank sum statistic. Such tables are given in Hollander and
Wolfe. The first value in the table is the sample size (excluding ties).
This value is referred to as n, which is the number of nonzero differences in
concentration values. Differences of zero indicate tide observations up and
downstream, which yield no information concerning the question of increases
in concentration. The second column referred to as the sample size (differ-
ence greater than zero) is the number of positive differences; it represents
the observations in which an actual increase in concentration was noted.
The third piece of information is the sum of the ranks associated with the
*Hollander, M., and D. Wolfe. 1973. Nonparametric Statistical Methods.
New York: John Wiley & Sons, Inc.
C-2
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
DRAF1
positive differences. It is this value that is the Wilcoxon test statistic.
Using these three pieces of information in table A5 of the Hollander and Wolfe
reference, the exact significance level of the test can be determined. This
information allows one to compute the exact significance level of the test.
The exact and the approximate significance levels existing in the table
differ very slightly, and conclusions based on the approximate signficance
level will not be altered when the exact significance level is determined.
C-3
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
DRAF1
UJ
u_
..J
l
§
5 j
IK ^
P h
Z .(.
UI 4
(J 0
Z u
U h
2
Z l
0 X
h
in i
UI II
H 1
U
Z '
2 i
Q
UJ
0
to
o
X
0
u
..J
3
I
a.
vJ
UI
g
1C
i
g
w
o
M
CO
^
•H
0
o
It
•r-4
Q
I-J
U
X
w
UI
Z
VI
in
- in •» -l>-"a)MBJ
QOMQMOQMQMZ
OOOOQOOOOQ<
glXOCEOCXCCOIXOU.
O Q£ O D! O O tt O Of *^
-j-io-jo-j-io-iori
xx-ix-ixx-ix-Jin
uuxuxuuxuxo
UIUIUIUIUUIUJUIUICJ 1 1
P-Q-XXOI-Q.XXO<<
E
(9
(2
in
UI
-S3
E t£ ce
U) X 0
13 (X ^
Q. ~— y
^ t/1 1
in z o
Z M Z
EC O 19
3 M a
u. a o
* * *
§££?8ggS5£o:sggS3S
f*i<7*p*»cy*orninfSi-/»<^'<^wooo«ooit
ooooooooooooooooo
liliillilllslllll
lit 1 1 1 1 II
lAooirtiAi^oooi/tatAooooo
mcu^«N-*i*
^-*cMr<^-*4
'f *y UI |A >O ^ K^ 1A ^* fO M Lft K^ O* CVJ Kl W
JJCMMI^^^inj-ca^eO^IOOCM^
UI UI UI
UI Z Z Z
O UI UI UI
M M N N
x a z z
O < UJ UI
UI IX OL \— ID (D
Z ui O 3 O O
zza a: x ooo
aaoz-i >- o x x
aJCK-'M XHMQUIX — 1 U O
zcoxKZuaoaoa XMM
a x x i — i CK »- i i i ixuim- z
1 UI
in M
ro ui
,:1
"2
m x
• u
* *
isss
o w o -*
0000
s^ss
o a e o
CM
-
co r» o
CM —
UJ
Z
UJ
ISuj
O >- UI UI
gx x z
K H> UI
x1.?^
< Z Z X
ui U U Z
(9
s
E
in
UI
h-
h-
X
a.
0
<
in
5
a
tn
i
UI
x
a.
x
2
o
o
1
a
PM
-
CM
JACENE
h&
x
f—
z
<
N
UI
OJ
* * *
CM O Kl
O O- O
O CM O
o e o
ooo
-» in m
•o o> in
in co CO «O
CO O O *
— ooo
o o o o
o o o o
•o in -r -o
o> >-i in o
O %T ** *O
0000
— f- o in
•* **
- -
CM CM
UI
Z
UI
cr
UJ
a.
— UJ
M Z
- UI
X X
- ui i— aj
(0 Z Z Z
— UJ < UJ
o in a a
N >- O 0
Z a: 3 3
UI X -J -J
CO U u, u.
C-4
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
DRAFT
UJ
u.
5
J
H
§
§
^
<
K
^
UJ
U
z
u
§
1-
in
UJ
H
2
K
a
Ul
2
M
in
I
0
u
_J
M
Z
I-
h
(.
4
a
u
H
2
l
h
C
u
V
1
u
V
4
i
= >
Q. W
«J i_J
i
y
in
Nl
si
^ 00
5
K +
w
BH -t-l
CO
i °
H ^
J
[
E j*^"^
- J" o
- M H
: in -r-i
j /^
5 _i
3 a. •
j r: pJ
1 < u
in c<
j ?S
1 W
C ^^^ ' '
POLLUTANT
NAME
i
3 in
-> tn
-j <
o —i
x *
co a r» CM o
5 3 2 o 5
-* O O O Ul
o o o o e
si^iis
o u) eg 10
«H
* "" ~ S —
IU
5
<
UJ X
IU X
IT £
0. It
-> >-
O X
U UJ
i uj x
IOENO(1,2,3
NAPHTHALENE
PHENANTHREN
PYRENE
BISIZ-ETHYL
«
*
Ul
e
i^
**
o
i
o
Kl
2
THALATE
x
0.
1
>•
H1
Z
1
e\j
£
o»
0
g
o
CM
1
0
**
in
§
DICHLOROPHE
•*tt4->?OcOOsO
o — PJSotoo?!
>O O
0 O 0 0 0 O
g JC5255
-^ O <*• ftj O* O
O W K» K» P4 *f
1 1 1
in i/i o ir> a i/i
CM — ( CJ —* Irt
*N *4 CM p^ CJ
B
•
C
H
c
4
n
u
CM ^ -i I
4
2
l
u
u
4
:
E
>. §
a a -J M
uj 5 uj z
a. < a o -J z
<
1
_i
<
Q!
O
y
N
i
^
^
oc
M
H
1— l
CO
O
P-i
UJ
N
W
in
_i
: in
^
>
:
e
j
t
:
j
>
".
5
_i
a.
^
5
-J
_J
^j
u
D>
^
0
CO
4-
O
^^
o
•H
j
u
X!
W
^~s
jj
in
^
u
ooooooooeeoo
oSSS2St5jSS2:S3!
^Joesoo — -i — ooo
it ii i i i i i
OOIAOOOOIAOOOO
'°SS3llo"'«e>*K;R?
~*
~*
+ « * « ^) - « « « «M ~ ~
z z z
IH >-l Z M Z
X X w X M
zz zooxox
Of OC ^ tt *t O O HH O ^
3 3 K 5 O: 1 1 O 1 Q
u. u. 5 u. 3 o. a. ID. i
i i u. i u. i i Q. i a.
o o i o i o o i o i
NNIOMONINONIO
ZZIMZNZZNZN
UJUJZUJZUJUJZUJZ
03 O3 UJ 03 UJ 03 03 UJ QJ U
MMCOMCQMMO3lHa3
§QMDMOOMa>-lZ
OOOOOOQOQ<
gaocEOaaoaou.
oaeoaeooaoac-j
— 1— IO— 'O— '— 1O-JOU
xx-ix-ixx-JX-Jin
uuxuxuuxuxo
oe^
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
BUR
UJ
u.
(ft
_l
h-
I
§
M
5
0! C
h- *
Z H
UJ L
Z 1
O u
U
C
§ r
4
i- :
in i
UJ U
1— «
4
i '
K
a
UJ
i
M
§
§
U
-4
M
2
|g
•*.j
IU
g
u
in
IM
So
g 03
< ,
a +
M
H -H
M Q
C/3
O
PJ
N 0
M II
-H
m Q
a •
r j
i w ><
* W
J v~'
(
c
L
C
J
i
c
i
j
i
c
\
K-
5
—1 UJ
a. z
z
2
D tO
-J tn
•j ^
O -1
a. u
siiisiiiii
oooooooooo
cMcjincoNoco-ocgo
III 1
OOOinOOOOIAO
•-t K» K1 ^ O t> 0
•^
in-oaoMCMoco
•^ ^4 *4 1^ <-* i^
Ul Ul
z z
UJ IU
M M
a Z
< Ul
g 5S
^ X gg
u ui a: -J
(- K iu x S SS
a a a o o E x IH
a o a o >- x o u a
i i i i x iu \— in < i
> - - -IUMXUIU -
a-on-H-CEo-ax —
8
o
in
i
o
"
NZENE
CHLOROBE
M
a
i
|
o
3
i
o
9-
NZENE
CHLOROBE
M
o
1
» * *
0000
sss-
1
o o o o
Kl O
••« V*
Ul
z
Ul
ui iu iu z
Z Z Z ui
IU Ul Ul ID
N IM M O
Z Z z a:
ui iu ui O
Sggx1
TRICHLOR
TRICHLOR
TRICHLOR
<»-TETRAC
iii>
«:«;««;
*
O O
2S
OM
O O
in o
"* "*
Ul
Z
IU
M
Z
Ul
to
O
a:
O
_i
x
u
<
a
h-
UJ UJ
t- Z
1 UJ
in N
5/1,2,3,
HLOROBEN
- u
*
o o
?§
1
in o
»<4
* °
"^
IU
Ul
LOROBENZ
CENE
x <
< X
ui Z
X <
ID
=)
in
IU
t-
X
a.
§
^
V)
5
a
v>
—i
i
UJ
x
n.
5
o
$
0
o
r-»ininin - — -J _) _i a uiZK
a -i < in o o o -j < o
— >-xm-izz -J iu>-x<
a XH--IOIUUI x ZXHCK
U UJXOZXX U < t- UJ 1-
ui i ui x a. 2 uj D. a. < xuiox
Z KIU1Z -IIUIXOO IE h-OUUI
Ul *ZUJ ^•JXO-Q£Q£ H£lUKO
X CMUJQ! X>-CLOOO Ul K £ O — ' C
UIKUJ--IX h-t-OOS-J-l I-OO-IX3
ZZZ — -OO2X2uj — ZXLJCEh-ZMCDOXQ: C
X^-IO-MIHI-IQ:IUUJXUI
-------
1
1
1
1
1
UJ
It
1 I
11
c
< *
CE C
fc t
IZ r
UJ (.
1 i
§
• - 3
• in i
• £ i
£ i
Is
0
i
w
(0
II
X
o
U
-J
M
1
1
1
1
1
1
1
* !« * * 1«
£1? (M^JOCVJOO — O
g ooo>ooor«o
< J oooooooo
1 iiiiiili
*N *"^ M O fO
-------
APPENDIX D
Appendix D provides supplemental information to table 4-7. It presents
the mean loading and the standard error of the means for each station, in
addition to the information given in table 4-8.
I
I
I
• Differential Loadings
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
IAFT
D-l
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
DRAFT
SUMMARY OF ESTIMATED DIFFERENTIAL LOADINGS FOR SCREENED POLLUTANTS
POLLUTANT
NAME
HEPTACHLOROO I BEMZO- FURAM
HEXACHLORCO I BENZO- FURAM
OCTACHIOROD I BEMZO- FURAM
POLLUTANT SAMPLE
NAME SIZE
ARSENIC 31
CHROMIUM 31
MERCURY 21
ZINC 31
POLLUTANT
NAME
ABHC
HEXACHLOROBENZENE
HEXACHLOROBUTAOIENE
NIREX
PCB'S
PENTACMLOROBENZENE
1,2,3-TRICLORBENZENE
1,2.3,4-TETRACHLOROBENZENE
1,2,4,5/1,2.3.5-TETRA
POLLUTANT
NAME
A CENTRE
BENZANTHRACENE
BENZO(A)PYRENE
8ENZO< A , H ) ANTHRACENE
BENZtKB+OFLUORANTHENE
BENZO(6,H,nPERYLENE
CHRYSENE
FLUORANTHENE
FLUORENE
PHENANTHRENE
.-•----- --GKuUfSEDIHeNT rKAUl 1UH kunruuw lire *viDCH£uruiuuiaiii/UAT j
SAMPLE MEAN STO DEV MEAN STO OEV MEAN STD OEV
SIZE (FE) MEAN(FE) (NOTL) MEAN(NOTL) (NOTL-FE) MEAN(NOTL-FE)
18 1.0988 0.5S13 1.3770 0.4195 0.2782 0.6909
18 0.2121 0.1211 1.1384 0.8354 0.9263 0.8586
18 0.6953 0.4134 6.9415 2.1414 6.2461 2.0739
CONFIDENCE
(NOTL-FE)
( -0.9239,
( -0.5677,
( 2.6376,
MEAN STD OEV MEAN STD DEV MEAN STD DEV CONFIDENCE
(FE) MEAN(FE) (NOTL) MEAN(NOTL) (NOTL-FE) ME AN (NOTL-FE) (NOTL-FE)
0.0392 0.0087 0.0633 0.0152 0.0241 0.0090 (
0.0347 0.0098 0.0655 0.0148 0.0309 0.0111 (
0.0004 0.0001 0.0009 0.0002 0.0005 0.0001 (
0.2677 0.0483 0.4326 0.0808 0.1649 0.0560 (
..__»».. f*9ft IDsCCn I fcfCUT CD AI^T f Oil f*f^Dn Itin TVDC •/••flf • yrt Out nDT UCO/b'P /ftA V A
SAMPLE MEAN STD OEV MEAN STD DEV MEAN STD DEV
SIZE (FE) MEAN(FE) (NOTL) MEAN(NOTL) (NOTL-FE) MEAN(NOTL-FE)
18 0.0034 0.0011 0.0096 0.0018 0.0061 0.0016
18 0.0041 0.0014 0.1526 0.0647 0.1485 0.0650
18 0.0036 0.0020 0.0389 0.0258 0.0353 0.0259
18 0.0002 0.0002 0.0104 0.0045 0.0102 0.0043
18 0.2376 0.1076 0.9889 0.2737 0.7513 0.2840
18 0.0010 0.0006 0.0781 0.0259 0.0771 0.0260
18 0.0004 0.0004 0.0088 0.0048 0.0084 0.0047
18 0.0034 0.0017 0.1210 0.0464 0.1176 0.0468
18 0.0022 0.0014 0.0526 0.0296 0.0504 0.0298
. pDmOvCcn f UCUT CDAPTII^U pmo/viun TVDC — oucum c DAMIC Aun DUTUAI *Tce/^/*/rs*v\
• -GKOUrm56DIWENT FRACTION COMPOUND TYPE =rH£NOL5, HAH'S AND rHTHALATES(KG/DAY)- - •
SAMPLE MEAN STO DEV MEAN STD DEV MEAN STD OEV
SIZE (FE) MEAN(FE) (NOTL) MEAN(NOTL) (NOTL-FE) MEAN(NOTL-FE)
9 0.0089 0.0089 0.9980 0.3714 0.9891 0.3702
17 1.9174 0.8067 6.6226 2.5514 4.7051 2.0657
17 1.2413 0.5942 5.0785 1.7046 3.8372 1.2513
9 0.7053 0.4099 4.0437 1.8583 3.3385 1.8304
17 2.4602 1.0035 11.5139 4.3796 9.0538 3.5417
6 0.0619 0.0619 2.1802 1.4074 2.1182 1.4225
17 0.7948 0.2733 4.9770 1.7779 4.1822 1.6261
17 2.8742 1.1017 14.5367 5.5172 11.6625 4.7726
9 0.0446 0.0446 0.8698 0.3638 0.8252 0.3605
9 2.0059 0.7769 9.4864 3.5002 7.4804 3.3765
0.0089, 0
0.0120, 0
0.0003, 0
0.0698, 0
CONFIDENCE
(NOTL-FE)
( 0.0033,
( 0.0354,
( -0.0098,
( 0.0027,
( 0.2570,
( 0.0319,
( 0.0003,
( 0.0362,
( -0.0015,
CONFIDENCE
(NOTL-FE)
( 0.3006,
( 1.0985,
( 1.6525,
( -0.0660,
( 2.8700,
( -0.7481,
( 1.3430,
( 3.3294,
( 0.1547,
( 1.2001,
INTERVAL
1.4803)
2.4203)
9.8547)
INTERVAL
.0393)
.0497)
.0007)
.2600)
INTERVAL
0.0090)
0.2616)
0.0803)
0.0178)
1.2455)
0.1223)
0.0165)
0.1990)
0.1023)
INTERVAL
1.6776)
8.3118)
6.0220)
6.7429)
15.2376)
4.9846)
7.0214)
19.9955)
1.4957)
13.7608)
D-2
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
fltor!
SUMMARY OF ESTIMATED DIFFERENTIAL LOADINGS FOR SCREENED POLLUTANTS (Continue)
-GROUP=SEDIMEMT FRACTION CONFOUND TYPE 'PHENOLS, PAH'S AND PHTHALATES(KG/DAY)-
POLLUTANT
PtWNE
SAMPLE
SIZE
17
MEAN
(FE)
2.1633
STO OEV
MEAN(FE)
0.8407
MEAN
(NOTL)
10.9963
STD DEV
NEAN(NOTL)
4.0834
MEAN
(NOTL-FE)
8.8325
STD DEV
MEAN(NOTL-FE)
3.5109
CONFIDENCE INTERVAL
(NOTL-FE)
( 2.7024, 14.9625)
POLLUTANT SAMPLE
NAME SIZE
ALUIIHILM EXTRACT 39 71
ARSENIC 39 Q
CKHMIUM 39 0
IR01 39 169
MANGANESE 39 4
HI OIL 39 0
ZINC 39 1
POLLUTANT SAMPLE
NAME SIZE
HEXAO10ROBENZENE 21
PENTACXLOROBENZENE 21
1,2,3-TRICLORBENZENE 21
1,2,3,4-TETRACHLOROBENZENE 21
1,2,4-TRICLORBENZENE 21
POLLUTANT SAMPLE
NAME SIZE
IDEKX1,2,3-CO)PYRENE 11
PHE1ANTHRENE 11
llKUUfVAIEK rKAlllUN UJnKUUNU 1 TPE =TW
MEAN STO DEV MEAN STO DEV
(FE) MEAN(FE) (NOTL) MEAN(NOTL)
.4870 18.2715 100.6337 20.3626
.3217 0.0196 0.3826 0.0232
.4914 0.0909 0.8421 0.1959
.2558 33.3672 275.3651 43.4303
.1883 1.0626 7.3392 1.2098
.6984 0.0957 0.9837 0.0997
ci*Latniu"/u/
MEAN
(NOTL-FE)
29.1467
0.0608
0.3508
106.1094
3.1509
0.2852
»TJ
STD DEV
MEAN(NOTL-FE)
12.0502
0.0194
0.1780
29.4393
0.7840
0.0801
.0424 0.1153 1.8062 0.2045 0.7638 0.1672
MEAN STD OEV MEAN STD OEV MEAN STD DEV
(FE) MEAN(FE) (NOTL) MEAN(NOTL) (NOTL-FE) MEAN(NOTL-FE)
0.0733 0.0178 0.1318 0.0444
0.0236 0.0100 0.2030 0.1001
0.1123 0.0597 1.1398 0.6730
0.0517 0.0229 0.7193 0.2940
0.7098 0.2845 4.5740 3.2814
•UATCD CDAPTfOAl (WD/^ IUH TVDC *DUCUfM C DAU
WATcK FRACTION COMPOUND TTPc aPHcNOL5, PAH
MEAN STD DEV MEAN STD OEV
(FE) MEAN(FE) (NOTL) MEAN(NOTL)
1.6798 1.6798 14.7949 3.9429
3.4517 1.8586 12.8536 5.1826
0.0585
0.1795
1.0275
0.6676
0.0477
0.0938
0.6603
0.2809
3.8642 3.1708
1C Aun DUTUAI ATCC/^r* /A A V \
MEAN STD OEV
(NOTL-FE)
13.1151
9.4019
MEAN(NOTL-FE)
4.1303
4.1941
CONFIDENCE INTERVAL
(NOTL-FE)
( 8.8180, 49.4754)
( 0.0281, 0.0936)
( 0.0504, 0.6512)
( 56.4453, 155.7734)
( 1.8282, 4.4736)
( 0.1501, 0.4203)
( 0.4818, 1.0458)
CONFIDENCE INTERVAL
(NOTL-FE)
( -0.0239, O.U08)
( 0.0177, 0.3413)
( -0.1115, 2.1666)
( 0.1829, 1.1522)
( -1.6054, 9.3338)
CONFIDENCE INTERVAL
(NOTL-FE)
( 5.6311, 20.5992)
( 1.8021, 17.0017)
D-3
------- |