LAKE ERIE DISSOLVED OXYGEN
MONITORING PROGRAM
TECHNICAL REPORT
Dissolved Oxygen and Temperature Profiles for the Open Waters of the
Central Basin of Lake Erie during Summer/Fall of 2016
SEPA
Urilsd Slates
Environmental Protection
Agency
March 2020
EPA 950-R-20-003
Great Lakes ,
RESTORATION {
Prepared By:
United States Environmental protection Agency
Great Lakes National Program Office
-------
LAKE ERIE DISSOLVED OXYGEN MONITORING PROGRAM TECHNICAL REPORT
ACKNOWLEDGMENTS
This work was supported with Great Lakes Restoration Initiative (GLRI) funding to the U.S.
Environmental Protection Agency (EPA) Great Lakes National Program Office (GLNPO) as part of EPA
Contract No. EP-C-15-012, Technical, Analytical, and Regulatory Mission Support for the Water Security
Division, with CSRA LLC, a General Dynamics Information Technology company (hereinafter referred
to as "CSRA"), under the direction of Louis Blume, Project Manager. The United States Geological
Survey (USGS) supported this work via the USGS Great Lakes Deepwater Program supplemented with
GLRI funding through Interagency Agreement DW 014 92453901. Special thanks to the captains and
crews of the EPA R/V Lake Guardian and the USGS R/V Muskie for field support. We gratefully
acknowledge the support of the following team members in the preparation of this Technical Report:
Affiliation
Team Members
EPA GLNPO
Jeffery May, Louis Blume
CSRA
Kathleen Marshall, Kenneth Miller
Cover photo: Wendy Luby.
Citation: U.S. EPA 2020. Lake Erie Dissolved Oxygen Monitoring Program Technical Report: Dissolved
Oxygen and Temperature Profiles for the Open Waters of the Central Basin of Lake Erie during
Summer/Fall of 2016. (EPA 950-R-20-003)
MARCH 2020
PAGE 11
-------
LAKE ERIE DISSOLVED OXYGEN MONITORING PROGRAM TECHNICAL REPORT
TABLE OF CONTENTS
Acknowledgments I
List of Tables II
List of Figures II
1 Executive Summary 1
2 Introduction 1
3 Methods 2
4 Quality Assurance and Quality Control 4
5 Results and Discussion 5
6 Conclusions 11
References 12
Appendix A - Quality Control Results A-l
Table 1. Acceptance criteria for DO and temperature data 4
Table 2. Mean water temperature (± SD) and DO for each survey in 2016 6
Table 3. Generalized linear model (GLM) results for the relationships between SurveyDay and
hypolimnion temperature, thickness and DO concentration 9
Table 3a. Overall GLM results 9
Table 3b. GLM fit statistics 9
Table 3c. GLM estimates of deviations in model intercept and slope used to calculate rate of
change in water temperature, thickness and DO concentration of the hypolimnion for years
2007-2015 compared to 2016 reference year 10
Table A-l. Quality control (QC) scorecard of 2016 CTD-collected temperature and dissolved
oxygen (DO) data not meeting acceptance criteria A-l
Figure 1. Map of GLNPO dissolved oxygen (DO) monitoring stations in the central basin of
Lake Erie 3
Figure 2. Example of a temperature and DO depth profile from Lake Erie central basin in
late summer 3
Figure 3. 2016 station means of hypolimnion DO concentrations in the central basin of Lake Erie 5
Figure 4. Hypolimnion temperatures in the central basin of Lake Erie from 2007-2016 7
Figure 5. Hypolimnion thicknesses in the central basin of Lake Erie from 2007-2016 8
Figure 6. Hypolimnion DO concentrations in the central basin of Lake Erie from 2007-2016 8
Figure 7. Dissolved oxygen depletion rate in the central basin of Lake Erie from 1970-2016 11
MARCH 2020
PAGE | II
-------
LAKE ERIE DISSOLVED OXYGEN MONITORING PROGRAM TECHNICAL REPORT
1 EXECUTIVE SUMMARY
The United States Environmental Protection
Agency (EPA) Great Lakes National Program
Office (GLNPO) Lake Erie Dissolved Oxygen
Monitoring Program monitors the oxygen and
temperature profiles at 10 stations in the central
basin of Lake Erie during the stratified season to
assess water quality trends and measure progress
made in achieving water quality improvements.
During the course of the 2016 sampling season
(June 6 - September 21):
• Six surveys were conducted during the 2016
sampling season using the EPA R/V Lake
Guardian and USGS R/V Muskie.
• Surface water temperatures increased from
15.1°C to 24.3°C, while hypolimnion
temperatures increased from 8.6°C to 16.2°C.
• Hypolimnion dissolved oxygen (DO)
concentrations during the sampling season
decreased from approximately 10.5 mg O2/L
to 0.2 mg O2/L.
• Low DO conditions (< 6 mg O2/L) were first
recorded at two stations on July 21-22, 2016.
• Hypoxic conditions (< 2 mg O2/L) were first
recorded at one station on August 11-12, 2016.
Low DO (< 6 mg O2/L) was observed at all
other stations during this sampling event.
• Nine of the 10 stations were anoxic (< 1 mg
O2/L) on September 20-21, 2016.
• The annual corrected DO depletion rate was
3.38 mg 02/L/month.
When compared to the previous 10-year record,
the hypolimnion at the beginning of the 2016
sampling season was similar in temperature, but
thicker and contained more DO. While surface
water values are centrally located within the
observed range for the last 10 years, the
hypolimnion temperature became one of the
warmest in recent years near the end of the season.
The corrected annual oxygen depletion rate
referenced above was slightly above the median
for the 2007-2016 time period and was similar to
the 46-year long term average from 1970-2016.
2 INTRODUCTION
Lake Erie has been severely impacted by
excessive anthropogenic loadings of
phosphorous resulting in abundant algal growth
and is a factor that contributes to dissolved
oxygen (DO) depletion in the bottom waters of
the central basin. Total phosphorus loads to
Lake Erie reached their peak in the late 1960s
and early 1970s with annual loads in excess of
20,000 metric tonnes per annum (MTA)
(Maccoux. et al.. 2016). In 1978, Canada and the
U.S. signed the Great Lakes Water Quality
Agreement (GLWQA) which sought to reduce
future phosphorus loadings to 11,000 MTA. In
order to determine if the areal extent or duration
of the oxygen-depleted area was improving or
further deteriorating, annual monitoring of the
water column for thermal structure and DO
concentration was needed throughout the
stratified season. The U.S. Environmental
Protection Agency (EPA) Great Lakes National
Program Office (GLNPO) established the Lake
Erie Dissolved Oxygen Monitoring Program in
1983. This program was designed to collect
necessary DO concentration data to calculate an
annual normalized rate of DO depletion in the
central basin of Lake Erie. Additionally, these
data could be used by federal and state water
quality agencies to assess the effectiveness of
phosphorus load reduction programs.
Numerous phosphorus reduction programs were
implemented in support of the GLWQA, and by
the early 1980s, the annual phosphorus load to
Lake Erie had been reduced to near targeted
amounts (Dolan. 1993). Correspondingly, the
load reduction resulted in the decrease of the
total area affected by low oxygenated waters
(Makarewicz and Bertram. 1991). By the mid-
1990s, the total extent of the hypoxic area (DO
levels below 2 mg/L) had decreased such that
the total impacted area was smaller (in km2)
than had been observed in previous decades.
However, by the 2000s the annual area affected
by hypoxia had increased, returning to the larger
areal extent seen in the late 1980s (Zhou, et al..
2013). The average hypoxic area in the central
basin since the early 2000s is approximately
4,500 km2 (1,737 mi2) (U.S.EPA. 2018). while
the largest hypoxic extent recorded in the past
MARCH 2020
PAGE | 1
-------
LAKE ERIE DISSOLVED OXYGEN MONITORING PROGRAM TECHNICAL REPORT
decade - 8,800 km2 (3,398 mi2) - occurred in
2012, following the record-setting algal bloom
in 2011 (U.S. EPA. 2018).
In 2012, the GLWQA was updated to enhance
water quality programs to ensure the "chemical,
physical and biological integrity" of the Great
Lakes (Canada and United States. 2012). As part of
Annex 4 (Nutrients Annex) of this agreement, the
governments of the U.S. and Canada are required to
adapt the following Lake Ecosystem Objectives:
• minimize the extent of hypoxic zones in the
waters of the Great Lakes associated with
excessive phosphorus loading, with particular
emphasis on Lake Erie;
• maintain the levels of algal biomass below the
level constituting a nuisance condition;
• maintain algal species consistent with healthy
aquatic ecosystems in the nearshore waters of
the Great Lakes;
• maintain cyanobacteria biomass at levels that
do not produce concentrations of toxins that
pose a threat to human or ecosystem health in
the waters of the Great Lakes;
• maintain an oligotrophic state, relative algal
biomass, and algal species consistent with
healthy aquatic ecosystems, in the open waters
of Lakes Superior, Michigan, Huron and
Ontario; and
• maintain mesotrophic conditions in the open
waters of the western and central basins of Lake
Erie, and oligotrophic conditions in the eastern
basin of Lake Erie.
GLNPO continues to monitor the thermal structure
and DO concentrations in the central basin of Lake
Erie throughout the stratified season each year.
This ensures that data are available to assess the
objectives put forth in the GLWQA, but also allow
for the evaluation of status and trends over time.
This report summarizes the results of the 2016
Lake Erie Dissolved Oxygen Monitoring Program
surveys and places those results within the context
of historical data.
3 METHODS
Annually, 10 sites (Figure 1) in the relatively
homogenous area of the central basin offshore
waters (Lesht. et al.. 2018) are sampled at
approximately 3-week intervals, during the
stratified season (June-October). Sampling usually
begins in early June, when the water column
begins to stratify, or separate, into a warmer upper
layer (epilimnion) and a cooler bottom layer
(hypolimnion) and typically concludes in late
September to mid-October just before the water
column seasonally destratifies, or "turns over,"
and assumes a uniform temperature profile. The
EPA R/V Lake Guardian is used as the sampling
platform whenever scheduling and other operating
constraints permit. In the event that the R/V Lake
Guardian is not available for one or more
scheduled sampling times, alternate vessel support
is used to conduct the sampling. For 2016, the
USGS R/V Muskie was used to conduct two
surveys. At each station visit, the thermal structure
of the water column is recorded by an electronic
profiling CTD (Conductivity, Temperature, Depth
(pressure) sensor) while DO concentrations are
measured and recorded by an additional oxygen
sensor integrated into the CTD instrument
package. For 2016, a SeaBird Scientific SBE
91 lplus CTD, SBE 19plus V2 SeaCAT Profiler
CTD and SBE 25plus Sealogger CTD were used
for collecting water temperature data, while a
SBE43 Dissolved Oxygen Sensor, which was
integrated into each of the SBE CTDs, was used
for collecting DO data. Comparison analyses using
the standard QC criteria for the DO program are
conducted to ensure comparable data are being
collected between different instrumentation
whenever more than one SBE CTD is used during
a given season. Samples from each instrument are
assessed. The resulting temperature and DO depth
profiles, which provide a visual display of the
thermal structure and DO content of the water
(Figure 2). are used for calculating the annual DO
depletion rate (U.S. EPA. 2016).
MARCH 2020
PAGE | 2
-------
LAKE ERIE DISSOLVED OXYGEN MONITORING PROGRAM TECHNICAL REPORT
ER30
ER43
ER36
Depth in Meters
Figure 1. Map of GLNPO dissolved oxygen (DO) monitoring stations in the central basin of Lake Erie.
Quality Assurance samples are collected at two of
the 10 stations during each survey and used to
confirm the accuracy of the sensor measurements.
Dissolved oxygen measurements from the sensor
are compared to those determined by the Winkler
micro-titration method (U.S. EPA. 2016) for water
samples collected at 2 meters below the surface
and at 1 meter above the lake bottom. Temperature
measurements from the sensor are compared to
surface water thermometer readings obtained from
the hull mounted transducer on the research vessel.
Temperature (deg C)
12 15 18
21
24
27
30
\ —
-
: r
— Temperature
— Dissolved Oxygen
10
Q.
Oi
Q
15
20
25
6 9
Dissolved Oxygen (mg/L)
12
15
Figure 2. Example of a temperature and DO depth profile from Lake Erie central basin in late summer.
MARCH 2020
PAGE j 3
-------
LAKE ERIE DISSOLVED OXYGEN MONITORING PROGRAM TECHNICAL REPORT
4 QUALITY ASSURANCE AND
QUALITY CONTROL
After each survey, water temperature and DO
concentration data from the CTDs are averaged for
the epilimnion and hypolimnion. A grand mean of
hypolimnion DO concentration is calculated for
each station to generate a map of bottom DO
concentrations for the central basin of Lake Erie at
the time of sampling.
To reduce the amount of inter-annual variability in
DO data from Lake Erie, an annual corrected
oxygen depletion rate is calculated using a
Microsoft Access program
(LakeErieDOv05.mdb). This software statistically
adjusts the data for vertical mixing and seasonable
variability and normalizes it to a constant
temperature and hypolimnion thickness according
to the procedures used by Rosa and Burns (1987).
The resultant or "corrected" annual rate of DO
depletion (mg 02/L/month) is artificial for any
given year, but permits the identification of time
trends with more precision.
For comparisons between years, results over a 10-
year period (2007-2016) were compared
statistically using a general linear model (GLM)
approach to test whether there is a significant
difference in the relationship between time
(expressed as Julian day minus 150 to place the y-
intercept near the beginning of the sampling
period; referred to as SurveyDay in Table 3) and
either hypolimnion temperature, thickness or DO
concentration (Tables 3a. 3b and 3c). This
approach assumes a constant rate of change per
day in the unadjusted measurements (i.e.,
hypolimnion temperature, thickness and DO) over
the full June to October sampling period within
each year, which differs slightly from the Rosa
and Burns (1987) method that only assumes a
constant rate of change between sampling events,
but not across the entire sampling period. The
GLM model includes a separate factor for the
sampling year, and a Julian day x year interaction
term, which is used to test whether the rate of
change in the hypolimnion temperature, thickness
or DO varies significantly between years (i.e.,
whether the estimated slope varies between years).
Statistical significance of the GLM model tests
was set at alpha=0.05. Statistical analysis was
performed using the GLM procedure in SAS
Version 9.4 (SAS Institute, Cary, NC).
GLNPO's DO monitoring surveys operate under
an approved Quality Management Plan (QMP), a
Quality Assurance Project Plan (QAPP), and
standard operating procedures (SOP) (U.S. EPA.
2014). In 2016, QAPP Revision 09, dated March
2016, was used. The overall quality objective for
this project is to acquire measurements of DO and
temperature at the central basin stations in Lake
Erie that are representative of the actual conditions
present at the time of sampling.
Acceptance criteria for DO and temperature
(Table 1) are based on the Relative Percent
Difference (RPD) between two independently
derived measurements. By definition, RPD is
the difference between two measurements
divided by the average of both and expressed as
a percent value.
The accuracy criteria for acceptable DO
measurements is an RPD of 10% between sensor
and averaged Winkler values or an absolute
difference between measurement methods of 0.5
mg/L when DO concentrations are less than 5
mg/L. A maximum RPD of 2% is the acceptable
accuracy for water temperature. Acceptable levels
of precision are defined as a maximum difference
of 0.2 mg/L between Winkler replicates and
agreement within 5% between sensor
measurements for DO. Acceptable precision for
water temperature was defined as agreement
within 2% between sensor measurements.
Table 1. Acceptance criteria for DO and
temperature data.
Parameter
Accuracy
criteria
Precision
criteria
Temperature
2% RPD
• 2% between sensor
measurements
Dissolved
oxygen
(> 5 mg/L)
10% RPD
• 0.2 mg/L between
Winkler replicates
• 5% between sensor
measurements
Dissolved
oxygen
(< 5 mg/L)
0.5 mg/L
absolute
difference
MARCH 2020
PAGE | 4
-------
LAKE ERIE DISSOLVED OXYGEN MONITORING PROGRAM TECHNICAL REPORT
For this project, completeness is the measure of
the number of samples obtained compared to the
number that was expected to be obtained under
normal conditions. The completeness goal is to
obtain DO and water temperature profiles within
accuracy and precision limits at 90% of all
designated stations during each survey.
5 RESULTS AND DISCUSSION
During the first survey (June 6-7, 2016), all
stations were stratified with an average
temperature difference of nearly 6.5°C between
the epilimnion and hypolinmion layers, and most
stations remained stratified throughout the
sampling period (Table 2). Over the sampling
season, average temperatures increased in the
epilimnion from 15.1°C to 24.3°C and in the
In po limn ion from 8.6°C to 16.2°C. Average
dissolved oxygen (DO) concentrations during the
sampling season decreased from approximately
10.5 mg O2/L during the first survey to 8.1 mg
O2/L in the epilimnion and to 0.2 mg O2/L in the
hypolimnion at the end of the sampling season.
Low DO concentrations (< 6 mg O2/L) in the
hypolimnion were first detected at the two western-
most sampling stations (ER42 and ER43) during
the July 21-22 cruise (Figure 3). By mid-August, all
stations had DO concentrations below 6 mg O2/L,
and one station had become anoxic (ER43, < 1 mg
O2/L). By the September 20-21 survey, all stations,
except one (ER30), were experiencing anoxic
conditions (Figure 3). However, during this survey,
the hypolimnion at four stations (ER30, ER32,
ER36 and ER43) was very thin (< 1.0 m), making it
difficult to position the CTD (Conductivity,
Temperature, Depth (pressure) sensor) within this
water layer. As such, ER30 may also have been
anoxic, but due to a very deep thermocline present
at this site, the hypolimnion (if one was present)
was not able to be sampled, resulting 111 no data.
August 11-12, 2016
September 20-21,2016
Dissolved Oxygen Concentrations
^ >6 mg/L ^ 4-6 mg/L Q 2-4 mg/L ^1-2 mg/L £ 0-1 mg/L ¦ ) No Data
Figure 3. 2016 station means of hypolimnion DO concentrations in the central basin of Lake Erie.
MARCH 2020
PAGE | 5
-------
LAKE ERIE DISSOLVED OXYGEN MONITORING PROGRAM TECHNICAL REPORT
Table 2. Mean water temperature (± SD) and DO for each survey in 2016.*
N
(#)
Epilimnion
Hypolimnion
2016 Survey Dates
CTD Used
Temperature
(°C)
DO
(mg/L)
Temperature
(°C)
DO
(mg/L)
Thickness
(m)
.Time 6-7
SBE911+
10
15.05 ±0.77
10.47 ±0.25
8.63 ±0.18
10.55 ±0.63
8.71 ± 1.54
.Time 27-28
SBE25
6
19.19 ±0.43
8.02 ± 0.43
9.82 ±0.37
8.52 ±0.65
5.51 ± 1.20
My 21-22
SBE911+
10
22.55 ±0.91
8.61 ±0.28
11.22 ±0.74
7.55 ±2.11
4.92 ±2.82
August 11-12
SBE911+
10
24.32 ± 1.18
8.36 ±0.28
11.53 ±0.82
3.94 ± 1.53
4.83 ± 1.18
September 6-7
SBE 19
0
ND
ND
ND
ND
ND
September 20-21
SBE911+
9
23.02 ±0.49
8.08 ±0.25
14.47 ±0.69
0.21 ±0.23
2.49 1.95
* N indicates the number of stations used to calculate survey averages;
COMPARISON TO HISTORICAL RESUL TS
Over the course of the summer, DO levels in the
bottom waters of Lake Erie's central basin steadily
decline (Burns, et al.. 2005). Variability in the rate
of DO depletion, its severity, and its duration are
related to year-to-year differences in the thickness
and temperature of the bottom water layer, as well
as winter ice coverage. Year-to-year differences in
the hypolimnion characteristics are determined by
the weather over Lake Erie in the spring (i.e.,
average air temperature and wind velocity).
Rapidly climbing air temperature with calm winds
will result in a thinner, warmer epilimnion and a
thicker, cooler hypolimnion that retains more DO
longer into the season. A cooler, windy spring will
permit the entire water column to warm before the
lake stratifies, resulting in a deeper thermocline
depth and a warm, thin hypolimnion that is more
prone to oxygen depletion earlier in the season
(Conrov. et al.. 2011). Furthermore, reduced ice
coverage over the winter will result in earlier
springtime mixing and a longer stratification
period, thus increasing the risk of oxygen
depletion in the hypolimnion (Perello. 2017).
In 2016, the hypolimnetic temperature remained
near the previous 10-year average through most of
the season; however, by the end of the season
temperature had increased to one of the highest
levels observed over this time period (Figure 4).
The rate of change in hypolimnion temperature
varied significantly between years (Table 3b). with
the hypolimnion temperature increasing
significantly faster in 2016 than in 2008, 2009,
2011, 2012 and 2014 (Table 3b. 3c).
At the start of the 2016 season, the hypolimnion
was one of the thickest observed over the 10-year
ND indicates no data available.
period. Not only does a thicker hypolimnion
contain a greater quantity of DO, but it has also
been shown to be associated with an overall
slower depletion rate (Charlton. 1980; Bouffard.
2013). However, by the end of the season, the
hypolimnion was one of the thinnest observed
(Figure 5). The rate of change in hypolimnion
thickness varied significantly between years
(Table 3b). with the hypolimnion thickness
decreasing faster in 2016 than for all years prior to
2013 other than 2010 (Table 3c).
The average hypolimnion oxygen concentration at
the start of the 2016 season was one of the highest
concentrations observed over the 10-year period
(Figure 6). It remained relatively high throughout
most of the season, dropping to 2 mg O2/L at the
third latest date over the 10-year period
(approximately September 1st). The rate of change
for unadjusted DO did not vary significantly
between years (Table 3b).
The corrected annual oxygen depletion rate for
2016 was 3.38 mg 02/L/month (Figure 7). This is
fairly typical, approximately 0.08 mg 02/L/month
above the median for the 2007-2016 time period.
The June 27-28 and September 6-7 survey data
were not included in the 2016 oxygen depletion
rate analysis.
In 2016, there were three surveys for which data
were either not collected or failed QA checks (as
described in Table A-l of Appendix A), which
limited their use in trend analysis. During the June
27-28 survey, no QC samples were collected.
Therefore, the SeaBird data values could not be
evaluated to ensure the instrumentation was
functioning properly and within acceptance limits.
No data were available for the September 6-7
MARCH 2020
PAGE | 6
-------
LAKE ERIE DISSOLVED OXYGEN MONITORING PROGRAM TECHNICAL REPORT
survey due to the instrumentation not functioning
properly. The new SeaBird used during this
survey was not set up correctly; the minimum
conductivity required to turn on the water pump
was set to a value appropriate for seawater. As
such, the water pump never turned on and no valid
data were collected. Data from the ER30 (Figure
I), during the September 20-21 survey could not
be included, as hypolimnion values were not
sampled due to the absence of the hypolimnion at
that station.
c
Julian Day
Figure 4. Hypolimnion temperatures in the central basin of Lake Erie from 2007-2016.
MARCH 2020
PAGE | 7
-------
LAKE ERIE DISSOLVED OXYGEN MONITORING PROGRAM TECHNICAL REPORT
14
12
10
"c/T
q3
O
•
• 2007
• •
• 2008
• 2009
Thickness (me
o kj CT> c
•
••
*
••
•
•
• •
• .
• 2010
• 2011
•
•
•
•
•
•
• •
• •
•• •.
••
• 2012
• 2013
• 2014
• 2015
• 2016
•
•
•
•
•
35
15 165 185 205 225 245 2i
Julian Day
Figure 5. Hypolimnion thicknesses in the central basin of Lake Erie from 2007-2016.
12
10
O)
E
(D
O)
>,
X
O
"O
(D
>
O
2007
>2008
>2009
>2010
>2011
>2012
>2013
>2014
>2015
>2016
145
165
185
205
Julian Day
225
245
265
Figure 6. Hypolimnion DO concentrations in the central basin of Lake Erie from 2007-2016.
MARCH 2020
PAGE | 8
-------
LAKE ERIE DISSOLVED OXYGEN MONITORING PROGRAM TECHNICAL REPORT
Table 3. Generalized linear model (GLM) results for the relationships between SurveyDay and hypolimnion temperature, thickness and DO concentration.
In the model, the SurveyDay term is defined as Julian day minus 150 to place the y-intercept near the beginning of the sampling period. The GLM model includes
a separate factor for the sampling year, and a Julian day x year interaction term, which is used to test whether the rate of change in the hypolimnion temperature,
thickness or DO varies significantly between years (i.e., whether the estimated slope varies between years). Statistical significance of the GLM model tests was set
at alpha=0.05.
Table 3a. Overall GLM results.
T emperature
Thickness
DO concentration
Source
DF
Sum of
Squares
Mean
Square
F
statistic*
p-value
R2t
Sum of
Squares
Mean
Square
F statistic
p-value
R2
Sum of
Squares
Mean
Square
F statistic
p-value
R2
Model
19
232.64
12.24
64
<.0001
0.9767
141.75
7.46
7.28
<.0001
0.8267
667.61
35.14
67.77
<.0001
0.9780
Error
29
5.55
0.19
29.72
1.02
15.04
0.52
Table 3b. GLM fit statistics.
T emperature
Thickness
DO concentration
Source
DF
Type III SS*
Mean Square
F statistic
p-value
Type III SS
Mean Square
F statistic
p-value
Type III SS
Mean Square
F statistic
p-value
SurveyDay§
1
107.93
107.93
564.15
<.0001
37.37
37.37
36.46
<.0001
469.39
469.39
905.27
<.0001
Year
9
26.26
2.92
15.25
<.0001
57.98
6.44
6.29
<.0001
32.02
3.56
6.86
<.0001
Interaction
(i.e., SurveyDay x year)
9
4.92
0.55
2.86
0.0153
44.17
4.91
4.79
0.0006
9.35
1.04
2
0.0757
Ratio of the Mean Squares to its Error (i.e., overall model significance)
t Estimate of the overall variability explained by the model
J Sum of Squares that includes the variation that is unique to the effect listed in that row (e.g., Temperature and SurveyDay) after adjusting for all other effects that are included in the model
§ Julian day minus 150
MARCH 2020
PAGE | 9
-------
LAKE ERIE DISSOLVED OXYGEN MONITORING PROGRAM TECHNICAL REPORT
Table 3c. GLM estimates of deviations in model intercept and slope used to calculate rate of change in water temperature, thickness and DO concentration of the
hypolimnion for years 2007-2015 compared to 2016 reference year.
Temperature ( C)
Thickness (m)
DO concentration (mg/L)
Parameter
Estimate
Standard Error
T statistic*
p-value
Estimate
Standard Error
T statistic
p-value
Estimate
Standard Error
T statistic
p-value
Intercept in 2016
7.7807
0.4243
18.34
<.0001
8.9320
0.9822
9.09
<.0001
11.9180
0.6986
17.06
<.0001
Slope in 2016
0.0621
0.0058
10.76
<.0001
-0.0596
0.0134
-4.46
0.0001
-0.1009
0.0095
-10.62
<.0001
Difference in intercept in 2007^
0.0063
0.5407
0.01
0.9908
-2.5838
1.2515
-2.06
0.048
-1.2254
0.8902
-1.38
0.1792
Difference in intercept in 2008
1.3196
0.5410
2.44
0.0211
-1.7398
1.2522
-1.39
0.1753
-0.4476
0.8907
-0.5
0.6191
Difference in intercept in 2009
1.7154
0.5425
3.16
0.0037
-4.2310
1.2557
-3.37
0.0021
-3.3592
0.8931
-3.76
0.0008
Difference in intercept in 2010
1.1866
0.5576
2.13
0.0419
1.5582
1.2905
1.21
0.237
-1.9196
0.9179
-2.09
0.0454
Difference in intercept in 2011
-0.3751
0.5707
-0.66
0.5161
-2.0792
1.3209
-1.57
0.1263
-1.2484
0.9395
-1.33
0.1943
Difference in intercept in 2012
3.3044
0.5609
5.89
<.0001
-5.6415
1.2983
-4.35
0.0002
-4.6477
0.9234
-5.03
<.0001
Difference in intercept in 2013
1.5574
0.5569
2.8
0.0091
-3.4113
1.2891
-2.65
0.013
-1.5496
0.9169
-1.69
0.1017
Difference in intercept in 2014
-1.6828
0.5373
-3.13
0.0039
-0.5629
1.2437
-0.45
0.6542
0.4576
0.8846
0.52
0.6089
Difference in intercept in 2015
0.6484
0.6544
0.99
0.3299
-0.9656
1.5145
-0.64
0.5288
-0.5763
1.0772
-0.53
0.5967
Difference in slope in 2007tt
-0.0114
0.0076
-1.5
0.1436
0.0407
0.0176
2.31
0.0279
0.0020
0.0125
0.16
0.8769
Difference in slope in 2008
-0.0319
0.0076
-4.19
0.0002
0.0373
0.0176
2.12
0.0428
-0.0031
0.0125
-0.24
0.8088
Difference in slope in 2009
-0.0227
0.0078
-2.91
0.0069
0.0608
0.0181
3.36
0.0022
0.0209
0.0129
1.62
0.1162
Difference in slope in 2010
-0.0162
0.0086
-1.9
0.0678
-0.0242
0.0198
-1.22
0.2308
-0.0037
0.0141
-0.26
0.7965
Difference in slope in 2011
-0.0251
0.0104
-2.41
0.0224
0.0755
0.0241
3.13
0.0039
0.0120
0.0171
0.7
0.4885
Difference in slope in 2012
-0.0252
0.0077
-3.25
0.0029
0.0708
0.0179
3.96
0.0005
0.0266
0.0127
2.09
0.0457
Difference in slope in 2013
-0.0105
0.0076
-1.37
0.1809
0.0298
0.0177
1.69
0.1026
-0.0034
0.0126
-0.27
0.7883
Difference in slope in 2014
-0.0203
0.0078
-2.59
0.0149
0.0283
0.0181
1.56
0.1293
0.0267
0.0129
2.07
0.0477
Difference in slope in 2015
-0.0136
0.0089
-1.53
0.137
0.0160
0.0206
0.78
0.4442
0.0044
0.0146
0.3
0.764
# Ratio of the Estimate to its Standard Error
TT Factors are for the difference in the intercept from the reference (i.e., 2016) and the specific year. The tests (i.e., T statistic and p-value) determine if there is a significant difference between the intercept in the reference
year (i.e., 2016) and the specific year. For example, in 2007, the estimated temperature intercept (i.e., estimated value on the 150th Julian day) is 7.7870 °C (7.7807 + 0.0063), and it is not significantly different from the
estimated temperature intercept in 2016 (i.e., 7.7807 °C) because the p-value is greater than alpha = 0.05.
Factors are for the difference in the slope from the reference (i.e., 2016) and the specific year. The tests (i.e., T statistic and p-value) determine if there is a significant difference between the slope in the reference year (i.e.,
2016) and the specific year. For example, in 2007, the estimated thickness slope is -0.0189 m/day (-0.0596 + 0.0407), and it is significantly different from the thickness slope in 2016 (i.e., -0.0596 m/day) because the p-
value is less than alpha = 0.05.
MARCH 2020
PAGE | 10
-------
LAKE ERIE DISSOLVED OXYGEN MONITORING PROGRAM TECHNICAL REPORT
Figure 7. Dissolved oxygen depletion rate in the central basin of Lake Erie from 1970-2016.
Data sources: Burns et al. (2005) and EPA GLNPO (unpublished).
6 CONCLUSIONS
The U.S. EPA GLNPO Lake Erie Dissolved Oxygen
Monitoring Program monitors the oxygen and
temperature profiles at 10 stations in the central basin
of Lake Erie to assess water quality trends and
measure progress made in achieving water quality
improvements. Six surveys were conducted in 2016
from June 6 to September 21 (Table 2); however,
data from two of these surveys did not meet the
QAPP requirement for completeness due to
instrumentation issues during those surveys (i.e.,
June 27-28 and September 6-7 surveys).
When compared to the previous 10-year record
(2007-2015), the hypolimnion at the beginning of
the 2016 sampling season was similar in
temperature, but thicker and contained more
dissolved oxygen (DO). Water temperatures
increased from 15.1°C to 24.3°C in the epilimnion
and from 8.6°C to 16.2°C in the hypolimnion
during the sampling season. These temperature
values are centrally located within the observed
range for the last 10 years (see Figure 4); however,
near the end of the season, the hypolimnion
temperature became one of the warmest in recent
years. The 2016 hypolimnion was the second
thickest at the start of the sampling season, but by
the end of the season it was one of the thinnest
(see Figure 5). Consequently, oxygen
concentrations decreased during the season from
approximately 10.5 mg O2/L to 8.1 mg O2/L in the
epilimnion and to 0.2 mg O2/L in the hypolimnion.
Low-oxygen conditions (< 6 mg O2/L) were
recorded in the hypolimnion at two western
stations in July and at all stations in August.
Hypoxic hypolimnion conditions (< 2 mg O2/L)
were recorded at one western station in August
and at all stations meeting acceptance criteria in
late September (n=9). Compared to the previous
10-year period (2007-2015), the average
hypolimnion oxygen concentration for 2016 was
the second highest at the beginning of the
sampling season and remained at higher levels
throughout the season, not dropping to 2 mg O2/L
until the third latest date for this time period (see
Figure 6). The corrected annual oxygen depletion
rate for 2016 was 3.38 mg 02/L/month, which was
slightly above the median for the 2007-2016 time
period and was similar to the 46-year long term
average from 1970-2016 (Figure 7).
MARCH 2020
PAGE | 11
-------
LAKE ERIE DISSOLVED OXYGEN MONITORING PROGRAM TECHNICAL REPORT
REFERENCES
Bouffard, D., J.D. Ackerman, and L. Boegman. 2013. Factors Affecting the Development and Dynamics
of Hypoxia in a Large Shallow Stratified Lake: Hourly to Seasonal Patterns. Water Resources
Research. 49(5): 2380-2394.
Bums, N.M., D.C. Rockwell, P.E. Bertram, D.M. Dolan and J.J. Ciborowski. 2005. Trends in Temperature,
Secchi Depth, and Dissolved Oxygen Depletion Rate in the Central Basin of Lake Erie, 1983-2002. J.
Great Lakes Res. 31 (Supplement 2): 35-49.
Canada and United States. 2012. Protocol Amending the Agreement Between Canada and the United
States of America on Great Lakes Water Quality, 1987.
Charlton, M.N. 1980. Hypolimnion Oxygen Consumption in Lakes: Discussion of Productivity and
Morphometry Effects. Can. J. Fish. Aquat. Sci. 37(10): 1531-1539.
Conroy, J.D., L. Boegman, H. Zhang, W.J. Edwards, and D. A. Culver. 2011. "Dead Zone" Dynamics in
Lake Erie: the Importance of Weather and Sampling Intensity for Calculated Hypolimnetic
Oxygen Depletion Rates. Aquat. Sci. (2011) 73:289-304.
Dolan, D.M. 1993. Point Source Loadings of Phosphorus to Lake Erie: 1986-1990. J. Great Lakes Res.,
19(1993). Pp. 212-223.
Lesht, B.M., R.P Barbiero and G.J. Warren. 2018. Using Satellite Observations to Assess the Spatial
Representativeness of the GLNPO Water Quality Monitoring Program. J. Great Lakes Res.
44(2018): 547-562.
Maccoux, M.J., A. Dove, S.M. Backus, and D.M. Dolan. 2016. Total and Soluble Reactive Phosporus
Loadings to Lake Erie. A Detailed Accounting by Year, Basin, Country and Tributary. J. Great
LakesRes. 42(2016): 1151-1165.
Makarewicz, J.C. and P.E. Bertram. Evidence for the Restoration of the Lake Erie Ecosystem - Water Quality,
Oxygen Levels, and Pelagic Function Appear to be Improving. Bioscience. 1991. 41(4), 216-223.
Perello, M. M., D. D Kane, P. Golnick, M.C Hughes, M.A Thomas, and J.D. Conroy. 2017. Effects of
Local Weather Variation on Water-Column Stratification and Hypoxia in the Western, Sandusky,
and Central Basins of Lake Erie. Water, 9(4), 279-291. https://doi.org/10.3390/w9040279
Rosa, F. andN.M. Burns. 1987. Lake Erie Central Basin Depletion Changes from 1929-1980. J. Great
LakesRes. 13(4):684-696.
U.S. Environmental Protection Agency (U.S. EPA). 2014. United States Environmental Protection
Agency Region 5, Quality Management Plan.
U.S. Environmental Protection Agency (U.S. EPA). 2016. Dissolved Oxygen and Temperature Profiles
for the Central Basin of Lake Erie Quality Assurance Project Plan Revision 09, dated March
2016. U.S. EPA Great Lakes National Program Office.
U.S. Environmental Protection Agency (U.S. EPA). 2018. U.S. Action Plan for Lake Erie. Retrieved from
https://www.epa.gov/sites/production/files/2018-03/documents/us dap final march l.pdf.
Zhou, Y., D.R Obenour, D. Scavia, T.H Johengen, and A.M. Michalak. 2013. Spatial and Temporal
Trends in Lake Erie Hypoxia, 1987-2007. Environ. Sci. Technol. 47(2): 899-905.
MARCH 2020
PAGE | 12
-------
LAKE ERIE DISSOLVED OXYGEN MONITORING PROGRAM TECHNICAL REPORT
APPENDIX A - QUALITY CONTROL RESULTS
A summary of 2016 results not meeting acceptance criteria is provided in the table below.
Table A-l. Quality control (QC) scorecard of 2016 CTD-collected temperature and dissolved oxygen (DO) data not meeting acceptance criteria.
Survey
Issue
Cause
Decision
Corrective Actions
June 27-28
Accuracy of CTD DO data was
not assessed
No water samples were collected.
Caution should be used with this
Adhere to QAPP and SOPs.
Incomplete dataset
Data were not collected at 4 stations due
to malfunctioning CTD.
dataset.
This CTD was not used during the rest of
the season.
Aug 11-12
DO accuracy check exceeded
QC criterion
The hypolimnion was less than 2m thick
at one of the stations sampled (ER32), so
thennocline/epilimnion waters may have
been in the water sample.
The surface water sample at this station
met the QC criterion.
All data are considered acceptable
because failures were likely caused
by water sampling and analytical
error that would not have impacted
CTD DO data.
Changes to sampling methodology are
being considered to move water sampling
container to the same height as CTD.
Currently the Niskin bottle on the Rosette
Sampler may be collecting water up to 1
m away from the CTD.
Winkler precision check
exceeded the QC criterion
An inexperienced technician ran Winkler
analyses on one station which may have
resulted in the greater variability in these
two samples.
Exploring the possibility of incorporating
an automatic titration system to reduce
subjectivity differences between
technicians.
Sept 6-7
Accuracy of CTD DO data was
not assessed
No water samples were collected.
No DO data are acceptable because
instrumentation did not function
properly.
Adhere to QAPP and SOPs.
Incomplete dataset
The DO pump on the CTD did not turn on
during any of the casts.
Minimum conductivity frequency for
pump turn on was changed to appropriate
freshwater value.
Sept 20-21
DO accuracy check exceeded
QC criterion
The hypolimnion thickness was only 0.7
m and 2.1 m at two of the sampled
stations (ER32 and ER42, respectively),
so thennocline/epilimnion waters may
have been in the water sample.
All data are considered acceptable
because failures were likely caused
by water sampling error that would
not have impacted CTD DO data.
Changes to sampling methodology are
being considered to move water sampling
container to the same height as CTD.
Currently the Niskin bottle on the Rosette
Sampler may be collecting water up to 1
m away from the CTD.
MARCH 2020
PAGE | A-l
------- |