I I
\ I
\ '
LAKE HURON INTENSIVE SURVEY, 1980: \
' i ' * i
1 // ' ^
ANALYSIS OF SAMPLES FROM TRE DETROIT MUNICIPAL W[ATFR INTAKE
1 ' ' ' i
1 , FOR ,USE IN MONITORING SOUTHERN LAKE HURON \
V by/
v' I
Russell Moll
V '
Under Contradt with,; ' \
1 ,;'. '
United States ^Environnietital P'rotection Agency
( t ; ^ I
Grant- Number R-&05510-03 -
' , Project Officer; , I
I /
} ' )
1 / \ David Cl Rockwell | ! >
, " Greet Lakes1 National Progr'am Office
tTnitied Graves Envirorimental, Protection Agency
?30 South Dearborn Street ' > '
l/ Chicago, Illinois. 60^04 \
1 ' . \ , i _
October 1986
I , ' ' / ,
EPA 950-R-86-009
-------
-------
LAKE HURON INTENSIVE SURVEY, 1980:
ANALYSIS OF SAMPLES FROM THE DETROIT MUNICIPAL WATER INTAKE
FOR USE IN MONITORING SOUTHERN LAKE HURON
by
Russell Moll
Under Contract with:
United States Environmental Protection Agency
Grant Number R-005510-03
Project Officer:
David C. Rockwell
Great Lakes National Program Office
United States Environmental Protection Agency
230 South Dearborn Street
Chicago, Illinois 60604
October 1986
-------
DISCLAIMER
The information In this document has been funded wholly or In part by
the United States Environmental Protection Agency under assistance agreement
number R-005510-03 to The University of Michigan; it has been subject to the,
Agency's peer and administrative review; and it has been approved for publi-
cation. The mention of trade names or commercial products does not consti'cut,&
endorsement or recommendation for use*
-------
CONTENTS
SUMMARY... v
INTRODUCTION 1
LAKE HURON-WATER INTAKE DATABASE, 1980.. 3
OBJECTIVES AND STATISTICAL METHODS 4
DISTRIBUTIONAL ANALYSIS. ". 7
COMPARISON OF MEANS AND DIFFERENCES BY STRATA 39
ANALYSIS OF VARIANCE 45
CORRELATION AND REGRESSION 49
TRANSFORMATIONS 54
COMPARISON OF INTAKE SAMPLES WITH OPEN LAKE HURON SAMPLES 56
DISCUSSION 60
REFERENCES 70
iii
-------
-------
SUMMARY
The overall conclusion from the analysis of the Detroit Municipal Water
Intake and crib samples is that for the most part intake samples can be a
useful tool for monitoring Lake Huron water quality. Most of the physical,
chemical, and biological variables did not change significantly between the
lake and the filtration plant. Changes were small and rarely did means of
lake and plant samples differ by more than 25%. Many of the differences were
less than 10%, especially during the third (May) field trip. However, there
were seasonal factors that entered into variability of lake and plant
differences. The most notable variation was in chlorophyll samples where the
difference between plant and lake means usually exceeded 30% for two of the
three field trips.
The consideration of dissolved and total metals data did not yield such
an optimistic set of results as the other variables. Differences between sev-
eral of the means exceeded 100% and in extreme cases 5001. Clearly, the use
of plant samples to monitor Cu and Zn is unacceptable. On some occasions the
same is true for Fe and Pb. Furthermore, variances associated with the metals
data were large and often positively correlated with the mean (variances in-
creased as did the means). Not all of the metals results fit this category.
For example, Mg and dissolved Mg values were identical (and invariant) between
lake and plant among all field trips. In general, caution must be used for
monitoring many of the metals with plant samples.
The Analysis of Variance (ANOVA) results appeared to contradict the con-'
elusions presented above. Many ANOVAs showed significant differences between
lake and plant samples for physical, chemical, and biological variables.
The opposite was true for the metals; many of the ANOVAs were non-significant.
v
-------
A slightly deeper examination of the data reveals the cause behind this
apparent discrepancy. The physical, chemical, and biological variables were
measured with relatively high precision. Thus each mean had a small standard
error. Because the ANOVAs are sensitive to both the mean and variance, these
small variances yielded several results that indicated lake and plant samples
were different. This result is real, but should be viewed in the context that
while the lake and plant samples are often different, their differences are
usually small. Thus if the investigator recognizes that difference and it is
acceptable, monitoring using intake samples can proceed. In contrast, the
metals data had large standard errors associated with the means. Thus before
intake samples are used for monitoring, those standard errors should be re-
duced and the statistical analyses repeated. The source of the large vari-
ances for many of the metals data could be either from analytical techniques
or a plant effect. If it is the latter, not much can be done to reduce the
variances and somewhat noisy data must be accepted.
Because the samples were collected as matched pairs (one lake sample
matched by a plant sample), several analyses could be conducted using the
paired data. Those analyses consisted of rank correlation, linear regression,
and direction (sign) of the differences between lake and plant data.
The intent was to determine if the lake and plant samples behaved in a similar
fashion, i.e., a high value for one was matched with a high value for the
other. The results clearly showed that the samples did not behave in a simi-
lar fashion. A high value of one variable was not necessarily matched by a
high plant sample and vice versa. All three analyses supported these find-
ings; rank correlations were generally low, regressions non-significant, and
differences not of uniform sign.
vi
-------
These findings can be interpreted as relatively neutral in regard to
monitoring Lake Huron using intake plant samples. On the positive side, there
was very little evidence of bias between plant and lake samples. This result
was true both on a field trip by field trip basis as well as for the entire
database. On the negative side, lake and plant samples showed little evidence
of direct correspondence. Evidently mixing processes within the intake pipe
and/or the plant served to homogenize water while it was in transit between
the lake and plant.
Several variables were transformed into natural logs in order to inves-
tigate the consequences of non-normal distributions within the original data.
The results showed that the transformed data yielded almost identical infer-
ences as the original data.
Finally the plant samples were compared to samples collected from a gen-
eral survey of Lake Huron in order to determine if the plant samples truly
matched open Lake huron water quality. A moderate degree of correspondence
between these two data sets was observed. That level of correspondence is
very encouraging considering all the factors which could lead to differences
between the two databases.
In sum, the results of the sampling programs and data analyses inferred
that many variables could be reasonably monitored with samples collected from
the Detroit Municipal Water Intake. Several variables cannot be monitored and
several yielded uncertain conclusions. In general, physical, chemical, and
biological variables were well represented by plant samples while metals
samples were not. The results point toward a program of monitoring of Lake
Huron with ample use of Detroit Municipal Water Intake samples and supple-
mented with field samples collected at least several times per year.
vii
-------
-------
INTRODUCTION
The concept of using municipal water intakes as a source of samples to
monitor the Great Lakes goes back to the turn of the century (Vorce 1881,
Strand 1973). Intakes along the Lake Michigan shore have proved particularly
well suited as a source of raw water from which a variety of analyses have
been conducted (Beeton and Strand 1975). These analyses have included routine
measurements such as water temperature and pH as well as more sophisticated
variables as algal species composition. The results have more often than not
proven rather provocative in revealing potential long-term shifts in the
status of the lakes (Shapiro and Swain 1983) or short-term phenomena (Mortimer
1971).
The advantages of utilizing municipal water intakes are apparent.
Samples may be obtained for analysis with minimal effort. Most water treat-
ment facilities conduct their own analyses on a routine basis at frequent
intervals (Strand 1973). In these cases the data are available at little or
no cost to interested scientists. Researchers requiring different types of
analyses and/or different intervals of data collection than provided by the
water treatment facility have found minimal difficulty in arranging for col-
lection of samples from the intake plant.
The convenience and ease with which the water intake samples can be ob-
tained and analyzed makes them an ideal choice for a monitoring tool. But,
use of such samples should not be without constraints for several reasons.
Most intakes draw their water from the nearshore environment and as a result,
samples cannot be considered representative of the lake as a whole. In addi-
tion, the intake pipes often impart certain characteristics to the water as it
-------
passes from the crib in the lake to the treatment plant. These characteris-
tics are known to change with the type of pipe (concrete, iron, etc.) as well
as the length of the pipe (Kubus and Egloff 1982). Furthermore, there is a
seasonal factor associated with the pipe effect; certain characteristics are
imparted to the water during one season and not the others. As a result, any
effort to use water intake samples to monitor a Great Lake should begin with
analysis of a series of matched samples, lake versus intake. Those matched
samples should be collected from all seasons of interest, analyzed for all
variables of interest, and compared to open lake samples.
As part of the effort to develop a basin-wide network to monitor the
Great Lakes, the Great Lakes National Program Office of the USEPA has sug-
gested collection of samples from several municipal water intakes. These sam-
ples would supplement the existing framework of ship-based sample collection
and analysis. Water intake samples would be collected throughout the year.
These samples are particularly valuable during those times of the year when
research vessels cannot collect samples (winter and early spring). Under
these conditions, water intake samples become a valuable addition to the moni-
toring network. In order to initiate this program the Great Lakes National
Program Office has called for analysis of several sets of matched samples from
different locations in the Great Lakes. This report discusses the analysis of
those samples collected from Lake Huron and the Detroit Municipal Water Intake
plant located near Port Huron, Michigan. The intent of these samples is to
determine if this intake can be used to monitor southern Lake Huron.
This type of study has precedent both for chemical analyses (Kubus and
Egloff 1982) and algal species composition (Ladewski et al. 1982). Both of
these studies used matched pairs of samples collected from the lake just above
-------
the intake crib and the water plant. Such a sampling design permitted testing
of specific hypotheses by statistical analyses, an approach followed by this
study. Many of the statistical procedures used by Kubus and Egloff (1982)
were followed in this study.
LAKE HURON-WATER INTAKE DATABASE, 1980
Samples were collected from Lake Huron just above the water intake crib
located 6.2 km (5 miles) east of the mouth of Milwaukee Greek at Lakeport,
Michigan. The water intake crib is 14 m (45 ft) below the surface at latitude
43° 07' 34.6" and longitude 82° 23' 27.2". Corresponding samples were
collected from the Detroit Municipal Water Intake plant approximately 37 km
(23 miles) southwest of the intake crib. Intake samples were collected from
the raw water tap in the basement at the east end of the plant. The transit
time for water flowing from Lake Huron to the intake plant was computed as
between 8 h 45 min and 9 h.
The database consists of results of sample analyses from field trips
covering three different seasons in 1980 and 1981. The field trips were con-
ducted during 19-26 August 1980, 7-9 October 1980, and 19 May 1981. All Lake
Huron samples were collected from 14 m depth and processed using USEPA-ap-
proved techniques. Sample collection and analyses were conducted by the
Michigan Department of Natural Resources (F. Horvath, Michigan Department of
Natural Resources, personal communication). Matching samples were collected
from the water intake plant after the appropriate time delay of 9 h. These
samples were handled in a manner identical to the Lake Huron samples to
minimize the sample analysis effect. A total of 88 samples was analyzed, 44
from the plant, and 44 from the lake. The breakdown of samples by cruise is
3
-------
16 each from the first two field trips and 12 from the last trip. Each sample
from the lake was matched by a sample from the intake plant. Furthermore,
when replicates were collected from the lake, they were also taken from the
raw water taps in the plant. The result in most cases is a balanced database
without missing cells.
Each water sample was analyzed for 41 different variables (Table 1).
These consisted of physical-chemical variables including dissolved nutrients,
particulate matter, total metals, and dissolved metals. The results were re-
ported to the highest level of accuracy or the limit of detection. Those data
were entered into the USEPA data storage and retrieval system, STORET. A data
tape was produced from STORET and used to create a database on the University
of Michigan computing systemj MTS. Database management and statistical
analyses were conducted with MIDAS, a University of Michigan statistical
package (Fox and Guire 1976) or BMDP, a University of California package.
OBJECTIVES AND STATISTICAL METHODS
The overall objective of the analyses was to determine if the samples
taken from the raw water tap in the treatment plant represent those collected
from the lake. This broad objective was broken down into five smaller objec-
tives which approximately matched the steps used in the analyses. Those five
objectives are:
1. Determine if the water intake facility had a significant effect on the 41
variables compared to the lake samples.
2. Determine if there are differences between the lake and intake samples, and
if those differences are consistent (bias) or random.
3. Determine if a simple transformation of variables with bias can remove the
significant difference between the lake and intake samples.
-------
TABLE 1. Variables which were measured from samples collected from the crib
and raw water tap of the Detroit Municipal Water Intake, 1980 and 1981.
Nltra te-ni trogen
Ammonium ion
Total Kjeldahl nitrogen
Total nitrogen
Total dissolved phosphorus
Soluble reactive phosphorus
Total phosphorus
Chloride
Soluble reactive silica
Sulfate
Alkalini ty
Dissolved oxygen (tng/L)
Chlorophyll
Turbidity
Conductivity
Total non-filterable residue
Total residue
Combustible residue
Calcium
Magnesium
Sodium
Potassium
Cadmium
Chromium
Copper
Nickel
Lead
Zinc
Iron
Dissolved Calcium
Dissolved Magnesium
Dissolved Sodium
Dissolved Potassium
Dissolved Cadmium
Dissolved Chromium
Dissolved Copper
Dissolved Nickel
Dissolved Lead
Dissolved Zinc
Dissolved Iron
Water temperature
pH
-------
4, Determine if samples collected from the intake can reliably predict the
open lake value of the variable in question. This objective must be ad-
dressed on a seasonal as well as composite basis.
5. Investigate the nature of the precision and accuracy for intake samples for
each variable.
A basic protocol for analyzing the 41 variables was developed which ad-
dressed these objectives. That protocol moved from relatively simple statis-
tical analyses to hypothesis testing.
The statistical analyses used to achieve the objectives of this study
fell into five basic categories. The first step was to obtain an understand-
ing of the database. This routine preliminary step is one of the most impor-
tant aspects of the analysis in that only by understanding the nature of the
database can the statistical results be put into perspective. This under-
standing was achieved with the use of descriptive statistics (minimum, maxi-
mum, mean, variance, skewness, kurtosis), histograms, scatter plots, and
distributional analysis; these analytical procedures were repeated both for
the entire database and separately for each field trip. The next step was
testing the hypothesis of differences between samples collected from Lake
Huron versus the water plant. This hypothesis was tested with one-way
Analysis of Variance (ANOVA) for each individual variable. One-way ANOVAs
were utilized for the entire database as well as for each field trip. The
third step was to determine the nature of the relationship between the lake
and treatment plant samples. This step used the samples as matched pairs with
a one-to-one correspondence between lake and plant samples. The statistical
approach used was linear regression, rank-correlation, and descriptive
statistics of differences between lake and plant samples. The fourth step was
to compare results from the lake or plant samples with those collected in
-------
adjacent portions of Lake Huron. The concept is to determine if either group
of samples represents open Lake Huron water quality. The fifth and final step
was to transform those variables with distributions that are highly non-normal
and repeat steps one to three. The transformation generally consisted of
converting the original data to natural logs.
The use of environmental data and a relatively small database raises the
concern of lack of normality (Lindgren 1976). In many cases the distribution
of the variables was highly skewed due to the presence of one or two large or
small values. This problem was handled in three ways. First, whenever
possible, techniques that do not require normal distributions were invoked.
Computation of descriptive statistics, scatter plots, rank correlations, and
analysis of differences of matched pairs are distribution-free procedures.
Second, whenever normally distributed data were required as in the case of
ANOVA and linear regression, extreme care was taken in the interpretation of
the results, i.e., the only results considered valid were from analyses with
large «.10) or very small (>.01) probabilities. Third, transformations were
employed in an attempt to convert distributions to something closer to normal.
DISTRIBUTIONAL ANALYSIS
The purpose of the distributional analyses was to develop an understand-
ing of the nature of the dispersion of each variable within the database.
The first step was to compute descriptive statistics (minimum, maximum, mean,
standard deviation, skewness, and kurtosis) for each variable for various
strata. The strata included: the entire database, Lake Huron samples, treat-
ment plant samples, each of the three field trips, and each of the six com-
binations of field trips and sample locations (Tables 2-13).
7
-------
TABLE 2. Descriptive statistics from all samples from the Lake Huron water intake database.
00
Variable
N02 + N03 (mg/L)
NH3 (mg/L)
Total Kjel N (mg/L)
Total N (mg/L)
Total Ortho P (mg P/L)
Soluble P (mg P/L)
Total P (mg P/L)
Chloride (mg/L)
Silicate (mg Si/L)
Sulfate (mg/L)
Alkalinity (mg CaCO«/L)
Diss Oxy (mg/L)
Chlorophyll (yg/L)
Turbidity (FTU)
Conductivity (y mho/ cm)
Tot non-F. Res. (mg/L)
Res Tot (mg/L)
Combust Res (mg C/L)
Ca (mg/L)
Mg (mg/L)
Na (mg/L)
K (mg/L)
Cd (yg/L)
Cr (Hg/L)
Cu (yg/L)
N
88
88
88
88
88
56
88
88
88
88
88
78
88
88
88
88
88
88
88
88
88
88
88
88
88
Minimum
.220
.001
.09
.33
.001
.001
.001
4.9
.35
12.
72.
8.3
.2
.4
196.
2.
80.
80.
25.
7.0
2.0
.80
.5
.5
1.
Maximum
.990
.017
.44
1.20
.004
.004
.013
6.4
.62
18.
81.
14.5
4.7
.9
220.
8.
180.
180.
30.
7.0
7.0
.90
2.0
7.0
33.
Mean
.286
.0067
.175
.461
.0011
.0011
.0051
5.52
.493
15.6
76.1
9.86
1.00
.52
203.4
3.0
128.6
127.4
26.5
7.00
3.72
.841
.91
1.91
6.7
Std Dev
.0884
.0027
.0505
.106
.00047
.00040
.0027
.36
.079
.98
2.3
1.45
.57
.10
4.1
1.1
17.4
17.4
1.9
1.25
.049
.23
1.48
5.7
Skewness
6.222
.663
2.397
4.267
5.619
7.281
.805
..263
-.089
-.881
.231
1.159
3.220
1.036
.804
1.209
.513
.591
1.002
1.402
.370
.106
1.258
1.741
Kurtosis
44.935
1.766
9.915
25.598
31.261
51.018
.489
-.698
-1.090
1.848
. -.978
.445
18.851
1.465
1.473
2.445
.649
.721
-.626
1.049
-1.86
4.437
.737
4.083
(continued).
-------
TABLE 2. (continued).
Variable
Nl (yg/L)
Pb (pg/L)
Zn (yg/L)
Fe (yg/L)
Diss Ca (rag/L)
Diss Mg (mg/L)
Diss Na (mg/L)
Diss K (mg/L)
Diss Cd (yg/L)
Diss Or (yg/L)
Diss Cu (yg/L)
Diss Ni (yg/L)
Diss Pb (yg/L)
Diss Zn (yg/L)
Diss Fe (yg/L)
Temp (°C)
pH (SU)
N
88
88
88
88
88
88
88
88
56
56
56
56
56
56
56
88
76
Minimum
.5
.5
1.
3.5
25.
7.0
2.0
.80
.1
.5
1.
.5
.5
.5
3.
6.7
7.20
Maximum
5.0
20.0
80.
60.0
30.
7.0
6.0
.90
1.0
7.0
7.
4.0
36.0
23.0
17.
22.0
9.60
Mean
1.65
1.89
19.0
21.40
26.3
7.00
3.45
.827
.77
2.92
3.9
1.45
5.42
8.61
6.8
14.98
8.184
Std Dev
.93
2.54
18.5
10.90
1.9
1.04
.0448
.26
1.45
1.7
.87
8.30
5.70
3.5
5.73
.563
Skewness
1.466
5.048
1.502
.823
1.202
1.299
1.021
-.390
-.121
.106
1.116
2.309
.805
.994
-.283
1.223
Kurtosis
2.273
29.913
1.712
1.333
-.120
.991
-.958
-1.422
-.391
-1.163
.323
4.295
-.154
-.062
-1.300
1.054
-------
TABLE 3. Descriptive statistics from intake crib samples from the Lake Huron water intake database.
o
Variable
NO 2 + N03 (mg/L)
NH3 (mg/L)
Total Kjel N (mg/L)
Total N (mg/L)
Total Ortho P (mg P/L)
Soluble P (mg P/L)
Total P (mg P/L)
Chloride (mg/L)
Silicate (mg Si/L)
Sulfate (mg/L)
Alkalinity (mg CaC03/L)
Diss Oxy (mg/L)
Chlorophyll (pg/L)
Turbidity (FTU)
Conductivity (ymho/em)
Tot non-F. Res. (mg/L)
Res Tot (mg/L)
Combust Res (mg C/L)
Ca (mg/L)
Mg (mg/L)
Na (mg/L)
K (mg/L)
Cd (yg/L)
Cr (yg/L)
Cu (yg/L)
N
44
44
44
44
44
28
44
44
44
44
44
42
44
44
44
44
44
44
44
44
44
44
44
44
44
Minimum
.240
.001
.13
.38
.001
.001
.003
4.9
.35
12.
72.
8.6
.4
.4
198.
2.
80.
80.
25.
7.0
2.0
.80
.5
.5
1.
Maximum
.990
.017
.44
1.20
.004
.004
.013
6.0
.62
17.
81.
14.5
4.7
.9
220.
8.
167.
164.
30.
7.0
7.0
.90
2.0
5.0
33.
Mean
.306
.0059
.192
.498
.0011
.0011
.0059
5.38
.481
15.7
76.6
10.27
1.15
.53
203.6
3.3
121.8
120.0
27.4
7.00
3.30
.852
.89
1.64
3.3
Std Dev
.118
.0029
.0589
.131
.00063
.00057
.0022
.28
.086
1.14
2.60
1.55
.69
.12
4.3
1.3
15.3
14.7
2.3
1.02
.051
.28
1.23
4.8
Skewness
4.760
1.240
2.486
3.875
4.364
5.004
1.048
.105
-.044
-1.275
-.171
.956
3.038
.919
1.422
1.169
.553
.690
.217
2.270
-.091
.766
1.297
5.528
Kurtosis
24.093
3.427
7.283
17.339
17.048
23.037
.958
-.808
-1.263
1.743
-1.275
-.296
13.557
.488
3.054
2.433
1.661
1.955
-1.798
4.730
-1.992
3.764
.338
31.290
(continued).
-------
TABLE 3. (continued).
Variable
Ni (yg/L)
Pb (yg/L)
Zn (yg/L)
Fe (yg/L)
Diss Ca (mg/L)
Diss Mg (mg/L)
Diss Na (mg/L)
Diss K (mg/L)
Diss Cd (yg/L)
Diss Cr (yg/L)
Diss Cu (yg/L)
Diss Ni (yg/L)
Diss Pb (yg/L)
Diss Zn (yg/L)
Diss Fe (yg/L)
Temp (°C)
pH (SU)
N
44
44
44
44
44
44
44
44
28
28
28
28
28
28
28
44
32
Minimum
1.0
.5
1.
3.5
25.
7.0
2.0
.80
.5
.5
1.
.5
.5
.5
3.5
7.0
7.50
Maximum
5.0
20.0
80.
60.0
30.
7.0
6.0
.90
1.0
5.0
4.0
3.5
36.0
12.0
14.0
21.0
9.60
Mean
2.00
2.49
9.4
19.6
27.1
7.0
3.14
.832
.71
2.75
2.5
1.71
6.04
4.89
7.1
14.66
8.278
Std Dev
1.08
3.44
13.8
13.3
2.3
.90
.0471
.25
1.57
.74
.88
8.74
3.09
3.4
5.38
.815
Skewness
1.040
3.610
3.854
.965
.404
2.018
.781
.289
-.400
-.425
.458
2.405
.764
.556
-.338
.719
Kurtosis
.595
14.285
15.534
.634
-1.703
4.587
-1.390
-1.917
-1.547
-.378
-1.076
4.737
-.343
-1.270
-1.311
-1.231
-------
TABLE 4. Descriptive statistics from intake plant samples from the Lake Huron water intake database.
Variable
N02 + N03 (mg/L)
NH3 (mg/L)
Total Kjel N (mg/L)
Total N (mg/L)
Total Ortho P (mg P/L)
Soluble P (mg P/L)
Total P (mg P/L)
Chloride (mg/L)
Silicate (mg Si/L)
Sulfate (mg/L)
Alkalinity (mg CaC03/L)
Diss Oxy (mg/L)
Chlorophyll (yg/L)
Turbidity (FTU)
Conductivity (y mho/ cm)
Tot non-F. Res. (mg/L)
Res Tot (mg/L)
Combust Res (mg C/L)
Ca (mg/L)
Mg (mg/L)
Na (mg/L)
K (mg/L)
Cd (pg/L)
Cr (pg/L)
Cu (yg/L)
N
44
44
44
44
44
28
44
44
44
44
44
36
44
44
44
44
44
44
44
44
44
44
44
44
44
Minimum
.220
.004
.09
.33
.001
.001
.001
5.0
.38
14.
72.
8.3
.2
.4
196.
2.
104.
104.
25.
7.0
2.0
.80
.5
.5
6.
Maximum
.410
.013
.22
.59
.002
.001
.013
6.4
.62
18.
80.
12.2
1.5
.7
210.
4.
180.
180.
30.
7.0
7.0
.90
1.0
7.0
20.
Mean
.266
.0075
.157
.423
.0010
.0010
.0044
5.65
.504
15.5
75.6
9.39
.86
.51
203.2
2.7
135.3
134.8
25.7
7.00
4.14
.830
.93
2.18
10.0
Std Dev
.0316
.0022
.0326
.0534
.00021
.0029
.38
.071
.79
1.87
1.16
.35
.074
4.0
.91
17.0
16.8
1.0
1.32
.046
.17
1.67
4.4
Skewness
2.303
.425
-.218
.820
4.364
1.158
-.087
.054
.134
.558
1.322
.092
.550
.048
.723
.498
.545
1.965
.968
.897
-2.119
1.048
.793
Kurtosis
8.199
-.155
-.393
1.449
17.048
.937
-.956
-1.153
1.005
-.039
1.108
-.844
.415
-.883
-1.382
.288
.465
4.742
-.075
-1.196
2.491
.167
-.794
(continued).
-------
TABLE 4. (continued).
Variable
m (pg/D
Pb (pg/L)
Zn (Ug/L)
Fe (yg/L)
Diss Ca (mg/L)
Diss Mg (mg/L)
Diss Na (mg/L)
Diss K (mg/L)
Diss Cd (yg/L)
Diss Cr (yg/L)
Diss Gu (yg/L)
Diss Ni (yg/L)
Diss Pb (yg/L)
Diss Zn (yg/L)
Diss Fe (yg/L)
Temp (°C)
pH (SU)
N
44
44
44
44
44
44
44
44
28
28
28
28
28
28
28
44
44
Minimum
.5
.5
8.
11.0
25.
7.0
2.0
.80
.1
1.0
4.
.5
.5
6.0
3.
6.7
7.20
Maximum
2.5
4.0
75.
51.0
27.
7.0
6.0
.90
1.0
7.0
7.
4.0
29.0
23.0
17.
22.0
8.40
Mean
1.30
1.28
28.7
23.1
25.6
7.00
3.77
.823
.83
3.1
5.43
1.18
4.80
12.32
6.6
15.29
8.116
Std De¥
.56
.69
17.7
7.5
.79
1.08
.042
.27
1.3
.92
.78
7.94
5.29
3.6
6.10
.251
Skewness
.855
2.129
1.028
1.247
.854
.916
1.302
-1.074
.528
-.077
2.204
2.159
.502
1.389
-.279
-1.315
Kurtosis
-.596
4.667
.358
2.912
-.830
-.287
-.306
-.184
1.070
-.824
4.820
3.276
-.997
1.093
-1.337
2.155
-------
TABLE 5. Descriptive statistics from the first field trip samples from the Lake Huron water intake
database.
Variable
NO. + N0_ (mg/L)
NH3 (mg/L)
Total Kjel N (mg/L)
Total N (mg/L)
Total Ortho P (mg P/L)
Soluble P (mg P/L)
Total P (mg P/L)
Chloride (mg/L)
Silicate (mg Si/L)
Sulfate (mg/L)
Alkalinity (mg CaC03/L)
Diss Oxy (mg/L)
Chlorophyll (yg/L)
Turbidity (FTU)
Conductivity (ymho/cm)
Tot non-F. Res. (mg/L)
Res Tot (mg/L)
Combust Res (mg C/L)
Ca (mg/L)
Mg (mg/L)
Na (mg/L)
K (mg/L)
Cd (yg/L)
Cr (yg/L)
Cu (yg/L)
N
32
32
32
32
32
0
32
32
32
32
32
30
32
32
32
32
32
32
32
32
32
32
32
32
32
Minimum
.260
.001
.13
.40
.001
.003
5.4
.35
15.
72.
8.3
.2
.4
196.
2.
108.
108.
25.
7.0
2.0
.80
1.0
1.0
1.
Maximum
.560
.013
.22
.77
.002
.013
5.9
.45
17.
77.
9.0
1.8
.9
210.
8.
148.
148.
30.
7.0
5.0
.90
1.0
2.0
20.
Mean
.294
.0063
.173
.467
.0011
.0064
5.59
.40
16.0
74.6
8.63
.91
.54
201.1
2.5
128.8
127.5
27.7
7.00
3.53
.847
1.00
1.03
8.5
Std Dev
.0577
.0034
.0300
.0714
.00025
.0025
.18
.027
.47
1.52
.26
.40
.12
3.3
1.2
11.2
11.4
2.5
.62
.051
.18
7.1
Skewness
3.452
.387
.371
2.620
3.615
.874
.241
.008
.106
-.074
.051
.188
.939
.691
3.301
.129
.229
-.125
-.119
.125
5.388
.160
Kurtosis
12.664
-.709
-1.237
8.435
11.067
.040
-1.302
-.843
1.557
-.952
-1.596
-.554
.710
-.211
12.071
-1.115
-1.076
-1.984
-.311
-1.984
27.032
-1.730
(continued).
-------
TABLE 5. (continued).
Variable
Ni (yg/L)
Pb (yg/L)
Zn (Hg/L)
Fe (yg/L)
Diss Ga (mg/L)
Diss Mg (mg/L)
Diss Na (mg/L)
Diss K (mg/L)
Diss Cd (yg/L)
Diss Cr (yg/L)
Diss Gu (yg/L)
Diss Ni (yg/L)
Diss Pb (yg/L)
Diss Zn (yg/L)
Diss Fe (yg/L)
Temp (°C)
pH (SU)
N
32
32
32
32
32
32
32
32
0
0
0
0
0
0
0
32
32
Minimum
1.0
1.0
3.
10.0
25.
7.0
2.0
.80
20.0
7.20
Maximum
3.0
3.0
75.
60.0
30.
7.0
5.0
.90
22.0
8.00
Mean
1.47
1.19
26.6
27.8
27.5
7.00
3.19
.806
21.23
7.738
Std Dev
.84
.59
23.1
10.6
2.5
.69
.0246
.79
.181
Skewness
1.253
2.787
.646
1.123
.000
.337
3.615
-.115
-.641
Kurtosis
-.366
5.770
-.867
1.172
-2.000
.219
11.067
-1.827
.732
-------
TABLE 6. Descriptive statistics from the second field trip samples from the Lake Huron water intake
database.
Variable
NO _ -f NO, (mg/L)
NH3 (mg/L)
Total Kjel N (mg/L)
Total N (mg/L)
Total Ortho P (mg P/L)
Soluble P (mg P/L)
Total P (mg P/L)
Chloride (mg/L)
Silicate (mg Si/L)
Sulfate (mg/L)
Alkalinity (mg CaC03/L)
i- Diss Oxy (mg/L)
Chlorophyll (pg/L)
Turbidity (FTU)
Conductivity (pmho/cm)
Tot non-F. Res. (mg/L)
Res Tot (mg/L)
Combust Res (mg C/L)
Ca (mg/L)
Mg (mg/L)
Na (mg/L)
K (mg/L)
Cd (Pg/L)
Cr (Pg/L)
Cu (Pg/L)
N
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
Minimum
.220
.003
.09
.33
.001
.001
.001
4.9
.49
12.
73.
9.5
.4
.4
200.
2.
80.
80.
25.
7.0
2.0
.80
.5
.5
3.
Maximum
.990
.017
.44
1.20
.004
.004
.013
6.0
.53
17.
80.
10.1
4.7
.7
220.
6.
180.
180.
30.
7.0
7.0
.90
2.0
7.0
10.
Mean
.275
.0077
.186
.462
.0012
.0011
.0045
5.16
.513
15.2
75.8
9.72
.86
.51
203.0
2.7
135.4
133.8
25.6
7.00
4.34
.806
.75
2.69
5.6
Std Dev
.135
.0024
.0745
.159
.00074
.00053
.0029
.20
.010
1.19
2.03
.13
.77
.078
4.0
1.1
20.7
20.9
1.3
1.79
.025
.34
1.86
2.1
Skewness
4.933
1.494
1.686
3.322
3.615
5.388
.901
2.338
-.342
-.420
.397
.556
4.113
.678
2.402
1.555
-.365
-.267
2.619
.328
3.615
1.620
.275
.201
Kurtosis
23.519
' 5.814
3.536
12.632
11.067
27.032
.570
7.331
-.367
.407
-.989
.855
17.520
.504
8.276
1.495
.254
.036
6.148
-1.552
11.067
3.694
-.798
-1.132
(continued).
-------
TABLE 6. (continued).
Variable
Ni (ug/L)
Pb (Ug/L)
Zn (Ug/L)
Fe (yg/L)
Diss Ca (mg/L)
Diss Mg (mg/L)
Diss Na (mg/L)
Diss K (mg/L)
Diss Cd (yg/L)
Diss Cr (yg/L)
Diss Cu (yg/L)
Diss Ni (yg/L)
Diss Pb (yg/L)
Diss Zn (yg/L)
Diss Fe (yg/L)
Temp (°C)
pH (SU)
N
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
Minimum
.5
.5
2.5
3.5
25.
7.0
2.0
.80
.1
.5
2.
.5
.5
.5
3.5
14.0
8.00
Maximum
5.0
20.0
31.0
51.0
27.
7.0
6.0
.80
1.0
7.0
7.
3.5
27.0
23.
17.0
15.0
9.60
Mean
1.84
3.09
14.17
18.59
25.3
7.00
3.97
.800
.60
2.52
4.3
1.56
3.70
7.88
6.47
14.80
8.581
Std Dev
1.21
3.89
9.81
11.87
.51
1.43
.23
1.59
1.5
.94
4.93
6.16
3.42
.33
.599
Skewness
1.224
2.999
.577
.374
1.875
.395
.870
.627
.530
.439
3.506
.808
1.550
-1.351
.813
Kurtosis
.854
9.608
-1.035
-.249
2.687
-1.394
.062
.087
-1.063
-1.091
13.842
-.368
1.619
.541
-1.032
-------
TABLE 7. Descriptive statistics from the third field trip samples from the Lake Huron water intake
database.
00
Variable
N02 + N03 (mg/L)
NH3 (mg/L)
Total Kjel N (mg/L)
Total N (mg/L)
Total Ortho P (mg P/L)
Soluble P (mg P/L)
Total P (mg P/L)
Chloride (mg/L)
Silicate (mg Si/L)
Sulfate (mg/L)
Alkalinity (mg CaC03/L)
Diss Oxy (mg/L)
Chlorophyll (yg/L)
Turbidity (FTU)
Conductivity (u mho/cm)
Tot non-F. Res. (mg/L)
Res Tot (mg/L)
Combust Res (mg C/L)
Ca (mg/L)
Mg (mg/L)
Na (mg/L)
K (mg/L)
Cd (yg/L)
Cr (ug/L)
Cu (yg/L)
N
24
24
24
24
24
24
24
24
24
24
24
16
24
24
24
24
24
24
24
24
24
24
24
24
24
Minimum
.270
.004
.12
.39
.001
.001
.002
5.5
.53
14.
75.
11.6
.9
.4
205.
4.
104.
104.
25.
7.0
3.0
.80
1.0
1.0
2.
Maximum
.350
.008
.20
.53
.001
.001
.009
6.4
.62
18.
81.
14.5
1.6
.7
210.
4.
180.
180.
28.
7.0
4.0
.90
1.0
5.0
33.
Mean
.290
.0059
.162
.451
.001
.001
.0044
5.89
.588
15.7
78.4
12.44
1.32
.50
207.1
4.0
119.0
118.8
26.3
7.00
3.13
.879
1.00
2.04
5.7
Std Dev
.0203
.0012
.0243
.0389
.0018
.25
.0317
.92
1.67
.64
.19
.093
2.5
15.5
15.5
.64
.34
.041
1.30
6.3
Skewness
1.392
.022
-.174
.012
1.033
.369
-.866
.013
-.687
2.233
-.566
.657
.338
2.630
2.631
.649
2.268
-1.436
.889
3.504
Kurtosis
1.647
-.779
-1.187
-.958
.419
-.757
-.856
.667
-.456
5.120
-.454
-.360
-1.886
8.134
8.093
.429
3.143
.063
-.608
12.789
(continued).
-------
TABLE 7. (continued).
Variable
Ni (yg/L)
Pb (yg/L)
Zn (yg/L)
Fe (Ug/L)
Diss Ca (rag/L)
Diss Mg (mg/L)
Diss Na (mg/L)
Diss K (mg/L)
Diss Gd (yg/L)
Diss Cr (yg/L)
Diss Cu (yg/L)
Diss Ni (yg/L)
Diss Pb (yg/L)
Diss Zn (yg/L)
Diss Fe (yg/L)
Temp (°C)
pH (SU)
N
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
12
Minimum
1.0
1.0
1.
11.
25.
7.0
3.0
.80
1.0
1.0
1.0
1.
1.0
4.0
3.
6.7
8.25
Maximum
2.0
3.0
80.
26.
27.
7.0
4.0
.90
1.0
4.0
6.0
4.
36.0
21.0
14.
7.0
8.40
Mean
1.63
1.21
15.3
16.6
26.2
7.00
3.13
.892
1.00
3.46
3.5
1.3
7.71
9.58
7.3
6.87
8.317
Std Dev
.49
.51
18.0
3.8
.83
.34
.028
1.02
1.9
.75
11.06
4.98
3.6
.14
.039
Skewness
-.516
2.378
2.346
.964
-.399
2.268
-3.015
-1.765
.120
2.632
1.429
1.174
.348
-.157
.581
Kurtosis
-1.733
4.788
5.524
.801
-1.386
3.143
7.091
1.650
-1.657
6.000
.506
.355
-1.440
-1.827
.120
-------
TABLE 8. Descriptive statistics from the first field trip crib samples from the Lake Huron water
intake database.
K3
o
Variable
N02 + NO- (mg/L)
NH3 (mg/L)
Total Kjel N (mg/L)
Total N (mg/L)
Total Ortho P (mg P/L)
Soluble P (mg P/L)
Total P (mg P/L)
Chloride (mg/L)
Silicate (mg Si/L)
Sulfate (mg/L)
Alkalinity (mg CaCO /L)
Diss Oxy (mg/L)
Chlorophyll (yg/L)
Turbidity (FTU)
Conductivity (y mho/ cm)
Tot non-F. Res. (mg/L)
Res Tot (mg/L)
Combust Res (ing C/L)
Ca (mg/L)
Mg (mg/L)
Na (mg/L)
K (mg/L)
Cd (pg/L)
Cr (yg/L)
Cu (yg/L)
N
16
16
16
16
16
0
16
16
16
16
16
14
16
16
16
16
16
16
16
16
16
16
16
16
16
Minimum
.260
.001
.13
.40
.001
.003
5.4
.35
16.
72.
8.6
.5
.4
198.
2.
108.
108.
30.
7.0
2.0
.80
1.0
1.0
1.
Maximum
.560
.013
.22
.77
.001
.011
5.6
.40
17.
76.
9.0
1.8
.9
210.
8.
144.
136.
30.
7.0
4.0
.90
. i.o
1.0
3.
Mean
.304
.0042
.174
.479
.0010
.0058
5.44
.379
16.3
73.9
8.87
1.10
.56
203.0
3.1
125.4
123.0
30.0
7.00
3.06
.863
1.00
1.00
1.7
Std Dev
.0716
.0029
.033
.083
.0023
.062
.015
.48
1.26
.11
.38
.15
3.3
1.5
9.8
8.9
.44
.0500
.6
Skewness
3.069
1.680
.225
2.824
.779
1.360
-.822
1.155
.240
-.976
.209
.771
.237
2.217
.116
.065
.354
-.516
.185
Kurtosis
8.564
2.998
-1.344
7.628
-.167
.775
-.489
-.667
-.923
1.031
-1.104
-.352
-.725
4.938
-.876
-1.298
2.228
-1.733
-.622
(continued).
-------
TABLE 8. (continued).
Variable
Ni (yg/L)
Pb (yg/L)
Zn (yg/L)
Fe (yg/L)
Diss Ca (mg/L)
Diss Mg (mg/L)
Diss Na (mg/L)
Diss K (mg/L)
Diss Gd (yg/L)
Diss Cr (yg/L)
Diss Cu (yg/L)
Diss Ni (yg/L)
Diss Pb (yg/L)
Diss Zn (yg/L)
Diss Fe (yg/L)
Temp (°C)
pH (SU)
N
16
16
16
16
16
16
16
16
0
0
0
0
0
0
0
16
16
Minimum
1.0
1.0
3.
10.0
30.
7.0
2.0
.80
20.0
7.50
Maximum
3.0
3.0
20.
60.0
30.
7.0
4.0
.90
21.0
7.70
Mean
1.88
1.25
7.0
30.31
30.0
7.00
2.88
.813
20.47
7.638
Std Dev
1.02
.68
4.6
13.72
.62
.034
.22
.089
Skewness
.252
2.268
1.768
.533
.054
2.268
-.354
-.801
Kurtosis
-1.937
3.143
2.231
-.503
-.314
3.143
2.228
-1.162
-------
TABLE 9. Descriptive statistics from the second field trip crib samples from the Lake Huron water
intake database.
Variable
N02 + NO^ (mg/L)
NH3 (mg/L)
Total Kjel N (mg/L)
Total N (mg/L)
Total Ortho P (mg P/L)
Soluble P (mg P/L)
Total P (mg P/L)
Chloride (mg/L)
Silicate (mg Si/L)
Sulfate (mg/L)
Alkalinity (mg CaCO»/L)
Diss Oxy (mg/L)
Chlorophyll (yg/L)
Turbidity (FTU)
Conductivity (y mho/ cm)
Tot non-F. Res. (mg/L)
Res Tot (mg/L)
Combust Res (mg C/L)
Ca (mg/L)
Mg (mg/L)
Na (mg/L)
K (mg/L)
Cd (yg/L)
Cr (yg/L)
Cu (Pg/L)
N
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
Minimum
.240
.003
.14
.38
.001
.001
.003
4.9
.49
12.
75.
9.5
.4
.4
200.
2.
80.
80.
25.
7.0
2,0
.80
.5
.5
3.
Maximum
.990
.017
.44
1.20
- .004
.004
.013
5.4
,52
17.
80.
10.1
4.7
7.0
220.
6.
167.
164.
30.
7.0
7.0
.90
2.0
4.0
10.
Mean
.312
.0071
.227
.539
.0014
.0012
.0064
5.08
.508
15.1
77.3
9.79
1.04
.54
201.9
3.1
124.2
122.0
25.6
7.00
3.75
.806
,69
2.06
4.0
Std Dev
.186
.0032
.0811
.196
.0010
.00075
.0025
.11
.010
1.6
1.7
.13
1.07
.089
5.1
1.3
21.4
21.2
1.7
1.57
.025
.40
1.49
1.7
Skewness
3.324
1.834
1.532
2.576
2.268
3.615
1.139
1.087
-.310
-.302
-.245
.052
2.756
.193
3.036
.914
.217
.362
2.268
.960
3.615
2.390
.201
2.878
Kurtosis
9.607
3.968
1.496
6.326
3.143
11.067
1.258
2.820
-.880
-.957
-1.183
1.354
6.884
-.618
8.174
-.478
.186
.075
3.143
-.545
11.067
5.108
-1.606
7.777
(continued).
-------
TABLE 9. (continued).
to
U>
Variable
Ni (yg/L)
Pb (yg/L)
Zn (yg/L)
Fe (Pg/L)
Diss Ca (mg/L)
Diss Mg (mg/L)
Diss Na (mg/L)
Diss K (mg/L)
Diss Gd (yg/L)
Diss Cr (yg/L)
Diss Cu (yg/L)
Diss Ni (yg/L)
Diss Pb (yg/L)
Diss Zn (yg/L)
Diss Fe (yg/L)
Temp (°C)
pH (SU)
N
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
Minimum
1.0
.5
2.5
3.5
25.
7.0
2.0
.80
.5
.5
2.
.5
.5
.5
3.5
14.0
8.00
Maximum
5.0
20.0
16.0
30.0
25.
7.0
6.0
.80
.5
5.0
4.
3.5
11.0
6.0
12.0
15.0
9.60
Mean
2.44
4.84
6.34
10.31
25.0
7.00
3.50
.800
.50
2.44
. 3.0
2.13
4.13
2.94
5.06
14.59
8.919
Std Dev
1.32
4.90
3.47
9.35
1.32
1.70
.37
.92
2.62
1.52
2.26
.36
.700
Skewness
.801
2.074
1.291
1.418
1.177
.006
0.
-.445
1.055
.464
2.272
-.302
-.402
Kurtosis
-.509
3.778
1.681
.321
.030
-1.656
5.000
-.851
1.155
-.579
4.056
-1.070
-1.581
-------
TABLE 10. Descriptive statistics from the third field trip crib samples from the Lake Huron water
intake database.
Variable
N02 -f N03 (mg/L)
NH3 (mg/L)
Total Kjel N (mg/L)
Total N (mg/L)
Total Ortho P (mg P/L)
Soluble P (mg P/L)
Total P (mg P/L)
Chloride (mg/L)
Silicate (nig Si/L)
Sulfate (mg/L)
Alkalinity (mg CaC03/L)
Diss Oxy (mg/L)
Chlorophyll (yg/L)
Turbidity (FTU)
Conductivity (ymho/cm)
Tot non-F. Res. (mg/L)
Res Tot (mg/L)
Combust Res (mg C/L)
Ca (mg/L)
Mg (mg/L)
Na (mg/L)
K (mg/L).
Cd (pg/L)
Cr (yg/L)
Cu (yg/L)
N
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
Minimum
.270
.005
.14
.41
.001
.001
.004
5.5
.53
15.
78.
11.6
.9
.4
205.
4.
104.
104.
26.
7.0
3.0
.90
1.0
1.0
2.
Maximum
.350
.008
.20
.53
.001
.001
.009
6.0
.62
16.
81.
14.5
1.6
.7
210.
4.
132.
132.
27.
7.0
3.0
.90
1.0
5.0
33.
Mean
.299
.0066
.170
.469
.0010
.0010
.0054
5.72
.582
15.7
79.3
12.53
1.38
.47
206.7
4.0
113.7
113.3
26.3
7.00
3.00
.900
1.00
1.92
4.7
Std Dev
.0239
.0011
.0191
.0350
.0018
.15
.0369
.49
.89
.72
.21
.098
2.5
7.3
7.3
.45
1.38
8.9
Skewness
.769
.001
-.164
-.113
.834
.244
-.637
-.707
.121
1.767
-1.273
1.304
.707
1.204
1.356
1.155
1.240
3.010
Kurtosis
-.304
-1.187
-.900
-.711
-.832
-.373
-1.450
-1.500
-.621
2.930
.878
.609
-1.500
1.352
1.655
-.667
.169
7.070
(continued).
-------
TABLE 10. (continued).
to
01
Variable
Ni (yg/L)
Pb (yg/L)
Zn (yg/L)
Fe (yg/L)
Diss Ca (mg/L)
Diss Mg (mg/L)
Diss Na (mg/L)
Diss K (mg/L)
Diss Cd (yg/L)
Diss Cr (yg/L)
Diss Gu (yg/L)
Diss Ni (yg/L)
Diss Pb (yg/L)
Diss Zn (yg/L)
Diss Fe (yg/L)
Temp (°C)
pH (SU)
N
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
0
Minimum
1.0
1.0
1.
12.0
25.
7.0
3.0
.90
1.0
1.0
1.
1.0
1.0
4.0
5.
7.0
Maximum
2.0
1.0
80.
26.0
27.
7.0
3.0
.90
1.0
4.0
2.
2.0
36.0
12.0
14.
7.0
Mean
1.58
1.00
16.5
17.83
25.9
7.00
3.00
.900
1.00
3.17
1.8
1.17
8.58
7.50
9.8
7.00
Std Dev
.51
25.0
4.47
.90
1.34
.45
.39
12.87
2.71
2.7
Skewness
-..338
1.715
.779
.161
-1.028
-1.155
1.789
1.284
.242
-.507
Kurtosis
-1.886
1.621
-.320
-1.636
-.829
-.667
1.200
-.088
-1.325
-.717
-------
TABLE 11. Descriptive, statistics from the first field trip plant samples from the Lake Huron water
intake database.
K>
Variable
N02 + N03 (mg/L)
NH3 (mg/L)
Total Kjel N (mg/L)
Total N (mg/L)
Total Ortho P (mg P/L)
Soluble P (mg P/L)
Total P (mg P/L)
Chloride (mg/L)
Silicate (mg Si/L)
Sulfate (mg/L)
Alkalinity (mg CaC03/L)
Diss Oxy (mg/L)
Chlorophyll (yg/L)
Turbidity (FTU)
Conductivity (y mho/ cm)
Tot non-F. Res. (mg/L)
Res Tot (mg/L)
Combust Res (mg C/L)
Ca (mg/L)
Mg (mg/L)
Na (mg/L)
K (mg/L)
Cd (yg/L)
Cr (yg/L)
Cu (yg/L)
N
16
16
16
16
16
0
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
Minimum
.260
.006
.14
.40
.001
,004
5.6
.38
15.
72.
8.3
.2
.4
196.
2.
114.
112.
25.
7.0
3.0
.80
1.0
1.0
11.
Maximum
.4100
.013
.22
.59
.002
.013
5.9
.45
16.
77.
8.8
1.3
.7
204.
2.
148.
148.
30.
7.0
5.0
.90
1.0
2.0
20.
Mean
.283
.0084
.172
.454
.0011
.007-0
5.75
.423
15.8
75.4
8.43
.73
.53
199.1
2.0
132.3
132.0
25.3
7.00
4.00
.831
1.00
1.06
15.3
Std Dev
.0389
.0024
.0281
.0569
.00034
.0027
.089
.0182
.40
1.4
.13
.35
.086
2.0
11.8
12.1
1.3
.37
.0479
.25
2.4
Skewness
2.513
.514
.547
1.175
2.268
.858
.577
-.550
-1.601
-.698
1.440
-.018
.164
.666
-.105
-.148
3.615
0.
.809
3.615
.472
Kurtosis
5.387
-1.193
-1.127
.388
3.143
-.303
-.667
.284
.564
.056
2.416
-.919
-.579
.339
-1.435
-1.392
11.067
5.000
-1.345
11.067
-.035
(continued).
-------
TABLE 11. (continued).
Variable
Ni (yg/L)
Pb (yg/L)
Zn (pg./L)
Fe (ME/L)
Diss Ca (mg/L)
Diss Mg (mg/L)
Diss Na (mg/L)
Diss K (mg/L)
Diss Cd (yg/L)
Diss Cr (yg/L)
Diss Cu (yg/L)
Diss Ni (yg/L)
Diss Pb (yg/L)
Diss Zn (yg/L)
Diss Fe (yg/L)
Temp (°C)
pH (SU)
N
16
16
16
16
16
16
16
16
0
0
0
0
0
0
0
16
16
Minimum
1.0
1.0
30.
20.0
25.
7.0
3.0
.80
22.0
7.20
Maximum •
2.0
3.0
75.
40.0
25.
7.0
5.0
.80
22.0
8.00
Mean
1.06
1.13
46.3
25.31
25.0
7.00
3.50
.800
22.00
7.838
Std Dev Skewness Kurtosis
.25 3.615 11.067
.50 3.615 11.067
16.2 .464 -1.269
5.62 1.332 1.269
.63 .816 -.333
.196 -2.240 5.097
-------
TABLE 12. Descriptive statistics from the second field trip plant samples from the Lake Huron water
intake database.
00
Variable
NO + NO^ (mg/L)
NH3 (mg/L)
Total Kjel N (mg/L)
Total N (mg/L)
Total Ortho P (mg P/L)
Soluble P (mg P/L)
Total P (mg P/L)
Chloride (mg/L)
Silicate (mg Si/L)
Sulfate (mg/L)
Alkalinity (mg CaCO /L)
Diss Oxy (mg/L)
Chlorophyll (yg/L)
Turbidity (FTU)
Conductivity (y mho/ cm)
Tot non-F. Res. (mg/L)
Res Tot (mg/L)
Combust Res (mg C/L)
Ca (mg/L)
Mg (mg/L)
Na (mg/L)
K (mg/L)
Cd (yg/L)
Cr (yg/L)
Cu (yg/L)
N
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
Minimum
.220
.006
.09
.33
.001
.001
.001
5.0
.50
14.
73.
9.5
.5
.4
200.
2.
124.
124.
25.
7.0
2.0
.80
.5
.5
6.
Maximum
.2400
.009
.19
.43
.001
.001
.008
6.0
.53
16.
76.
9.8
1.0
.5
205.
4.
180.
180.
27.
7.0
7.0
.90
1.0
7.0
9.
Mean
.239
.0083
.145
.384
.0010
.0010
.0025
5.25
.518
15.2
. 74.3
9.66
.68
.48
204.1
2.3
146.6
145.6
25.6
7.00
4.94
.806
.81
3.31
7.1
Std Dev
.0050
.00093
.0363
.035
.0018
.24
.0086
.54
1.0
.081
.15
.044
2.0
.6
12.7
12.7
.7
1.84
.025
.25
2.02
.8
Skewness
-3.615
-1.024
-.409
-.395
1.962
2.089
-.164
.170
.134
.177
.535
-1.155
-1.601
1.708
.661
.886
.851
-.236
3.615
-.516
-.064
.567
Kurtosis
11.067
.130
-1.352
-1.299
4.153
4.476
-.579
.043
-1.036
-.501
-.583
-.667
.564
1.813
1.420
1.640
-.556
-1.616
11.067
-1.733
-.820
.185
(continued).
-------
TABLE 12. (continued).
Variable
Ni (yg/L)
Pb (Ug/L)
Zn (yg/L)
Fe (yg/L)
Diss Ca (mg/L)
Diss Mg (mg/L)
Diss Na (mg/L)
Diss K (mg/L)
Diss Cd (yg/L)
Diss Cr (yg/L)
Diss Cu (yg/L)
Diss Ni (yg/L)
Diss Pb (yg/L)
Diss Zn (yg/L)
Diss Fe (yg/L)
Temp (°C)
pH (SU)
N
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
Minimum
.5
.5
12.
19.0
25.
7.0
2.0
.80
.1
1.0
4.
,5
.5
6.0
4.
15.0
8.10
Maximum
2.5
4.0
31.
51.0
27.
7.0
6.0
.80
1.0
7.0
7.
2.0
27.0
23.0
17.
15.0
8.40
Mean
1.25
1.34
22.0
26.88
25.5
7.00
4.44
.800
.69
2.59
5.5
1.00
3.28
12.81
7.9
15.00
8.244
Std Dev
.71
.87
7.5
7.58
.6
1.41
.30
1.53
1.0
.55
6.56
4.92
3.9
.089
Skewness
.949
1.830
.044
2.119
.816
-.235
-.177
1.519
0.
.943
3.231
.513
1.073
.775
Kurtosis
-.547
3.443
-1.718
4.582
-.333
-1.449
-1.160
2.352
-1.063
-.333
9.196
-.604
.161
-.449
-------
TABLE 13. Descriptive statistics from the third field trip plant samples from the Lake Huron water
intake database.
o
Variable
N02 + NO, (mg/L)
NH3 (mg/L)
Total Kjel N (mg/L)
Total N (mg/L)
Total Ortho P (mg P/L)
Soluble P (mg P/L)
Total P (mg P/L)
Chloride (mg/L)
Silicate (mg Si/L)
Sulfate (mg/L)
Alkalinity (rig CaCO/L)
Diss Oxy (mg/L)
Chlorophyll (ug/L)
Turbidity (FTU)
Conductivity (pmho/cm)
Tot non-F. Res. (mg/L)
Res Tot (mg/L)
Combust Res (mg C/L)
Ca (mg/L)
Mg (mg/L)
Na (mg/L)
K (mg/L)
Cd (vg/L)
Cr (pg/L)
Cu (yg/L)
N
12
12
12
12
12
12
12
12
12
12
12
4
12
12
12
12
12
12
12
12
12
12
12
12
12
Minimum
.270
.004
.12
.39
.001
.001
.002
5.8
.55
14.
75.
12.1
1.0
.4
205.
4.
104.
104.
25.
7.0
3.0
.80
1.0
1.0
6.
Maximum
.290
.007
.20
.49
.001
.001
.005
6.4
.62
18.
80.
12.2
1.5
.7
210.
4.
180.
180.
28.
7.0
4.0
.90
1.0
4.0
11.
Mean
.280
.0053
.154
.433
.0010
.0010
.0033 '
6.07
.593
15.7
77.5
12.18
1.26
.53
207.5
4.0
124.3
124.3
26.4
7.00
3.25
.858
1.00
2.17
6.7
Std Dev
.00953
.0011
.0271
.0353
.0011
.21
.0257
1.2
1.8
.050
.17
.078
2.6
19.6
19.6
.79
.45
.0515
1.27
1.4
Skewness
.000
-.030
.252
.167
.683
.012
-.823
.057
-.000
-1.155
-.140
.581
-.000
1.962
1.962
.283
1.155
-.338
.518
2.539
Kurtosis
-1.800
-1.342
-1.427
-1.517
-.789
-1.114
-.833
-.571
-1.269
-.667
-.648
.120
-2.000
3.610
3.610
-.252
-.667
-1.886
-1.322
5.273
(continued).
-------
TABLE 13. (continued).
Variable
Ni (yg/L)
Pb (yg/L)
Zn (yg/L)
Fe (yg/L)
Diss Ca (mg/L)
Diss Mg (mg/L)
Diss Na (mg/L)
Diss K (mg/L)
Diss Cd (yg/L)
Diss Cr (yg/L)
Diss Cu (yg/L)
Diss Ni (yg/L)
Diss Pb (yg/L)
Diss Zn (yg/L)
Diss Fe (yg/L)
Temp (°C)
pH (SU)
N
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
Minimum
1.0
1.0
8.
11.0
25.
7.0
3.0
.80
1.0
3.0
4.
1.0
1.0
6.0
3.
6.7
8.25
Maximum
2.0
3.0
24.
19.0
27.
7.0
4.0
.90
1.0
4.0
6.
4.0
29.0
21.0
12.
6.8
8.40
Mean
1.67
1.42
14.1
15.3
26.5
7.00
3.25
.883
1.00
3.75
5.3
1.42
6.83
11.67
4.8
6.73
8.317
Std Dev
.49
.67
6.9
2.7
.7
.45
.0389
.45
.78
1.00
9.40
5.91
2.5
.05
.039
Skewness
-.707
1.267
.598
-.060
-.930
1.155
-1.789
-1.155
-.626
1.968
1.422
.602
2.252
.707
.581
Kurtosis
-1.500
.412
-1.470
-1.423
-.240
-.667
1.200
-.667
-.960
2.148
.551
-1.264
4.039
-1.500
.120
-------
The most obvious aspect of these data is the presence of several invari-
ant variables. Invariant variables are those that do not change (have no
variance) for one or more strata. Table 14 lists the invariant variables by
location and field trip.
Generally the invariant variables included total dissolved phosphorus
(Phos T) and/or soluble reactive phosphorus (SRP). The cause of the invariant
phosphorus distribution was always due to reporting values as .001 mg PL ,
the analytical limit of detection. In effect these results suggest that all
samples analyzed fell below the limit of detection and thus were reported as
.001 mg PL . Under these conditions there is no point in further analysis.
The important aspect of the phosphorus data is to determine when and where
samples were above the limit of detection; this analysis does not require
statistics. Table 15 indicates where phosphorus values exceeded the limit of
detection when invariant distributions were not encountered (First field trip
- dissolved-phosphorus at the plant, Second field trip - dissolved phosphorus
and soluble reactive silica from the lake).
The majority of the invariant distributions were either total metals or
total dissolved metals (Table 14). In the case of magnesium (Mg) the concen-
tration of the total metal did not change from 7.00 mg L throughout the
course of the study. Similarly, the concentration of dissolved Mg did not
change from 7.00 mg L throughout the course of the study. Generally the
invariant metal distributions were the same for the lake and plant samples
within one field trip. Metals which followed this pattern Included: First
field trip - cadmium (Cd), Second field trip - dissolved potassium (Diss K),
and Third field trip - total and dissolved cadmium (Cd and Diss Cd). These
results show that plant samples were identical to lake values. In a few
32
-------
TABLE 14. Invariant variables from the three field trips.
Crib
Total dissolved P
Ca
Mg
Cd
Cr
Dissolved Ca
Dissolved Mg
Crib
Mg
Dissolved Ca
Dissolved Mg
Dissolved K
Dissolved Cd
Crib
Cruise 1 - 19-26 August 1980
Plant
Total non-filterable residue
Mg
Cd
Dissolved Ca
Dissolved Mg
Dissolved K
Water temperature
Cruise 2 - 7-9 October 1980
Plant
Total dissolved P
Soluble reactive P
Total non-filterable residue
Mg
Na
K
Cd
Dissolved Mg
Dissolved Na
Dissolved K
Dissolved Cd
Water temperature
Total dissolved P
Soluble reactive P
Mg
Dissolved Mg
Dissolved K
Water temperature
Cruise3- 19 May 1981
Plant
Total dissolved P .
Soluble reactive P
Total non-filterable residue
Mg
Cd
Dissolved Mg
Dissolved Cd
33
-------
TABLE 15. Invariant variables from the three field trips.
are the same as Table 2.
Units for means
Cruise 1 -
Variable ,
Total dissolved P
Total non-filterable residue
Ca
Mg
Cd
Cr
Dissolved Ca
Dissolved Mg
Dissolved K
Water temperature
Cruise 2 -
Variable
Total dissolved P
Soluble reactive P
Mg
Dissolved Ca
Dissolved Mg
Dissolved K
Dissolved Cd
Water temperature
Cruise 3
Variable
Total dissolved P
Soluble reactive P
Total non-filterable residue
Mg
Na
K
Cd
Dissolved Mg
Dissolved Na
Dissolved K
Dissolved Cd
Water temperature
19-26 August 1980
Crib
.001
3.06*
30.0
7.00
1.00
1.00
30.0
7.00
.813*
20.5*
7-9 October 1980
Crib
.00138*
.00119*
7.00
25.0
7.00
'8.00
.500
14.6*
- 19 May 1981
Crib
.001
.001
4.00
7.00
3.00
.900
1.00
7.00
3.00
.900
1.00
7.00
. Plant
.0011*
2.00
25.3*
7.00
1.00
1.06*
25.0
7.00
.800
22.0
Plant
.001
.001
7.00
25.5*
7.00
8.00
.694*
15.0
Plant
.001
.001
4.00
7.00
3.25*
.858*
1.00
7.00
3.25*
.883*
1.00
6.73*
* - Variables are not invariant.
matching invariant means.
Means have been listed for comparison to
34
-------
cases, lake samples were invariant while the plant samples showed minor
changes; in these cases the means of the metal concentration from the two
sites were very close. These results show that the plant samples were almost
a perfect mirror of the lake samples with the exception of one or two samples
out of 12 or 16. Metals which followed this pattern were: First field trip -
total and dissolved calcium (Ca and Diss Ca) and chromium (Cr), Second field
trip - dissolved calcium (Diss Ca) and dissolved Cd, and Third field trip - Na
and dissolved Na and potassium (K) (Table 15).
Two additional variables were associated with invariant distributions,
total non-filterable residue and water temperature. Total non-filterable
*—l
residue was invariant and reported at 4.00 mg L (the limit of detection)
during the third field trip. During the first field trip it was invariant at
«•"!
the plant at a level of 2.00 mg L (also listed as the limit of detection).
In both cases, the true value of total non-filterable residue was unknown
except that it fell below 4.00 or 2.00 mg L . Plant water temperatures were
invariant at 15°C during the second field trip while lake temperatures were
slightly cooler and variable (Table 15). Lake water temperatures were
invariant at 7°C during the third field trip while plant temperatures were
slightly cooler and variable (Table 15).
In sum, of a total of 41 variables analyzed on three different field
trips (123 cruise-variables), 27 were associated with invariant distributions.
This is almost 22% of the total database. Added to that total are another
nine variables that were not sampled during the field trips (Table 16). As a
result, a substantial portion of the database (29.3%) was not available for
statistical analyses.
35
-------
TABLE 16. Variables which were not measured (missing observations) from
intake database.
Cruise 1 - 19-26 August 1980
Crib Plant
Soluble reactive P Soluble reactive P
Dissolved Cd Dissolved Cd
Dissolved Cr Dissolved Cr
Dissolved Cu Dissolved Gu
Dissolved Ni Dissolved Ni
Dissolved Pb Dissolved Pb
Dissolved Zn ' Dissolved Zn
Dissolved Fe Dissolved Fe
Cruise 2 - 7-9 October 1980
Crib • Plant
none missing none missing
Cruise 3-19 May 1981
Crib Plant
pH none missing
36
-------
The distributions of the remaining variables were evaluated for confir-
mation to normality. The descriptive statistics in Tables 2-13 provide useful
insight into the distribution of each variable as considered by different
strata. In particular the statistics gl and g2 which are derived from the
third and fourth moments give an indication of skew and kurtosis respectively
(Sokal and Rohlf 1981). Table 17 lists those variables that had distributions
with excessive skew and/or kurtosis beyond a normal distribution. The cri-
terion used for this conclusion was a gl and/or g2 statistic in excess of +2.5
(Sokal and Rohlf 1981). Considered on a cruise-by-cruise basis, of the 123
cruise-variables (41 variables analyzed -over three cruises), 36 were identi-
fied as either invariant or not measured. Of the remaining 87 cruise-vari-
ables, 25 have excessive skew and/or kurtosis, leaving 62 normally distributed
cruise-variables, or almost exactly half of the original total.
In almost every case of the 25 non-normally distributed cruise-vari-
ables, the presence of one or two observations well outside the range of the
other observations (outliers) created the lack of normality. This distribu-
tion was particularly common among the total metals and total dissolved
metals. The elimination of these data values as outliers would convert the
distribution to invariant or normal.
Histograms confirmed the presence of the data points well outside of the
range of the other observations. The histograms also indicated that in the
case of many of the total metals and total dissolved metals the data distribu-
tions were not continuous. In other words, the values reported were in dis-
crete steps rather than as a continuance. For these reasons, conclusions con-
cerning the difference between lake and plant samples for metals and dissolved
37
-------
TABLE 17. Variables with non-normal distributions from the water intake
database.
Cruise1 - 19-26 August 1930
Crib Plant
Ni tra te~ni trogen Ni tra te-ni trogen
Total nitrogen Total dissolved P
Total non-filterable residue Ca
Pb Gr
Dissolved K Ni
Pb
PH
Cruise 2 - 7-9 October 1980
Crib Plant
Nitrate-nitrogen Nitrate-nitrogen
Ammonium ion Total phosphorus
Total nitrogen Chloride
Total dissolved P K
Soluble reactive P Pb
Chlorophyll Fe
Conductivity Dissolved Pb
Ca
K
Cu
Pb
Dissolved Fe
Cruise 3 - 19 May 1981
Crib Plant
Dissolved oxygen Total residue
Cu Dissolved 180
Cu
Dissolved Ni
Dissolved Fe
38
-------
metals primarily relied on distribution-free statistical methods rather than
linear models.
An additional form of distributional analysis was conducted employing
the transformation of each variable to standard deviates and plotting as a
cumulative distribution. In this technique, each variable is standardized
(subtract mean and divide by standard deviation) and then normalized
(converted to a deviation from a normal distribution) (Natter and Wasseraan
1974). Cumulative plots of these transformed data quickly show deviations
from a normal distribution because they deviate from a straight line. The re-
sults confirmed the findings from the descriptive statistics in Table 17 and
the histograms. Again, the presence of several "outlier" type data points
generated most of the deviation from a normal distribution.
COMPARISON OF MEANS AND DIFFERENCES BY STRATA
The first set of results, comparison of lake and plant means, is rela-
tively easy to interpret. There are no underlying assumptions required of the
data. The difference of the means simply demonstrates the degree to which
each variable changes between the lake and the plant. That difference can
then be expressed as a percent of the mean of the lake observations. This
analysis was conducted for the entire database as well as for each cruise.
The results are presented in Tables 18-21.
Considered across all three field trips, the results show that with a
few exceptions the lake and plants means rarely differed by more than 20%
(Table 18). The largest difference was for Cu, dissolved Cu, Zn, and
dissolved Zn which differed by 199%, 120%, 207%, and 152% respectively. In
all of these cases metal concentrations were much higher in the plant samples
39
-------
TABLE 18. Means for the lake (intake crib) and plant samples from all three
field trips. Difference between the means for each variable are presented as
well as the percent of the lake mean represented by the difference.
Units for means are the same as Table 2,
Variable
Nitrate-N
Ammonium ion
Total Kjeldahl N
Total N
Total dissolved P*
Soluble reactive P*
Total P
Chloride
Soluble reactive Si
Sulfate
Alkalinity
Dissolved oxygen
Chlorophyll
Turbidity
Conductivity
Total non-filterable
Combustible residue
Dissolved 180
Ca
Mg
Na
K
Cd
Cr
Cu
Ni
Pb
Zn
Fe
Dissolved Ca
Dissolved Mg
Dissolved Na
Dissolved K
Dissolved Cd
Dissolved Cr
Dissolved Cu
Dissolved Ni
Dissolved Pb
Dissolved Zn
Dissolved Fe
Water temperature
PH
Crib
.306
.00591
.193
.498
.00114
.00111
.00593
5.38
.481
15.68
76.61
10.27
1.15
.527
203.6
residue* 3.32
121.8
120.0
27.39
7.00
3.30
.852
.886
1.64
3.34
2.00
2.49
9.35
19.64
27.07
7.00
3.14
.830
.714
2.75
2.46
1.71
6.04
4.89
7.07
14.66
8.28
Plant
.266
.00750
.157
.423
.00105
.00100
.00436
5.65
.504
15.55
75.57
9.39
.856
.509
203.2
2.66
135.3
134.8
25.71
7.00
4.14
.830
.932
2.18
9.98
1.30
1.28
28.66
23.16
25.59
7.00
3.77
.823
.825
3.09
5.43
1.18
4.80
12.32
6.57
15.29
8.12
Diff.
.0395
.00159
.0360
.0752
.00009
.000108
.00157
.273
.0227
.136
1.05
.347
.298
.0182
.386
.659
13.55
14.84
1.68
0,00
.841
.0227
.0455
.545
6.64
.705
1.21
19.31
3.52
1.48
0.00
.636
.00910
.111
.339
2.96
.536
1.23
7.43
.500
.632
.238
% of Lake
12.9
26.9
18.1
15.1
7.9
9.7
26.5
5.1
4.7
0.9
1.4
3.4
25.9
3.5
0.2
19.8
11.1
12.4
6.1
-
25.5
2.7
5.1
33.2
198.7
35.3
48.4
206.5
18.0
5.5
-
20.3
1.1
15.5
12.3
120.3
31.3
20.4
151.9
7.1
4.3
2.9
* Includes limit of detection values.
40
-------
TABLE 19. Means for the lake (intake crib) and plant samples from the first
field trip (19-26 August 1980). Difference between the means for each vari
able are presented as well as the percent of the lake mean represented by
the difference. Units for means are the same as Table 2.
Variable
Nitrate-N
Ammonium ion
Total Kjeldahl N
Total N
Total dissolved P*
Soluble reactive P
Total P
Chloride
Soluble reactive Si
Sulfate
Alkalinity
Dissolved oxygen
Chlorophyll
Turbidity
Conductivity
Total non-filterable
Total residue
Combustible residue
Ca
Mg
Na
K
cd
Cr
Cu
Ni
Pb
Zn
Fe
Dissolved Ca
Dissolved Mg
Dissolved Na
Dissolved K
Dissolved Cd
Dissolved Cr
Dissolved Cu
Dissolved Ni
Dissolved Pb
Dissolved Zn
Dissolved Fe
Water temperature
PH
Crib
.304
.00419
.174
.479
.00100
NS
.00581
5.44
.379
16.25
73.9
8.87
1.10
.556
203.0
residue* 3.06
125.4
123.0
30.00
7.00
3.06
.863
1.00
1.00
1.69
1.88
1.25
7.00
30.3
30.0
7.00
2.88
.813
NS
NS
NS
NS
NS
NS
NS
20.5
7.64
Plant
.283
.00844
.172
.454
.00113
NS
.00700
5.75
.423
15.83
75.4
8.43
.731
.525
199.1
2.00
132.3
132.0
25.31
7.00
4.00
.831
1.00
1.06
15.3
1.06
1.13
46.3
25.3
25.0
7.00
3.50
.800
NS
NS
NS
NS
NS
NS
NS
22.0
7.84
Diff.
.0213
.00425
.00250
.0244
.000125
_
.000119
.313
.0438
.438
1.50
.443
.365
.0313
3.88
1.06
6.82
9.00
4.69
0.00
.940
.0312
0.00
.063
13.63
.813
.125
39.3
5.00
5.00
0.00
.625
.0125
_
-
-
-
-
-
_
1.53
.200
% of Lake
7.0
101.4
1.4
5.1
12.5
_
20.5
5.7
11.5
2.7
2.0
5.0
33.3
5.6
1.9
34.7
5.4
7.3
15.6
30.6
3.6
-
6.3
807.4
43.3
10.0
560.7
16.5
16.7
21.7
1.5
-
-
-
-
-
-
-
7.5
2.6
NS = No sample.
* Includes limit of detection values.
41
-------
TABLE 20, Means for the lake (intake crib) and plant samples from the second
field trip (7-9 October 1980). Difference between the means for each vari
able are presented as well as the percent of the lake mean represented by
the difference. Units for means are the same as Table 2.
Variable
Nitrate-N
Ammonium ion
Total Kjeldahl N
Total N
Total dissolved P*
Soluble reactive P*
Total P
Chloride
Soluble reactive Si
Sulfate
Alkalinity
Dissolved oxygen
Chlorophyll
Turbidity
Conductivity
Total non- filterable
Total residue
Combustible residue
Ca
Mg
Na
K
Cd
Cr
Cu
Ni
Pb
Zn
Fe
Dissolved Ca
Dissolved Mg
Dissolved Na
Dissolved K
Dissolved Cd
Dissolved Cr
Dissolved Cu
Dissolved Ni
Dissolved Pb
Dissolved Zn
Dissolved Fe
Water temperature
pH
Crib
.312
.00713
.227
.539
.00138
.00119
.00644
5.08
.508
15.13
77.3
9.79
1.04
.544
201.9
residue* 3.06
124.2
122.0
25.6
7.00
3.75
.806
.686
2.06
4.00
2.44
4.84
6.34
10.3
25.0
7.00
3.50
.800
.500
2.44
3.00
2,13
4.13
2.94
5.06
14.6
8.92
Plant
.239
.00825
.145
.384
.00100
.00100
.00250
5.25
.518
15.19
74.3
9.66
.681
.475
204.1
2.31
146.6
145.6
25.6
7.00
4.94
.806
.813
3.31
7.13
1.25
1.34
22.00
26.9
25.5
7.00
4.44
.800
.694
2.59
5.50
1.00
3.28
12.8
7.88
15.0
8.24
Diff.
.0731
.00113
.0819
.156
.000375
.000188
.00394
.175
.0100
.063
3.00
.131
.359
.688
2.19
.750
22.4
23.64
.0625
0.00
1.19
0.00
.125
1.25
3.13
1.19
3.50
15.7
16.6
.500
0.00
.938
0.00
.194
.156
2.50
1.13
.844
9.88
2.81
.406
.675
% of Lake
23.4
17.5
36.1
28.9
27.3
15.8
61.1
3.4
2.0
0.4
3.9
1.3
34.6
12.6
1.1
24.5
18.1
19.3
0.2
-
31.7
_
18.2
60.6
78.1
48.7
72.3
246.8
160.6
2.0
-
26.8
-
38.8
6.4
83.3
52.9
20.5
336.2
55.6
2.8
7.6
Includes limit of detection values.
42
-------
TABLE 21. Means for the lake (intake crib) and plant samples from the third
field trip (19 May 1981). Difference between the means for each variable
are presented as well as the percent of the lake mean represented by the
difference. Units for means are the same as Table 2.
Variable
Nitrate-N
Ammonium ion
Total Kjeldahl N
Total N
Total dissolved P*
Soluble reactive P*
Total P
Chloride
Soluble reactive Si
Sulfate
Alkalinity
Dissolved oxygen
Chlorophyll
Turbidity
Conductivity
Total non-filterable
Total residue
Combustible residue
Ca
Mg
Na
K
Cd
Cr
Cu
Hi
Pb
Zn
Fe
Dissolved Ca
Dissolved Mg
Dissolved Na
Dissolved K
Dissolved Cd
Dissolved Cr
Dissolved Cu
Dissolved Ni
Dissolved Pb
Dissolved Zn
Dissolved Fe
Water temperature
*pH
Crib
.299
.00658
.170
.469
.00100
.00100
.00542
5.72
.582
15.7
79.3
12.5
1.38
.467
206.6
residue* 4.00
113.7
113.3
26.3
7.00
3.00
.900
1.00
1.92
4.67
1.58
1.00
16.5
17.83
25.9
7.00
3.00
.900
1.00
3.17
1.75
1.17
8.58
7.50
9.75
7.00
NS
Plant
.280
.00525
.154
.433
.00100
.00100
.00333
6.07
.593
15.7
77.5
12.2
1.26
.533
207.5
4.00
124.3
124.3
26.4
7,00
3.25
.858
1.00
2.17
6.67
1.67
1.42
14.1
15.33
26.5
7.00
3.25
.883
1.00
3.75
5.33
1.42
6.83
11.7
4.83
6.73
8.13
Diff.
.019
.00133
.0158
.0358
0.00
0.00
.00208
.350
.0117
0.00
1.83
.358
.126
.0667
.833
0.00
10.7
11.0
.167
0.00
.250
.0417
0.00
.250
2.00
.0833
.417
2.41
2.50
.583
0.00
.250
.0167
0.00
.583
3.58
.250
1.75
4.17
4.92
.267
—
% of Lake
6.4
20.2
9.3
7.6
-
_
38.5
6.1
2.0
-
2.3
2.9
9.1
14.3
0.4
-
9.4
9.7
0.6
-
8.3
4.6
-
13.0
42.9
5.3
41.7
14.6
14.0
2.3
-
8.3
1.9
-
18.4
204.8
21.4
20.3
55.6
50,4
3.8
•"*
NS = No sample.
* Includes limit of detection values.
43
-------
than the lake samples. The next three largest differences were for Pb, Ni,
and Cr. These results strongly suggest that as the water passed through the
intake pipe, pumps, and other parts of the plant, metal contamination
occurred. Other variables with lake versus plant mean differences that
exceeded 20% were: ammonium ion, total phosphorus, chlorophyll, Na, dissolved
Na, dissolved Ni, and dissolved Pb,
When considered on a cruise-by-cruise basis, the results are relatively
similar to the overall database differences discussed above. The main differ-
ence is that among the three field trips are several large differences between
lake and plant means that are not present in the overall database. Table 19
shows the differences for the first field trip in August 1980. Similar to the
total database, the largest differences between lake and plant samples were Cu
and Zn at 807% and 561%, respectively. Other variables with differences that
exceeded 20% between lake and plant samples were ammonium ion, Ni, total non-
filterable residue, chlorophyll, Na, dissolved Na, and total P. Dissolved Cu
and dissolved Zn were not measured on the first field trip.
In general, there were larger differences between lake and plant samples
for the the second field trip compared to the first field trip (Table 20).
The second field followed a similar pattern to the first field trip, with the
greatest differences due to total metals and dissolved metals. The variables
that differed the most were Zn, Fe, Cu, dissolved Gu, and dissolved Zn.
Differences between plant and lake samples for all but two of the total metals
and dissolved metals exceeded 20%. Large differences also existed for several
of the physical-chemical variables including chlorophyll, Kjeldahl nitrogen,
total nitrogen, nitrate-nitrogen, total dissolved phosphorus, and total non-
filterable residue.
44
-------
Differences between lake and plant means were much smaller for the third
field trip (May 1981) compared to the first two field trips. The largest dif-
ferences were associated with Cu, Pb, dissolved Cu, dissolved Pb, dissolved
Zn, and dissolved Fe (Table 21). Beyond those five variables the only lake-
versus-plant means that differed by more than 20% were dissolved Ni, total P,
and ammonium ion.
An additional aspect of the the difference between lake and plant sam-
ples is the direction of difference or the sign of the differences. If the
differences were always in one direction, they all had the same sign. For ex-
ample, dissolved Cu values were always higher in plant samples than lake
samples; thus all differences between lake and plant samples had the same
sign. This type of result shows a consistent pattern and can be readily
interpreted as bias. Dissolved Cu was the only variable that showed such a
pattern over all three field trips. Variables which displayed unidirectional
trends for one field trip were: water temperatures during the first field trip
were always higher in the plant. Total nitrogen concentrations were always
lower in the plant during the second field trip while dissolved Zn values were
higher in the plant. During the third field trip water temperatures were
always lower in the plant as were total P concentrations while chloride values
were higher in the plant. These unidirectional differences between lake and
plant values are extremely important in revealing possible linear relation-
ships between lake and plant samples as discussed below.
ANALYSIS OF VARIANCE
Although the analysis of means by different strata as presented above
provides a large amount of insight into the differences between lake and plant
45
-------
samples, it does not considered the overall variability of the data. In order
to incorporate the full measure of variability of the database, Analysis of
Variance (ANOVA) was used to compare lake and plant samples. The hypothesis
tested by ANOVA was: did a significant difference exist between the lake and
plant samples (H : L = P vs. H : L not= P)? These were one-way two-group
ANOVAs which are essentially the same as t-tests (Sokal and Rohlf 1981). The
major aspect of the ANOVAs was that within-group variation is considered in
contrasting the two groups, i.e., means and variances of each group are com-
pared for differences (Sokal and Rohlf 1981). These ANOVAs were repeated for
the total database and each of the three field trips.
The results stand somewhat in contrast to the analysis of the means when
considered for the total database (Table 22). Whereas the the largest differ-
ence between means was for metals and dissolved metals, the ANOVAs did not
always show significant differences among the lake and plant for these vari-
ables. Only 7 of the 14 ANOVAs concerning the metals showed a significant
difference between the lake and plant. Five of those seven were highly sig-
nificant (P<.001). In contrast, of the 17 physical, chemical, and biological
variables analyzed, 11 were significantly different between the lake and plant
samples. Five of these 12 were highly significant (P<.001).
The results of the analysis of means and ANOVA suggest that the follow-
ing occurred between the lake and plant. Physical, chemical, and biological
variables were generally measured with high precision and were fairly similar
between the lake and plant. As a result, the differences for these variables
between the two locations was small. But, because the samples were measured
with precision, the within-group variance (variance about the lake or plant
means) was small. Thus comparing the two groups of data and "their associated
46
-------
TABLE 22. Analysis of Variance results from the Lake Huron Intake database.
?ariable
Nitrate-N
Ammonia
Total Kjeldahl N
Total N
Total phosphorus
Dissolved phosphorus
Total P
Chloride
Silicate
Sulfate
Alkalinity _.
Dissolved oxygen (mg L )
Chlorophyll
Turbidity
Conductivity
Total non-filterable residue
Total residue
Combustible residue
Ca
Mg
Na
K
Cd
Cr
Cu
Ni
Pb
Zn
Fe
Ca dissolved
Mg dissolved
Na dissolved
K dissolved
Cd dissolved
Cr dissolved
Cu dissolved
Nl dissolved
Pb dissolved
Zn dissolved
Fe dissolved
Water temperature
PH
Total
*
AA
AAA
AAA
NA
NA
#A
AAA
n.s.
n.s.
*
AA
*
n.s.
n.s.
NA
AAA
AAA
AAA
NA
AA
*
n. s.
n. s.
AAA
AAA
A
AAA
n.s.
AAA
NA
AA
n. s.
n.s.
n.s.
JU«***U
*
n.s.
AAA
n.s.
n.s ,
n. s .
Cruise 1
n.s.
AAA
n.s.
n.s.
NA
NA
n.s.
AAA
AAA
AA
AA
AAA
AA
n.s.
AAA
NA
n.s.
*
AAA
NA
AAA
n.s.
NA
n.s.
***
A*
n.s.
A**
n.s.
NA
NA
AA
n.s.
NA
NA
NA
NA
NA
NA
NA-
AAA
AAA
Cruise 2
n.s.
n.s.
AAA
AA
NA
NA
AAA
A
AA
n.s.
AAA
AA
n.s.
AA
n.s.
NA
AA
AAA
n.s.
NA
n.s.
n.s.
n.s.
n.s.
AAA
A*
AA
AAA
AAA
AA
NA
n.s.
n.s.
A
n.s.
AAA
AAA
n.s.
AAA
A
AAA
AAA
Cruise 3
A
AA
n.s.
A
NA
NA
AA
AAA
n.s.
n.s.
AA
n.s.
n.s.
n.s.
n.s.
NA
n.s.
n.s.
n.s.
NA
n.s.
A
NA
n.s.
n.s.
n.s.
A
n.s.
n.s.
n.s.
NA
n.s.
n.s.
NA
n.s.
AAA
n.s.
n.s .
A
AAA
NA
NA
Key: n.s. = non-significant, * = P<.05, **
AAA = .OOKP, NA = no analysis
= .01
-------
variances yielded a statistically significant result. In contrast, the metals
and dissolved metals were measured with lower precision. The high within-
group variance obscured differences between lake and plant metals concentra-
tions.
The ANOVA results can somewhat be attributed to the behavior of the sta-
tistical procedures. But, this does not suggest that the ANOVA results should
be ignored. The results suggest that improvement in the analysis of metal
concentrations is necessary before the intake should be considered as a moni-
toring tool. With improvement of analytical techniques, within-group vari-
ances will shrink and more significant differences between groups can be ex-
pected. The physical, chemical, and biological variables in contrast were
measured with sufficient precision. Small but real differences existed be-
tween lake and plant samples. Those differences do not appear to have bias,
but randomly change among samples. If that small amount of difference intro-
duced by the Intake can be accepted, then monitoring of Lake Huron water can
be carried out from intake samples.
The pattern established by the ANOVAs for the entire database persisted
through the analysis of the results of each field trip (Table 22). Generally
metal and dissolved metal concentrations were not significantly different be-
tween the lake and plant while the physical, chemical, and biological
variables were significant. There was also a trend toward fewer significant
results on the second and third field trips compared to the first. This trend
may have resulted from improved sample handling on the later field trips. The
results from the third field trip clearly indicate the feasibility of using
the water intake for monitoring purposes. Only 10 of 30 ANOVAs showed a
48
-------
significant difference between lake and plant samples with only 3 of the 10
highly significant (P-C.001).
CORRELATION AND REGRESSION
The sampling program employed an approach where each sample collected
from the lake had a matching sample collected from the water treatment plant.
Approximate times for a parcel of water to travel from the intake crib in the
lake to the plant were computed, and samples were collected accordingly.
Following the techniques of Kubus and Egloff (1982), a linear regression ap-
proach can be used to determine the correspondence between matched" lake and
plant samples. The theory behind such an approach is as follows: If the
intake pipe and plant do not affect water quality, there should be a perfect
1:1 correspondence between lake and plant samples. That perfect
correspondence could be reflected in a linear regression of plant on lake
values. The regression would have a slope of 1.00 and a y-intercept of 0.00.
In other words, high values for lake samples would yield high values for plant
samples and, conversely, low values would yield low values. Furthermore,
deviations from the regression should indicate if a bias exists and the nature
of that bias, e.g., are all plant values lower or higher than lake values?
The assumption of using this technique is that the data are normally
distributed, have independent variances, etc. (Netter and Wasserman 1974).
Similar to the analysis of means and ANOVAs, the regression approach was
used for two different strata, the total database and by field trips. The re-
sults of the linear regression analysis (Table 23) show that very few of the
regressions of lake onto plant data were statistically significant. This was
true for the total database as well on an individual field trip basis. Of the
49
-------
TABLE 23. Linear regression results from the Lake Huron Intake database.
Variable
Nitrate-N
Ammonia
Total Kjeldahl N
Total N
Total phosphorus
Dissolved phosphorus
Total P
Chloride
Silicate
Sulfate
Alkalinity
Dissolved oxygen (mg L )
Chlorophyll
Turbidity
Conductivity
Total non-filterable residue
Total residue •
Combustible residue
Ca
Mg
Na
K
Cd
Cr
Cu
Ni
Pb
Zn
Fe
Ca dissolved
Mg dissolved
Na dissolved
K dissolved
Cd dissolved
Cr dissolved
Cu dissolved
Ni dissolved
Pb dissolved
Zn dissolved
Fe dissolved
Water temperature
PH
Total
n.s.
n.s.
**(-.741)
n.s.
NA
NA
n.s.
***(.662)
***(!. 18)
n.s.
**(.535)
***(!. 08)
n.s.
n.s.
**(4.14)
NA
n.s.
n.s.
n.s.
NA
**(.349)
*(.350)
n.s.
**(.345)
n.s.
n.s.
n.s .
n.s.
n. s.
***(-!. 46)
NA
**(.349)
***(.882)
**(.538)
n.s.
n.s.
n.s.
**(.528)
n.s.
*(.378)
***(.881)
***(2.11)
Cruise 1
n.s.
n.s.
* (-.703)
n.s.
NA
NA
n.s.
*(.417)
n.s.
n.s.
*(.462)
n.s.
*(-.643)
n.s.
**(1.00)
NA
n.s.
n.s.
NA
NA
n.s.
n.s.
NA
NA
*(.132)
n.s .
n.s .
n.s.
n.s.
NA
NA
n.s.
NA
NA
NA
NA
NA
NA
NA
NA
NA
n.s.
Cruise 2
n.s.
n.s.
n.s.
n.s.
NA
NA
n.s.
n.s.
***(.909)
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
NA
n.s.
n.s.
n.s.
NA
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
**(-.322)
n.s.
NA
NA
n.s.
NA
NA
*<-.574)
*(.188)
n.s.
n.s.
n.s.
n.s.
NA
n.s.
Cruise 3
n.s.
n.s.
n.s.
n.s.
NA
NA
***(!. 45)
***{.636)
**<1.16)
n.s.
n.s.
n. s.
n. s.
n. s.
*(.667)
NA
n.s.
n. s.
*(.398)
NA
NA
n.s.
NA
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n. s.
NA
NA
n.s.
NA
n.s.
n.s.
***(.382)
*(.873)
n.s.
n. s.
NA
NA
Key: n.s. = non-significant, * = P<.05, ** = ,01
-------
total of 126 linear regression analyses, only 34 yielded slopes that were sta-
tistically different from 0.00. Those analyses included variables with highly
non-normal distributions. Of the 34 significant regressions, only 11 were
significant at the P<.001 probability level. Most (seven) of those highly
significant regressions were from analysis of the total database.
Scatter plots of the plant versus lake samples give a relatively clear
indication of the relationship between the two sets of samples. In effect,
there is a high degree of within-group variation, especially for the metals
and dissolved metals. This concept emerged from the ANOVAs as well as from
the regression analysis. For the most part there were small differences in
the means between lake and plant data. Those small differences combined with
the high within-group variances did not produce a clear relationship between
lake and plant results. Rather, the lake results appeared to vary about a
lake mean and the plant results about a plant mean. The two sets of data
overlapped in such a way as to produce very low correlations. In effect, the
regression analysis determined that if a lake sample had a low value for a
particular variable, one should not expect a matching low value from the
plant. But, the plant value should be very close to the lake value. There
are many explanations for this phenomenon, but the most obvious is that during
the 9-hour passage of water through the intake pipe and pumps, a large amount
of mixing occurs. Thus parcels of water lose their integrity and become well
mixed before they reach the plant. This appeared true during any one field
trip. The occurrence of this relationship across the entire database is
somewhat surprising in that seasonal factors should disrupt this pattern.
Sample analysis problems are the other possible cause behind the lack of
significant regressions.
51
-------
Because of the problem of non-normal distributions for many variables, a
non-parametric approach was used in addition to the linear regression. The
rationale behind using this technique was that the non-parametric approach
should not be so readily influenced by the non-normal distributions (Conover
1980). The approach used was the rank correlation analysis. This technique
takes the two sets of data and ranks all the results from low to high values.
Orders are then assigned to each value with the highest value and the lowest
value receiving the highest and lowest ranking respectively (Sokal and Rohlf
1981). Rankings within the two sets of data are then compared to determine if
the highest ranked value in one data set matches the highest ranked item in
the other set, etc. An expression of the correspondence is the rank-order
correlation coefficient.
The rank-order correlation results were very similar to the linear re-
gression and parametric correlation results (Table 24). Of the 126 analyses
conducted, only 35 yielded significant rank correlations. Of the 35 sta-
tistically significant rank-correlations, 10 were highly significant at the
P<.001 level. Out of the 126 analyses, there were only 10 differences between
the regression and the rank -correlation results. This high level of corre-
spondence between the two types of analyses gives support to both sets of
analyses. Similar to the linear regression analyses, the majority of signifi-
cant rank correlations were associated with the total database analyses. Very
few of the rank correlations were statistically significant when considered on
an individual field trip basis. The net conclusion of the rank correlation
analysis matches that from the linear regression results; there were small
differences between the mean lake and plant values, but the samples do not
match on a 1:1 basis.
52
-------
TABLE 24. Rank correlation results from the Lake Huron Intake database.
Variable
Nitrate-N
Ammonia
Total Kjeldahl N
Total N
Total phosphorus
Dissolved phosphorus
Total P
Chloride
Silicate
Sulfate
Alkalinity
Dissolved oxygen (mg L )
Chlorophyll
Turbidity
Conductivity
Total non-filterable residue
Total residue
Combustible residue
Ca
Mg
Na
K
Cd
Cr
Cu
Ni
Pb
Zn
Fe
Ca dissolved
Mg dissolved
Na dissolved
K dissolved
Cd dissolved
Cr dissolved
Cu dissolved
Ni dissolved
Pb dissolved
Zn dissolved
Fe dissolved
Water temperature
pH
Total
***(.590)
n.s.
n.s.
n.s.
NA
NA
n.s.
***(.886)
***(,948)
n.s.
*(.367)
***(.895)
n.s.
n.s.
*(.338)
NA
n.s.
n.s.
n.s.
NA
**(.432)
*(.320)
n.s.
**(.395)
*(-.368)
n.s.
n.s.
n. s.
n.s.
**(-.456)
NA
n. s.
***(.794)
**(.592)
n.s .
n. s.
n.s.
n.s.
n. s .
**(-.560)
***( . 959 )
***(.779)
Cruise 1
n.s.
n.s.
*<-.546)
n.s.
NA
NA
n, s.
*(.592)
n.s.
n.s.
*(.534)
n. s.
*<-.561)
n.s.
n.s.
NA
n.s.
n.s.
NA
NA
n.s.
n. s.
NA
NA
*(.568)
n.s.
n.s.
n.s.
n.s.
NA
NA
n. s.
NA
NA
NA
NA
NA
NA
NA
NA
NA
n, s.
Cruise 2
n.s.
n.s.
n.s.
n.s.
NA
NA
n.s.
n.s.
***(.824)
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
NA
n.s.
n.s.
n.s.
NA
*(.482)
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
**(-.644)
n.s.
NA
NA
n.s.
NA
NA
*(-.586)
*(.517)
n.s.
*(-.511)
n.s.
n.s.
NA
n.s.
Cruise 3
n.s.
n.s.
n.s.
n.s.
NA
NA
**(.777)
***(.924)
**(.783)
n.s.
n.s.
n.s.
n.s.
n. s.
*(.707)
NA
n.s.
n.s.
*(.689)
NA
NA
NA
NA
n.s.
n.s.
n.s.
n.s.
*(-.614)
n.s.
n.s.
NA
NA
NA
NA
n.s.
n.s.
***(.996)
n.s.
n. s.
n.s.
NA
NA
Key: n.s. = non-significant, * = P<.05, ** = .01
-------
TRANSFORMATIONS
The object of the transformations was to change the distribution of
those variables which were obviously non-normal. After the transformation,
the statistical analyses are repeated to determine if the results change.
Results that do not change between original and transformed data are a strong
indication that the non-normal distribution has not substantially affected the
analytical procedure. In contrast, those results that change can be
considered as highly influenced by the distribution of the data. Analyses
with these data cannot be considered reliable.
The type of transformation used depends greatly on the perceived factors
creating the non-normal distribution. For example, data that are originally
recorded as counts lend themselves to a square root transformation (Sokal and
Rohlf 1981). Percentages or proportions are commonly subject to a arcsine
transformation (Netter and Wasserman 1974). The most common transformation is
converting the original data to logs which is carried out when the variance is
positively correlated with the mean (Sokal and Rohlf 1981). The Lake Huron
water intake data set was primarily plagued with extreme skew due to occa-
sional large values. Those types of transformation generally defy transforma-
tion. Five variables, nitrate-nitrogen, total nitrogen, total phosphorus,
chlorophyll, and chloride, were transformed into natural logs. Table 25 shows
that three of the five variables had much lower skew and kurtosis after trans-
formation, while the distributions of nitrate-nitrogen and total nitrate did
not change.
The ANOVA and linear regression analyses were repeated for these five
variables to determine if the results changed much from the original data.
54
-------
TABLE 25. Descriptive statistics for five variables after a natural log transformation.
Ul
Ul
Variable
Log NO + NO
Log total N
Log total P
Log chlorophyll
Log conductivity
N
88
88
88
88
88
Minimum
-1.514
-1.109
-6.908
-1.561
5.278
Maximum
-.010
.182
-4.343
1.548
5.394
Mean
-1.2766
-.7936
-5.4153 •
-.1196
5.3150
Std Dev
.1932
.1784
.5798
.5066
.0202
Skewness
3.926
2.220
-.716
-.268
.722
Kurtosis
21.024
9.565
.553
1.010
1.164
-------
The rank correlations will not change if only the scale of the data is
changed; the relative ranks remain unaffected by the transformation,
'The results of the ANOVA and linear regression analyses with the trans-
formed data showed very little difference from the results with the untrans-
formed data (Table 26). In only one case of the 40 analyses (20 ANOVAs and 20
regressions) was there a difference of interpretation. That case was that
lake and plant nitrate values were considered statistically different for the
second field trip (October 1980), while they were not for the untransformed
data. However, the level of significance for the transformed data was P<.05,
which is not very high considering the non-normal distribution of the nitrate
data. There were five analyses where the probability levels of significant
statistical results increased substantially, and one case where it was reduced
(Table 26). Overall, the conclusion was that the transformations did not
alter the interpretation of the results.
COMPARISON OF INTAKE SAMPLES WITH OPEN LAKE HURON SAMPLES
Perhaps the most fundamental question in addressing the issue of use of
water treatment samples for monitoring is the ability of those samples to re-
flect conditions in the lake. Even if plant and intake crib samples are per-
fectly matched, there will be no value toward monitoring if the samples do not
represent the location of interest. In this case, the area of interest is
southern Lake Huron. Presumably the Detroit Municipal Water Intake crib
should be far enough offshore to adequately sample southern Lake Huron water.
But, a fortunate set of circumstances allows more than a simple presumption.
During 1980, an intensive study of Lake Huron was conducted by a,joint
American-Canadian program. These results were used to characterize Lake Huron
56
-------
TABLE 26. Comparison of results using transformed and untransformed data for
the ANOVA and linear regression analyses. Original data were transformed
into natural logs.
Analysis
of Variance
Transformed
Variable
Nltrate-N
Total N
Total P
Chlorophyll
Conductivity
T 1
** n.s.
*** n.s. *
*** n.s. *
* ** n
n.s. *** n
Linear
2
*
•rJU^U
:**
.S.
.S.
3
*
*
***
n.s.
n.s.
Un transformed
T 1 2
* n.s. n.s.
*** n.s. **
** n.s, 5t**
* ** n.s.
n.s. *** n.s.
3
•k
*
**
n.s.
n.s.
Regression
Transformed
Variable
Nitrate-N
Total N
Total P
Chlorophyll
Conductivity
T 1
n.s. • n . s . n
n.s. n.s, n
n.s. n.s. n
n.s. * n
5JrsV * n
2
.3.
,S.
.S.
. S.
.S.
3
n.s.
n.s.
**
n.s.
*
Un trans f o rme d
T 1 2
n.s, n.s. n.s.
n.s. n.s. n.s.
n.s . n.s . n.s.
n.s. * n.s.
** * * n.s.
3
n.s.
n.s.
***
n. s.
-A-
Key: T = total database, 1 = first field trip, 2 = second field trip,
3 = third field trip. Significance levels the same as Table 22.
57
-------
in a variety of ways, including water quality by region of the lake (Moll et
al. 1985). Southern Lake Huron water masses were identified and means and
variances computed for each water mass. These values allow comparison of the
plant results to the open lake conditions. Some compromises were necessary to
match the open Lake Huron data with the water intake plant data. Most of
those compromises were in comparing samples collected from a few weeks apart.
The first field trip of 19-26 August 1980 was matched with open lake samples
collected from 11-15 September. The second field trip of 7-9 October 1980 was
matched with open lake samples from 22-27 October. The third field of 19 May
1981 was matched with samples from 10-11 May 1980. In addition, the number of
comparisons between water treatment samples and open lake samples was limited
by the overlap in the number of variables measured. Furthermore, the success
of any matches assumes no difference in analytical methods, which is very
unlikely.
Southern Lake Huron water quality is compared to the intake plant sam-
ples in Table 27 through the use of means from each field trip. Only epilim-
netic samples were used to compute the open lake means, although the southern
basin was hornothermous during the October and May field trips. The results
show that a moderate degree of bias due to sampling from the water can be ex-
pected. Eleven variables were compared among three field trips with the fol-
lowing results: Total dissolved phosphorus and soluble reactive silica were
always substantially lower at the plant compared to the open lake. Ammonium
ion, dissolved oxygen, and chloride were always slightly lower in the plant
compared to the open lake. Nitrate-nitrogen, soluble reactive phosphorus,
conductivity, and alkalinity were generally relatively similar between the
plant and lake, although occasional large differences were observed. Total
58
-------
TABLE 27. Comparison of southern Lake Huron means (from Moll et al, 1985)
to water intake plant means for three field trips.
Field Trip
Augus t
Variable
NO + NO
NH3
Total Ortho P
Soluble P
Total P
Silicate
Dissolved oxygen
pH
Conductivity
Chloride
Alkalinity
Lake
.240
2.60
2.10
0,66
3.90
0.89
9.5
8.38
197.1
5.90
78.5
Plant
.283
.844
1.13
NS
4.36
.423
8.43
7.84
199.1
5.75
75.4
October
Lake
.255
5.20
2.80
0.72
4.80
1.09
10.8
8.31
210.7
5.60
80.8
Plant
.239
.825
1.00*
1.00*
2,50
.518
9.66
8.24
204.1
5.25
74.3
May
Lake Plant
.284
1.80
2.70
0.80
4.00
1.50
14.2
8.02
203.1
NS
76.3
.280
.525
1.00*
1.00*
3.33
.593
12,2
8.12
207.5
6.07
77.5
NS = no sample.
* Includes limit of detection values.
59
-------
phosphorus was usually but not always in higher concentrations in the lake
than the plant. The differences between the open lake and the plant samples
were probably as much a result of the location of the water intake as due to
any effect from the plant itself. Differences between the crib and open lake
samples appeared as large as between the plant and open lake. In effect, the
use of the intake plant for monitoring purposes appeared as much constrained
by the location of the crib as by the effect of pumping water through the
intake pipe.
DISCUSSION
The results of the sample analysis indicate that several factors af-
fected water quality as the water moved from Lake Huron to the intake facil-
ity. Only some of these factors were attributed directly to the intake pipe
itself. Other factors included seasonal!ty in the sense that some variables
changed significantly during one season, but not others. Also, seasonal!ty
was important because of the "sampling effect" in regard to the intake crib.
This last effect was generated by the location of the crib in relation to the
physical regime of Lake Huron. Finally, in a few instances, bias also con-
tributed toward the difference between the plant and lake samples.
Tables 28 to 31 summarize the results presented above and provide some
overview to the response of the different variables investigated. On a
generic level, the overall differences between the plant and lake samples were
much higher for the total and dissolved metals than for the other variables.
This effect was persistent throughout the three field trips, although some
question of analytical technique emerges. Basically, the metals results sug-
gest somewhat noisy data with large variances for both the treatment plant and
60
-------
TABLE 28, Summary statistics for all analyses for the entire database.
Statistics include: means for the lake (intake crib) and plant samples and
per cent of the crib means represented by the differences between the means.
Also shown are ANOVA, rank correlation, and linear regression results.
Variable
Nitrate-N
Ammonium ion
Total KJeldahl N
Total N
Total Dissolved P
Soluble reactive P
Total P
Chloride
Soluble reactive Si
Sulfate
Alkalinity
Dissolved oxygen
Chlorophyll
Turbidity
Conductivity
Total non-fil. residue
Total residue
Combustible residue
Ca
Mg
Na
K
Cd
Cr
Cu
Ni
Pb
Zn
Fe
Dissolved Ca
Dissolved Mg
Dissolved Na
Dissolved K
Dissolved Cd
Dissolved Cr
Dissolved Cu
Dissolved Ni
Dissolved Pb
Dissolved Zn
Dissolved Fe
Water temperature
pH
Crib
.306
.00591
.193
.498
.00114
.00111
.00593
5.38
.481
15.68
76.61
10.27
1.15
.527
203.6
3.32
121.8
120,0
27.39
7.00
3.30
.852
.886
1.64
3.34
2.00
2,49
9.35
19.64
27.07
7.00
3.14
.830
.714
2.75
2.46
1.71
6.04
4.89
7.07
14.66
8.28
Plant
.266
.00750
.157
.423
.00105
.00100
.00436
5.65
.504
15.55
75.57
9.39
.856
.509
203.2
2.66
135,3
134.8
25.71
7.00
4.14
.830
.932
2.18
9.98
1.30
1.28
28.66
23.16
25.59
7.00
3.77
.823
.825
3.09
5.43
1.18
4.80
12.32
6.57
15.29
8.12
% of
Crib
12.9
26.9
18.1
15.1
7.9
9,7
26.5
5.1
4.7
0.9
1.4
3.4
25.9
3.5
0.2
19.8
11.1
12.4
6.1
-
25.5
2.7
5.1
33.2
198.7
35.3
48.4
206.5
18.0
5.5
-
20.3
1.1
15.5
12.3
120.3
31.3
20.4
151.9
7.1
4.3
2.9
ANOVA
*
**
***
***
NA
NA
**
***
n.s.
n.s.
*
**
*
n.s.
n.s.
NA
***
***
***
NA
**
*
n.s.
n.s.
***
***
-------
TABLE 29. Summary statistics for all analyses for the first field trip.
Statistics include: means for the lake (intake crib) and plant samples and
per cent of the crib means represented by the differences between the means.
Also shown are ANOVA, rank correlation, and linear regression results.
Variable
Ni tra te-N
Ammonium ion
Total Kjeldahl N
Total N
Total dissolved P
Soluble reactive P
Total P
Chloride
Soluble reactive Si
Sulfate
Alkalinity
Dissolved oxygen
Chlorophyll
Turbidity
Conductivity
Total non-fil. residue
Total residue
Combustible residue
Ca
Mg
Na
K
Cd
Cr
Cu
Ni
Pb
Zn
Fe
Dissolved Ca
Dissolved Mg
Dissolved Na
Dissolved K
Dissolved Cd
Dissolved Cr
Dissolved Cu
Dissolved Ni
Dissolved Pb
Dissolved Zn
Dissolved Fe
Water temperature
PH
Crib
.304
.00419
.174
.479
.00100
NA
.00581
5.44
.379
16.25
73.9
8.87
1.10
.556
203.0
3.06
125.4
123.0
30.00
7.00
3.06
.863
1.00
1.00
1.69
1.88
1.25
7.00
30.3
30.0
7.00
2.88
.813
NA
NA
NA
NA
NA
. NA
NA
20.5
7.64
Plant
.283
.00844
.172
.454
.00113
NA
.00700
5.75
.423
15.83
75.4
8.43
.731
.525
199.1
2.00
132.3
132.0
25.31
7.00
4.00
.831
1.00
1.06
15.38
1.06
1.13
46.3
25.3
25.0
7.00
3.50
.800
NA
NA
NA
NA
NA
NA
NA
22.0
7.84
% of
Crib
7.0
101.4
1.4
5.1
12.5
-
20.5
5.7
11.5
2.7
2.0
5.0
33.3
5.6
1.9
34.7
5.4
7.3
15.6
-
30.6
3.6
-
6.3
807.4
43.3
10.0
560.7
16.5
16.7
_
21.7
1.5
-
_
-
-
-
-
-
7.5
2.6
ANOVA
n.s.
***
n.s.
n.s.
NA
NA
n.s.
***
***
**
**
***
**
n.s.
***
NA
n.s.
*t»
***
NA
***
n.s.
NA
n.s.
***
fc*
n.s.
***
n. s.
NA
NA
**
**
NA
NA
NA
NA
NA
NA
NA
**iV
***
Rcorr
n.s.
n.s.
*
n.s.
NA
NA
n.s.
*
n.s.
n.s.
*
n.s.
*
n.s.
n.s.
NA
n.s.
n.s.
NA
NA
n.s.
n.s.
NA
NA
*
n.s.
n.s.
n.s.
n.s.
NA
NA
n.s.
NA
NA
NA
NA
NA
NA
NA
NA
NA
n.s.
Regress
n.s.
n.s.
*
n.s.
NA
NA
n.s.
*
n.s.
n.s.
*
n.s.
*
n.s.
**
NA
n.s.
n.s.
NA
NA
n.s.
n.s.
NA
NA
*
n.s.
n.s.
n.s.
n.s.
NA
NA
n.s.
NA
NA
NA
NA
NA
NA
NA
NA
NA
n.s.
Key: n.s. = non-significant, * - P<.05,-** = .01
-------
TABLE 30. Summary statistics for all analyses for the second field trip.
Statistics include: means for the lake (intake crib) and plant samples and
per cent of the crib means represented by the differences between the means.
Also shown are ANOVA, Rank correlation and linear regression results.
Variable
Nitrate-N
Ammonium ion
Total Kjeldahl N
Total N
Total dissolved P
Soluble reactive P
Total P
Chloride
Soluble reactive Si
Sulfate
Alkalinity
Dissolved oxygen
Chlorophyll
Turbidity
Conductivity
Total non-fil. residue
Total residue
Combustible residue
Ca
Mg
Na
K
Cd
Cr
Cu
Ni
Pb
Zn
Fe
Dissolved Ca
Dissolved Mg
Dissolved Na
Dissolved K
Dissolved Cd
Dissolved Cr
Dissolved Cu
Dissolved Ni
Dissolved Pb
Dissolved Zn
Dissolved Fe
Water temperature
pH
Crib
.312
.00713
.227
.539
.00138
.00119
.00644
5.08
.508
15.13
77.3
9.79
1.04
.544
201.9
3.06
124.2
122.0
25.6
7.00
3.75
.806
.686
2.06
4.00
2.44
4.84
6.34
10.3
25.0
7.00
3.50
.800
.500
2.44
3.00
2.13
4.13
2.94
5.06
14.6
8.92
Plant
.239
.00825
.145
.384
.00100
.00100
.00250
5.25
.518
15.19
74.37
9.66
.681
.475
204.1
2.31
146.6
145.6
25.6
7.00
4.94
.806
.813
3.31
7.13
1.25
1.34
22.0
26.9
25.5
7.00
4.44
.800
.694
2.59
5.50
1.00
3.28
12.8
7.88
15.0
8.24
% of
Crib
23.4
17.5
36.1
28.9
27.3
15.8
61.1
3.4
2.0
0.4
3.9
1.3
34.6
12.6
1.1
24.5
18.1
19.3
0.2
-
31.7
_
18.2
60.6
78.1
48.7
72.3
246.8
160.6
2.0
-
26.8
-
38.8
6.4
83.3
52.9
20.5
336.2
55.6
2.8
7.6
ANOVA
n. s.
n.s.
#**
#•*
NA
NA
***
*
A"?V
n.s.
***
**
n.s.
itit
n.s.
NA
#*
***
n. s .
NA
n.s.
n.s .
n, s.
n.s.
***
•kit
**
itit-is
***
**
NA
n.s.
n. s.
*
n.s.
***
***
n.s.
***
*
***
.JWWU
Rcorr
n.s.
n.s.
n.s.
n.s.
NA
NA
n.s.
n.s.
***
n.s.
n. s.
n.s.
n.s.
n.s.
n.s.
NA
n.s.
n.s.
n.s.
NA
"i\
n.s.
n. s .
n.s.
n.s.
n.s.
n.s.
**
n.s.
NA
NA
n.s.
NA
NA
*
*
n.s.
*
n. s .
n. s.
NA
n.s.
Regress
n.s.
n.s.
n.s.
n.s.
NA
NA
n.s.
n.s.
***
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
NA
n.s.
n.s.
n.s.
NA
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
**
n.s.
NA
NA
n.s.
NA
NA
*.'*
*
n.s.
n. s.
n.s.
n.s.
NA
n.s.
Key: n.s. = non-significant, * = PC05, ** = .01
-------
TABLE 31. Summary statistics for all analyses for the third field trip.
Statistics include: means for the lake (intake crib) and plant samples and
per cent of the crib means represented by the differences between the means.
Also shown are ANOVA, rank correlation, and linear regression results.
% of
Variable
Nitrate-N
Ammonium ion
Total Kjeldahl N
Total N
Total dissolved P
Soluble reactive P
Total P
Chloride
Soluble reactive Si
Sulfate
Alkalinity
Dissolved oxygen
Chlorophyll
Turbidity
Conductivity
Total non-fil. residue
Total residue
Combustible residue
Ca
Mg
Na
K
Cd
Cr
Cu
Ni
Pb
Zn
Fe
Dissolved Ca
Dissolved Mg
Dissolved Na
Dissolved K
Dissolved Cd
Dissolved Cr
Dissolved Cu
Dissolved Ni
Dissolved Pb
Dissolved Zn
Dissolved Fe
Water temperature
PH
5
15
79
12
1
206
4
113
113
26
7
3
1
1
4
1
1
16
17
25
7
3
1
3
1
1
8
7
9
7
NA
Key: n.s. = non-significant
*** = .OOKP, NA
= no
Crib
.299
.00658
.170
.469
.00100
.00100
.00542
.72
.582
.7
.3
.5
.38
.467
.6
.00
.7
.3
.3
.00
.00
.900
.00
.24
.67
.58
.00
.5
.83
.97
.00
.00
.900
.00
.17
.75
.17
.58
.50
.75
.00
Plant
.
.
.
.
.
.
.
6.
.
15.
77.
12.
1.
.
207.
4.
124.
124.
26.
7.
3.
.
1.
2.
6.
1.
1.
14.
15.
26.
7.
3.
.
1.
3.
5.
.1.
6.
11.
4.
6.
8.
, * = P<.05
280
00525
154
433
00100
00100
00333
07
593
7
5
2
26
533
5
00
3
3
4
00
25
858
00
17
67
67
42
1
33
5
00
25
883
00
75
33
42
83
7
83
73
32
, ** =
Crib
6.
20.
9.
7.
-
-
38.
6.
2.
-
2.
2.
9.
14.
0.
-
9.
9.
0.
-
8.
4.
-
13.
42.
5.
41.
14.
14.
2.
-
8.
1.
-
18.
204.
21.
20.
55.
50.
3.
.OK
4
2
3
6
5
1
0
3
9
1
3
4
7
4
6
3
6
0
9
3
7
6
0
3
3
9
4
8
4
3
6
4
8
P<
ANOVA
*
**
n.s.
*
NA
NA
**
***
n.s.
n.s.
**
n.s.
n.s.
n.s.
n.s.
NA
n.s.
n.s.
n.s.
NA
n.s.
*
NA
n.s.
n.s.
n.s.
*
n.s.
n. s.
n.s.
NA
n.s.
n.s.
NA
n.s.
***
n.s.
n.s.
*
***
NA
NA
.05,
Rcorr
n.s.
n.s.
n.s.
n.s.
NA
NA.
**
***
**
n.s.
n.s.
n.s.
n.s.
n.s.
*
NA
n.s.
n.s.
*
NA
NA
NA
NA
n.s.
n.s.
n.s.
n.s.
*
n. s.
n.s.
NA
NA
NA
NA
n.s.
n.s.
***
n.s.
n.s.
n.s.
NA
NA
Regress
n.s
n.s
n.s
n.s
NA
NA
.
,
.
.
***
***
**
n.s
n.s
n.s
n.s
n.s
*
NA
n.s
n.s
*
NA
NA
n.s
NA
n.s
n.s
n.s
n.s
n.s
n.s
n.s
NA
NA
n.s
NA
n.s
n.s
,
*
,
.
,
^
m
,
.
,
,
,
,
.
.
*
f
f
***
*
n.s
n.s
NA
NA
t
analysis.
64
-------
intake and between the two locations. As a result, the metals data did not
show a significant plant versus lake effect with ANOVA for about half of the
of metals. In contrast, the physical-chemical data appeared to have smaller
variances, especially at either the intake crib or plant for one field trip.
Viewed on an individual basis, several variables showed patterns that
indicated a very strong seasonal interaction. For example, chlorophyll con-
centrations were generally lower in the plant versus the corresponding lake
values. The difference was about 25% of the mean (Tables 28-31). Yet, only
during the first (August) field trip are the differences between the plant and
lake significant as indicated by the ANOVA. The lower chlorophyll values in
the plant probably reflected the loss of phytoplankton during the 9-hour
transit through the dark intake pipe. The magnitude of that loss appeared
similar between the first and second field trips, yet only the first trip
showed a statistical difference. The inference from these analyses is that
only at certain times of the year (spring and fall) should chlorophyll be
monitored without concern as to pipe effects. The floristic analysis of
Ladewski et al. (1982) supported this conclusion; certain species of algae
were sensitive to transit through the pipe while other species were not
affected. Because of the seasonal dominance of different species of phyto-
plankton, it appears reasonable that the plant effect is seasonal.
Another variable that showed an extreme seasonal pattern was total phos-
phorus (TP). Overall, TP was significantly lower in the plant sample than the
lake samples. The mechanism for this phosphorus loss was not apparent.
Further, there was a seasonal component to the changes of TP between plant and
lake. During the summer (first field trip), TP concentrations did not change
between plant and lake. But, during the second and third field trips, TP con-
65
-------
centrations were much lower in the plant compared to the lake. The seasonal
loss mechanism is not apparent; discretion must be used when monitoring TP
using water intake data.
A third example of the season by variable interaction is for soluble
reactive silica (SRS). During the first field trip (August), the plant silica
concentrations were significantly higher that the lake values. Although that
trend continued for the remaining two field trips, the magnitude of the dif-
ferences became smaller until the third field trip (May) when the differences
were not statistically significant. Because SRS is a crucial nutrient for
diatom growth (Ladewski et al. 1982), concentrations are lower in the
epilimnion during the summer. Thus, the location of the intake crib in rela-
tion to the thermocline can play a large role in the concentration of SRS ar-
riving at the water treatment plant. Because the thermocline oscillates
(Mortimer 1971), the intake crib may alternatively draw from epilimnetic and
hypolimnetic water (silica poor and silica rich). During the winter and early
spring the thermocline, if it exists, is above the intake at 14 m below the
surface. In the fall, the thermocline is below the depth of the crib. Thus,
the potential for SRS differences between plant and lake is minimal. But,
during the summer a mixing of epilimnetic and hypolimnetic waters can affect
the SRS concentrations at the plant. The mixing effect of the intake pipe is
a feature that was evident in several variables,
Several other examples of the season by location interaction exist in
the database. As a result, each variable should be carefully examined for its
usefulness in a monitoring program. But, a general categorization is possible
from the analyses discussed above: variables that did not change between lake
and plant, changed a small amount, a large amount, and a very large amount.
66
-------
Table 32 provides that categorization and indicates that the intake plant ef-
fects are not restricted to one type of variable (e.g. metals), but cut across
all types of variables. Approximately two fifths of the total database dis-
played large or very large effects. The remaining variables should be useful
for monitoring. Also, several of the variables can be readily monitored from
the plant during certain seasons; for example, as discussed above, chlorophyll
during the non-summer season.
One additional aspect of the sampling program is the use of a fixed
location "sampling device." The lack of flexibility in the location of the
collection of the sample can be both a benefit and a problem. As mentioned
above, during the period of thermal stratification, a concern is the type of
water drawn into the intake structure. This is a problem only during the
relatively brief period when the thermocline is within 5 meters of the intake
structure depth. Perhaps a more persistent problem is the nature of the water
mass sampled. Moll et al. (1985) showed that distinct water masses exist
during much of the year for Lake Huron. The location of those water masses
changes among the seasons. Thus it is probable that the type of water
arriving at the water treatment plant will vary depending on the water mass
overlying the intake crib. Identification of that water mass will be diffi-
cult without analysis of samples from other locations in Lake Huron. In con-
trast to the uncertainty of the water type entering the intake pipe is the
fixed and known location of the intake crib. Unlike the samples collected
from ships, the location and depth of the intake is precisely known. This
provides a good reference if water intake samples are included in the monitor-
ing program.
67
-------
TABLE 32. Categorization of variables in regard to changes generated by
passage through the intake pipe for the entire database.
Little or No Change
Soluble reactive silica
Turbidity
Total dissolved phosphorus
Soluble reactive phosphorus
Mg
Pe
Dissolved Cr
Dissolved Fe
Dissolved Mg
Sulfate
Conductivity
PH
Water temperature
Cd
Dissolved Cd
Dissolved Pb
Dissolved K
Small Changes
Nitrate
Dissolved oxygen
Alkalinity
K
Large Changes
Ammonium ion
Chlorophyll
Total non-filterable residue
Chloride
Nay
Cr
Dissolved Na
Total phosphorus
Total residue
Combustible residue
Ca
Cd
Pb
Very Large Changes
Total Kjeldahl nitrogen
Cu
Zn
Dissolved Cu
Dissolved Ni
Total nitrogen
Ni
Dissolved Ca
Dissolved Zn
68
-------
In sum, the use of the Detroit Municipal Water Intake as a source of
water for monitoring Lake Huron is a worthwhile concept. The advantages
include a reliable source of water at all times of the year, particularly when
sampling by ship is not feasible. Further, the cost of collecting the samples
is trivial compared to collection with a research vessel. An additional
advantage is the knowledge of precisely where the sample was collected. In
contrast, several disadvantages are associated with analysis of water intake
samples. The largest disadvantage is the change imparted to the water when it
is pumped into the treatment plant. For many variables the change is small
and insignificant, while for others it renders the samples useless for
monitoring. The most difficult problem arises with those samples which are
affected during some seasons but not others. An additional disadvantage is
the uncertainty of the nature of the water drawn into the intake crib,
especially during the period of thermal stratification. The results point
toward a program of monitoring of Lake Huron with ample use of Detroit
Municipal Water Intake samples, supplemented with field samples collected at
least several times per year.
69
-------
REFERENCES
Beeton, A.M., and J. Strand. 1975. Sources of water quality, lake level, ice
water temperature and meteorological data for the St. Lawrence Great
Lakes. Center for Great Lakes Studies, Spec. Rept. No. 20 and Sea Grant
Advisory Rept. No. 10.
Conover, W.J. 1980. Practical Nonparametric Statistics. Wiley, New York.
Fox, D.J., and K.E. Guire. 1976. Documentation for MIDAS, University of
Michigan Statistical Research Laboratory.
Kubus, J.J., and D.A. Egloff. 1982. Monitoring water quality parameters from
municipal water intakes. J. Water Poll. Con. Fed. 541: 1592-1598.
Ladewski, B.G., R.G. Kreis, and E.F. Stoermer. 1982. A comparative analysis of
Lake Huron phytoplankton assemblages after entrainment at selected water
intake facilities. Spec. Rept. No. 92, Great Lakes Res. Div., Univ. of
Michigan.
Lindgren, B. 1976. Statistical Theory. Macmillan Press, New York.
Moll, R.A., R. Rossmann, D.C. Rockwell, and W.Y.B. Chang. 1985. Lake Huron
intensive survey, 1980. Spec. Report No. 110, Great Lakes Res. Div,,
Univ. of Michigan.
Mortimer, C.H. 1971. Large-scale oscillatory motions and seasonal temperature
changes in Lake Michigan and Lake Ontario. Parts I & II. Spec. Rept.
No. 12, Univ. Wisconsin-Milwaukee.
Netter, J., and W. Wasserman. 1974. Applied Linear Statistical Models.
Richard D. Irwin, Inc., Homewood, 111.
Shapiro, J., and E.B. Swain. 1983. Lessons from the silica "decline" in Lake
Michigan. Science 221: 457-459.
Sokal, R.R., and F.J. Rohlf. 1981. Biometry. W.H. Freeman and Co.,
San Francisco.
Strand, J.W. 1973. Inshore-offshore chemical gradients and reliability of
chemical analyses of municipal intakes in southern Lake Michigan.
Unpublished Report. Center for Great Lakes Studies, University of
Wisconsin, Milwaukee.
Vorce, C.M. 1881. Forms observed in water of Lake Erie. Proc. Am. Soc.
Microscopists 4:51-90.
70
-------
------- |