Mystic River Watershed Alternative TMDL Development for
Phosphorus Management - Final Report

January 2020

Prepared for:

U.S. Environmental Protection Agency
Region 1 - Office of Ecosystem Protection
5 Post Office Square
Boston, MA 02109

Prepared by:

Eastern Research Group, Inc.
100 Hartwell Ave.

Lexington, MA 02421

Subcontractors:

Horsley Witten Group
Paradigm Environmental
Nigel Pickering, PhD
PG Environmental

EPA Contract No. EP-C-16-003

Reviewer:

Dr. Jeff Walker

Walker Environmental Research, LLC

CoVtr Photo: The Tufts University sailing program practicing on Upper Mvstic Lake in Medford with informal swimming at the
Bacow Sailing Pavilion. Upper Mystic Lake is a popular destination with additional access at Massachusetts Department of
Conservation and Recreation's Shannon Beach, and the Medford and Winchester Boat Clubs. Photo credit unknown.


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Executive Summary

The Mystic River Watershed, located a few miles north of Boston, is a 76-square mile area that
drains into Boston Harbor. Encompassing all or portions of 22 urban and suburban communities,
the watershed is highly developed and faces multiple water quality impairments. The Massachusetts
Department of Environmental Protection's (MassDEP's) water quality assessment indicates that
nutrients and pathogens are the primary causes of "use impairment" in the freshwater portion of
Mystic River watershed, the focus of this report. Cultural eutrophication—the degradation of aquatic
environments by nutrient pollution caused by human activity and urban development—is a major
cause of impairments in the watershed as evidenced by excessive algal and macrophyte growth and
harmful cyanobacteria blooms. Regular occurrences of severe algal blooms during the summer
months reduce water clarity and contribute to anoxic bottom waters that do not support aquatic life.
Algal blooms and macrophyte growth degrade tile aesthetic quality of the river, reduce water clarity,
and impair designated uses such as fishing and boating.

Photo Above: The Mystic River Run and Paddle (May 2016). This annual race celebrates the return of the river herring
and draws the public to the Mystic River. This image looks downstream from Route 16 to the Blessing of the Bay
Boathouse in Somerville with the Boston skyline in the background. Photo credit: Ram Subramanian.

Clean Water Act Requirements

Section 303(d) of the Clean Water Act and the U.S. Environmental Protection Agency's (EPA's)
Water Quality Planning and Management Regulations (Title 40 of the Code of Federal Regulations
[CFR] Part 130) require states to develop Total Maximum Daily Loads (TMDLs) for impaired water
bodies. A TMDL establishes the amount of a pollutant that a water body can assimilate without
exceeding the applicable water quality standard. A TMDL consists of the sum of individual waste

2


-------
Mystic River Watershed TMDL Alternative Development — Final Report

load allocations for point sources and load allocations for nonpoint sources and natural background
conditions. The Massachusetts Water Quality Standards (WQS), codified at 314 CMR 4.00, identify
the Mystic River as a Category 5 water body on the Massachusetts 303(d) "List of Impaired Waters"
(2014) for phosphorus, arsenic, chlordane, chlorophyll, DDT (dichlorodiphenyltrichloroethane),
dissolved oxygen, E. coli, PCBs (Polychlorinated biphenyls) [in fish tissue], Secchi depth, and
sediment bio-chronic toxicity.

This report addresses those impairments associated with excessive nutrient loading including
phosphorus, chlorophyll, dissolved oxygen, and secchi depth (water clarity).

Alternative TMDL Process

In 2013, the EPA announced a new framework (Vision) for prioritizing and implementing TMDLs
and pollution control strategies. The guidance for this Vision (found here:

https://www.epa.gov/sites/production/files/2015-07/documents/vision 303d program dec 2013.pdf
allows states to adopt strategies tailored to their water quality program goals and priorities. The
Vision acknowledges that alternative restoration approaches may be more immediately beneficial or
practical in achieving water quality standards than a traditional TMDL. The Vision calls on states to
strategically focus efforts and demonstrate progress over time.

EPA is supporting MassDEP in piloting an alternative TMDL designed to address nonattainment of
nutrient related water quality standards over a period of time. The approach, based on rigorous data
gathering, scientific analysis, and modeling, provides guidance to communities based on a scientific
understanding of conditions. The agencies have already begun working with communities to develop
stormwater management (SWM) strategies to begin progress on implementing effective stormwater
control measures (SCMs) to restore the river and degraded lakes and ponds. This "adaptive
management" approach for the Mystic will be an iterative process of implementing control actions
over an extended period of time while progress is monitored, and new information is gathered to
further inform management needs for attaining water quality standards.

Study and Analysis

The objectives of this technical analysis, conducted between 2017 and 2019, were to: estimate annual
loadings of phosphorus; relate phosphorus loads to response variables in critical surface water
reaches of the watershed; estimate the load reductions needed to improve water quality and attain
water quality standards; and introduce a pilot Opti-Tool analysis that demonstrates cost-effective
and opportunistic stormwater load reduction strategies that communities can consider adopting.

Much of the scientific research needed to document existing conditions occurred prior to this
project, during which time EPA Region 1 collaborated with the Mystic River Watershed Association
(MyRWA), MassDEP, U.S. Geological Survey (USGS) and the Massachusetts Water Resources
Authority (MWRA). Baseline water quality monitoring included the collection of composite samples
linked to streamflow. This project builds on past analyses to develop target reductions in
phosphorus inputs in order to improve water quality in the Mystic River.

To assist with developing phosphorus budgets and recommending load reductions, EPA convened a
Technical Steering Committee (TSC) to provide data, expertise and advice. The Technical Steering
Committee met three times annually. In addition to the Technical Steering Committee, the
consultant team also benefited from expert review provided by Dr. Jeff Walker (Walker
Environmental Research LLC). Further refinements to analyses were conducted based on Dr.

3


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Walker's input following consultation with the Technical Steering Committee. This interactive
approach resulting in the project team providing the most comprehensive and complete assessment
possible at this time and one that the Technical Steering Committee determined is suitable to
support an adaptive management process for die Mystic River watershed. Technical Steering
Committee members and our expert reviewer are listed in the Acknowledgements.

The components of the analytical approach, discussed in detail in the report, are listed below:

•	Develop conceptual model of hydrology and nutrient dynamics

•	Evaluate existing water quality monitoring data

•	Review modeling endpoint approaches

•	Estimate watershed phosphorus loading

•	Evaluate combined sewer overflow (CSO) and sanitary sewer overflow (SSO) data

•	Conduct BATHTUB modeling and calibrate results

•	Determine critical period of interest for phosphorus load reduction analysis

•	Evaluate watershed phosphorus load reduction analysis

•	Develop nutrient stormwater management strategies using Opti-Tool

The pollutant of concern for this study is phosphorus because it is directly causing or contributing
to the excessive algal biomass. Since there are no numeric criteria available for phosphorus (i.e., no:
specific concentration of phosphorous that represents a violation of standards), a surrogate water
quality target was needed to calculate pollutant load reductions to the river. Chlorophyll-^ was
chosen as the surrogate water quality target. Chlorophyll-^ is the photosynthetic pigment found in
algae and is, therefore, a direct indicator of algal biomass. EPA and the Technical Steering
Committee determined that a seasonal average chlorophyll-a concentration of <10 j-ig/L would be
protective of narrative eutrophication standards in the watershed, from which associated total

Photo Left:
Cyanobacteria bloom
in the freshwater
segment of the Mystic
River between
Arlington and Medford
in June of 2017. Photo
credit: Jack Bitney,

4


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Results

Watershed analyses conducted during this study demonstrate that inadequately controlled
stormwater (SW) runoff from developed landscapes are the predominant source of nutrient loads—
specifically phosphorus loads—to the surface waters of the Mystic River watershed. Under existing
conditions, this study estimated that to meet the selected chlorophyllwater quality target for
attaining water quality standards in the most impacted segment, the lower Mystic River, will require a
67 percent reduction of stormwater phosphorus loadings from the watershed However, this
estimate assumes all reduction would be achieved through stormwater control measures.

Load reduction estimates were also modeled for future conditions to account for key variables: wet
vs. dry years; future control of combined sewer overflows and sanitary sewer overflows and
sediment load reduction. Overall, the analysis showed that elimination of combined sewer overflows
and sanitary sewer overflows had minimal impact compared to reducing stormwater loads and
internal loads released from bottom sediments of the river system. The difference between wet vs.
dry years is significant, with much greater difficulty meeting water quality targets during dry years.
The stormwater load reductions required to meet water quality targets under future conditions
(which account for baseline stormwater management, combined sewer overflows/sanitary sewer
overflows controls and an estimate of associated reductions in internal loads) were between 59 and
62 percent.

The Path Ahead

Knowing how and where to site cost-effective stormwater controls to reduce phosphorus loads
from stormwater runoff will be critical for meeting state and federal water quality regulations. The
Opti-tool analysis included in this report shows that by optimizing sizing and location of BMPs,
significant cost savings can be realized. ERG and the project team is currently working with
communities in the watershed to develop cost-effective stormwater best management practices with
a focus on green infrastructure solutions and changes to local bylaws/ordinances to streamline the
process. .EPA envisions a sustained collaborative process of working with the communities to
develop realistic and effective strategic stormwater management approaches to effectively advance
watershed restoration efforts.

Photo: Left: River
herring above the fish
ladder in Upper Mystic
Lake in 2012. This was
the first year since the
late 1800s that had
significant passage of
anadromous fish into
the Upper Mystic Lake.
Photo credit: Patrick
Herron, Mystic River
Watershed Association.

5


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Acknowledgements

The following individuals participated on the technical steering committee and EPA thanks them for
their valuable contribution to this project.

Name

Agency/Affiliation

Lise Marx

MWRA

David Wu

MWRA

Chris Goodwin

MWRA

Maret Smolow

MWRA

David Taylor

MWRA

Toby Stover

EPA

Erik Beck

EPA

Suzanne Warner

EPA

Mark Voorhees

EPA

Shutsu Wong

EPA

Newton Tedder

EPA

Barbara Kickham

MassDEP

Matt Reardon

MassDEP

Patrick Herron

MyRWA

Andy Hrycyna

MyRWA

Arleen O'Donnell

ERG

Braden Rosenberg

ERG

Matt Reusswig

PG Environmental (ERG)

Adriane Garneiter

PG Environmental (ERG)

Jennifer Relstab

Horsley Witten Group

Nigel Pickering

Nigel Pickering

Jeff Walker

Walker Environmental Research

Jamie Houle

University of New Hampshire

Khali d Alvi

Paradigm

Ryan Murphy

Paradigm

Photo Above: Rowers practicing on the Maiden River, a tributary to the Lower Mystic River in Medford. Photo credit:
Greig Cranna.

6


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table of Contents

Executive Summary	2

Clean Water Act Requirements	2

Alternative TMDL Process	3

Study and Analysis	3

Results	5

The Path Ahead	5

Acknowledgements	6

I.	Introduction	16

LA. Background	16

I.B. Purpose	16

I.C.	Organization and Overview	17

II.	Conceptual Model of the Hydrology and Nutrient Dynamics in the Mystic River Basin	18

IIA. Watershed Overview	18

II.B.	Upper Watershed	20

II.C. Central Watershed	20

II.D. Streamflow Impact of Mystic Lakes Dam	21

II.E. Water Quality Impact of Mystic Lakes Dam	21

II.F. Lower Watershed	22

II.G. Hydrologic and Hydraulic Impacts of Amelia Earhart Dam	23

II.H. Water Quality Gradient Along the Mystic River	23

ILL Long Term Changes in Water Quality	24

II.J.	Impact of Aquatic Vegetation on Water Quality	25

III.	Review of Existing Water Quality Monitoring Data	26

IIIA. Data Gaps and Recommendations for Future Sampling Efforts	26

III.A.l. Ecological/Biological Indicators of Over-Enrichment	26

III.A.2. Streamflow	26

III.A.3. Sediment	26

III.B.	MyRWA and MWRA Monitoring Data	27

III.B.l. Data Characterization	27

III.B.2. Data Vivien'	37

III.C. USGS Flow Data	38

III.D. GIS Datasets	38

IV.	Review of Modeling Endpoint Approaches	39

7


-------
Mystic River Watershed TMDL Alternative Development — Final Report

IV.A. Water Quality Standards Applicable to the Mystic River Watershed	

IV.B. Existing Regional/Local Targets	

IV.C. Reference Water Body Conditions	

IV.D. Stressor-Response Relationships	

IV.E. Mechanistic Models	

IV.F.	Recommendations	

V.	Watershed Phosphorus Loading Estimates	

VA. Stormwater Loads	

V.B.	Groundwater Loads	

V.C. Sediment Loads	

V.D.	Observed Loads	

V.D. 1. Adjustments of Streamflow	

V.D.2. Adjustments of Total Phosphorus Concentrations	

V.D.3. Observed Total Phosphorus Loads	

V.D.4. Calibrated Streamflow	

V.D.5.	Calibrated Total Phosphorus Loads	

VI.	Evaluation of Combined Sewer Overflow and Sanitary Sewer Overflow Data for the Mystic
River Watershed	

VIA. Data Types and Sources	

VI.	A. 1. Spatial Data	

VI.A.2. I Tolnmetric Data	

VI.A.3. Precipitation Data	

VI.A.4. Nutrient Concentrations	

VI.B.	CSO Data Analyses	

VI.B.l. CSO Analyses	

VI.B.2. CSO Statistical Outliers	

VI.C. SSO Data Analyses	

VI.C.l. SSO Data Rei'ieii'.	

VI.C.2. SSO Data Processing	

VI.C.3. SSO Trend Analyses	

VI.C.4. SSO Statistical Outliers	

VI.C.5. SSO Missing Datasets	

VI. C. 6. /otal Phosphorus and Total Nitrogen Loads for Model Calib ratio n	

VII.	BATHTUB Modeling Approach	

VILA. Model Selection	

,40

,40

,41

,43

,46

,47

,48

,48

,56

,57

,58

,58

,59

,61

,62

,62

,66

,66

,66

,67

,67

,67

,68

,68

,68

,71

,72

,73

,73

,74

,74

,74

,80

,80

8


-------
Mystic River Watershed TMDL Alternative Development — Final Report

VII.B. Model Setup	81

VII.B.l. Segmentation	81

VII.B.2. Model Options	82

VII.B.3. Atmospheric Fluxes	83

VII.B.4. Modeled Land Loads	84

VII.B. 5. External Loads	85

VII.B.6. Internal Loads	85

VII.C.	Model Calibration and Validation	86

VII.C. 1. Water Quality Data Availability	86

VII.C.2. Calibration and I Talidation Periods	86

VII.C.3. Receiving Water Parameters	87

VII.C.4. Dispersion	87

VII.C. 5. Nutrient Availability Factors	88

VII.C.6. Internal Loads	88

VII.C.7. Chl-a Model.	88

VII.C.8. Model Calibration	89

VII.C.9. Model I Talidation	94

VII.C.10. Model Sensitivity Analysis	98

VIII.Critical	Period of Interest for Phosphorus Load Reduction Analysis	100

VIII.A.	Water Quality Data	100

VIII.B. Combined Sewer and Sanitary Sewer Overflow Data	100

VIII.C. Rainfall Data	100

VIII.D. Critical Period Selection	102

VIII.E.	Extreme Rainfall Years	102

IX.	Evaluation of Watershed Phosphorus Load Reduction Analysis	103

IX.A.	Phosphorus Loading Estimates for Critical Period of Interest	103

IX.A. 1. Existing Conditions	103

IX.A.2. Future Conditions	103

IX.B. Scenarios for Evaluation of Phosphorous Reduction	105

IX.C. Modeling Methodology	106

IX.C.l. Model Setip	106

IX.C.2. Model Inputs	106

IX.C.3. Analysis ivith Wet and Dry Year Data	107

IX.D. Model Results with Average Annual Data	107

9


-------
Mystic River Watershed TMDL Alternative Development — Final Report

IX.E. Analysis with Wet and Dry Year Data	110

IX.F.	Discussion	113

X.	Broad-Based Nutrient Stormwater Management Strategies for the Mystic River Watershed
using Opti-Tool	116

X.A.	Study Objectives	116

X.B. Pilot Sub-Watershed Selection	116

X.C. Technical Approach	123

X.C.I. S form water Management Categories Development	123

X.C.2. Estimating BMP Footprints and Impervious Drainage Areas	127

X.C.3. Opti-Tool Setup	128

X.D. Management Scenarios	130

X.D. 1. Results: Scenario 1	131

X.D.2. Results: Scenario 2	135

X.E. Summary	138

X.F. Example Projects	140

X.F.I. Berry Brook Project, Dover, New Hampshire	140

X.F.2. The Advancing Green Infrastructure Program, New Haven, Connecticut	141

XI.	References	144

Appendix A: Water body and Monitoring Location Summary Table	148

Appendix B: Water Quality Parameters Included in Monitoring Programs	151

Appendix C: Descriptive Figures of Water Quality Data by Water body	152

Appendix D: Identified Extreme Values	159

Appendix E: Modeled Stormwater TP Load and Rainfall-Runoff Results	160

Appendix F: Baseflow Estimates for Aberjona River and Alewife Brook	168

Appendix G: Maps Depicting CSO Drainage Basins	169

Appendix H: Water Quality Data Used in Calibration and Validation of the Bathtub Model	173

Appendix I: Bathtub Model Inputs for Calibration	175

Appendix J: Bathtub Model Inputs and Outputs for Scenarios	179

Appendix K. BMP Design Parameters Used in the Pilot Watershed	187

10


-------
Mystic River Watershed TMDL Alternative Development — Final Report

List of Tables

Table III-1. Water Quality Parameters Included in Each Monitoring Program	28

Table III-2. Temporal Range and Duration of Monitoring within Water Bodies	30

Table III-3 Number of Seasons Wiere Few (N < 3) Samples were Collected for a Waterbody-

Parameter Combination	34

Table III-4. Summary of Documented Data Quality Issues	35

Table III-5 Summary of Non-Detect Data.1	35

Table III-6. Summary of USGS Monitoring Stations	38

Table III-7. GIS Data Sources Available for the Mystic River Watershed	39

Table IV-1. Existing Water Quality Targets	41

Table IV-2. Eco-regional Criteria and Observed 25th Percentile (2000 — 2016) Mystic River

Watershed Values	43

Table IV-3. Simple Linear Regression on Paired TP & Chlorophyll-a Monitoring Results Collected

April - October	44

Table IV-4. Simple Linear Regression on Average TP and Average or 90th Percentile Chlorophyll-a1

	45

Table V-l. Opti-Tool-SWMM HRU Model Climatological Input Datasets	49

Table V-2. Land-Use Categories	50

Table V-3. DCIA Adjustment Factors from MA MS4 Permit	51

Table V-4. Opti-Tool Export Rates by HRU	51

Table V-5. GIS Layers Used to Develop HRUs	52

Table V-6. Estimated Reach Detention Times and Attenuation Factors	64

Table VI-1. Volumetric Data Sources for CSOs and SSOs	67

Table VI-2. Statistical Outlier Analysis for CSO Datasets	71

Table VI-3. CSO Volumetric Datasets for Alewife and Mystic River Drainage Basins	71

Table VI-4. Review of SSO Data — Frequency, Duration and Discharge Points	72

Table VI-5. Review of SSO Data — Volumes	73

Table VI-6. Statistical Outlier Analysis for Annual SSO Discharge Volumes (Gallons/yr.)	76

Table VI-7. Annual SSO Discharge Volumes for All Sub watersheds (Gallons/yr.)	76

Table VI-8. Estimated Annual Total Phosphorus and Total Nitrogen CSO Loads (lbs./yr.) for

Alewife and Mystic Sub watersheds	77

Table VI-9. Estimated Annual Total Phosphorus SSO Loads (lbs./yr.) for all Sub watersheds	78

Table VI-10. Estimated Annual Total Nitrogen SSO Loads (lbs./yr.) for all Sub watersheds	78

Table VII-1. Model Options	83

Table VII-2. Upstream Reach and Local Sub-basin Contributions	85

Table VII-3. Average Water Quality Data Available by Segment, Site and Year	86

Table VII-4. Total Phosphorus Loads by Segment for Calibration Period	92

Table VII-5. Results of BATHTUB Sensitivity Analysis	99

Table VIII-1. Standard Precipitation Index (SPI) Reference Values	101

Table VIII-2. Summary of Rainfall Data Analyses	101

Table IX-1. Modeled Scenarios for Phosphorus Load Reduction Evaluations	106

Table IX-2. BATHTUB Model Runsawith Average Annual Data	107

Table IX-3. Scenario Results3 with Average Annual Data	109

Table IX I. Model Runsawith Wet and Dry Year Data	Ill

Table IX-5. Scenario Results3 with Wet and Dry Year Data	111

Table IX-6. Total Phosphorus Load Reductions for Scenario 2A 	114

Table X-l. Ground Slope Classification in Pilot Sub-Watershed, Mystic River	118

11


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table X-2. Hydrologic Soil Group Classification in Pilot Sub-Watershed, Mystic River	119

Table X-3. Land Use Classification in Pilot Sub-Watershed, Mystic River	123

Table X-4. Potential Stormwater Management Categories and BMP Types in Opti-Tool	125

Table X-5. Potential BMP Opportunity Areas (Maximum Footprints) in the Pilot Sub-Watershed,

Mystic River	125

Table X-6. BMP-Treated Impervious Area (Drainage Area) Distribution by Land Use Category

Group in the Pilot Sub-Watershed, Mystic River	127

Table X-7. BMP Area (Footprints) Distribution by Land Use Category Group Required to Treat 1

Inch of Runoff from the Impervious Surface in the Pilot Sub-Watershed, Mystic River	128

Table X-8. Scenario 1: BMP Types and BMP Sizes for The Selected Target Solution 1 (Existing

Condition) in Opti-Tool	133

Table X-9. Scenario 1: BMP Types and BMP Sizes for the Selected Target Solution 2 (Future

Condition 1) in Opti-Tool	133

Table X-10. Scenario 1: BMP Types and BMP Sizes for the Selected Target Solution 3 (52% Target)

in Opti-Tool	134

Table X-ll. Scenario 2: BMP Types and BMP Sizes for the Selected Target Solution 1 (Existing

Condition) in Opti-Tool	136

Table X-12. Scenario 2: BMP Types and BMP Sizes for the Selected Target Solution 2 (Future

Condition 1) in Opti-Tool	137

Table X-13. Scenario 2: BMP Types and BMP Sizes for the Selected Target Solution 3 (52% Target)

in Opti-Tool	138

Table X-14. BMP Scenarios Comparison in Opti-Tool	139

12


-------
Mystic River Watershed TMDL Alternative Development — Final Report

List of Figures

Figure II-I Schematic Diagram and Map of Mystic River Watershed	19

Figure II-II Instantaneous Streamflow Above and Below Mystic Lakes	21

Figure II-III Annual Distributions of Monthly TP Concentrations at the Aberjona River Outlet

(ABR006), Mystic Lake Dam (UPL001), and Start of the Mystic River Mainstem (MYR071)	22

Figure II-IV Relative Water Surface Elevation at USGS Gages on Alewife Brook and Mystic River

(Sept. 1 - 10, 2016)	23

Figure II-V Chlorophyll-a and TP Concentrations Along the Mystic and Maiden Rivers, 2015 — 2016

	24

Figure II-VI Annual Mean, Median, 90th Percentile of Chlorophyll-a and TP at Three Stations on

the Mystic River, 2000 — 2016 (June — Oct. only)	25

Figure III-I. Mystic River Watershed Drainage Basins and Monitoring Locations	29

Figure III-II. Number of Years in Which Observations Were Made for Eutrophication Related

Parameters in the Watershed	31

Figure III-III Number of Years in Which Observations Were Made for Eutrophication Related

Parameters in The Mystic River	32

Figure III-IV. Temporal Coverage (both number of days and Percentage of Total Available Days

from 2000 — 2016) of Water Quality Parameters by Water Body	33

Figure IV-I. Reference Condition Approach for Establishing Numeric Water Quality Targets (EPA,

1998)	42

Figure IV-II. April — October Total Phosphorus vs. Chlorophyll-a Concentrations in the Upper

Lobe of the Upper Mystic Lake (Monitoring Station UPLUPL)	44

Figure IV-III. April — October Average Total Phosphorus vs. Average Chlorophyll-a Concentrations

in the Mystic River	46

Figure IV-IV. April — October Average Total Phosphorus vs. 90th Percentile Chlorophyll-a

Concentrations in the Mystic River	46

Figure V-I. Estimated Annual Runoff Volume and TP Load for the Mystic River	54

Figure V-II. Mystic River Watershed Sub-Basin Delineation and Schematic Diagram for Final Model

	55

Figure V-III. Baseflow Estimates and Streamflow Measurements at Aberjona River for 2016 (USGS

Gage 01102500)	56

Figure V-IV. Baseflow Estimates and Streamflow Measurements at Alewife Brook for 2016 (USGS

Gage 01103025)	57

Figure V-V. Regression of Flow Measurements at Aberjona River (01102500) and Alewife Brook

(01103025) USGS Stations	58

Figure V-VI. Regression of Flow Measurements at Aberjona River (01102500) and Alewife Brook

(01103025) USGS Stations	59

Figure V-VII. Correction of TP Values from Method 365.1 to 365.4	60

Figure V-VIII. Adjusted Total Phosphorus Concentrations in the Aberjona River	60

Figure V-IX. Adjusted Total Phosphorus Concentrations in the Mystic River	61

Figure V-X. Adjusted Total Phosphorus Concentrations in Alewife Brook	61

Figure V-XI. Phosphorus Attenuation Curves	63

Figure V-XII. Modeled Land vs. Attenuated Reach Phosphorus Loads (2007-2016)	65

Figure VI-I. Annual CSO Volumes over Time	69

Figure VI-II. Annual CSO Volumes versus Annual Rainfall	69

Figure VI-III. Normalized Alewife Annual CSO Volumes versus Annual Precipitation	70

13


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Figure VI-IV. Normalized Mystic Annual CSO Volumes versus Annual Precipitation	70

Figure VI-V. Annual SSO Discharge Volumes by Sub-watershed versus Time	74

Figure VII-I. BATHTUB Model Schematic	81

Figure VII-II. BATHTUB Segmentation for the Mystic River	82

Figure VIITII. Annual Precipitation and Lake Evaporation	84

Figure VIITV. Dispersion Effect on Modeled Phosphorus Concentrations	88

Figure VII-V. Predicted vs. Observed TP by Segment for Calibration Period	90

Figure VII-VI. Predicted vs. Observed TP Relationship for Calibration Period	90

Figure VII-VII. Predicted vs. Observed Chl-a by Segment for Calibration Period	91

Figure VII-VIII. Predicted vs. Observed Chl-a Relationship for Calibration Period	91

Figure VII-IX. Calibration 2015 - Total Phosphorus Loads for Mystic River	93

Figure VII-X. Validation 2016 - Predicted vs. Observed TP by Segment	95

Figure VII-XI. Validation 2016 - Predicted versus Observed Chl-a by Segment	95

Figure VII-XII. Validation 2017 - Predicted vs. Observed TP by Segment	96

Figure VII-XIII. Validation 2017 - Predicted vs. Observed Chl-a by Segment	96

Figure VII-XIV. SSO Outlier Replacement 2010 - Predicted versus Observed TP	97

Figure VII-XV. SSO Outlier Replacement 2014 - Predicted versus Observed TP	97

Figure VIIIT. Standard Precipitation Index (SPI) between 2000-2017	102

Figure IX-I. Predicted Phosphorus Concentrations with Average Annual Data	110

Figure IX-II. Predicted Chlorophyll-a Concentrations with Average Annual Data	110

Figure IX-III. Predicted Phosphorus Concentrations with Wet and Dry Year Data	113

Figure IX-IV. Predicted Chlorophyll-a Concentrations with Wet and Dry Year Data	113

Figure X-I. Location Map of Pilot Sub-Watershed, Mystic River (Highlighted in Yellow)	118

Figure X-II. Ground Slope Map of Pilot Sub-Watershed, Mystic River	119

Figure X-III. Soil Map (Hydrologic Soil Group) of Pilot Sub-Watershed, Mystic River	120

Figure X-IV. Land Use Map of Pilot Sub-Watershed, Mystic River	121

Figure X-V. Land Cover Map of Pilot Sub-Watershed, Mystic River	122

Figure X-VI. Stormwater Management Categories Map of Pilot Sub-Watershed, Mystic River	126

Figure X-VII. Treated Impervious Areas by BMP Type and Land Use Type in the Pilot Sub-

Watershed, Mystic River	128

Figure X-VIII. Opti-Tool Model of Pilot Sub-Watershed, Mystic River	130

Figure X-IX. Scenario 1: Opti-Tool Cost-Effectiveness Curve (Optimize Locations Only) of TP

Annual Average Load Reduction for the Pilot Watershed	132

Figure X-X. Scenario 2: Opti-Tool Cost-Effectiveness Curve (Optimize Locations and BMP Sizes)

of TP Annual Average Load Reduction for the Pilot Watershed	136

Figure X-XI. Green Infrastructure Retrofits for Berry Brook Project in Dover, New Hampshire ..141
Figure X-XII. Homeowners and Common Ground High School students plant perennials in a New
Haven Bioswale	143

14


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Abbreviations

cso

Combined sewer overflow

Chl-a

Chlorophyll-a

CSS

Combined sewer separation

DCIA

Directly connected impervious area

DCR

Massachusetts Department of Conservation and Recreation

EPA

U.S. Environmental Protection Agency

ERG

Eastern Research Group

HRU

Hydrologic Response Units

HUC

Hydrologic Unit Code

LLRM

Lake Loading Response Model

MassDEP

Massachusetts Department of Environmental Protection

MS4

Municipal Separate Storm Sewer System

MWRA

Massachusetts Water Resources Authority

MyRWA

Mystic River Watershed Association

SSO

Sanitary sewer overflow

TMDL

Total Maximum Daily Load

TN

Total nitrogen

TP

Total phosphorus

USGS

U.S. Geological Survey

WQS

Water quality standards

15


-------
Mystic River Watershed TMDL Alternative Development — Final Report

I. Introduction

LA. Background

The Mystic River Watershed is a 76-square mile watershed that drains into Boston Harbor. Located
in the Greater Boston, Massachusetts metropolitan area, it encompasses all or portions of 22 urban
and suburban communities. The watershed faces multiple water quality impacts related to cultural
eutrophication including excessive algal growth, harmful cyanobacteria blooms, and invasive
macrophyte growth. Sources of pollutants from the watershed include stormwater runoff, combined
sewer overflows (CSO), sanitary sewer overflows (SSO), non-point runoff, contaminated sediment,
and three Superfund sites. The watershed suffers from many legacy pollutants as well as present day
pollutant loadings. Several environmental justice communities are located within the watershed, and
although most developable land is built upon, there is high development and re-development
pressure throughout the watershed.

The Mystic River is listed as a Class B water with a Category 5 water quality rating in the
Massachusetts 303(d) "List of Impaired Waters" (2014) for phosphorus, arsenic, chlordane,
chlorophyll, DDT, dissolved oxygen, E. coli, PCB in fish, Secchi depth, and sediment bio-chronic
toxicity. Due to the multiple stressors present in this watershed, development of a traditional Total
Maximum Daily Load (TMDL) to address all pollutants would be a lengthy and complicated task,
especially considering resource limitations. Instead, the Massachusetts Department of
Environmental Protection (MassDEP), U.S. Environmental Protection Agency (EPA) Region 1
(serving New England) and the Mystic River Watershed Association (MyRWA) have embarked on a
plan to pilot an alternate but equally rigorous method ("Alternative TMDL") for determining how to
address impairments, including the nutrient water quality studies documented herein. These studies
as well as other efforts have determined that effective nutrient management will likely go a long way
towards addressing sources of other impairments in the watershed such as bacteria and sediment-
bound contaminants.

I.B. Purpose

The purpose of this project is to support elements of EPA's TMDL Vision process by providing
technical support for watershed restoration efforts to address phosphorus load reduction needed to
meet water quality targets in the Mystic River Watershed. This project provides an opportunity to
achieve multiple TMDL "vision" goals:

•	Estimate the load reduction needed to meet water quality targets in critical water bodies
within the watershed;

•	Engage with communities, state and regional governmental agencies and the local watershed
group;

•	Inform and guide load reduction implementation by municipalities in the watershed;

•	Integrate actions needed to address multiple Clean Water Act programs, such as point and
non-point pollution.

Findings will inform the development of analytical tools for EPA Region 1 to estimate phosphorus
load reductions that are needed to attain applicable Massachusetts surface water quality standards
(WQS) related to cultural eutrophication. Another project goal is to strengthen regional
collaborations for enhanced watershed nutrient management approaches.

16


-------
Mystic River Watershed TMDL Alternative Development — Final Report

I.C. Organization and Overview

This report is a compilation of nine related technical memoranda that Eastern Research Group
(ERG), its sister company PG Environmental, and subcontractors Horsley Witten Group (HW),
Paradigm Environmental and Dr. Nigel Pickering, have developed over the past two years.
Components of the study are listed and described below:

•	Phase 1

o Conceptual Model of Hydrology and Nutrient Dynamics in the Mystic River Basin
(Section II)

o Review of Existing Water Quality Monitoring Data (Section III)
o Review of Modeling Endpoint Approaches (Section IV)
o Watershed Phosphorus Loading Estimates (Section V)

•	Phase 2

o Evaluation of Combined Sewer Overflow and Sanitary Sewer Overflow Data for the

Mystic River Watershed (Section VI)
o BATHTUB Modeling Approach and Calibration Results (Section VII)
o Critical Period of Interest for Phosphorus Load Reduction Analysis (Section
VII.C.9)

o Evaluation of Watershed Phosphorus Load Reduction Analysis (Section IX)
o Broad-Based Nutrient Stormwater Management Strategies for the Mystic River
Watershed using Opti-Tool (Section X)

Phase 1 of the project began with a review of the hydrographic and geographic features of the
Mystic River in 2017-2018, and of the vegetation management practices used in the river in the
preceding two decades. This review informed the development of a conceptual model of flows and
nutrient dynamic processes within the Mystic River. Concurrently, the investigators performed an
exploratory review of the available flow and water quality data within the watershed, identification of
gaps in the data record, and identification of potentially useful GIS information. The result was the
identification of information and data sources that might be usefully exploited in planning modeling
of the watershed, presented in Sections II and III of this report. In addition, these sections
document the region of interest (i.e., the freshwater portion of the Mystic River above Amelia
Earhart Dam) used for later modeling activities.

Under the federal Clean Water Act, Massachusetts statutes, and applicable regulations, narrative
nutrient standards are applicable to the watershed. Section IV of this report presents and discusses
several options for translating the applicable narrative water quality standard to one or more numeric
endpoints. A numeric endpoint provides a standard for water quality modelers to ascertain that
water quality conditions comply with applicable standards. This section includes a review of all
approaches considered by EPA, the consultant team and the Technical Steering Committee (TSC),
including approaches that were not ultimately utilized in later modeling and analyses.

Using available water quality data and GIS information as inputs, the project team developed a
modeling approach to estimate an annual time series of total phosphorus and streamflow within the
watershed modeling domain. The methodology employed for estimating nutrient load and flow time
series from combined sewer overflows (CSOs) and from sanitary sewer overflows (SSOs) is

17


-------
Mystic River Watershed TMDL Alternative Development — Final Report

documented; time series based on all other sources (e.g., precipitation-driven overland flow,
groundwater, etc.) are also documented in this report.

Early in Phase I, EPA and the project team identified a discrepancy between measurements of total
phosphorus at two different labs used by local monitoring programs to measure in-stream water
quality. The source of the discrepancy—which was consistent and predicable between the two
datasets—was not resolvable during the Phase I or Phase II development period and, therefore,
EPA determined to adjust the observed total phosphorus measurements to make the data from both
laboratories mutually consistent. This rendered both datasets usable for model calibration purposes
and will allow future adjustments to the dataset and model results should EPA make a final
determination regarding the source of the discrepancy. The basis for the total phosphorus data
adjustment is documented in Section V.D.

Sections VII through X cover work completed under Phase II including the setup and calibration of
BATHTUB model, which was used to model water quality response for select reaches of the
watershed. The information and rationale used in selecting a critical period of interest for the model
is described as well as the modeling scenarios evaluated (e.g., baseline conditions, current conditions,
future loading conditions) and the resulting load reduction targets needed to meet water quality
standards.

II. Conceptual Model of the Hydrology and Nutrient
Dynamics in the Mystic River Basin

This section presents a conceptual model of the hydrology and water quality dynamics in the Mystic
River Watershed. The purpose of this conceptual model is to describe the primary sources and sinks
of nutrients, and to highlight the important features (e.g., dams/impoundments) and dynamics (e.g.,
macrophytes) that affect water quality conditions in the Mystic River. The conceptual model is
intended to serve as a foundation for evaluating alternative target endpoints (Section IV) and
modeling strategies (Section V).

II.A. Watershed Overview

The freshwater portion of the Mystic River Watershed—the focus of this study — has a total drainage
area of 63 sq. miles. The watershed can be divided into three sub-watersheds referred to as the
upper, central, and lower watersheds. The delineation of these sub-watersheds was based on the
locations of U.S. Geological Survey (USGS) streamflow gages as well as the impacts of the Mystic
Lakes on both the hydrology and water quality of the river.

Figure II-I shows a schematic diagram and corresponding map of the major sub-basins, water
bodies, streamflow gages, and dams across the watershed.

18


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Central
Watershed

^ USGS Flow Gage
\y Dam

Major Tributary
[^j Minor Tributaries/Direct Runoff
j] Receiving Water

Boston Harbor

Note this figure does not include all flow gages or dams across the watershed.

Figure ll-l Schematic Diagram and Map of Mystic River Watershed

Legend

V Dam

# USGS Flow Gage

I I Upper Watershed
I I Central Watershed
I I Lower Watershed

19


-------
Mystic River Watershed TMDL Alternative Development — Final Report

II.B. Upper Watershed

The upper watershed contains the Aberjona River basin, which has a total drainage area of 25 sq.
miles (40 percent of the total freshwater Mystic River Watershed). A USGS streamflow gage (Station
01102500) is located near the outlet of the Aberjona just before it flows into Upper Mystic Lake.
There is currently one major impoundment on the Aberjona River at the Center Falls Dam in
Winchester, which is located about 0.5 mile upstream of the USGS flow gage. Historically, several
smaller dams were also constructed to support mill operations (Knight, 2016). The exact number
and locations of existing small dams along the Aberjona River requires further research. There are
also numerous impounded ponds (e.g., Horn Pond, Wedge Pond) along the tributaries to the
Aberjona River. These impoundments likely reduce the phosphorus loads originating from the land
surface through settling and vegetative uptake.

The primary source of nutrient loading in the upper watershed is stormwater runoff. There are no
combined sewer areas or wastewater treatment facility discharges. However, much of the watershed
is served by separate sanitary sewer systems and storm sewer drainages systems. Illicit discharges of
sanitary sewage to separate storm drainage systems are not uncommon in suburban/urban
watersheds and are known to exist within the upper watershed, although programs to eliminate them
are underway. Additionally, SSOs are also known to occur infrequently, typically only during major
storm events. Internal loading from sediment fluxes is likely not significant but may occur in some
impounded areas such as above the Center Falls Dam in Winchester.

II.C. Central Watershed

The central watershed encompasses the Upper and Lower Mystic Lakes, which have a total drainage
area of 9 sq. miles (15 percent of total freshwater watershed) excluding the upper watershed. The
two lakes are separated by the Upper Mystic Lake Dam, which was recently rebuilt in 2012. Both
lakes have a maximum depth of about 80 ft. While the Upper Mystic Lake undergoes seasonal
stratification from spring to fall, Lower Mystic Lake is perennially stratified due to entrapment of
saltwater that prevents complete turn over in the fall (Ludlam and Duval, 2001). Water quality and
streamflow data show that the two lakes have a significant effect on both the hydrology and water
quality of water flowing from the mouth of the Aberjona River to the head of the Mystic River.

The primary source of nutrients to Upper Mystic Lake is the discharge from the Aberjona River in
addition to direct runoff from the local drainage basin. Discharge from the Aberjona River first
enters what is referred to as the upper lobe of Upper Mystic Lake. Due to the nutrient loading from
the Aberjona River, the upper lobe, which is shallow and not likely to stratify, frequently experiences
eutrophic conditions, including cyanobacteria blooms and excessive aquatic vegetation.

Outflow from Upper Mystic Lake passes through the dam into Lower Mystic Lake, which also
receives inflow from direct runoff of its local drainage basin and from Mill Brook (drainage area of
5.5 sq. miles). Because of the stratification in the two lakes, internal loading from sediment fluxes
may not be a significant source of nutrients during the growing season because the bottom water
(hypolimnion) is thought to not fully mix with the surface water (epilimnion); however, this question
is still under investigation.

The Massachusetts Department of Conservation and Recreation (DCR) operates the Mystic Lakes
Dam. Information about dam operations such as release schedules or target water levels are not
currently available. However, prior to storm events, additional water is often released to increase
available storage in Upper Mystic Lake and prevent flooding along its shoreline. Because of these

20


-------
Mystic River Watershed TMDL Alternative Development — Final Report

operations,, the dam has a significant effect on the flow and water quality dynamics between the
Aberjona and Mystic Rivers.

II.D. Streamflow Impact of Mystic Lakes Dam

Figure II-II shows the instantaneous streamflow over a 7-day period in November 2016 at the
Aberjona River gage above Upper Mystic Lake (01102500) and at the Mystic River gage just below
Lower Mystic Lake (01103010). Lhe Aberjona River gage shows a typical storm hydro-graph during
the night of November 15. Lhe Mystic River gage also shows evidence of an increase in flows due to
stormwater; however, the shape and magnitude of the hydrograph is significantly altered. Lhe peak
flow is lower below the lakes and occurs about 6 hours later. There is also a significant diurnal
pattern in flows at the Lower Mystic Lake outlet due to the release of water during low tides at the
Amelia Earhart Dam located about 5 miles downstream (discussed below) near the mouth of the
Mystic River. Lastly, the total volume of flow during storm events is in fact higher at the Aberjona
River gage than at the Mystic River gage likely due to some of the flow being stored in Upper Mystic
Lake.

150-

Nov 15

Nov 17

Nov 19

Nov 21

	Aberjona (01102500)

	Mystic @ Lower Lake Outlet (01103010)

Figure ll-li Instantaneous Streamflow Above and Below Mystic Lakes
I I.E. Water Quality Impact of Mystic Lakes Dam

In addition to the change in the streamflow hydrograph, the lakes also have a significant effect on
water quality., Figure II-III shows the distribution of monthly total phosphorus (LP) concentrations
measured by the MyRWA Baseline Monitoring Program at the Aberjona River streamflow gage
(ABR006), the Mystic Lakes dam (UPL001) at the outlet to the Upper Mystic Lake, and below the
outlet of the Lower Mystic Lake at the mouth of the Mystic River mainstem (MYR071). LP
concentrations are highest coming out of the Aberjona River but drop significantly at the outlet of
Upper Mystic Lake likely due to settling of particulate phosphorus and uptake by aquatic plants. In
Lower Mystic Lake, concentrations increase from the levels coming through the dam due to loads
from Mill Brook and other tributaries. Lhis figure demonstrates the critical role the lakes have in
"resetting" the water quality from the upper watershed (Aberjona River) to the Mystic River

21


-------
Mystic River Watershed TMDL Alternative Development — Final Report

mainstem. Upper Mystic Lake is thus a major sink of nutrients due to the long residence time that
promotes settling of particulate phosphorus and uptake by aquatic vegetation and algae.

0.100-

o.ooo

Location ID
E^3 ABR006
E^3 UPL001
MYR071

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016

Year

Figure ll-lll Annual Distributions of Monthly TP Concentrations at the Aberjona
River Outlet (ABR006), Mystic Lake Dam (UPL001), and Start of the Mystic River

Mainstem (MYR071)1

II.F. Lower Watershed

The lower watershed includes the Mystic River, Alewife Brook, Maiden River and numerous smaller
tributaries, which have a total drainage: area of 28 sq. miles (45 percent of the freshwater portion of
the watershed) excluding the central and upper watersheds. Land use in this area is heavily
urbanized, especially in areas around the lower section of the Mystic River. A portion of this area
contributes to the MWRA combined sewer system, which has numerous CSO outfalls along Alewife
Brook and the Mystic River. Over the past decade, many of these outfalls have been closed. Among
the remaining open outfalls, the frequency and magnitude of CSO discharges has been drastically
reduced. For the period of 2000 to 2016, CSO mitigation projects by the MWRA and the cities of
Cambridge and Somerville have reduced annual CSO discharge volumes to the freshwater portion
of the Mystic River by approximately 88 percent (59 to 7 million gallons) for the typical rainfall year.

The primary sources of nutrient loads are from the outflow of Lower Mystic Lake, stormwater
runoff, CSOs, illicit discharges, and internal loading. Due to the long history of CSO discharges to
the low gradient Alewife Brook and impounded Mystic River, the sediments in Alewife Brook and
the Mystic River are likely highly organic with elevated phosphorus levels, which could cause
significant internal loading. The legacy sediments are likely to also drive a high sediment oxygen
demand, especially in Alewife Brook where the entire water column has been observed to become
hypoxic (and occasionally anoxic) during hot, dry periods in the summer.

1 Boxplot hinges are 25th, 50th, and 75th percentiles of the concentration distribution. Upper and lower whiskers
represent largest and smallest values inside 1.5 x Interquartile Range.

22


-------
Mystic River Watershed TMDL Alternative Development — Final Report

II.G. Hydrologic and Hydraulic Impacts of Amelia Earhart Dam

Discharge from Mystic River is controlled by the Amelia Earhart Dam, which is operated by DCR.
The dam includes 3 locks, which unintentionally allow saltwater intrusion from Boston Harbor. The
dam is typically operated by pumping water out of the lower basin prior to storm events to increase
available storage and prevent flooding along the shoreline. Water is also typically released each day at
low tide. The exact operational schedule and targets for this dam are unknown.

Along the entire Mystic River, as well as most of Alewife Brook and the lower section of the Maiden
River, there is a very small gradient in elevation of both the sediment and water surfaces (about 1
ft.). As a result, flow velocities are very low in these water bodies and the entire system acts as one
large impoundment.

Figure II-IV shows the relative water surface elevation at three USGS stations along the Mystic
River and Alewife Brook during an 11-day period from Sept 1, 2016 to Sept 10, 2016. The relative
water surface elevation was computed by subtracting the mean elevation at each station from the
instantaneous value. This was necessary because stage data at the Alewife station is not reported
relative to the same vertical datum as the other stations. The rapid drop in elevation on Sept 6 was
likely due to pumping at the Amelia Earhart Dam in preparation for an upcoming storm. All three
stations reflect the same change in elevation indicating that the entire Mystic River and Alewife
Brook behave as a single impoundment.

Relative Water Surface Elevation at USGS Stations in Mystic River Basin

Relative WSE(t) = WSE(t) - mean(WSE)

Sep 02	Sep 04	Sep 06	Sep 08	Sep 10

Date

Figure II-IV Relative Water Surface Elevation at USGS Gages on Alewife Brook

and Mystic River (Sept. 1-10, 2016)

II.H. Water Quality Gradient Along the Mystic River

Water quality in the upper reach ot the Mystic River in the lower watershed is relatively good quality
due to the reduced phosphorus levels in the outflow from Lower Mystic Lake. Phosphorus and
chlorophyll-a levels then gradually increase going downstream along the Mystic River towards the
Amelia Earhart Dam. In the lower reaches ot the lower Mystic River Watershed, excessive
macrophyte growth due to nutrient enrichment constitutes major water quality impairment, as
farther described below.

Figure II-Y shows chlorophyll-a and TP concentrations at five stations along the Mystic River and
one station in the lower basin of the Maiden River during 2015 and 2016. MyRWA collected the

23


-------
Mystic River Watershed TMDL Alternative Development — Final Report

data for its phosphorus loading study. The high concentrations, in late summer of 2016, are most
likely caused by an herbicide application on the Mystic River to remove aquatic vegetation. The
result of this application was an instant release of phosphorus to the water column, which,
combined with an increase in light availability, likely spurred significant phytoplankton growth. A
cyanobacteria bloom was also observed during this period.

	6—

MYR43

^ *



125

—X

100



75

3

19

50

ฃ
O

25



0



0.3





CT>

0.2

k

a.

0.1





0.0















-*>>

•*A

•V .'v.



v'-V \





r*- ir>
c c

N ID IT)

ฃ 5 3

Figure ll-V Chiorophyll-a and TP Concentrations Along the Mystic and Maiden

Rivers, 2015-2016

11.1. Long Term Changes in Water Quality

Over the long term, there has been a gradual decline in chlorophyll-a and TP concentrations along
the Mystic River based on the MWRA data, except for 2016, due to the herbicide treatment
mentioned above. More information on the 2016 data is presented in Section II.J below. Figure
II-VI shows the annual mean, median, and 90th percentile of chlorophyll-a and TP at the three
MWRA stations based on data from June — October of each year. The decline in chlorophyll-a is
most pronounced at the lower-most station, MWRA167. In 2012-2014, 90th percentile of
chlorophyll-a was below 25 ppb, which is slightly greater than the target 20 ppb in the Lower
Charles River TMDL. As discussed below, the declines in TP and chlorophyll-a in the lower Mystic
coincide with a steady increase in growth and coverage of aquatic macrophytes during this period.

24


-------
My stic River Watershed TMDL Alternative Development — Final Report

MWRA083

MWRA066

MWRA167

100

75

50-

25-

0-

0.20-

_ 0.15-

ni

E

0.10-

0.05-

0.00-



2000 2005 2010 2015 2000 2005 2010 2015 2000 2005 2010 2015

Year

Statistic

Mean

Median

Q90

Figure II-VI Annual Mean, Median, 90th Percentile of Chlorophyll-a and TP at
Three Stations on the Mystic River, 2000 - 2016 (June - Oct. only)

11. J. Impact of Aquatic Vegetation on Water Quality

Aquatic vegetation is a major nuisance on the Mystic River. Over the past 10 years, the coverage of
water chestnut and water hyacinth has grown dramatically. MyRWA currently spends significant
resources to manually remove this vegetation to allow boat passage. Treatment of the vegetation
includes mechanical and manual harvesting, as well as herbicidal treatments, which began more
recently. The vegetation is likely having a significant impact both on the hydraulics of the river by
increasing drag and forcing flow through a narrow channel, as well as on water quality through
nutrient uptake, increased water column shading, and increased particulate settling. Consequently,
the increase in aquatic vegetation in the lower Mystic is likely the primary cause for the declining
trend in chlorophyll-a and TP concentration described above.

In 2016, a major dose of herbicide was applied to the Mystic River to remove vegetation. Shortly
after tins treatment, both phosphorus and chlorophyll-a levels spiked (see Figure II-II-6). This
response indicates that aquatic vegetation is likely a major control on phytoplankton growth along
the Mystic River mainstem and in the lower basin. Presently, most of the aquatic vegetation found in
the lower Mystic can be rooted or free floating and is capable of taking up nutrients from both
bottom sediments and the water column. The combination of increased flow resistance due to the
rooted plants, which increases particulate settling rates, and the uptake of nutrients from the water
column is likely the primary cause for the declining trend in TP. At the typical TP levels observed in
the lower Mystic (<100 (Jg/L), decreases in TP in open water areas (i.e., free of excessive; aquatic
plants) will likely result in less phytoplankton growth and lower chlorophyll-a. Also, the abundance
of vegetation is likely to be a contributing cause to suppressing phytoplankton growth by reducing

25


-------
Mystic River Watershed TMDL Alternative Development — Final Report

light availability through shading and causing light limitation. Thus, it is reasonable to infer that the
long-term decline in chlorophyll-a and TP levels shown above in Figure II-VI (excluding 2016 due
to the herbicide treatment) are likely caused by the increasing abundance of aquatic vegetation
removing phosphorus from the water column and limiting light penetration.

III. Review of Existing Water Quality Monitoring Data

The purposes of this section are as follows:

•	Describe and summarize the known, available water quality monitoring data.

•	Review the data for precision, accuracy, representativeness, comparability, and completeness.

According to available data, water quality surveillance of the watershed has been ongoing since at
least 1989. Consistent with the scope of this analysis, the examination will be restricted to data
collected from 2000 through 2016. Data for 2017 were not available until late in the project and
were only included for BATHTUB model validation.

III.A. Data Gaps and Recommendations for Future Sampling
Efforts

Based on the review of the available data and discussions with MyRWA and EPA Region 1, the
following data gaps have been identified which could be addressed through future monitoring
efforts.

III.A.1.Ecological/Biological Indicators of Over-Enrichment

Currently, little data is available on excess vegetative growth. Measurements are limited to
chlorophyll-a and do not include macrophyte abundance, percent cover, or broader measures of
species richness. MyRWA and EPA should consider including, at a minimum, percent of
macrophyte cover in the water body during monitoring events for baseline and phosphorus loading.

III.A.2.Streamflow

As discussed below, in the section of this memo on the available USGS flow data, there are few
locations in the watershed where it is currently feasible to make direct flow measurements. To
develop reliable estimates of nutrient loads through the watershed, measurements or reliable
estimates of flows in the watershed will be needed. This task is further complicated by multiple
impoundments. Should methods for reliable direct measurement prove infeasible, other approaches
for estimating flow based on well-established modeling techniques (e.g., using climatological, land
use, and soil type data available in GIS databases) may be explored to estimate precipitation driven
flows.

III.A.3.Sediment

Sediment attributes (e.g., total phosphorus concentrations, sediment oxygen demand) would be
useful for future modeling but was not available for the modeling portion of the project, and it is
recommended to include these attributes in future watershed surveillance efforts, if feasible.

26


-------
Mystic River Watershed TMDL Alternative Development — Final Report

III.B. MyRWA and MWRA Monitoring Data
III.B.1.Data Characterization

The MyRWA provided water quality monitoring data to ERG on January 31, 2017 and provided
supplementary data on February 11, 2017. The dataset was composed of samples collected under:

•	MyRWA's baseline water quality monitoring.

•	MyRWA's phosphorus loading monitoring survey.

•	MWRA Boston Harbor water quality monitoring.

•	MWRA's combined sewer overflow event monitoring.

The baseline monitoring program has been in operation since 2000 and is used to monitor a
variety of trends in watershed water quality. Collected constituents include pathogen indicators,
nutrients, and physical-chemical water quality parameters (e.g., total suspended solids, pH, etc.).

The phosphorus loading monitoring program has been conducted since 2015 and is used to
collect information on parameters that contribute to eutrophication impairments (e.g., phosphorus)
and response parameters, which could potentially be used as indicators of nutrient over enrichment.

The MWRA water quality monitoring in general started in 1989, with the beginning of the CSO
monitoring program. The Boston Harbor monitoring in the Harbor proper began in 1993, and in
the rivers in 1995. This program was created to establish long-term water quality trends in the
Harbor and tributary watersheds for pathogen indicators, nutrients, and physical-chemical water
quality parameters.

CSO monitoring is conducted to evaluate water quality risks associated with the discharge of
untreated sewages and stormwater runoff into the watershed during CSO events. Monitoring is
conducted on an ongoing basis in Alewife Brook, Chelsea River, Little River, and the Mystic River.
Note that monitoring is not restricted to CSO discharge events. The CSO monitoring program
collects data on pathogen indicators and on physical-chemical water quality parameters.

Table III-1 summarizes the water quality parameters collected under each program included in the
dataset.

27


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table 111-1. Water Quality Parameters Included in Each Monitoring Program2

Parameter

Monitoring Program

Baseline

Boston
Harbor

cso

Phos.
Loading

Nitrogen

Total Nitrogen

yj

yj





Particulate Nitrogen



yj





Nitrogen, total dissolved



yj





Ammonia-nitrogen

yj

yj





Nitrite

yj







Inorganic nitrogen (nitrate and nitrite)

yj

yj





Nitrate

yj







Phosphorus

Total Phosphorus

sj

sj



yj

Phosphorus, Particulate Organic



yj





Dissolved Phosphorus



yj





Orthophosphate



yj



yj

Biological

Chlorophyll-a



yj



yj

Pheophytin a3



yj





Oxygen

Dissolved Oxygen

sj

sj

yj

yj

Dissolved Oxygen (Perc. Saturation)

yj

yj

yj

yj

Water Clarity

Attenuation Coefficient



yj





Secchi Disk Depth



yj

yj



Turbidity



yj

yj

yj

Other Physical/Chemical

Carbon, Total Particulate



yj





PH

yj

yj

yj



Salinity

yj

yj

yj



Specific conductance

yj

yj

yj

yj

Total suspended solids

yj

yj





Water Temperature

yj

yj

yj

yj

Pathogen Indicators

Escherichia coli

yj

yj

yj



Enterococcus

yj

yj

yj



Fecal Coliform4

yj

yj

yj



28


-------
Mystic River Watershed TMDL Alternative Development — Final Report

The watershed is subdivided into nine sub-basins5. As illustrated in Figure III-I, water quality
monitoring stations: are present in each of the sub-basins and are typically located at or near
confluences between the drainage areas.



















l

rt /



Bed1ฐ Woburn Sloneham

Reservation \~n-i '



Aberjona River







^ | t 'V >• 'ซ/

-' • ฃ

3' Saugus - ฆ ^



^ jง . J •' Middlesex Melrose







' f Winchester Reservation .

/





ฃ ฉ ~ % J







Vs ' ' ~



'-S ' ' Rumnev

type

Upper Mystic Lake





• Fresh

~

ซ sT *** (So



Reservation

A Saline

Lower Mystic Lake



\ 0ฐZ)'

category

. A 2. . (9~



J

\ * • + •Medford 4
j Arlington • # # €





• Estuary



ฎ I 1

• Lake

ฆc Mystic River (FreS^a|den River

~ Alewife Brook

TuftS ^ VJS)
ซ • University % \ # ^



Revere i

A ฐ ^

• River/Stream







* . ฉ \

Somerville MysticJRiver (Salt)

Chelsea

fe ' "--or./ A
~



„ ป ^ — O ^'

% „ 0 Cambridge

a. (f'
H

Boston Logan _^JnthrฐP ,
International QiS)



Figure III-!, Mystic River Watershed Drainage Basins and Monitoring Locations

Most water bodies in the watershed have been sampled all years from 2000 — 2016 (Table III-2). Six
water bodies, mostly located in the upper portions of the watershed, have been sampled in 2015 and
2016 as part of the phosphorus loading survey but have not been monitored otherwise. Two water
bodies in the lower portion of the watershed, Mill Creek and the Belle Island Inlet, have been
monitored for approximately the past decade.

2	Particulate nitrogen, particulate phosphorus, particulate carbon, total dissolved phosphorus, and total dissolved
nitrogen are only tested at brackish/saltwater locations downstream of the Amelia Earhart dam.

3	Phaeophytin is not currently tested using our WQ sondes: though there are older data in the dataset.

4	Fecal colifonn is not currently tested in any of the freshwater locations for either MWRA monitoring program, but
like phaeophytin there will be older data in the dataset. It is currently tested at locations downstream of the dam.

' Note that this delineation is provided for general information purposes and to provide the reader with an
approximate sense of patterns of drainage within the watershed and is not intended to provide a definitive
delineation of drainage within the basin.

29


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table 111-2. Temporal Range and Duration of Monitoring within Water Bodies

Water Body

Sample Range

No. Years Monitored

Min
Year

Max
Year

All
Programs

Program Subtotals

Baseline

Phos.
Loading

Boston
Harbor

cso

Aberjona River

2000

2016

17

17

2

-

-

Horn Pond

2015

2015

1

-

1

-

-

Horn Pond Brook

2015

2015

1

-

1

-

-

Winter Pond

2015

2015

1

-

1

-

-

Wedge Pond

2015

2016

2

-

2

-

-

Upper Mystic Lake

2000

2016

17

17

2

-

-

Lower Mystic Lake

2015

2015

1

-

1

-

-

Mill Brook

2000

2016

17

17

1

-

-

Spy Pond

2015

2015

1

-

1

-

-

Winns Brook

2000

2016

17

17

-

-

-

Little River

2000

2016

17

-

-

-

17

Alewife Brook

2000

2016

17

17

-

-

17

Mystic River (Fresh)

2000

2016

17

17

2

-

17

Meetinghouse
Brook

2000

2016

17

17

1

-

-

Maiden River

2000

2016

17

17

2

15

-

Mystic River (Salt)

2000

2016

17

9

-

17

17

Mill Creek

2008

2016

9

9

-

-

-

Chelsea River

2000

2016

17

9

-

-

15

Belle Isle Inlet

2009

2016

8

8

-

-

-

Restricting the examination to parameters that are likely to be of the greatest significance when
assessing conditions related to eutrophication, it is apparent that total phosphorus and dissolved
oxygen data are temporally and spatially well represented, and Secchi depth data are available for
most of the water bodies (see Figure III-III). In particular, multiple monitoring locations on the
main stem of the Mystic River downstream of Lower Mystic Lake possess large historical datasets
for most parameters of potential interest for developing in-stream water quality targets (Figure
III-IV).

30


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Aberjona River	Alewife Brook	Belle Isle Inlet	Chelsea River	Horn Pond

I

I



ll



1 1

Lower Mystic Lake

1

|

ฆ



Horn Pond Brook

ฆ



Little River

ฆ

1

1

Maiden River

ฆ

Meetinghouse Brook

_	I _ Jill I

**— u
o

S?	Mill Brook	Mill Creek	Mystic River (Fresh)	Mystic River (Salt)	Spy Pond

fl I. iHillIIi

A*. ^ ^ is

Upper Mystic Lake	Wedge Pond	Winns Brook	Winter Pond	^ ^ (j

J l_ . I J_

d f / ^	cf ? ,oฐ cF f J~ ,


-------
Mystic River Watershed TMDL Alternative Development — Final Report

MWRA056	MWRAQ57	MWRA059	MWRA066

II II II u

o

MWRA067	MWRAQ83	MWRA177	MYR071

II Mi. II

i	i	i	i	i	i	i	i	i	i	i	i

E	MYR33	MYR43	f? cf ^ ^

15-	O ฃ	O ฃ

ฆS	-X

10-
5-

o-

Refer to Appendix B for additional detail on water quality parameter codes.

Figure Ill-Ill Number of Years in Which Observations Were Made for
Eutrophication Related Parameters in The Mystic River

As shown in Figure III-III, data for water quality parameters relevant to nutrient impairment have
been collected throughout the watershed—particularly in the main waterways—and at a relatively
high frequency. Sampling frequencies associated with the Boston Harbor monitoring and
phosphorus loading programs occurred bi-weekly on a seasonal basis, with much of the monitoring
for the baseline monitoring program occurring monthly. Refer to Appendix A for additional detail
on sampling frequency broken down by water body, monitoring location, and monitoring program.

32


-------
My stic River Watershed TMDL Alternative Development — Final Report

Number of Sample Days

Winter Pond -
Winns Brook-
Wedge Pond-
Upper Mystic Lake-
Spy Pond-
Mystic River (Salt)-
Mystic River (Fresh)-
Mill Creek-
Mill Brook-
Meetinghouse Brook-
Maiden River-
Lower Mystic Lake -
Little River-
Horn Pond Brook-
Horn Pond-
Chelsea River-
Belle Isle Inlet-
Alewife Brook-
Aberjona River-

No. of Days

3000

2000

1000

i of Days

Refer to Appendix B for additional detail on water quality parameter codes.

Figure lil-IV. Temporal Coverage (both number of days and Percentage of Total
Available Days from 2000 - 2016) of Water Quality Parameters by Water Body

In terms of seasonal coverage, the number of observations was relatively evenly split between
summer (June - October) and winter (November — May).: The fraction of summer observations
ranged from approximately 35 percent to 70 percent within the watershed and showed typical values
of 53 +/- 19 percent (average +/- standard deviation) for all water bodies with at least 10
observations for a given parameter.

While temporal and geographic coverage is robust within the dataset, specific
season/parameter/water body combinations do exist where relatively limited data coverage is
available. Table III-3 identifies specific instances where data limitations exist for nutrient response
variables at a seasonal level.

33


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table 111-3 Number of Seasons Where Few (N < 3) Samples were Collected for a

Waterbody-Parameter Combination

Parameter

No. of Seasons with n<3*

Jun-Oct

Nov-May

Alewife Brook

Secchi Depth

3

9

Turbidity

-

5

Little River

Dissolved Oxygen

7

7

Dissolved Oxygen (% Saturation)

7

7

PH

7

5

Specific Conductivity

8

7

Water Temperature

7

7

Turbidity

2

3

Maiden River

Dissolved Oxygen

1

19

Dissolved Oxygen (% Saturation)

-

10

PH

-

9

Secchi Depth

-

12

Specific Conductivity

-

13

Water Temperature

-

12

Turbidity

2

20

Mystic River (Fresh)

Secchi Depth

-

2

Total Dissolved Nitrogen

1

4

Total Dissolved Phosphorus

1

4

Turbidity

-

1

Mystic River (Salt)

Attenuation Coefficient

6

1

Chlorophyll-a

2

-

Ammonia-Nitrogen

2

-

Inorganic Nitrogen

2

-

Phaeophytin

2

-

Orthophosphate

2

-

Total Dissolved Nitrogen

2

-

Total Dissolved Phosphorus

1

-

Total Particulate Carbon

1

-

Total Particulate Nitrogen

2

-

Total Particulate Phosphorus

3

-

Total Suspended Solids

3

-

Turbidity

2

3

Where a water body-parameter combination is absent, or a is listed, then n > 3 observations.

Appendix C includes a series of boxplots that summarize the observed characteristics of the dataset
(i.e., minimum, 25th percentile, median, 75th percentile, and maximum). Additionally, a small number
of unusual outlier values have been noted on the figures. Appendix D includes a table of the outlier
observations with any documented data quality issues noted.

34


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Dataset attributes that were assessed included the frequency of observations flagged for quality
concerns and the frequency of censored observations (i.e., measurement results below the analytical
method detection limit). Data quality documentation is currently available for only two of the
monitoring programs—Baseline and Phosphorus Loading. As shown in Table III-4, approximately
98 percent of the observations from these two programs were free of documented data quality
issues.

Table 111-4. Summary of Documented Data Quality Issues

Flag

Flag Description

Program Subtotals

Grand Total

Baseline

Phos. Loading

No.

Percent of
Total

B

Analyte detected in the blank

-

6

6

0.023

E

Instrument error

-

7

7

0.027

F

Field replicate quality control
failure

308

-

308

1.186

H

Holding time issues

-

15

15

0.058

J

Detected above method
detection limit but below
quantitation limit—result is an
estimate

-

11

11

0.042

K

pH calibration error

-

101

101

0.389

L

Lab duplicate relative percent
difference exceeded

168

2

170

0.655

0

Other unspecified issues

2

-

2

0.008

No
Flag

No documented data quality
issue

22,652

2,697

25,349

97.613

The majority of the non-detect numbers were total suspended solids data and nitrogen data. In some
cases (e.g., total suspended solids and nitrites), non-detect values make up a substantial portion of
the observations. In these instances, accurate estimates of water quality parameters may be difficult
to develop with a high degree of confidence. See Table III-5 for a summary of non-detect data.

Table 111-5 Summary of Non-Detect Data.1

Water body

No of Samples

Method Detection
Limit

Unit

s

Detecte
d

Non-
Detect

Total Suspended Solids

Aberjona River

275

283

2 - 11.4

mg/L

Alewife Brook

148

41

1-7.7

mg/L

Chelsea River

77

4

5-5

mg/L

Maiden River

119

63

1-20

mg/L

Meetinghouse
Brook

70

120

o

1

1

1

mg/L

Mill Brook

106

81

2-20

mg/L

Mystic River (Fresh)

829

116

0.24-10.9

mg/L

Mystic River (Salt)

1048

6

0.24 -10

mg/L

Upper Mystic Lake

50

133

1-8.3

mg/L



35


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Winns Brook

92

96

1-12

mg/L

Total Particulate Nitrogen

Mystic River (Salt)

439

1

0.0027

mg/L

Total Particulate Carbon



Mystic River (Salt)

441

1

0.016

mg/L

Total Phosphorus

Aberjona River

582

13

0.05 - 0.05

mg/L

Chelsea River

92

3

0.01-0.025

mg/L

Maiden River

253

2

0.05 - 0.05

mg/L

Meetinghouse
Brook

199

3

0.05 - 0.05

mg/L

Mill Brook

221

3

0.05 - 0.05

mg/L

Mill Creek

92

1

0.01

mg/L

Mystic River (Fresh)

1041

6

0.05-0.11

mg/L

Mystic River (Salt)

175

2

0.01-0.012

mg/L

Spy Pond

16

1

0.005

mg/L

Upper Mystic Lake

233

6

0.005 - 0.05

mg/L

Total Dissolved Phosphorus



Mystic River (Salt)

455

1

0.0034

mg/L

Orthophosphate

Aberjona River

0

7

0.005 - 0.005

mg/L

Alewife Brook

13

3

0.005 - 0.005

mg/L

Horn Pond Brook

1

5

0.005 - 0.005

mg/L

Maiden River

0

14

0.005 - 0.005

mg/L

Mill Brook

3

4

0.005 - 0.005

mg/L

Mystic River (Fresh)

744

35

0.00031 - 0.005

mg/L

Mystic River (Salt)

903

4

0.00031 - 0.030

mg/L

Spy Pond

0

7

0.005 - 0.005

mg/L

Upper Mystic Lake

0

10

0.005 - 0.005

mg/L

Wedge Pond

0

7

0.005 - 0.005

mg/L

Nitrate

Maiden River

4

1

0.1

mg/L

Mill Brook

5

1

0.1

mg/L

Mystic River (Salt)

7

1

0.1

mg/L

Winns Brook

5

1

0.1

mg/L

Inorganic Nitrogen (Nitrite plus Nitrate)

Belle Isle Inlet

43

38

0.1-0.5

mg/L

Chelsea River

60

27

0.1-0.5

mg/L

Mystic River (Salt)

1023

52

0.00028-0.5

mg/L

Nitrite

Aberjona River

6

12

0.1-0.1

mg/L

Alewife Brook

0

6

0.1-0.1

mg/L

Chelsea River

0

6

0.05-0.1

mg/L

Maiden River

0

5

0.1-0.1

mg/L

36


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Meetinghouse
Brook

0

6

0.1-0.1

mg/L

Mill Brook

0

6

0.1-0.1

mg/L

Mill Creek

1

4

0.1-0.1

mg/L

Mystic River (Fresh)

0

6

0.1-0.1

mg/L

Mystic River (Salt)

0

6

0.05-0.1

mg/L

Upper Mystic Lake

0

6

0.1-0.1

mg/L

Winns Brook

0

6

0.1-0.1

mg/L

Ammonia

Mystic River (Salt)

865

42

0.00039 - 0.010

mg/L

Upper Mystic Lake

56

1

0.00039

mg/L

Winns Brook

56

1

0.00039

mg/L

1. Locations with zero non-detect observations have been excluded.

III.B.2.Data Review

The investigators reviewed the available data for accuracy/precision, representativeness,
comparability, and completeness and concluded that the data are largely acceptable for use in
assessing appropriate nutrient endpoint targets in the watershed.

Precision measures the reproducibility of repeated measurements and accuracy measures the
"correctness" of an estimate. Overall, the dataset exhibited satisfactory levels of accuracy and
precision where quality was documented. A relatively small fraction of the dataset included
documented data quality issues (see Table III-4), though this conclusion may be revised upon receipt
of quality control data for the CSO and Boston Harbor monitoring programs. Where data quality
concerns have been documented in the Baseline and Phosphorus Loading data, excluding flagged
data is recommended, with the possible exception of the J-flagged data (i.e., observations where the
parameter was positively detected in the water, but at a level where quantitation is less precise than is
usual).

The data are largely representative, both spatially and temporally, of the watershed. Most water
bodies in the watershed have been monitored for the entire duration of interest (2000 — 2016). Two
water bodies have been monitored for a majority of the period of interest, and the remainder has
been intensively sampled in 2015 — 2016. In addition, the dataset also captures seasonal variation in
parameters. Table III-3 documents specific water bodies where the available data are limited. In
addition, Table III-5 documents limited instances—particularly with TSS and nitrites—-where a large
fraction of the data are censored and which may influence the ability to draw accurate inferences
regarding water quality parameters (e.g., averages, standard deviations) for these constituents.
However, these limitations are unlikely to present difficulties when developing nutrient modeling
endpoints, as other constituents are present in the dataset and are likely to be more informative for
purposes of developing protective in-stream water quality targets.

Comparability is an expression of the confidence with which one data set can be compared to
another. Based on the available information, the data from the different programs, which compose
the dataset, appear comparable.

37


-------
Mystic River Watershed TMDL Alternative Development — Final Report

The available data appears sufficiently complete for use in developing water quality targets. The vast
majority of the data are valid and composed of point estimates (i.e., detect values) that can be used
to develop reference conditions or stressor-response relationships for use in establishing modeling
endpoints and protective instream water quality targets.

III.C. USGS Flow Data

Valid daily flow data are available from two USGS monitoring stations for the Aberjona River and
Alewife Brook (Table III-6). These stations collectively drain 8.96 square miles of the watershed,
which accounts for approximately 12 percent of the watershed's land area. The two stations possess
historical records, which extend prior to 2000 and to 2007, respectively.

Table 111-6. Summary of USGS Monitoring Stations

Gauge ID
No.

Description

Location
(Lat; Lon)

Discharge Data
Availability

HUC

Net Drainage
Area (sq. mi.)

01102500

Aberjona River at
Winchester, MA

42ฐ26'50.5";
71ฐ08'18.9"

<2000 - 2/2017

01090001

24.7

01103010

Mystic River at
Arlington, MA

42=25' 14";

71ฐ08'33"

6/2016 - 2/20171

01090001

Undetermined

01103025

Alewife Brook Near
Arlington, MA

42ฐ24'25";
71ฐ08'04"

10/2007 - 2/2017

01090001

8.36

01103038

Maiden River at
Maiden, MA

42ฐ25'04";
71ฐ04'23"

7/2016 - 2/20171,2

01090001

Undetermined

01103040

Mystic River at RT16
at Medford, MA

42ฐ24'20.6";
71ฐ05'45.6"

10/2015 - 2/20171

01090001

Undetermined

01103050

Mystic River at
Amelia Earhart Dam

42ฐ23'44";
71ฐ04'32"

None3

01090001

62.7

1.	USGS has indicated that all or some of the discharge data collected at these stations are unreliable.

2.	Stream temperature data are also available.

3.	Gauge height data are available.

Three additional monitoring stations are equipped to estimate discharge volumes, however, USGS
has indicated their measurements may be unreliable or inaccurate due to a combination of the
shallowness of the river and low stream velocities. Currently, it is unclear if all discharge data from
these stations are unusable or if some portion of the streamflow data might be utilized.

III.D. GIS Datasets

Geospatial data are available from the sources listed in Table III-7, below. The data layers that were
considered for calculating phosphorus loads in the watershed include: impervious cover, land use,
and hydrologic soil types. In addition, phosphorous loads may be estimated for the Mystic River
using similar export rates developed for the Massachusetts Municipal Separate Storm Sewer System
(MS4) Permit and which are available from EPA Region 1.

38


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table 111-7. GIS Data Sources Available for the Mystic River Watershed

Data layer

Source

Description

Land use/land
cover

Massachusetts Office of
Geographic Information
(MassGIS) or Multi-
Resolution Land
Characteristics Consortium
(MLCD)

Land use/land cover (LULC) layers contain information on the
physical land type, e.g., forest, wetlands, of an area as well as
information on how people are using the land, e.g., row crops,
low-intensity development. Land cover can be determined in
the field or by interpreting remotely sensed imagery (i.e.,
satellite imagery, aerial photos). Two sources for LULC datasets
are MassGIS's Land Use (2005) or NLCD 2011.

Impervious
surface

MassGIS

Impervious surfaces are surfaces that do not allow water to
penetrate, forcing water to runoff. As water runs off, it can
carry pollution from waterways and other surfaces into water
bodies. Common impervious surfaces include roads, parking
lots, rooftops, driveways and sidewalks, and compacted soils.

NRCS HUC
Basins

(8,10,12) (Sub-
watersheds)

MassGIS

A hydrologic unit code (HUC) is the number assigned to a
hydrologic unit, which is a drainage area that nests in a multi-
level drainage system. Its boundaries are defined by
hydrographic and topographic criteria that delineate an area of
land upstream from a specific point on a river, stream or similar
surface water. HUCs are identifiers as assigned to basin
polygons by the USGS.

Soils

MassGIS (NRCS SSURGO) or
NRCS Web Soil Survey

Information on underlying soils can help determine how much
water can be absorbed or how much will runoff. There are
specific hydrologic soil groups identified for areas that are
based on estimates of runoff potential. Soils are assigned to
one of four groups according to the rate of water infiltration
when the soils are not protected by vegetation, are thoroughly
wet, and receive precipitation from long-duration storms. The
soils in the United States are assigned to four groups (A, B, C,
and D) and three dual classes (A/D, B/D, and C/D).

Sewer-shed

Cambridge and Somerville

Drainage system for the local storm sewer (separate and/or
combined) discharging into the Mystic River Watershed. Data
available for Cambridge and Somerville only.

IV. Review of Modeling Endpoint Approaches

Water quality segments within the Mystic River Watershed are currently impaired for eutrophication-
related parameters—high phosphorus, nitrogen, and chlorophyll-a levels have been documented
throughout the watershed. Over enrichment has resulted in algae blooms and periods of excessive,
nuisance vegetation growth. The purpose of selecting water quality targets is to establish a set of
modeling and water quality endpoints (i.e., target water quality conditions) that meet Massachusetts'
water quality standards and are protective of the designated uses established in the WQS (314 CMR
4.00). Attainment of appropriate targets could eventually result in acceptable levels of algal growth
and the cessation of use impairments caused by excessive macrophyte growth.

This section reviews the following approaches for establishing targets in the watershed:

•	Use of existing regional/local targets.

•	Use of reference water body conditions.

39


-------
Mystic River Watershed TMDL Alternative Development — Final Report

•	Development of targets based on stressor-response relationships.

•	Development of targets based on mechanistic models.

IV.A. Water Quality Standards Applicable to the Mystic River
Watershed

Once a tidal river, the Lower Watershed currently functions as a large impoundment due to a dam
located at the basin outlet. Another significant impoundment on the main stem of the river exists at
the outlet of the Upper Mystic Lake (Figure IV-I and Figure IV-II).

In 314 CMR 4.00, Massachusetts establishes two eutrophication related standards applicable to the
watershed: a narrative nutrient standard and a numeric dissolved oxygen standard.

•	Nutrients. The narrative standard prohibits discharges containing nutrients in
concentrations that would " cause or contribute to cultural eutrophication, including excessive growth of
aquatic plants or algae, and othenvise render water unsuitable for designated uses." [314 CMR 4.05(5)]

•	Dissolved Oxygen. For Class B waters, the concentration of dissolved oxygen shall be
greater than or equal to 5 mg/L at all times. [314 CMR 4.05(3) (a)(1)]

IV.B. Existing Regional/Local Targets

Several examples exist that establish protective water quality targets for either (1) local or regional
water bodies, or (2) specific water body types. These include the Total Maximum Daily Toadfor
Nutrients in the Tower Charles River Basin, Massachusetts (Charles River TMDL), Massachusetts' numeric
standards and impairment assessment criteria (Massachusetts Department of Environmental
Protection [DEP], 2016), and the criteria included in EPA's Quality Criteria for Water (1986);
informally known as the "Gold Book." These targets are summarized in Table IV-1.

40


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table IV-1. Existing Water Quality Targets

Parameter

Numeric Target

Target Duration/
Frequency

Source

Total
Phosphorus

<100 ng/L
(Free flowing rivers)

Instantaneous

Gold Book (1986)

<50 ng/L
(Entering
lakes/impoundments)

Instantaneous

<25 ng/L
(Exiting
lakes/impoundments)

Instantaneous

Dissolved
Oxygen

>5 mg/L

Instantaneous

MA Surface Water Quality Standards
[314 CMR 4.05(3)(a)(l)]

Dissolved
Oxygen,
Saturation

<125%

Instantaneous

Upper/Middle Charles TMDL (MA DEP,
2016)

Chlorophyll-a

<10 ng/L

Seasonal Average

Lower Charles TMDL (MA DEP & EPA,
2007) and Upper/Middle Charles
TMDL (MA DEP, 2016)

<18.9 ng/L

90th Percentile

IV. C. Reference Water Body Conditions

The reference water body condition method utilizes observations of water quality conditions (e.g.,
total phosphorus and chlorophyll-a concentrations) in water bodies with limited anthropogenic
impacts—or, at least, limited eutrophication—to establish "natural" or background nutrient
conditions in regional waters. In principle, reference condition targets should approximate the best
possible attainable water quality in the absence of human activity or if human impacts are entirely
controlled (Dodds and Oakes, 2004).

EPA suggests several methods for establishing reference nutrient conditions for a water body (Buck,
et al., 2000). Of those applicable to the Mystic River Watershed, the first requires the identification
of reference water bodies comparable to the Mystic River water bodies but which display limited or
no human influences on water quality. The 75th percentile water quality condition of the reference
water bodies is estimated, and this percentile is applied as a target in the water body of interest. The
second approach calculates the 25th percentile concentration of the general population of water
bodies—including water bodies with clear human impacts—to develop a target. Figure IV-I
conceptually illustrates these two approaches.

41


-------
My stic River Watershed TMDL Alternative Development — Final Report

1	1	1	1	1	1	1 t	1	1	1	1	1	!	1	1	1"

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80
Total phosphorus (jig/L)

Figure IV-I. Reference Condition Approach for Establishing Numeric Water

Quality Targets (EPA, 1998)

EPA has utilized this reference: approach to develop eco-region-based nutrient criteria for
application through the United States. In eco-region XIV sub-region 59 (EPA, 1998), where the
Mystic River is located? EPA was unable to identify suitable reference water bodies and,
consequently, based their eco-regional criteria on the 25th percentile: of the general population of
water bodies, To confirm that no new, appropriate reference water bodies have been identified
following the publication of the eco-regional criteria, the project team reviewed recent survey efforts
undertaken as part of EPA's National Aquatic Resources Survey. The project team was unable to
identify any water bodies that could serve as appropriate reference water bodies for the Mystic River.

Table IV-2 summarizes EPA's eco-region XIV sub-region 59 criteria and, for comparative purposes,
the 25th percentile parameter values for Upper Mystic Lake, the Upper Mystic River, and for the
entire watershed. Monitoring data for the water bodies of interest display 25th percentile
observational values ranging from 20 [Jg/L - 34 (Jg/L for total phosphors and 1.9 (Jg/L - 6.6 (Jg/L
for chlorophyll a.

42


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table IV-2. Eco-regional Criteria and Observed 25th Percentile (2000 - 2016)

Mystic River Watershed Values

Type/Location

Total
Phosphorus
(Hg/L)

Total Nitrogen
(mg/L)

Chlorophyll-a1
(Hg/L)

Secchi Depth
(meters)

Turbidity
(NTU)

Eco-region XIV, Sub-region 59

Rivers &
Streams

23.75

0.59

0.442

-

1.683

Mystic River Watershed 25th Percentile Observations (2000 - 2016)

Upper Mystic
Lake4

20

-

6.6

-

1.7

Upper Lobe of
Upper Mystic
Lake5

25

-

6.0

-

5.1

Mystic River
(Lower Basin)4

30

1.0

1.9

0.84

3.8

Total Mystic

River
Watershed4

34

1.0

2.0

0.70

3.9

1.	Fluorometric method

2.	Aggregated by subregion

3.	Average of reported and calculated values

4.	Aggregate of all monitoring station located in the water body

5.	Measured at monitoring station UPLUPL

IV.D. Stressor-Response Relationships

Water quality targets developed based on stressor-response relationships utilize empirical
relationships between nutrients and response variables (e.g., chlorophyll-a, excessive macrophyte
growth) to estimate protective numeric, water body-specific targets. This technique includes four
steps: (1) develop a conceptual model, (2) assemble and explore water quality data, (3) develop
statistical relationships between variables, and (4) derive protective targets based on those
relationships.

A benefit of the stressor-response approach is that nutrient targets are based on functional
relationships between nutrients and attainment of designated uses. This reduces risk associated with
developing excessively stringent targets, as can happen with reference condition approaches.
However, there is substantial risk that, after analysis, no reliable relationship will be discernable. This
can occur when the water body of interest is extremely impaired—resulting in a saturated response
signal—or when multiple confounding effects, which cannot be sufficiently controlled for,
contribute to the impairment.

For purposes of demonstration, the project team performed a preliminary stressor-response analysis
using Mystic River Watershed data described in Section III of the report. As a preliminary analysis,
the direct relationship was assessed between variables, while not controlling for covariate effects and
other confounding phenomenon (e.g., flow, phosphorus uptake by macrophytes, etc.). In addition,

43


-------
Mystic River Watershed TMDL Alternative Development — Final Report

the project team restricted the analysis to Upper Mystic Lake and the main stem of the Mystic River
in the Lower Watershed, and to total phosphorus concentration values as a stressor variable.

Table IV-3 presents the results of the relationship between total phosphorus and chlorophyll-a
concentrations in the upper lobe of Upper Mystic Lake, all sites within Upper Mystic Lake, and in
the Mystic River. Simple linear regressions were performed on paired concentration values (i.e., total
phosphorus and chlorophyll-a on a given day and monitoring location). Based on the resulting
relationship, a total predicted phosphorus concentration was calculated to produce a 10 (Jg/L
chlorophyll-a concentration (i.e., the seasonal average target for the Lower and Upper/Middle
Charles River TMDLs). Data was restricted to the period April — October when evidence of algae
growth was greatest. Predicted total phosphorus concentrations varied from 15 (Jg/L to 29 (Jg/L.
Regressions for the Mystic River and the total Upper Mystic Lake were not significant; however,
regression parameters for the upper lobe of the Upper Mystic Lake were significant (p < 0.05;

Figure IV-IV).

Table IV-3. Simple Linear Regression on Paired TP & Chlorophyll-a Monitoring

Results Collected April - October

Water body

Linear Regression Parameters

Total Phosphorus1
(Hg/L)

Slope

Intercept

r-squared

Upper Lobe of Upper Mystic Lake2

1.433

-27.89

0.43

26

Upper Mystic Lake

0.5872

1.092

0.37

15

Mystic River

0.1332

6.164

0.12

29

1.	Total phosphorus concentration implied by linear regression. Corresponds to a chlorophyll-a concentration of 10
Hg/L.

2.	Monitoring station UPLUPL.

























_J

7 40-

>ป

1-
CL
O



































O
J=

O

20-

o
o
O







•















* • • •



•







Q.

<

o-







	 •

•

•

•



























20



30 40
April-October Total Phosphorus (|ig/L)



50

Figure IV-II. April - October Total Phosphorus vs. Chlorophyll-a Concentrations in
the Upper Lobe of the Upper Mystic Lake (Monitoring Station UPLUPL)

Regressions based on seasonal averages in the Mystic River resulted in implied seasonal average total
phosphorus concentrations of approximately 33 and 41 (Jg/L to meet Charles River TMDL targets.
However, like the regressions in Table IV-3 for the Mystic River, these relationships were not

44


-------
Mystic River Watershed TMDL Alternative Development — Final Report

statistically significant. Table IV-4, Figure IV-III and Figure IV-IV show regressions of seasonal
average total phosphorus concentrations and chlorophyll-a as seasonal average and a seasonal 90th
percentile value, respectively (shaded area indicates 95% confidence interval). Data was aggregated
based on April — October measurements in each year. Upper Mystic Lake was not included in this
analysis since only 2 years of data were available.

Table IV-4. Simple Linear Regression on Average TP and Average or 90th

Percentile Chlorophyll-a1

Water body

Linear Regression Parameters

Chlorophyll-a
Target (ng/L)

Total
Phosphorus2
(Hg/L

Slope

Intercept

r-squared

Mystic River:

Seasonal Average Chlorophyll-a

0.16929

4.43453

0.14

10

33

Mystic River:

Seasonal 90th Percentile

Chlorophyll-a

0.3679

3.7722

0.19

18.9

41

1.	Averages and 90th percentile values aggregated by year for the period April - October.

2.	Concentration predicted by regression equation for Charles River TMDL target.

45


-------
Mystic River Watershed TMDL Alternative Development — Final Report

20-

30	40	50	60	70

April-October Average Total Phosphorus (ng/L)

Figure IV-III. April - October Average Total Phosphorus vs. Average Chlorophyll-a

Concentrations in the Mystic River

30	40	50	60	70

April-October Average Total Phosphorus (ng/L)

Figure IV-IV. April - October Average Total Phosphorus vs. 90th Percentile
Chlorophyll-a Concentrations in the Mystic River

IV.E. Mechanistic Models

Mechanistic models use systems of equations to represent ecological and hydrodynamic processes
within a water body. These can be used to predict changes in eutrophication-related processes in
response to changes in the level of water body enrichment. Like stressor-response models,
mechanistic water quality models can be used to develop nutrient targets based on functional
relationships and using site-specific empirical data. While stressor-response relationships are treated
like statistical "black box" processes, mechanistic models are largely deterministic models of system
behavior (U.S. EPA, 2001).

Mechanistic models have substantial input data requirements and would require a higher level of
effort relative to the options discussed above. Models of this type are not in widespread use for

46


-------
Mystic River Watershed TMDL Alternative Development — Final Report

developing modeling endpoints or protective nutrient targets, though they have been successfully
implemented in the past (Paul et al., 2011).

IV.F. Recommendations

The project team recommends the use of existing water quality targets designed for nearby water
bodies. While the Mystic River Watershed possesses some unique hydrologic features, the Charles
River shares many similar features (climate, hydrologic features) and the targets developed in its
TMDL are protective of the applicable Massachusetts water quality standards. Thus, the project
team decided to use the lOug/L target for chl-a from the Charles River TMDL and apply it to the
load reduction analysis for the Mystic River. During modeling and calibration, it was determined that
the lOug/L target was not easily applied to stormwater management BMPs. Instead, the team
decided to use a TP target of 30ug/L based on the Charles River TMDL, which references a TP
target of 30ug/L to achieve water quality conditions that correlate to lOug/L for chl-a.

A mechanistic model is not recommended due to high resource and data requirements. Stressor-
response relationships require the selection of a response variable target (e.g., macrophyte cover and
chlorophyll-a) and, consistent with the results of the preliminary analysis documented in this
memorandum, may not produce results of sufficiently robust statistical significance on which to base
the alternative TMDL analysis. A reference condition approach—particularly in the absence of
reliable reference water bodies—may prove to be unreliable, as the resultant target does not reflect a
functional relationship between nutrients and over-enrichment conditions.

47


-------
Mystic River Watershed TMDL Alternative Development — Final Report

V. Watershed Phosphorus Loading Estimates

Average annual phosphorus loading was estimated for the watershed area sub-basins tributary to the
freshwater portion of the Mystic River. These estimates will be used to estimate loading to critical
receiving water bodies that will be modeled using the Lake Loading Response Model (LLRM,
Wagner, 2009) and BATHTUB water quality models (please note, LLRM analysis is ongoing and
results are not presented in this report). Output from the receiving water model will then be used in
subsequent analyses to identify the reductions in phosphorus loads necessary to bring the watershed
into compliance with selected water quality targets (see Table IV-1 for TP and chl-a targets). In
addition, observed loading data corresponding to several USGS gage stations in the watershed have
been compiled for use in calibration of the land load estimates.

V.A. Stormwater Loads

Stormwater loads to the receiving water system are carried across the land surface by runoff during
most precipitation events. Regional pollutant loading export rates (PLERs) are commonly used as a
way to estimate annual phosphorus loads from various Hydrologic Response Units (HRUs), where
each HRU is comprised of a unique combination of land use, cover type (impervious or pervious),
and soil type. PLERs define the TP mass per unit area per time (i.e., lbs./acre/year) that is exported
from each type of HRU. HRUs are a common modeling method to categorize areas of land that
function similarly in terms of their hydrologic fluxes and pollutant loads.

To develop stormwater loads, EPA Region l's Opti-Tool modeling package was utilized. Opti-Tool
incorporates model generated time-series of hourly stormwater runoff volumes and nutrient runoff
quality that reflects pollutant build-up/wash-off processes and that has been calibrated to
stormwater quality and climatic data representative of the New England region. The Opti-Tool
package includes companion HRU Storm Water Management Models (SWMM) that were used to
dynamically simulate rainfall-runoff events for land-use based impervious and pervious HRU
categories that reflect land cover characteristics in the Mystic River Watershed. Using the SWMM
HRU models, Opti-Tool can be used to simulate runoff volumes and pollutant loads (e.g., annual
loads) for defined sub-basins within the watershed for any period of interest.

Opti-Tool includes SWMM HRU models for each HRU category that collectively are used to
represent key watershed characteristics in the Mystic watershed related to surface runoff. Individual
HRU models applied to the Mystic include land use specific for impervious cover (e.g., commercial
impervious) and land use specific for pervious cover and hydrological soil group (HSG) A, B, C, and
D (e.g., high density residential pervious HSG B). Hourly precipitation data representative of the
Mystic watershed and daily maximum and minimum temperatures are used as inputs to conduct
continuous HRU model simulations to generate annualized stormwater runoff volumes and loads.
The HRU model outputs produce land-use category specific PLERs for impervious cover and
pervious cover (according to HSG) and rainfall-runoff total (in/year). The unit PLER and flow can
be generated for any period of interest, including on an annual basis for each individual year, or as
overall average annual values for the period of record analyzed.

Using the resultant unit export rates, the annual stormwater load and/or flow for a given HRU
category within the watershed area of interest is generated by multiplying the total area represented
by the HRU type (e.g., commercial impervious) by the corresponding HRU specific PLER.

OptiTool loads and flows for each land use were used as input to the land loading models. That

48


-------
Mystic River Watershed TMDL Alternative Development — Final Report

loading model accumulated loads and flows and then attenuated the loads to provide values at the
outlet of each sub-basin. The basin values were used as the input to the BATHTUB model.

Table V-l presents the climatological datasets used as inputs to the Opti-Tool model to generate
annual total stormwater phosphorus loads and flows.

Table V-1. Opti-Tool-SWMM HRU Model Climatological Input Datasets

Type

Description

Source

Precipitation

Hourly time series of precipitation values at
Boston Gen E Logan International Airport
(WSAF-WBAN ID 725090 14739) from 1992 -
2016

https://gis.ncdc.noaa.gov/maps/ncei/cdo
/hourly

Temperature

Daily time series of maximum and minimum
temperature values at Boston Gen E Logan
International Airport (WSAF-WBAN ID 725090
14739) from 1992 - 2016

https://gis.ncdc.noaa.gov/maps/ncei/cdo
/hourly

The delineation of sub-basins for quantifying flows and loads through HRU accounting within the
freshwater portion of the Mystic River Watershed was initially based on an existing sub-basin
delineation available from the Massachusetts Office of Geographic Information (MassGIS).
Modifications and refinements were made to further sub-divide several basins to better reflect
watershed routing processes and more closely align with critical water body assessment locations.
Some of these modifications were made using a digital elevation model and delineation for quality
control. Further modifications to several impaired lake basins were made based on consultation with
MassDEP staff and a review of previously developed procedurally generated delineations for the
lake basins. Ultimately, all sub-basin delineations were visually compared to FEMA derived flood
plain delineations as an added quality control measure.

Land-use categories within each sub-basin were delineated based on the intersection of three GIS
layers: land use, impervious cover, and soil type. Table V-2 lists the land-use categories used in Opti-
Tool and provides a crosswalk with the GIS layer categories.

49


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table V-2. Land-Use Categories

Opti-Tool Land-Use

MassGIS ID Code

MassGIS Description

Agriculture

1

Cropland

Agriculture

2

Pasture

Agriculture

23

Cranberry Bog

Agriculture

26

Golf Course

Agriculture

36

Nursery

Commercial

15

Commercial

Commercial

29

Marina

Commercial

31

Urban Public/Institutional

Forest

3

Forest

Forest

4

Non-Forested Wetland

Forest

35

Orchard

Forest

37

Forested Wetland

Forest

40

Brushland/Successional

High Density Residential

10

Multi-Family Residential

High Density Residential

11

High Density Residential

Highway

18

Transportation

Industrial

5

Mining

Industrial

16

Industrial

Industrial

19

Waste Disposal

Industrial

39

Junkyard

Low Density Residential

13

Low Density Residential

Low Density Residential

38

Very Low Density Residential

Medium Density Residential

12

Medium Density Residential

Open Land

6

Open Land

Open Land

7

Participation Recreation

Open Land

8

Spectator Recreation

Open Land

9

Water-Based Recreation

Open Land

17

Transitional

Open Land

24

Powerline/Utility

Open Land

25

Saltwater Sandy Beach

Open Land

34

Cemetery

Water

14

Saltwater Wetland

Water

20

Water

The total impervious area (TIA) associated with each Opti-Tool land-use category was adjusted to
directly connected impervious area (DCIA) or effective impervious area using the following formula
(refer to Table V-3 for adjustment factors):

DCIA = C x TIA

50


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Unconnected impervious area (i.e., TIA minus DCIA) was then distributed proportionally among
the pervious soil types within the land-use (e.g., if 70 percent of the soil was assigned as type A soils,
then 70 percent of the unconnected impervious area was re-assigned to type A soils). Table V-3
shows DCIA adjustment factors from MA MS4 Permit.

Table V-3. DCIA Adjustment Factors from MA MS4 Permit

Opti-Tool Land-Use

Adjustment Factor (C)

Agriculture

0.004

Commercial

0.570

Forest

0.001

High Density Residential

0.360

Highway

0.440

Industrial

0.670

Low Density Residential

0.110

Medium Density Residential

0.160

Open Land

0.080

Table V-4 integrates the land-use categories with soil and impervious cover categories. Table V-5
lists the data sources for the layers discussed above, all of which were obtained from the MassGIS.

Table V-4. Opti-Tool Export Rates by HRU

Opti-Tool HRU

Average Annual

PLER
(Ibs./acre/year)1

Average Annual
Rainfall-Runoff Rate
(in/year)

Average Annual
Flow-weighted TP
SW concentration
(mg/L)

Agriculture Impervious

1.49

39.51

0.17

Forest Impervious

1.49

39.51

0.17

Highway Impervious

1.38

39.51

0.15

Industrial Impervious

1.79

39.51

0.20

Commercial Impervious

1.79

39.51

0.20

High Density Residential Impervious

2.36

39.51

0.26

Medium Density Residential
Impervious

1.95

39.51

0.22

Low Density Residential Impervious

1.49

39.51

0.17

Open Land Impervious

1.49

39.51

0.17

Agriculture Pervious

0.44

2.50

0.78

Forest Pervious

0.11

2.50

0.19

Developed2 Pervious A

0.03

0.46

0.29

Developed2 Pervious B

0.11

2.50

0.19

Developed2 Pervious C3

0.21

5.64

0.16

Developed2 Pervious C/D

0.30

7.54

0.18

Developed2 Pervious D

0.37

10.30

0.16

1. Based on simulations spanning 1992 — 2017.

51


-------
Mystic River Watershed TMDL Alternative Development — Final Report

2.	Developed Pervious land categories include commercial, industrial, residential, highway, and open land (see Table 1-
2 of Attachment F to 2016 MA MS4 General Permit).

3.	Areas with undefined or unknown soil types were assumed to be soil type C.

Table V-5. GIS Layers Used to Develop HRUs

Type

Description

Source

Land Use (2005)

The Land Use (2005) data layer is a
Massachusetts statewide seamless digital
dataset of land cover/land use, created
using semi-automated methods, and based
on 0.5-meter resolution digital ortho
imagery captured in April 2005.

http://www.mass.gOv/a nf/research-and-
tech/it-serv-and-support/application-
serv/office-of-geographic-information-
massgis/datalayers/lus2005.html

Impervious Cover
(2005)

The Impervious Surface raster layer
represents impervious surfaces covering
the Commonwealth of Massachusetts. The
surfaces were acquired in April 2005 as
part of the Color Ortho Imagery project.

http://www.mass.gOv/a nf/research-and-
tech/it-serv-and-support/application-
serv/office-of-geographic-information-
massgis/datalayers/im pervioussurface.html

NRCS SSURGO-
Certified Soils
(2014)

The SSURGO-certified soils dataset is
generally the most detailed level of soil
geographic data developed by the National
Cooperative Soil Survey. The data include a
detailed, field-verified inventory of soils
and miscellaneous areas that normally
occur in a repeatable pattern on the
landscape and that can be cartographically
shown at the scale mapped.

http://landscapeteam.maps.arcgis.com/ap

ps/SimpleViewer/index.html?appid=4dbfec

c52fl442eeb368c435251591ec

Drainage Sub-
Basins (2007)

MassGIS has produced a statewide digital
data layer of the approximately 2,300 sub-
basins as defined and used by the USGS
Water Resources Division and the
Massachusetts Water Resources
Commission and as modified by Executive
Office of Environmental Affairs (EOEA)
agencies.

http://www.mass.gOv/a nf/research-and-
tech/it-serv-and-support/application-
serv/office-of-geographic-information-
massgis/datalayers/subbas.html

Elevation
(Topographic)
Data (2005)

These data represent the "bare earth"
elevation of the terrain surface without
vegetation and artificial features. As a
requirement for the orthorectification
process of the 1:5,000 Color Ortho Imagery
(2005), elevation points were compiled
photogrammetrically by human operators
from imagery acquired by Sanborn, Inc. in
April 2005.

http://www.mass.gOv/a nf/research-and-
tech/it-serv-and-support/application-
serv/office-of-geographic-information-
massgis/datalayers/elev2005.html

Sub-basins delineated within the Mystic River Watershed were based on the connectivity to four
critical reaches and seven ponds. The four critical reaches chosen were modeled using the receiving
water model while ponds selected were those identified for future TMDL modeling using the LLRM
model.

52


-------
Mystic River Watershed TMDL Alternative Development — Final Report

The three critical reaches are:

•	Upper Lobe of Upper Mystic Lake.

•	The Main-body of Upper Mystic Lake.

•	Upper Mystic Basin

•	Lower Mystic Basin.

The major lakes/ponds are:

•	Blacks Nook Pond (MA71005), Cambridge.

•	Horn Pond (MA71019), Woburn.

•	Judkins Pond (MA71021), Winchester.

•	Mill Pond (MA71031), Winchester.

•	Spy Pond (MA71040), Arlington.

•	Wedge Pond (MA71045), Winchester.

•	Winter Pond (MA71047), Winchester.

The sub-basin delineation within the freshwater portion of the Mystic is presented in Figure V-l
(note: the stippled area denotes the area served by a combined sewer). Figure V-2 shows the
expected routing scheme through the sub-basins in the Mystic River Watershed from the headwaters
to the outlet. The delineation allows modelers to characterize flow and pollutant loading to each of
the three critical reaches, seven major lakes/ponds, and at mainstream segments (the Aberjona
River, the Maiden River, and Alewife Brook).

Sub-basins were named for the pond or water body to which they drained (e.g., the upper reaches of
the Aberjona River drainage area were labeled the Judkins Pond sub-basin since this is the water
body located at the bottom of the drainage area) or to main stream segment in which they
encompass. The exceptions to this naming convention are the Aberjona River 1 and Aberjona River
2 sub-basins which are split around the USGS streamflow gauge for the river.

Estimated annual stormwater (SW) phosphorus loads and runoff volumes (1992 — 2017) for the
entire Mystic River watershed are presented in Figure V-I. Details for the three critical water quality
segments and seven impaired ponds within the watershed are presented in Appendix E.

53


-------
My stic River Watershed TMDL Alternative Development — Final Report

>
c

1

u

CO
O)

5

o
c

cc:

1,SQO,000
1,600,000
1,400,000
1,200,COO
1,000,000
BOO,000
600,0G0
400,COO
200,000
0

1992 1994 1996 1998 20CO 2002 2004 2006 200B 2010 2012 2014 2016
5W	GW ~ota)

•Tl CT| rr. Ci CTl en CTl CTl OOOOOOOOOO^HtHrHt-frHr-J^lrH

aa?)99aaaRRRRRRRRRRRRRRRRRR

ฆ 5W

. GW

rota I

Figure V-l. Estimated Annual Runoff Volume and TP Load for the Mystic River

54


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Upper Mystic/
Aberjona River

Upper Lobe of
Mystic Lake

Upper Mystic
Lake

Alewife Brook

Lower Mystic
Lake

Subbasin

Subbasin

Subbasin
(includes Mill R)

Subbasin

Maiden River

Upper Mystic
Basin

Lower Mystic
Basin

Subbasin*

Tributary*

Massachusetts
Bay

Critical

Modeled



Mill



River

Lower Mystic
Lake

Basin

\

Upper Mystic
Lake

Upper Lobe of
Mystic Lake

Subbasin

Modeled

		

Critical

•Subbasin loads are unattenuated while tributary loads are attenuated by the river network

Figure V-II. Mystic River Watershed Sub-Basin Delineation and Schematic

Diagram for Final Model

55


-------
Mystic River Watershed TMDL Alternative Development — Final Report

V.B. Groundwater Loads

Groundwater loads to the river system are those that result from water that infiltrates through the
soil and moves via groundwater flow through the underlying aquifer. In general, phosphorus
movement is retarded through soils and aquifers by years or decades because it is highly adsorbed by
clay particles in either media.

Groundwater flow was estimated by analysis of available streamflow records and separating these
components into stormflow and baseflow (e.g., USGS, 1996; Arnold and Allen, 1999, Arnold, et al.,
2005). Baseflow approximates groundwater flow assuming that the riparian evapotranspiration is
minimal.

Streamflow data suitable for estimating stormflow and baseflow components is available at the
Aberjona River and Alewife Brook USGS gages (Figure V-III). Using the Soil & Water Assessment
Tool (SWAT) Baseflow (Bflow) program, baseflow was estimated at both locations and annual
baseflow / total streamflow fractions were computed (Appendix F),. Figure V-III and Figure V-IV
display baseflow and total streamflow for the Aberjona River and Alewife Brook for 2016.

Streamflow		Baseflow

u
4)

ro

160
140
120
100
80
60
40
20

0 	~~	

01/2016 02/2016 04/2016 05/2016 07/2016 09/2016 10/2016 12/2016

Figure V-III. Baseflow Estimates and Streamflow Measurements at Aberjona River

for 2016 (USGS Gage 01102500)

56


-------
My stic River Watershed TMDL Alternative Development — Final Report

50
45
40

0 	

01/2016 02/2016 04/2016 05/2016 07/2016 09/2016 10/2016 12/2016

Streamflow		Baseflow

Figure V-IV. Baseflow Estimates and Streamflow Measurements at Alewife Brook

for 2016 (USGS Gage 01103025)

From 1992 — 2017 in the Aberjona River, the average annual baseflow contribution to total
streamflow was approximately 65 percent. From 2006 — 2016, the average in Alewife Brook was
approximately 70 percent. For each HRU in these basins, baseflow is assumed to be an equivalent
fraction,. For HRU's not in either basin, an average (68 percent) of the two estimated values were
applied. Annual groundwater flow contributions were computed for each HRU using the following
equation:

Groundwater Flow = (f x Stormwater Flow) / (1 — f)

Where "Stormwater Flow" was the HRU rainfall-runoff total estimated using the Opti-Tool model
and f is the groundwater fraction of total streamflow.

Groundwater phosphorus concentration can be estimated from well sampling data; however, such
data is unavailable. It is difficult to estimate different groundwater concentrations for each HRU
without a well sampling effort devoted to that end. A well sampling effort undertaken in the Boston
metropolitan area found an average region TP concentration of 0.008 mg/L (Flanagan, et al., 2001).
Based on this study, a groundwater TP concentration of 0.008 mg/L was assumed for all basins.

Using groundwater concentration (i.e., 0.008 mg/L) and baseflow for each sub-basin, the
groundwater phosphorus load from the sub-basin was calculated as the product of the two inputs.

Estimated annual groundwater (GW) phosphorus loads and runoff volumes (1992 — 2017) for the
entire Mystic River watershed were previously presented in Figure V-I. Details for the three critical
water quality segments and seven impaired ponds within the watershed are presented in Appendix
E.

V.C. Sediment Loads

Ponds and impoundments typically act as sinks for nutrients, but they can become net sources to
downstream waters when internal nutrient stores are mobilized and exported (Powers et al., 2015).
In the Mystic River Watershed, there have.: been no direct measurements of nutrient release: rates
from the sediments, which represent a major data gap in this watershed.

57


-------
Mystic River Watershed TMDL Alternative Development — Final Report

In the Upper/Middle Charles Nutrient TMDL (EPA/DEP, 2007), which used measured nutrient
release rates from sediments, EPA/DEP estimated that about 22 percent of upstream land loads
were retained on net in sediments. As there have been no direct measurements of nutrient release
rates from the sediments in the Mystic River Watershed, a net attenuation factor will be assumed
through the watershed based on a reach detention time. The initial attenuation factor will be used as
a calibration parameter (see Section VII for discussion of attenuation).

V.D. Observed Loads

The estimation of observed loads was made at sites that have concurrent flows and water quality
data to allow the most accurate estimation of the annual phosphorus load. These sites are the USGS
gages at the Aberjona River (1939-2017), Alewife Brook (2005-2017), and Mystic River (2015-2017),
which all have reliable daily flow records.

V.D.1. Adjustments of Streamflow

Data gaps on the order of several days exist in the streamflow records for the Aberjona River and
for Alewife Brook. In order to develop flow estimates for missing days, a series of regression
relationships were developed between the USGS sites (Figure V-V and Figure V-VI). These
regressions were used to fill missing days in records at the two sites and to develop modeled
estimates of daily average flow rates at the Alewife Brook and Mystic River for dates outside their
respective periods of record.

October 2005 - December 2017

250
200
150
100
50

y-0.1494X+3.8475









.

R1

-0.7653



•





• **

•



•

•

•

•

•

.*>ฆ***

•

•





..

•• •
kl'/

•

•

•*** •





•







•











200 400 600 800 1000 1200 1400 1600
Aberjona Flow, CFS

Figure V-V. Regression of Flow Measurements at Aberjona River (01102500) and

Alewife Brook (01103025) USGS Stations

58


-------
Mystic River Watershed TMDL Alternative Development — Final Report

June 2016 - December 2017

350
300
250

E 200

1 ISO

u

1. 100

50
0

y-0.9545x+11.264







....

R1

-0.8412

















•

• • #











• 1
li







•



• • •









ฆ M























50 100 150 200 250 300 350
Aberjona Flow, CFS

Figure V-VI. Regression of Flow Measurements at Aberjona River (01102500) and

Alewife Brook (01103025) USGS Stations

V.D.2. Adjustments of Total Phosphorus Concentrations

Differences in the TP concentrations from different laboratories for the same water body were
known by MyRWA prior to commencing this study in early 2017. These differences were previously
discussed in the memo "Dataset Assessment for Development and Calibration of the Mystic River Watershed
Loading and Receiving Water Quality Models" (Walker, September 15, 2017).

The TP differences had to be reconciled prior to proceeding with the Alternative TMDL analysis.
The options for reconciliation were: (1) choosing one of the two data sources as the preferred data,
or (2) finding a way to convert from one source to another. Option 2 was preferred since it would
allow more instances where TP and chl-a were both measured at the same site, which is a
requirement for choosing a suitable model calibration period.

To reconcile the data, we went back to the original data in the memo by EPA and MyRWA entitled
"Mystic Raver TP Laboratory Split-Study Results and Discussion (Hrycyna, September 7, 2017). They
reported TP values measured using EPA-approved methods from the two laboratories: MWRA
Deer Island (MWRA) and EPA Region 1 in Chelmsford (EPA). TP results for side-by-side field
samples in the Mystic River Watershed were systematically higher at MWRA (Method 365.4) than
EPA (Method 365.1). Consultation with other laboratory experts in water quality analysis led to the
conclusion that the differences between labs for field samples is a result of better
conversion/digestion of TP to orthophosphate for Method 365.4 versus 365.1. Because Method
365.1 was considered to be have an incomplete conversion to orthophosphate, it is likely an
underestimate of TP.

For this study, Method 365.4 was considered to be "true" value of TP. Although there were some
seasonal differences between April and December in the relationship between the two methods, we
opted to use a single equation for all the data (see Figure V-VII). This approach results in a lower
slope than Walker's (2017) equation because only the April data were available for that analysis. The
final equation to convert TP data using Method 365.1 to Method 365.4 was:

Y = 1.15 X + 21.7

59


-------
Mystic River Watershed TMDL Alternative Development — Final Report

where:

Y = MWRA TP concentrations using method 365.4
X = EPA TP concentrations using method 365.1

Biweekly and monthly sampling of phosphorus data at the Aberjona River Mystic River Watershed
Association (MyRWA) monitoring site (ABR006), Mystic River site (MYR071), and Alewife Brook
(ALB003, ALB006) were linearly interpolated on a daily basis in order to produce an estimated daily
time series of TP concentrations (Figure V-VII through Figure V-X).

140

M 120

3

in 100
to
00

T3

O 80

+-ป

(U

CuO

_c

*!7)

3

CL 40
I-

<
cc

^ 20
0

0	20	40	60	80	100

EPA TP using Method 365.1 (|ig/L)

Figure V-VII. Correction of TP Values from Method 365.1 to 365.4

___ Aberjona River

	i

"m 600

5 500

0

ro 400
+-ป

ฃ 300

1	200 .









1

12/1999 09/2002 05/2005 02/2008 11/2010 08/2013

05/2016

Figure V-VIII. Adjusted Total Phosphorus Concentrations in the Aberjona River

60

























~~

~ >/y

= 1.1543x + 21.7





/ #

R2 = 0.8582



~





4

~


















-------
Mystic River Watershed TMDL Alternative Development — Final Report

Mystic River

^ 200
o 150

H 12/1999 09/2002 05/2005 02/2008 11/2010 08/2013 05/2016

Figure V-IX. Adjusted Total Phosphorus Concentrations in the Mystic River

__	Alewife Brook

_i

400

H 12/1999 09/2002 05/2005 02/2008 11/2010 08/2013 05/2016

Figure V-X. Adjusted Total Phosphorus Concentrations in Alewife Brook.
V.D.3. Observed Total Phosphorus Loads

Observed loads for TP were calculated for the river reaches where USGS flow gages and
measurements of nutrient concentrations coincide. The measured TP sites and flow measurements
were both available at USGS flow gages, namely, the Aberjona River (USGS 1102500, ABR006),
Lower Mystic Lake (USGS 1103010, MYR071), and Alewife Brook (USGS 1103025,
ALB003/ALB006) gages. At each of the three gages, estimates of concurrent streamflow and total
phosphorus concentration were used to calculate the daily total phosphorus load (daily flow x daily
concentration) and was summed to give a time series of annual loads.

61


-------
Mystic River Watershed TMDL Alternative Development — Final Report

V.D.4. Calibrated Streamflow

A comparison of land-based and observed flows was performed for the two-gauge sites that have
long-term flow data (Aberjona River, Alewife Brook). The analysis of 1992-2017 daily flow data
revealed a close match to the observed average flow at Aberjona River but an underestimate of
observed average flow at Alewife Brook. The average observed flow at Aberjona River was
reasonable (21.3 in/yr.) since it compares well with typical flows for eastern Massachusetts rivers
(20-25 in/yr.). In contrast, the average observed flow at Alewife Brook was very low (15.3 in/yr.).
We have no explanation for this apparent low average flow at Alewife Brook but decided not to use
it. Comparison of average annual streamflows from land-based estimates and observed values at the
Aberjona River gauge were reasonable (R2=0.72) with no visible bias. Given the uncertainty of the
Alewife Brook flow data and the decent fit of modeled land-based loads with observed flows for the
Aberjona gauge, we decided not to calibrate streamflow for this study.

V.D.5. Calibrated Total Phosphorus Loads

Observed loads within a reach are often less than the land loads because there has been some
attenuation (or retention) of nutrients within the river system. Attenuation of nutrients in river
reaches occurs because of biological and chemical changes, plant uptake, particulate settling, and
organic settling from algae or aquatic plant senescence. Higher residence time, or detention time,
usually means more nutrient attenuation occurs.

Estimation of attenuated land loads was performed for the period 2007-2016 since this is the
modeling period (see Section VII.C.3). Attenuated cumulative TP loads for each reach were initially
determined from the modeled land values by using an estimate of instream attenuation factors based
on reach detention time (volume/flow, days). These estimates were calibrated to match observed
reach loads (see below).

A review of the literature on phosphorus attenuation in impoundments revealed two similar curves;
Figure 7 of Powers et al. (2015) and Figure 4 of Koiv et al. (2011). Both of these curves are shown in
Figure V-XI. The equation derived from the Powers et al. (2015) data, which indicates slightly higher
attenuation rates than Koiv et al. (2011) was chosen for the Mystic watershed because the calibration
process indicated that high attenuation factors were needed. The Powers equation for the
attenuation factor is:

Attenuation Factor = 0.5598 + 0.1278+loge (Rd/365), Rd = detention time (days)

62


-------
Mystic River Watershed TMDL Alternative Development — Final Report

1.0
0.9
0.8

200	400	600

Detention Time (d)

• Koiv (2011) —•- Powers (2015)

800

1000

Figure V-XI. Phosphorus Attenuation Curves

Detention time (days) was computed from the estimated reach volume (m3) and the modeled annual
flow rate (m3/d). Impoundment volumes were estimated from various pond studies (WH, 1987,
1988; DFW, no date; ENSR, 2000; DEP, 2010; and EPA, 2018). For river reaches (i.e. not pond
reaches) that did not have information to compute an attenuation factor, the value was set to a
nominal value of 0.05 (see Table VII-4).

Predicted TP land loads were calibrated to measured loads at three calibration sites (Aberjona,

Lower Mystic, and Alewife USGS gauges) by changing the reach attenuation factors. The calibration
objective was to minimize the error between observed and attenuated TP loads as closely as possible
at the three sites while still maintaining reasonable attenuation coefficients (0.05 to 0.9) that were
also reasonably close to the initial estimates. The process was conducted sequentially down the
watershed, starting at the most upstream calibration point first. All initial attenuation factors above
this point were adjusted by the same factor to best match the observed data. This process was
repeated downstream. Attenuation in all the reaches contributing to Alewife Brook was increased
further to meet the observed load for that calibration site.

With this process, the investigators were able to reduce the error in the annual average loads to zero
at the three sites and still have reasonable attenuation factors, although the ones in the Alewife
Brook watershed seem a little high, possibly because the flow at this gauge is unreasonably low (see
Section V.D.1X). The error at all sites except the three gauges is unknown because there were no TP
load estimates at these other sites. The estimated and final attenuation factors for TP are given in
Table V-6 and a comparison of modeled land versus attenuated sub-basin loads is given in Figure
V-XII. There might also be some monthly error at the two-gauge sites, but this is not relevant to the
receiving water model which is based on annual load inputs.

63


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table V-6. Estimated Reach Detention Times and Attenuation Factors

Reach Name

Reach Volume
(m3)

Cumulative
Reach Flow

(mWr.)1

Detention Time
(d)

Estimated
Attenuation
Factor (-)

Final TP
Attenuation
Factor (-)

Winter
Pond

58,768

254,236

84

0.37

0.70

Horn
Pond

2,951,250

10,898,619

99

0.39

0.85

Wedge
Pond

277,555

11,968,307

8

0.08

0.20

Judkins
Pond

9,065

37,176,274

0

0.05

0.05

Mill
Pond

6,475

37,395,435

0

0.05

0.05

Aberjona
River 1



37,704,020



0.05

0.05

Calibration Site is USGS Streamflow Gauge at Aberjona River (1102500



Aberjona
River 2



38,177,963



0.05

0.05

Upper
Lobe

219,434

38,368,707

2

0.05

0.05

Upper Mystic
Lake

7,385,437

40,449,476

67

0.34

0.26

Lower Mystic
Lake

3,529,612

49,102,911

26

0.22

0.19

Calibration Site is USGS Streamflow Gauge at Lower Mystic Lake (1103010)

Blacks Nook
Pond

15,110

5,969

924

0.68

0.90

Spy
Pond

1,690,000

2,110,121

292

0.53

0.80

Alewife Brook

-

17,560,488

-

0.05

0.48

Calibration Site is USGS Streamflow Gauge at Alewife Brook (1103025)

1 From land loading model

64


-------
Mystic River Watershed TMDL Alternative Development — Final Report

9,000

8,000

^ 7,000

— 6,000
ฆa

o 5,000

ฆ Land Attenuated

Figure V-XII. Modeled Land vs. Attenuated Reach Phosphorus Loads (2007-2016)

65


-------
Mystic River Watershed TMDL Alternative Development — Final Report

VI. Evaluation of Combined Sewer Overflow and Sanitary
Sewer Overflow Data for the Mystic River Watershed

The following describes the data types, sources and approaches used to develop estimates of CSO
and SSO volumes and phosphorous loads in the Mystic River Watershed. CSO and SSO data were
quantified in an effort to provide a more accurate estimate of all loads entering the waterbodies
within the watershed. The land loads provided in Section V will be added to the CSO and SSO
loads and modeled in Section VII. As discussed in Section V, annual loads are the expected input
for the receiving water models; therefore, load estimates have been developed on an annual basis.

VI.A. Data Types and Sources

Four primary data types were required to evaluate CSO and SSO contributions:

•	Spatial data (CSO and SSO drainage areas).

•	Volumetric data (annual CSO and SSO discharge volumes).

•	Annual precipitation data.

•	CSO and SSO discharge concentrations for total phosphorus and total nitrogen.

The CSO and SSO data used for this analysis included GIS data, Excel spreadsheets, reports and
literature. The sources of these data are noted in the sections below.

VI.A. 1. Spatial Data

There are two sub watersheds in the Mystic River Watershed that contain CSO drainage areas:
Alewife Brook and Mystic River. The outfalls included in these analyses are (location in parenthesis):

•	Alewife Brook: CAM001 (Cambridge), CAM002 (Cambridge), MWR003 (Cambridge),
CAM004 (Cambridge), CAM400 (Cambridge), CAM401A (Cambridge), CAM401B
(Cambridge), SOMOOIA (Somerville).

•	Mystic River: SOM007A/MWR205A (Somerville). MWR205 (Somerville), which is located
downstream of the Amelia Earhart dam, was evaluated to compare total discharges from the
Somerville Marginal CSO Facility.

The MWRA CSO map is presented in Appendix G (Figure G-XI-1). The data collected for the CSO
drainage basins included a GIS polygon file provided by the City of Cambridge showing areas
contributing to Alewife Brook with a corresponding attribute table noting the year those areas were
separated. Maps showing the City of Cambridge's CSO drainage basins from 2000 and 2017 are
presented in Appendix G (Figure G-XI-2 and Figure G-XI-3). Additional GIS data on CSO
drainage areas were provided by the City of Somerville in Appendix G (Figure G-XI-4). Portions of
Somerville's CSO drainage areas discharge to Alewife Brook while others discharge to the lower
Mystic River directly, depending on the size of the storm event (see attached map).

There were no GIS data available for SSO drainage areas. A spreadsheet from the Massachusetts
Water Resources Authority (MWRA) provided estimated volumetric discharge data and latitude and
longitude for each SSO. Those locations were converted into a GIS point shapefile that was overlaid
onto the Mystic River sub watersheds to assign SSOs volumes to their recipient sub watersheds.

66


-------
Mystic River Watershed TMDL Alternative Development — Final Report

VI.A.2. Volumetric Data

The data sources for volumetric data gathered are shown in Table VI-1.

Table VI-1. Volumetric Data Sources for CSOs and SSOs

Data Type

Source

Years

Data Comments

CSO data

MWRA

2000 to 2017

Modeled data; from Annual Reporting



City of Cambridge

2006 to 2017

Annual NPDES Reports



City of Somerville

2016

Annual NPDES Report

SSO data

MWRA

2000 to 2017

Data from MWRA directly; no SSOs for
2016 or 2017. Includes
latitudinal/longitudinal information.



Massachusetts Department of
Environmental Protection
(MassDEP)

2000 to 2017

Data between 2000-2016 received from
Mystic River Watershed Association;
spreadsheet format. No spatial data
provided. Data includes all known
discharges within the Mystic River
Watershed (including MWRA data).
2017 data were digitized from forms
provided by MassDEP for this project.

The data were reviewed to determine completeness (e.g., data gaps, including missing values or no
data) and for consistency (e.g., extreme data). For CSOs, data provided by the cities were also cross
checked with the MWRA data. In the annual CSO reports, modeled CSO activation durations and
volumes were compared to the reported CSO activations. In some instances, the reported data were
not consistent with the modeled data (model was either over or under predicting), which appeared
to be due to either the modeling of the CSO system/outfall or the resolution of the metered data
versus modeled data. Based on the annual reports, improvements to the model appeared to have
been made over time to address some of these issues.

For SSOs, MWRA and MassDEP noted that the reporting of SSO volumes was updated in 2012.
Prior to 2012 the form only included SSO volume ranges, which generally were: <10,000 gallons,
10,000 to 100,000 gallons, 100,000 gallons to 1,000,000 gallons and > 1,000,000 gallons, limiting the
upper limit of reported SSO volumes. Starting in 2012, the form was updated to allow an estimate of
SSO volume and method for estimating.

VI.A.3. Precipitation Data

Hourly precipitation data for Boston Logan International Airport was extracted from the Opti-Tool
model for 1992 to 2016. Additional hourly data for 2017 was downloaded from NOAA's website
(NOAA, 2018).

VI.A.4. Nutrient Concentrations

Nutrient concentrations for CSO and SSO discharges were estimated to facilitate the calculation of
annual phosphorus and nitrogen loads. TP and total nitrogen (TN) concentrations for CSOs were
based on data from Breault et. al. (2012). The document reported average CSO TP and TN
concentrations of 3.1 mg/L and 9.3 mg/L, respectively, for samples collected by MWRA. SSO TP
and TN concentrations were based on the average annual influent wastewater concentrations for
2016 sampled at the Deer Island Sewage Treatment Plant of 5.23 mg/L and 41.8 mg/L,

67


-------
Mystic River Watershed TMDL Alternative Development — Final Report

respectively.6 The analysis used in this memorandum assumes that that SSOs discharges have similar
concentrations to untreated wastewater.

VI.B. CSO Data Analyses

After discussions with MWRA and EPA Region 1, the MWRA CSO data were further analyzed to
evaluate how precipitation and time influenced the CSO data and how average CSO volumes
compared to the annual data. Two data analyses were performed on the datasets: regression analyses
and evaluation of statistical outliers. The purpose was to determine if the average annual CSO
volume for the evaluation of phosphorus load reduction estimates (Section IX) was appropriate
and/or if modifications to the annual CSO volumes would allow a more representative average. In
particular the analysis focused on years where no CSOs were reported (modeled discharge estimates
noted as '0') and CSO data extremes.

VI.B.1. CSO Analyses

CSO volumetric data were plotted against time and annual precipitation depth to evaluate potential
relationships. The volumetric data were also normalized by acreage of CSO drainage basins
contributing to the CSO outfalls. Figure VI-I through Figure VI-IV represent these comparisons.
There were significant linear relationships with time (years) and annual rainfall for the Alewife CSO
drainage basin (p-value < 0.05). There were no similar significant relationships for the Mystic River
CSO drainage basin, which might be due to the complex connection of Somerville's CSO system to
Alewife Brook, the Mystic River, and other sewer systems.

VI.B.2. CSO Statistical Outliers

A statistical analysis was conducted on each of the CSO datasets to determine if there were any
outliers that could be having excessive influence on the volumetric averages. This process first
involved defining the upper bounds of the annual volumetric data to identify the outliers using the
following two equations:

Upper Bound: Q3 + (1.5*IQK)

where Q3 is the third quartile value and IQR is the interquartile range or the difference between the
first and third quartiles (Q1 and Q3). Outliers were defined as values higher than the upper bound.
The results are presented in Table VI-2. The mean of the datasets was also calculated.

6 SSO TP and TN concentrations were based on MWRA's North System influent only, which includes communities
in the Mystic Watershed.

68


-------
My stic River Watershed TMDL Alternative Development — Final Report

90,000,000

80,000,000

70,000,000

10,000,000

CSO Volumes in the Alewife and Mystic River basins

~ Mystic River
ฆ Alewife

—	Linear (Mystic River)

—	Linear (Alewife)

2004 2006 2008 2010 2012 2014 2016 2018

Year

Figure VI-I. Annual CSO Volumes over Time

CSO Volumes in the Alewife and Mystic River basins

90,000,000

80,000,000

70,000,000

= 50,000,000
ru

3B

E 40,000,000
o

O 30,000,000

to

u

20,000,000
10,000,000

y = 3E+06x- 1E+08
R2= 0.6355

~ Mystic River
ฆ Alewife

—	Linear (Mystic River)

—	Linear (Alewife)

20	30	40

Rainfall (inches/year)

Figure Vl-ll. Annual CSO Volumes versus Annual Rainfall

69


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Alewife CSO Volumes by Area (gal/ac)

>

y
ro

1
00

01
Q.
HI
ฃ
_D
O
>
O

i/i

u

60,000
50,000
40,000
30,000
20,000
10,000

ฆ Alewife CSO

	Linear (Alewife CSO)





ฆ

ฆ _







ฆ



y = 1722.4x -57102





RJ = 0.6337







ฆ .







ฆ

ฆ
ฆ



10

20

30

Rainfall (in/yr)

40

50

60

Figure Vl-lll. Normalized Alewife Annual CSO Volumes versus Annual

Precipitation

Mystic CSO Volumes by Area (gal/ac)

30,000
.> 25,000

u
nj

"rc 20,000

ฆSB

S 15,000

g. 10,000
5,000

Q.

cu

E

o
>

O

to
VJ

~ Mystic CSO

Linear (Mystic CSO)



~

y = 154.52x +1283.4



~

~

<

R2 = 0.0281
~









—	~



i i

~

i i

* , .

10

20	30	40

Rainfall (in/yr)

50

60

Figure VI-IV. Normalized Mystic Annual CSO Volumes versus Annual Precipitation

70


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table VI-2. Statistical Outlier Analysis for CSO Datasets

Statistical Data Type

Alewife CSO Volumes (gal)

Mystic River CSO Volumes (gal)

First Quartile (Ql)

10,590,000

3,660,000

Third Quartile (Q3)

36,220,000

10,085,000

Interquartile Range (IQR)

25,630,000

6,425,000

Upper Bound

74,665,000

19,722,500

Mean (all data)

27,208,333

8,013,636

Mean (without outliers)

22,656,364

6,580,000

Two data points were identified as outliers: CSO volumes from 2010 in the Mystic River Drainage
Basin and 2008 in the Alewife Drainage Basin. However, the volumetric mean was not modified
because data points were within 10 percent of the upper bound, so was not considered to have
undue influence on the volumetric mean. The CSO volumes from 2006 were not available for the
Mystic River and was replaced with the mean using all data to complete the dataset from 2006 to
2017. The final datasets are provided in Table VI-3.

Table VI-3. CSO Volumetric Datasets for Alewife and Mystic River Drainage

Basins

Year

Alewife CSO Volumes (gal)

Mystic River CSO Volumes (gal)

2006

61,540,000

8,013,636a

2007

15,320,000

5,750,000

2008

77,280,000

10,420,000

2009

12,310,000

920,000

2010

63,590,000

22,350,000

2011

27,780,000

9,240,000

2012

21,830,000

11,760,000

2013

5,430,000

9,750,000

2014

22,450,000

5,120,000

2015

12,620,000

9,360,000

2016

1,300,000

1,280,000

2017

5,050,000

2,200,000

a) CSO volume was not available. CSO volume replaced with mean value of all data (2006-2017).

VI.C. SSO Data Analyses

The SSO data were first reviewed and summarized to characterize the raw data, including discharge
frequency and duration, discharge locations and volumes. Then the data were processed to be able
to assign SSO volumes by sub watershed for further evaluation. Finally, similar to CSO data, SSO
data analyses were completed to understand how precipitation and time influenced the annual
volumes and how average SSO volumes compared to annual data. The purpose of the analyses was
to determine if the average annual SSO volume for the evaluation of phosphorus load reduction
estimates (Section IX) was appropriate and/or if modifications to the annual SSO volumes would
allow a more representative average.

71


-------
Mystic River Watershed TMDL Alternative Development — Final Report

VI.C.1. SSO Data Review

A more in-depth review of the data entries was completed to understand the frequency and duration
of SSOs, their discharge points and the magnitude of the reported volumes. The data are
summarized below in Table VI-4 and

Table VI-5.

Overall, of the 774 total SSOs reported during the period of available data, only a small percentage
of them were identified as discharging to a catch basin or directly to a waterbody (60 in all, 8% of
total) despite the large number of them being a result of a rain event (528 in all, 68% of total). The
locations, durations and volumes ranged significantly, which suggests that SSO discharges are
dependent on the type and location of event (i.e., there is not a singular threshold event which will
result in SSO discharges in all sub watersheds). For example, note that both 2010 and 2014 appear as
years with much more frequent SSOs, accounting for approximately 68% and 8%, respectively, of
the total discharges over the period from 2006 to 2017.

Further discussions with MWRA and MassDEP on the SSO data indicate that there can be
variability in the data reporting. Data varies in the way that durations are reported (e.g., what is
identified as the 'start' of the SSO event may be assumed or may be at the time it is 'found'),
volumes are calculated (e.g., some locations are monitored, others are estimated), and number of
SSOs identified (e.g., only 'found' SSOs are reported, may be others not being reported). Therefore,
while the data is representative, it may not be consistent or accurate in all cases.

Table VI-4. Review of SSO Data - Frequency, Duration and Discharge Points

Year

Total

Number of

SSOs

reported

Average
duration of
SSOs (hrs.)

Min/Max
Duration of
SSOs (hrs.)

Community with
Highest SSOs
Reported/# of
SSOs

Number of
SSOs Directly
Discharging to a
Catch Basin

Number
Directly
Discharging

to a
Waterbody

2006

24

N/A

N/A

Arlington
(6)/Winchester (6)

—

—

2007

21

N/A

N/A

MWRA (15)

-

-

2008

24

N/A

N/A

MWRA (15)

-

-

2009

11

N/A

N/A

MWRA
(2)/Lexington (2)

—

—

2010

524

N/A

N/A

MWRA (107)

-

-

2011

22

N/A

N/A

N/A

-

-

2012

26

N/A

N/A

MWRA (4)

-

-

2013

25

4.3

0.3/22

Wakefield (10)

3

1

2014

65

9.6

0.3/52

Medford (29)

7

43

2015

10

8.9

0.2/72.8

Lexington (2)

-

2

2016

8

14.5

0.5/66.5

Burlington (3)

1

3

2017

14

7.1

0/71.8

Cambridge (5)

-

-

Totals

774

-

-

-

11

49

72


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table VI-5. Review of SSO Data - Volumes

Year

Total

Number of

Number of

Average

Minimum

Maximum



Number of

SSOs as a

entries where

reported SSO

Reported SSO

Reported



SSOs

result of a

SSO volume

volume (MG)

volume (gal)

SSO Volume



reported

rain event

was estimated





(MG)

2006

24

17

18

2.10

10,000

26.9

2007

21

17

19

0.20

50

1

2008

24

20

20

0.30

1,000

1

2009

11

2

9

0.07

4,000

0.6

2010

524

404

303

0.30

3.5

9

2011

22

11

20

0.30

2

2.6

2012

26

1

22

0.20

10

1.9

2013

25

0

23

950

5

<0.1

2014

65

47

32

2.40

10

24.6

2015

10

3

10

0.01

10

<0.1

2016

8

0

8

<0.01

2

0.1

2017

14

6

11

0.09

10

<0.1

Totals

774

528

495

-

-

-

VI.C.2. SSO Data Processing

The two available SSO volumetric datasets (see Table VI-1) were merged before proceeding with the
statistical analyses. The following analysis steps were used to process the MassDEP SSO data:

•	Remove all MWRA data from the MassDEP datasets.

•	Identify data points from the MassDEP data located in Mystic River Watershed.

•	Convert volumetric ranges (see Section VI.A.2) to a single volume. For example, data
identified as "<10,000 gallons" was converted to 10,000 gallons. Similarly, ">1,000,000
gallons were converted to 1,000,000 gallons. The intermittent range, "100,000 gallons to
1,000,000 gallons" was converted to a midpoint value (500,000 gallons).

•	Assign SSO volumes to the sub watersheds.

In contrast to the MWRA data, the MassDEP datasets were not geospatially located, so the data
were filtered and summarized by the communities in the Mystic River. A unit discharge
(gallons/acre) was calculated for each community and then allocated to each sub watershed by area
to estimate the total SSO volumes by sub watershed. Finally, MWRA and MassDEP SSO volumes
were combined by sub watershed.

VI.C.3. SSO Trend Analyses

SSO volumetric data were plotted over time and normalized by the annual rainfall data to transform
the data for evaluating trends. The data are graphed in Figure VI-V and Figure VI-VI. These plots
highlight the lack of observable or statistical relationships between the rainfall and SSO data in any
of the sub watersheds.

73


-------
Mystic River Watershed TMDL Alternative Development — Final Report

VI.C.4.

SSO Statistical Outliers

Similar to the CSO outlier analysis presented in Section VI.B.2, an evaluation of SSO outliers was
completed by comparing the dataset to upper volumetric bounds. The results are presented in Table
VI-4. This evaluation identified several outliers; all sub watersheds had a statistical outlier in 2010,
while other watersheds had outliers in 2006 and 2014. After reviewing the data, these outliers were
removed and replaced by the upper bound value. The final datasets are provided in Table VI-5.

VI.C.5. SSO Missing Datasets

Similar to the CSO datasets, missing (or no data) was evaluated to determine if the values should be
replaced or not. Based on the evaluation of the trend analyses and the statistical outliers, it seemed
likely that some sub watersheds with missing data may have zero annual SSO discharge. Therefore,
no missing data was replaced.

VI.C.6.

Total Phosphorus and Total Nitrogen Loads for Model Calibration

The annual watershed phosphorus and nitrogen loading estimates were developed using the final
CSO and SSO discharge volumes and the TP and TN concentrations noted in Section VI.A.4. The
final TN and TP load estimates are presented in Table VI-6, Table VI-7, Table VI-8,

Table VI-9, and Table VI-10.

SSO Volumes

ro
01
>ฆ

40,000,000
35,000,000
30,000,000
25,000,000
20,000,000

C

_o

"S

01

E

3

o 15,000,000
O

i/>



10,000,000

5,000,000

2004

~

||i
2006 2008

2010 2012 2014
Year

2016

~	Aberjona River 1

ฆ	Aberjona River 2
A Alewife

Blacks Nook
> Horn

~	Judkins

i Lower Mystic Lake

-	Maiden

-	Mill

~	Mystic River

ฆ	Spy Pond
Upper Lobe
Upper Mystic Lake
Wedge

Winter

2018

Figure Vl-V. Annual SSO Discharge Volumes by Sub-watershed versus Time

74


-------
Mystic River Watershed TMDL Alternative Development — Final Report

SSO Volumes

TO
0)
>•

l/l
ฃ
_o

.25
o
E

_3

O
>

O

i/>
i/)

40,000,000
35,000,000
30,000,000
25,000,000
20,000,000
15,000,000
10,000,000
5,000,000

10

20

X

~ . +
I *-~

—งฆฆ ii ii—jh-M-

30	40	50

~	Aberjona River 1

ฆ	Aberjona River 2
A Alewife

X Blacks Nook
> Horn

~	Judkins

t Lower Mystic Lake
- Maiden
- Mill

~	Mystic River

ฆ	Spy Pond
Upper Lobe
Upper Mystic Lake
Wedge

Winter

60

Rainfall (inches/year)

Figure VI-VI. Annual SSO Discharge Volumes by Sub watershed versus Rainfall

75


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table VI-6. Statistical Outlier Analysis for Annual SSO Discharge Volumes
(Gallons/yr.)

Statistical Data
Type

Mystic

Alewife

Aberjona
River 1

Aberjona
River 2

Blacks
Nook

Horn

Judkins

Lower
Mystic Lake

Maiden

Mill

Spy Pond

Upper
Lobe

Upper
Mystic
Lake

Wedge

Winter

1st quartile (25%)

273

8,622

0

0

0

912

460

3,296

213

0

0

0

0

0

0

3rd quartile (75%)

2,591,755

181,440

46,771

168,117

75

1,311,220

847,633

1,764,945

128,257

25,145

262,014

56,666

459,882

114,294

47,818

Interquartile
range

2,591,482

172,818

46,771

168,117

75

1,310,308

847,174

1,761,649

128,044

25,145

262,014

56,666

459,882

114,293

47,818

Upper bound
(Q3+1.5*IQR)

6,478,979

440,668

116,927

420,293

187

3,276,683

2,118,394

4,407,418

320,324

62,863

655,035

141,665

1,149,705

285,733

119,544

Mean (all data)

4,454,583

241,727

31,859

2,290,032

249

3,205,967

3,687,084

3,252,328

621,826

17,128

286,506

38,599

339,345

83,825

36,425

Mean (w/o
outliers)

1,494,411

67,331

47,810

84,824

52

519,601

261,804

640,451

50,047

25,704

189,330

57,925

301,668

73,825

31,069

Table VI-7. Annual SSO Discharge Volumes for All Sub watersheds (Gallons/yr.)

Year

Mystic

Alewife

Aberjona
River 1

Aberjona
River 2

Blacks
Nook

Horn

Judkins

Lower
Mystic
Lake

Maiden

Mill

Spy
Pond

Upper
Lobe

Upper
Mystic
Lake

Wedge

Winter

2006

111,630

212,747

90,077

164,790

0

3,987,702

2,172,54
5

3,887,253

588,314

48,428

356,831

109,134

795,749

226,172

95,998

2007

462,267

17,056

0

45,000

0

798,129

256,280

59,757

450,000

0

11,752

0

4,200

360

232

2008

2,144,904

150,133

72,754

178,100

0

1,026,864

826,009

732,833

21,010

39,115

235,038

88,147

623,718

177,714

74,334

2009

363

9,945

0

0

55

426,822

131,302

17,436

18,069

0

0

0

0

0

0

2010

9,701,117

475,150

138,580

428,612

219

3,987,702

2,172,54
5

6,177,132

588,314

74,504

722,475

167,899

1,286,235

338,584

141,642

2011

84,268

215,806

38,109

114,719

62

2,164,288

912,507

1,057,509

382

20,489

342,942

46,172

405,270

93,154

38,979

2012

3,932,309

33,861

0

0

181

36

60

0

611

0

0

0

0

0

0

76


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Year

Mystic

Alewife

Aberjona
River 1

Aberjona
River 2

Blacks
Nook

Horn

Judkins

Lower
Mystic
Lake

Maiden

Mill

Spy
Pond

Upper
Lobe

Upper
Mystic
Lake

Wedge

Winter

2013

4

20,537

0

0

113

508

519

1

215

0

0

0

0

7

4

2014

7,596,241

5,573

34,645

428,612

3

764,894

462,123

6,177,132

9,939

18,626

85

41,975

257,041

84,733

35,467

2015

2

7,299

0

0

1

653

212

4,437

28

0

0

0

0

0

0

2016

1

0

0

0

0

998

280

656

11

0

0

0

0

0

0

2017

1,653

348

3,464

6,388

2

12,814

28,746

4,176

205

1,863

0

4,197

25,701

8,463

3,540

Table VI-8. Estimated Annual Total Phosphorus and Total Nitrogen CSO Loads
(Ibs./yr.) for Alewife and Mystic Sub watersheds

Year

Alewife CSO TP

Mystic River CSO TP

Alewife CSO TN

Mystic River TN

2006

1,591.1

207.3

4,776.3

622.0

2007

396.3

148.8

1,189.0

446.3

2008

1,999.3

269.6

5,997.9

808.7

2009

318.5

23.8

955.4

71.4

2010

1,645.1

578.2

4,935.4

1,734.6

2011

718.7

239.0

2,156.1

717.1

2012

564.8

304.2

1,694.3

912.7

2013

140.5

252.2

421.4

756.7

2014

580.8

132.5

1,742.4

397.4

2015

326.5

242.2

979.5

726.5

2016

33.6

33.1

100.9

99.3

2017

130.6

56.9

391.9

170.7

77


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table VI-9. Estimated Annual Total Phosphorus SSO Loads (Ibs./yr.) for all Sub

watersheds

Year Mystic Alewife Aberjona Aberjona Black Horn Judkins Lower Maiden Mill Spy Upper Upper Wedge Winter

River 1 River 2 s Mystic Pond Lobe Mystic Lake
Nook Lake

2006

4.9

9.3

3.9

7.2

0.0

174.0

94.8

169.7

25.7

2.1

15.6

4.8

34.7

9.9

4.2

2007

20.2

0.7

0.0

2.0

0.0

34.8

11.2

2.6

19.6

0.0

0.5

0.0

0.2

0.0

0.0

2008

93.6

6.6

3.2

7.8

0.0

44.8

36.1

32.0

0.9

1.7

10.3

3.8

27.2

7.8

3.2

2009

0.0

0.4

0.0

0.0

0.0

18.6

5.7

0.8

0.8

0.0

0.0

0.0

0.0

0.0

0.0

2010

423.4

20.7

6.0

18.7

0.0

174.0

94.8

269.6

25.7

3.3

31.5

7.3

56.1

14.8

6.2

2011

3.7

9.4

1.7

5.0

0.0

94.4

39.8

46.2

0.0

0.9

15.0

2.0

17.7

4.1

1.7

2012

171.6

1.5

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

2013

0.0

0.9

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

2014

331.5

0.2

1.5

18.7

0.0

33.4

20.2

269.6

0.4

0.8

0.0

1.8

11.2

3.7

1.5

2015

0.0

0.3

0.0

0.0

0.0

0.0

0.0

0.2

0.0

0.0

0.0

0.0

0.0

0.0

0.0

2016

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

2017

0.1

0.0

0.2

0.3

0.0

9.3

1.3

0.2

0.0

0.1

0.0

0.2

1.1

0.4

0.2

Table VI-10. Estimated Annual Total Nitrogen SSO Loads (Ibs./yr.) for all Sub

watersheds

Year Mystic Alewife Aberjona Aberjona Blacks Horn Judkins Lower Mystic Maiden Mill Spy Pond Upper Upper Wedge Winter

River 1 River 2 Nook Lake Lobe Mysti

c Lake

2006

38.9

74.2

31.4

57.5

0.0

1,390.1

757.9

1,356.0

205.2

16.9

124.5

38.1

277.6

78.9

33.5

2007

161.3

5.9

0.0

15.7

0.0

278.2

89.4

20.8

157.0

0.0

4.1

0.0

1.5

0.1

0.1

2008

748.2

52.3

25.4

62.1

0.0

358.2

288.1

255.6

7.3

13.6

82.0

30.7

217.6

62.0

25.9

2009

0.1

3.5

0.0

0.0

0.0

148.9

45.8

6.1

6.3

0.0

0.0

0.0

0.0

0.0

0.0

2010

3,384.1

165.8

48.3

149.5

0.1

1,391.1

757.9

2,154.8

205.2

26.0

252.0

58.6

448.7

118.1

49.4

2011

29.4

75.3

13.3

40.0

0.0

755.0

318.3

368.9

0.1

7.1

119.6

16.1

141.4

32.5

13.6

2012

1,371.7

11.8

0.0

0.0

0.1

0.0

0.0

0.0

0.2

0.0

0.0

0.0

0.0

0.0

0.0

2013

0.0

7.2

0.0

0.0

0.0

0.2

0.2

0.0

0.1

0.0

0.0

0.0

0.0

0.0

0.0

2014

2,649.9

1.9

12.1

149.5

0.0

266.8

161.2

2,154.8

3.5

6.5

0.0

14.6

89.7

29.6

12.4

78


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Year

Mystic

Alewife

Aberjona
River 1

Aberjona
River 2

Blacks
Nook

Horn

Judkins

Lower Mystic
Lake

Maiden

Mill

Spy Pond

Upper
Lobe

Upper
Mysti
c Lake

Wedge

Winter

2015

0.0

2.5

0.0

0.0

0.0

0.2

0.1

1.5

0.0

0.0

0.0

0.0

0.0

0.0

0.0

2016

0.0

0.0

0.0

0.0

0.0

0.3

0.1

0.2

0.0

0.0

0.0

0.0

0.0

0.0

0.0

2017

0.6

0.1

1.2

2.2

0.0

4.5

10.0

1.5

0.1

0.6

0.0

1.5

9.0

3.0

1,2

79


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

VII. BATHTUB Modeling Approach

This section summarizes the modeling approach and calibration and validation results for the
BATHTUB receiving water model of the Mystic River. The BATHTUB model uses instream
measurements of TP, TN, orthophosphate (OP), inorganic nitrogen (IN), chlorophyll-a (chl-a), and
oxygen depletion in the hypolimnetic and metalimnetic layers (if the data are available) for
calibrating the model.

The calibrated BATHTUB model was used to run nutrient reduction scenarios to identify the
reductions necessary to bring the critical receiving water reaches (see Section V.A.I) into compliance
with water quality targets selected for this project.

VILA. Model Selection

Prior to modeling of water quality of the Mystic River, three receiving water models were reviewed,
namely, the LLRM (Wagner, 2009), BATHTUB (Walker, 2004), and AQUATOX (Clough, 2014)
models (see memo "Options for Modeling the Mystic River Watershed" (ERG/Pickering/PGE, May 26,
2017). These models were selected based on the conditions, data, and modeling needs in the Mystic
River Watershed. They all predict receiving water quality (nitrogen, phosphorus, chlorophyll-a, etc.)
to water inputs and nutrient loads (nitrogen, phosphorus) from all contributing sources (land, point,
and atmospheric loads) in the watershed. They are all scientifically sound and well suited for this
task.

The LLRM and BATHTUB models are both regression-based models that are similar in complexity
and effort, whereas the AQUATOX model is a much more mechanistic model. All three models
predict the nutrient water quality responses in multiple reaches, lakes, and impoundments. The
AQUATOX model was eliminated from consideration because it requires significantly more time
and effort. The other two models are similar in complexity and level of effort however, the
BATHTUB model offers a number of extra features like additional water quality inputs and outputs,
the ability to link a number of reaches or sub-reaches together, and an easy-to-use interface.

The BATHTUB model is an appropriate choice for this study because it is a semi-empirical model
that computes both the mass balance of each segment and utilizes empirical relationships between
the water quality variables (see Figure VII-I). Those empirical relationships have been calibrated to a
large dataset from Army Corps of Engineers (US-ACOE) reservoirs across the country (Walker,
1982; Walker, 1985). The tool is also appropriate to the level of available water quality data in the
Mystic River since it has a limited number of calibration factors that can be adjusted thus avoiding
over-calibration.

Based on input from the technical team and the TSC, the BATHTUB model was selected.
BATHTUB Version 6.2 was used for this modeling task. BATHTUB has been used in in many
other similar studies (Walker, 1996; Walker 2004). In particular, the BATHTUB model was used in
the development of the Lake Champlain Phosphorus Total Maximum Daily Load study (TetraTech,
2015; EPA, 2016).

Since nitrogen was ultimately not used in the final calibration, discussion of nitrogen has been
minimized in the rest of the BATHTUB modeling section.

80


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Figure Vll-I. BATHTUB Model Schematic
VII.B. Model Setup
VII. B. 1. Segmentation

The BATHTUB model allows for a number of segmentation options in the set up. A basic setup
would use a single upstream tributary to a single water body. The model also allows for river reaches
to be linked together, each with single or multiple tributary inputs, and multiple sub-reaches within a
reach. The chosen configuration for this project is discussed below.

The final BATHTUB model was divided into 5 segments (see Figure VII-II) as described in the list
below. Each segment is numbered from upstream to downstream, with abbreviated names in
parentheses, and the critical reaches for water quality attainment identified:

1.	Upper lobe of Upper Mystic Lake (Upper Lobe, critical).

2.	Main body of Upper Mystic Lake (Upper Lake, critical).

3.	Lower Mystic Lake (Lower Lake, not critical).

4.	Upper part of the Lower Mystic Basin (Upper Basin, critical).

5.	Lower part of the Lower Mystic Basin (Lower Basin, critical).

81


-------
Mystic River Watershed TMDL Alternative Development — Final Report

VII.B.2. Model Options

The BATHTUB model allows the user to select a number of model options to represent the
receiving water body response to the estimated input loads. The nutrient related options in
BATHTUB include the following:

82


-------
Mystic River Watershed TMDL Alternative Development — Final Report

•	Phosphorus Sedimentation Models

•	Nitrogen Sedimentation Models

•	Chlorophyll-a Models

•	Secchi Depth Models

•	Longitudinal Dispersion Models

•	Application of Nutrient Availability Factors

•	Application of Various Calibration Factors

More information on these options and the different equations used is given in the BATHTUB user
manual (Walker, 2004).

The final choice of model options is primarily a function of the availability of water quality data. For
example, Secchi depth and transparency data were limited in the segments and years modeled, so
those options were dropped. In addition, nitrogen data were sparse and very different from
predicted incoming land-based concentrations, so although nitrogen was included in initial testing of
the model, it was not used in the final BATHTUB setup. The final model options used in this study
are given in Table VII-1.

Table VII-1. Model Options

Model Option

Default Choice

Final Choice

Calibrated

Nitrogen Model

Not computed

Not computed*

-

Phosphorus Model

2nd order, available P

Same as default

No

TN Calibration

Sedimentation rates

Not computed*

No

TP Calibration

Sedimentation rates

Same as default

No

Nutrient Availability Factors

Not included

Not included

No

Chl-a Model

P, light, flushing

Same as default

Yes

Secchi Depth Model

Chi a and turbidity

Not computed

-

Transparency Model

Chl-a, turbidity

Not computed

-

Longitudinal Dispersion

Fischer-Numeric

Same as default

Yes

Internal Loading

Not included

Included

Yes

* See reasons in Section VII.C.3.

VII.B.3. Atmospheric Fluxes

Atmospheric TP loading was considered to have a minor effect on the total watershed load (see
VII.C.10), so we used the BATHTUB default TP value (0.27 lb./ac/yr.) split evenly between organic
and inorganic fractions.

Annual precipitation (PREC, in/yr.) for Logan Airport was derived from the NCEI (2018) daily data
downloaded previously (see Section V.A.1).

Annual lake evaporation (ETw, in/yr.) was more difficult to estimate for the reasons below:

• The world-wide adopted FAO Penman-Monteith method is the best method for estimating
potential and lake evaporation but it uses solar radiation that is no longer collected by US
Class I national weather stations

83


-------
Mystic River Watershed TMDL Alternative Development — Final Report

•	Available MassGIS data on potential and lake evaporation derived using similar radiation-
based methods does not apply to the modeling period, and only has monthly averages not a
time series of yearly values.

•	Methods that do not include solar radiation (like Blaney-Criddle) are usually biased high or
low and would have to be calibrated to a more reliable method.

Since annual lake evaporation is a minor variable in this model that only affects the total flow for
each reach slightly, we used an alternative simpler approach for estimating yearly lake evaporation.
The 1961-2005 annual time series of precipitation and lake evaporation data from the HSPF model
in the Upper/Middle Charles TMDL (2011) was used to create a regression equation (PETW =
41.06 -0.0952+PREC, R2=0.2) to predict annual lake evaporation from precipitation. Although the
R2 value for this relationship is low, it is based on the best available data and evaporation methods
and it does reflect higher lake evaporation for drier years and vice versa (see Figure VII-III
comparison for 2007 to 2017).

Figure VII-III. Annual Precipitation and Lake Evaporation

For the calibration period (2015), the values for annual precipitation and average lake evaporation
were 34.8 and 37.7 in/yr., respectively. In comparison, for the scenario period (2007-2016), the
average values for annual precipitation and lake evaporation were 43.0 and 37.0 in/yr., respectively.
These two time periods represent the calibration and critical period for scenarios (see Sections
VII.C.2 and VIII.D).

VII.B.4. Modeled Land Loads

As described in Section V, cumulative annual flow (in/yr.) and TP loads (lb./yr.) were developed for
each sub-basin and corresponding reach for the period 1992-2017. The annual land-based flows and
loads include contributions from stormwater, baseflow, and CSOs/SSOs. The initial sub-basin
delineation for the Lower Mystic Basin (8.63 mi2) was split into the Upper (6.08 mi2) and Lower
Basins (2.55 mi2) using available elevation, and a spatial understanding of connectivity between
combined sewer areas (CSAs) and the associated receiving waters based on information from
community and DEP knowledge (see Section VI.B).

84


-------
Mystic River Watershed TMDL Alternative Development — Final Report

VII.B.5. External Loads

In this section, we estimate the total external loads for the BATHTUB model. The model allows the
input of loads using flow and concentration. We used the BATHTUB "Tributary" feature to input
the combined total load from the upstream reach and the local sub-basin using the scheme laid out
in Figure VII-I. Because land loading is part of the "Tributary" input, explicit export coefficients
were not used for the BATHTUB model.

External load was specified as a flow-weighted average concentration (load/flow) and an annual
streamflow using the Tributary Input option in BATHTUB. The total external load is the sum of the
attenuated upstream reach load plus the unattenuated land load from the local sub-basin (see Table
VII-2). The total segment loads were then converted to an average concentration (i.e. load/flow)
associated with the average annual streamflow for input into the BATHTUB model.

Table VII-2. Upstream Reach and Local Sub-basin Contributions

Segment Name

Reach Name

Sub-basin Name



(attenuated load)

(unattenuated load)

Upper Lobe

Aberjona 2

Upper Lobe

Upper Lake

-

Upper Lake

Lower Lake

-

Lower Lake

Upper Basin

Alewife

Upper Basin

Lower Basin

Maiden

Lower Basin

The BATHTUB model requires the total external contributions to be divided into organic and
inorganic nutrient parts. Because the Mystic River had few nutrient measurements from tributaries,
we used the average measured concentration ratio (60 percent inorganic TP) from the
Upper/Middle Charles Nutrient TMDL (DEP-EPA, 2011). These ratios were used to split the total
concentration into organic and inorganic concentrations. This assumption is justified because the
Charles and Mystic tributary watersheds have very similar land use patterns and potential nutrient
sources.

VII.B.6. Internal Loads

The BATHTUB model can also use available sediment TP release rates for estimating the internal
load instead of adjusting the sedimentation rates. Internal load in river reaches results from nutrient
release from the accumulated organic sediments. Since there have been no direct measurements of
nutrient release rates from the sediments in the Mystic River Watershed, measured release rates from
the Upper/Middle Charles Nutrient TMDL (DEP-EPA, 2011) were used to constrain the initial
model inputs. Average TP release rates from impoundments in the Upper/Middle Charles TMDL
were less than 6 mg/m2/d. TP values were set to a maximum of 6 mg/m2/d during the calibration
process according to the presence of soft sediments (i.e., not actual sediment release rates) as
detected by field monitoring by EPA (2018). Reaches with large areas of sediments were assigned a
high value (6 mg/m2/d) while those with no sediments or no data (e.g. Upper Lake and Upper
Lobe) were assigned a low value (1 mg/m2/d). As part of the calibration process, these internal
loads were adjusted with the expected range (0-6 mg/m2/d) to better match observed water body
TP concentrations.

85


-------
Mystic River Watershed TMDL Alternative Development — Final Report

VII.C. Model Calibration and Validation
VII.C.1. Water Quality Data Availability

The average water quality data available for the modeled five segments in all calibration years and
validation years are summarized in the following table. The unaveraged values for water quality data
for the modeled segments are given in Appendix H. All values for TP concentration were previously
adjusted to correct for differences in laboratory methods (see Section V.D.2).

Table VII-3. Average Water Quality Data Available by Segment, Site and Year

Year

Segment

WQ Site IDs

AvgTP

Avg Chl-a

2010

Upper Lobe

UPLUPL





2010

Upper Lake

UPLCTR / UPL001

0.029



2010

Lower Lake

MYR071

0.036



2010

Upper Basin

MWRA083 / MWRA066

0.043

7.086

2010

Lower Basin

MYR33 / MAR003 / MWRA167

0.054

14.582

2014

Upper Lobe

UPLUPL





2014

Upper Lake

UPLCTR / UPL001

0.038



2014

Lower Lake

MYR071

0.034



2014

Upper Basin

MWRA083 / MWRA066

0.045

8.701

2015

Upper Lobe

MYR33 / MAR003 / MWRA167

0.052

16.790

2015

Upper Lake

UPLCTR / UPL001

0.032

8.929

2015

Lower Lake

MYR071

0.036

4.749*

2015

Upper Basin

MYR43

0.056

17.944

2015

Lower Basin

MYR33 / MAR003 / MWRA167

0.059

23.534

2016

Upper Lobe

UPLUPL

0.060

17.103

2016

Upper Lake

UPLCTR / UPL001

0.029

8.709

2016

Lower Lake

MYR071

0.036



2016

Upper Basin

MYR43

0.072

21.273

2016

Lower Basin

MYR33 / MAR003 / MWRA167

0.089

30.636

2017

Upper Lobe

UPLUPL

0.053

13.351

2017

Upper Lake

UPLCTR / UPL001

0.036

13.257

2017

Lower Lake

MYR071

0.038



2017

Upper Basin

MYR43

0.062

22.476

2017

Lower Basin

MYR33 / MAR003 / MWRA167

0.066

27.264

* Estimated from adjacent segments and other variables

VII.C.2. Calibration and Validation Periods

The calibration period was determined by the availability of good quality and representative instream
data in the critical receiving water bodies. Originally, the period 2015-2017 was recommended for
calibration. However, 2015 was found to be a better choice because it is the only common period
with good quality data for all five segments that have typical water quality without the interference
from the macrophytes herbicide treatments in 2016 and 2017. Although this year is drier than
average (see Figure VII-III), it serves as a critical-conditions period because the dry, sunny
conditions provide more ideal conditions for phytoplankton growth due to increased direct sunlight,

86


-------
Mystic River Watershed TMDL Alternative Development — Final Report

higher temperatures, and lower flows (increased residence times). Since one of the chl-a attainment
targets was the 90th percentile, it is important to capture optimal growth conditions for model
calibration. For these reasons, 2015 was used as the calibration period for the joint calibration of all
segments.

VII.C.3. Receiving Water Parameters

The BATHTUB model requires the input of the following physical parameters for each receiving
water body: surface area, mean depth, length, epilimnion depth, hypolimnion depth, and non-algal
turbidity. Since most ponds in Massachusetts do not stratify if less than 10 ft deep, epilimnion depth
was estimated as 10 ft or the actual depth if less. Any remaining depth was assigned to the
hypolimnion or set to zero if fully assigned to the epilimnion. Non-algal turbidity is an inverse
measure of Secchi depth that represents the portion of light extinction that is due to factors other
than algae (inorganic suspended solids, color). A value of 1 per 6 ft was used since Secchi depth
measurements for the non-algal season in the segments and years modeled were not available. The
value of 6 feet was an average determined for the non-algal season (outside of May-Sep) in non-
modeled river segments for the years modeled.

VII.C.4. Dispersion

The longitudinal dispersion rate was calculated in the model using the BATHTUB model default
dispersion method (Fischer et.al., 1979; adapted by Walker, 1985). Longitudinal dispersion is a result
of the mixing effect among adjacent segments, both between reaches and within reaches if there are
sub-reaches. High dispersion rates create high mixing conditions and a low range of the values
among segments whereas low dispersion rates create low mixing conditions with the segments all
having distinctly different values.

Observed TP data showed more of a range of data values among the segments than the default
model predicted. Therefore, the longitudinal dispersion rate was calibrated by using a multiplier of
0.2. This approach gave more differentiation of the TP values among the segments than using the
default multiplier of 1.0, allowing the modeled TP (see Figure VII-VFigure VI-V) to better reflect
the U-shaped pattern of observed TP values (see Figure VII-VI).

87


-------
Mystic River Watershed TMDL Alternative Development — Final Report

70

Upper Lobe Upper Lake Lower Lake Upper Basin Lower Basin

ฆ Calibrated (Dispersion*0.2)	Default (Dispersion*!.0)

Figure VII-IV. Dispersion Effect on Modeled Phosphorus Concentrations
VII.C.5. Nutrient Availability Factors

The BATHTUB model allows the use of organic and inorganic fractions to drive the nutrient
sedimentation model through the use of a Nutrient Availability Factor (see Table VII-1). Organic
phosphorus (OP) controls sedimentation and can influence the predicted TP values allowing more
options for calibration. The OP fraction of TP was set to 40-50% percent for inflow to all segments
based on observed data from the Upper Charles TMDL (DEP-EPA, 2011). Sensitivity trials that
varied this ratio resulted in no model prediction improvement. Ultimately, this model option was not
used in the final calibration run.

VII.C.6. Internal Loads

The TP predictions in the five modeled segments were generally lower than measured values
indicating that there is some internal loading of TP from the sediments. Average TP release rates
from impoundments in the Upper/Middle Charles TMDL averaged less than 19.5 lb./ac/yr. for
aerobic and anaerobic conditions, respectively. (DEP-EPA, 2011). The final calibrated values for the
BATHTUB model were set to 19.5, 3.3, 3.3, 13.0 and 13.0 lb./ac/yr. for segments 1 to 5,
respectively (see Section VII.B.l for names).

VII.C.7. Chl-a Model

Predicted chl-a values using the default chl-a method (P, light, flushing) in BATHTUB model were
always too high (16-58% high with the percent errors larger downstream) even though TP was being
accurately predicted.

The linear chl-a method in BATHTUB (P only) worked slightly better. That approach is consistent
with the chl-a versus TP regressions developed in Table IV-3 (R2 = 0.12 to 0.43).

88


-------
Mystic River Watershed TMDL Alternative Development — Final Report

In the final model, we used a chl-a calibration factor. The best calibration for chl-a used the default
chl-a model with a global calibration factor of 0.6 applied to all segments. An atypical relationship
between TP and chl-a in ponds can be the result of light limitation (Filstrup and Downing, 2017).

Calibration factors provide a means for adjusting model predictions to account for site-specific
conditions. These modify the coefficients of the empirical models within the BATHTUB model.
They are usually set to 1.0 and should be modified only with extreme caution and site-specific data.
We justify the use of a calibration coefficient based on some evidence of light limitation the Mystic
River. One possible explanation is that the natural tan color of the Mystic River limits light
penetration into the water column. Another explanation is that the submerged and floating aquatic
plants (see Section II.J) create a similar light penetration issue.

VII.C.8. Model Calibration

In summary, the default BATHTUB model was calibrated to the Mystic River conditions using the
following sequential approach:

•	Changing the default multiplier for longitudinal dispersion from 1.0 to 0.2

•	Adding internal TP loading to each modeled segment, and

•	Changing the default multiplier for the chl-a model from 1.0 to 0.6.

The 2015 data were used to calibrate the model since it had the most available water quality data and
that year was unaffected by transient nutrient and chlorophyll-a changes from herbicide applications
in 2016 and 2017 used for macrophyte control in the Lower Basin. Despite this good set of data, the
chl-a value for the Lower Lake had to be estimated from adjacent segments and other variables.

Since this segment is not a critical reach, it is not a critical value.

A total of 22 calibration runs were performed and tracked both visually and for computed goodness-
of-fit parameters for TP and chl-a. A high correlation coefficient (R2) represents a strong linear
relationship between predicted and observed values while a low root-mean square error (RMSE,
Mg/L) indicates good overall fit to the observed data. The best calibration run was a compromise
between TP and chl-a runs with the goal of having similar R2 values for both parameters. The final
calibration run had an R2 and RMSE for TP of 0.86 and 3.8 (Jg/L, and an R2 and RMSE for chl-a of
0.84 and 1.5 (Jg/L, respectively. Excluding the estimated chl-a value for Lower Lake (not a critical
segment) slightly improved the model fit for chl-a.

Predicted versus observed values of TP are given in Figure VII-V and Figure VII-VI. Similar plots
are for chl-a are given in Figure VII-VII and Figure VII-VIII. The error bars in these plots are +/-
one standard deviation.

Table VII-4 gives a breakdown of the mass balance for TP in each modeled segment excluding
transfers from one model segment to the next. Since the focus of this mass balance is the five
modeled segments of the BATHTUB model, we report attenuated land loads in this table. A pie
chart of total loads to the Mystic River Watershed for the calibration period is given in Figure
VII-IX.

A complete set of the BATHTUB inputs is provided in Appendix I.

89


-------
Mystic River Watershed TMDL Alternative Development — Final Report

80

^ 70

3 60

ฃ

O

+3 50
ro

c

OJ
u
c
o
u

O.

40

30

20

10

i



1

i

Upper Lobe Upper Lake Lower Lake Upper Basin Lower Basin
• Predicted • Observed

Figure Vll-V. Predicted vs. Observed TP by Segment for Calibration Period

Observed TP (|ig/L)

• Predicted	1:1 Line

Figure VII-VI. Predicted vs. Observed TP Relationship for Calibration Period

90


-------
Mystic River Watershed TMDL Alternative Development — Final Report

35

^ 30
—

3
c
o
5
to

ฃ 20

0)
u
c

ฐ 1c
U 15

25

to

IE
u

10















I

(

I

)









t t

c

J

Upper Lobe Upper Lake	Lower Lake Upper Basin

• Predicted C Observed

Lower Basin

Figure VII-VII. Predicted vs. Observed Chl-a by Segment for Calibration Period

25

bo
3
to

20

-a

a>

15

T3
0)

10

10	15

Observed Chl-a (|ig/L)

20

25

• Predicted

1:1 Line

Figure VII-VIII. Predicted vs. Observed Chl-a Relationship for Calibration Period

91


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table VII-4. Total Phosphorus Loads by Segment for Calibration Period

Name

External Load*
(lb./yr.)

Internal Load
(lb./yr.)

Atmospheric Load
(lb./yr.)

Total Load
(lb./yr.)

Upper Lobe*

2734.6

676.5

9.3

3420.4

Upper Lake

218.3

459.1

37.7

715.1

Lower Lake

1000.8

306.0

25.1

1332.0

Upper Basin

2250.2

740.9

15.2

3006.4

Lower Basin

2937.1

1610.8

33.1

4580.9

Mystic River

9141.0

3793.3

120.4

13054.8



Name

External Load

(%)

Internal Load

(%)

Atmospheric Load
(%)

Total Load

(%)

Upper Lobe*

80.0

19.8

0.3

100.0

Upper Lake

30.5

64.2

5.3

100.0

Lower Lake

75.1

23.0

1.9

100.0

Upper Basin

74.8

24.6

0.5

100.0

Lower Basin

64.1

35.2

0.7

100.0

Mystic River

70.0

29.1

0.9

100.0

* External load includes input from Upper Mystic/Aberjona River, other segments exclude load transfers

92


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Loads are in lb./yr. External load - stormwater + groundwater +_CSO/SSO load

Figure VII-IX. Calibration 2015 - Total Phosphorus Loads for Mystic River

93


-------
Mystic River Watershed TMDL Alternative Development — Final Report

VII.C.9. Model Validation

The 2016 and 2017 years were used to validate the BATHTUB that was model calibrated to the
2015 data. These years were excluded from the calibration because of herbicide treatment of
macrophytes in the Upper/Lower Basin segments of the model. These treatments generally kill the
vegetation, release nutrients into the water column, open up the water surface to allow greater light
penetration, and result in high algal growth (see Section II.J). According to MyRWA (pers. comm.,
2019), the treatment in 2016 was a contact herbicide that has rapid results, while the treatment in
2017 was a systemic herbicide that is slower acting but more prolonged. In the light of this
information, we expected the predicted TP and chl-a values to be lower than the observed results
for the Upper and Lower Basin, with more of a difference in 2016 than 2017.

In 2016, the predicted values were similar to observed values in the upper segments but lower in the
Upper and Lower Basin segments. Differences were in the range of 13-23 (jg and 7-13 (Jg/L for TP
and chl-a, respectively. (Figure VII-VIII and Figure VII-IX). In 2017, the predicted values were
higher that observed values in the most segments but much larger in the Upper and Lower Basin
than in 2016, in the range of 9-13 (Jg/L for TP and 8-11 (Jg/L for chl-a, respectively (Figure VII-X
and Figure VII-XI). This was expected because the systemic herbicide used in 2017 acts more slowly
than the contact herbicide used in 2016. The reason for under prediction of TP and chl-a for the
upper model segments in 2017 is unclear.

The 2010 and 2014 years were used to validate the outlier replacement technique used to develop
the annual flow and loads for SSOs (see Section VI.C). These two years had significant occurrences
and volumes of SSOs occur throughout the Mystic River Watershed. The BATHTUB was run in
two modes, with and without SSO outlier replacement, to test which outlier approach gave predicted
results closer to the observed TP values. In 2010, the predicted TP values using outlier replacement
closely matched the observed values (-1.1 to 3.9 (Jg/L), while when outliers were included gave
higher than observed TP values (3.1 to 8.1 (Jg/L) (Figure VII-12). In 2014, the results were similar
but had more spread. Predicted TP values using outlier replacement matched the observed values on
average (-5.8 to 8.3 ug/L), while with outliers included gave higher than observed TP values (-1.9 to
14.7 (Jg/L) (Figure VII-XIII).

These 2016 and 2017 validation runs confirm that the calibrated model was able to perform as
expected for those years. The 2010 and 2014 validation runs confirm that the outlier replacement
approach used for SSOs was not an inappropriate approach.

94


-------
Mystic River Watershed TMDL Alternative Development — Final Report

00
3

C

o
5
to

100
90
80
70

60	•

= 50	•

0)

u

ฃ 40

o
u
O. 30

ฐ I	• s

o

20
10
0

Upper Lobe Upper Lake Lower Lake Upper Basin Lower Basin
• Predicted O Observed

Figure Vll-X. Validation 2016 - Predicted vs. Observed TP by Segment

35

^ 30

—

M
2.

"c 25

.ฐ •

ฆฃ •

ฃ 20

4->

c	•

CD	ฆ

u

C 15
O

u
to

— 10

0

Upper Lobe Upper Lake Lower Lake Upper Basin Lower Basin
• Predicted • Observed

Figure VII-XI. Validation 2016 - Predicted versus Observed Chl-a by Segment

95


-------
Mystic River Watershed TMDL Alternative Development — Final Report

70

^ 60
—

H.	•

"c 50
o

to

ฃ 40
C
0)

C	•	•

o 30
u

O.

I_ 20

10



Upper Lobe Upper Lake Lower Lake Upper Basin Lower Basin
• Predicted O Observed

Figure VII-XII. Validation 2017 - Predicted vs. Observed TP by Segment

30

< 25

M
3
c

O 20
'+ฆป
ro

4-ป

c

u 15	•

ง
u

? 10

Upper Lobe Upper Lake Lower Lake Upper Basin Lower Basin
• Predicted C Observed

Figure VII-XIII. Validation 2017 - Predicted vs. Observed Chl-a by Segment

96


-------
Mystic River Watershed TMDL Alternative Development — Final Report

70

T 60

3
c
o
5
to



50



ฃ 40
c

0)
u
c

O 30
u

O.

I—

20

10



Upper Lobe Upper Lake Lower Lake Upper Basin Lower Basin
ฉObserved • Predicted (outliers replaced) Predicted (with outliers)

Figure VII-XIV. SSO Outlier Replacement 2010 - Predicted versus Observed TP

70

—T 60

3
c
o

to

50

i: 40
ฃ

0)
u

o 30
u

Q.

I—

20

10







9

Upper Lobe Upper Lake Lower Lake Upper Basin Lower Basin
ฉObserved • Predicted (outliers replaced) Predicted (with outliers)

Figure VII-XV. SSO Outlier Replacement 2014 - Predicted versus Observed TP

97


-------
Mystic River Watershed TMDL Alternative Development — Final Report

VII.C.10. Model Sensitivity Analysis

A model sensitivity analysis was performed on the calibrated BATHTUB model by varying six
important variables (model parameters and TP input loads). Although BATHTUB has a built-in
sensitivity analysis tool, we chose not to use it because it only allows analysis of two parameters, it
does not allow the variation of input loads, and it does not have a consistent sensitivity range for all
variables to allow cross-comparison of sensitivity among all items.

The sensitivity analysis that we performed multiplied the values by 50% or 200% of the original
calibrated values of the following parameters/loads. Note that only the first two parameters (*) are
included in the built-in tool in BATHTUB.

•	Longitudinal dispersion*

•	Sedimentation rate of TP+

•	Atmospheric deposition TP load

•	Internal/sediment load of TP

•	Segment TP load (by varying concentration)

•	Segment TP load (by varying flow)

To conduct this sensitivity analysis, two additional runs were necessary for each calibrated variable,
performed by multiplying the test value by 50% or 200%. For each parameter/load, the output TP
and chl-a values were recorded for the calibrated run and the two sensitivity runs. A normalized
sensitivity range was calculated using the following formula:

Sensitivity Factor = (Value 200% — Value 50%) / Value calibrated

Table VII-5 gives the results of the sensitivity analysis. The table lists the predicted TP and chl-a
results for the three runs for each parameter/load and also shows the sensitivity factor both
numerically and graphically.

From these results, we can conclude the sensitivity for tested parameters/loads was the following:

•	Longitudinal dispersion — low

•	Sedimentation rate of TP — high

•	Atmospheric deposition TP load — very low

•	Internal/sediment load of TP - moderate

•	Segment TP load (by varying concentration) - high

•	Segment TP load (by varying flow) - moderate

This analysis provides confirmation that several professional judgements made in calibrating the
BATHTUB model consistently chose the correct approach:

•	Use of default values for atmospheric deposition loads

•	Calibration of longitudinal dispersion before internal load

•	Use of internal load instead of sedimentation rate to adjust segment TP

•	Not calibrating segment input flow to observed gauge values because of data ambiguity

The result for varying segment input TP load by concentration versus flow is worth additional
discussion. Lowering the flow, and consequently the load, has a muted response on lowering TP and

98


-------
Mystic River Watershed TMDL Alternative Development — Final Report

chl-a. This outcome is likely because flow has a negative feedback effect via residence time. Even
though the load is lower, the lower flow results in a higher residence time in each segment, such that
the TP/chl-a concentrations do not decrease as much compared to same change in load by lowering
the input concentrations.

Table VII-5. Results of BATHTUB Sensitivity Analysis

Longitudinal Dispersion

50%

100%

200%

50%

100%

200%

Sen sitivity

Sensitivity

Sensitivity

Sensitivity

Segment

TP

TP

TP

Chl-a

Chl-a

Chl-a

TP

Chl-a

TP

Chl-a

Upper Lobe

57.6

49.2

42.9

19.0

16.2

14.0

-29.9%

-30.3%



1

Upper Lake

33.1

33.4

33.7

7.9

7.9

8.0

1.7%

1,6%







Lower Lake

32.5

32.6

35.4

7.7

7.7

8.4

9.1%

8,9%



1

Upper Basin

49.6

49.7

47.9

18.0

18.1

17.4

-3,4%

-3.7%





1

Lower Basin

66.1

66.1

65.0

22.7

22.7

22.3

-1,7%

-1.6%





1























Sedimentation Rate

50%

100%

20 D%

50%

100%

200%

Sen sitivity

Sensitivity

Sensitivity

Sensitivity

Segment

TP

TP

TP

Chl-a

Chl-a

Chl-a

TP

Chl-a

TP

Chl-a

Upper Lobe

54.1

49.2

44.5

17.8

16.2

14.6

-19.4%

-19.8%





1

Upper Lake

42.4

33.4

25.4

9.9

7.9

6.0

-50.8%

-49.3%





ฆ

Lower Lake

42.4

32.6

24.3

9.9

7.7

5.7

-55.7%

-54.3%





ฆ

Upper Basin

56.7

49.7

43.3

20.8

18.1

15.5

-26.9%

-29.1%





I

Lower Basin

73.7

66.1

58.1

25.1

22.7

20.0

-23.6%

-22.296





1























Atmos Deposition

50%

100%

200%

50%

100%

200%

Sensitivity

Sensitivity

Sensitivity

Sensitivity

Segment

TP

TP

TP

Chl-a

Chl-a

Chl-a

TP

Chl-a

TP

Chl-a

Upper Lobe

49.1

49.2

49.4

16.2

16.2

16.3

0.6%

0,6%







Upper Lake

33.3

33.4

33.6

7.9

7.9

8.0

1.1%

1,1%







Lower Lake

32.4

32.6

32.9

7.7

7.7

7.8

1.4%

1,4%







Upper Basin

49.5

49.7

50.0

18.0

18.1

18.2

1.096

1,1%







Lower Basin

65.9

66.1

66.5

22.6

22.7

22.9

1.096

0,9%





























Internal Load

50%

100%

200%

50%

100%

200%

Sen sitivity

Sensitivity

Sensitivity

Sensitivity

Segment

TP

TP

TP

Chl-a

Chl-a

Chl-a

TP

Chl-a

TP

Chl-a

Upper Lobe

44.6

49.2

58.1

14.6

16.2

19.1

27.5%

27.7%



1

1

Upper Lake

30.4

33.4

38.9

7.2

7.9

9.2

25.5%

24.5%



1

1

Lower Lake

30.0

32.6

37.2

7.1

7.7

8.8

21.8%

21.2%



1

1

Upper Basin

44.7

49.7

59.4

16.1

18.1

21.8

29.6%

31.7%



1

ฆ

Lower Basin

57.7

66.1

82.3

19.9

22.7

27.6

37.3%

34,096



ฆ

ฆ























Segment Load (cone)

50%

100%

20 D%

50%

100%

200%

Sen sitivity

Sensitivity

Sensitivity

Sensitivity

Segment

TP

TP

TP

Chl-a

Chl-a

Chl-a

TP

Chl-a

TP

Chl-a

Upper Lobe

32.2

49.2

80.7

10.2

16.2

25.6

98.6%

95,096







Upper Lake

25.1

33.4

46.3

5.9

7.9

10.7

63.3%

59.9%





ฆ

Lower Lake

24.7

32.6

44.8

5.8

7.7

10.4

61.7%

59.1%





ฆ

Upper Basin

34.2

49.7

77.7

11.8

18.1

28.3

87.6%

91.5%



m



Lower Basin

46.4

66.1

101.4

15.9

22.7

32.6

83.36

73.7%





m























Segment Load (flow)

50%

100%

200%

50%

100%

200%

Sensitivity

Sensitivity

Sensitivity

Sensitivity

Segment

TP

TP

TP

Chl-a

Chl-a

Chl-a

TP

Chl-a

TP

Chl-a

Upper Lobe

54.6

49.2

47.7

18.4

16.2

15.1

-14.1%

-20.3%

1



1

Upper Lake

32.8

33.4

36.3

8.0

7.9

8.1

10.3%

1.3%



1



Lower Lake

31.0

32.6

36.5

7.5

7.7

8.2

17.096

7,9%



1

1

Upper Basin

54.8

49.7

49.2

20.4

18.1

17.3

-11.296

-17.1%

1



1

Lower Basin

76.1

66.1

62.4

26.5

22.7

20.6

-20.8%

-25.9%

1



1

99


-------
Mystic River Watershed TMDL Alternative Development — Final Report

VIII. Critical Period of Interest for Phosphorus Load
Reduction Analysis

This section documents the approach for evaluating and selecting the critical period of interest for
the phosphorus load reduction analyses in the BATHTUB model discussed in Section IX. The
approach involved evaluating available water quality data availability; identifying events or actions
that may influence the water quality data; and identifying a range of representative climatic
conditions.

VIII.A. Water Quality Data

As discussed in Section III and noted in Table III-2 water quality monitoring data was available
between 2000 and 2017. Six water bodies (located primarily in the upper watershed) were sampled
more intensively in 2015 through 2017. The review of eutrophic related parameters, Chlorophyll-a
(chl-a) and TP, including statistical analyses completed in Section IV indicated that there has been a
gradual decline in chl-a and TP concentrations in the Mystic River. This appears to be primarily due
to the growth of macrophytes, particularly in the last 10 years, which appear to be removing
phosphorus from the water column and/or suppressing phytoplankton growth due to reduced light
availability. The only exceptions to this trend are the instant releases caused by the use of herbicide
applications in 2016 and 2017.

Based on these considerations, data between 2000 and 2015 was considered to be most
representative of the available water quality datasets.

VIII.B. Combined Sewer and Sanitary Sewer Overflow Data

As noted in Table VII-1 in Section VII, CSO and SSO data was available for the years 2000 to 2017,
with the exceptions noted for missing 2006 CSO data in the Mystic River basin and missing SSO
event data.

VIII.C. Rainfall Data

Annual rainfall data was available from NCEI (2018). As noted in Section VII, the annual lake
evaporation was not able to be directly estimated, so the 1961-2005 annual time series of
precipitation and water surface evaporation data (PETW) from the HSPF model in the
Upper/Middle Charles TMDL (2011) was used. Evaluations of the rainfall data and potential
evapotranspiration were reported in inches per year and meters per year.

A review of the annual precipitation conditions was conducted using the standardized precipitation
index, which is a statistical method for assessing rainfall (SPI; McKee, 2993). The SPI normalizes the
data to provide a better understanding of whether a year was wet (positive SPI values, greater than
average precipitation) or dry (negative SPI values, less than average precipitation). A summary table
of the SPI values and the representative condition is provided below in Figure VIII-I.

Table VIII-1. The complete summary of the rainfall data between 1990 and 2017 is provided in
Table VIII-2 SPI between 2000 and 2017 is shown in Figure VIII-I.

100


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table VIII-1. Standard Precipitation Index (SPI) Reference Values

SPI Value

Drought/Wetness Condition

2 and above

Extremely wet

1.5 to 1.99

Severely wet

1.0 to 1.49

Moderately wet

-0.99 to 0.99

Near normal

-1.0 to-1.49

Moderately dry

-1.5 to-1.99

Severely dry

-2.0 and less

Extremely dry

Table VIII-2. Summary of Rainfall Data Analyses

Year Precip PETW SPI value Condition
(in/yr.) (in/yr.)

1990

46.5

36.6

0.5

Near normal

1991

42.3

37.0

-0.1

Near normal

1992

43.7

36.9

0.1

Near normal

1993

43.2

36.9

0.0

Near normal

1994

47.6

36.5

0.7

Near normal

1995

35.1

37.7

-1.2

Moderately dry

1996

48.7

36.4

0.9

Moderately wet

1997

28.3

38.4

-2.2

Severely dry

1998

51.3

36.2

1.3

Very wet

1999

37.8

37.5

-0.8

Near normal

2000

45.6

36.7

0.4

Near normal

2001

30.8

38.1

-1.8

Severely dry

2002

41.1

37.1

-0.3

Near normal

2003

44.4

36.8

0.2

Near normal

2004

44.6

36.8

0.2

Near normal

2005

43.7

36.9

0.1

Near normal

2006

52.9

36.0

1.4

Moderately wet

2007

39.5

37.3

-0.5

Near normal

2008

54.5

35.9

1.7

Very wet

2009

43.5

36.9

0.0

Near normal

2010

49.7

36.3

1.0

Moderately wet

2011

52.4

36.1

1.4

Moderately wet

2012

36.8

37.6

-1.0

Moderately dry

2013

40.4

37.2

-0.4

Near normal

2014

45.3

36.7

0.3

Near normal

2015

34.8

37.7

-1.2

Moderately dry

2016

33.1

37.9

-1.5

Severely dry

2017

43.5

36.9

0.0

Near normal

101


-------
My stic River Watershed TMDL Alternative Development — Final Report

5PI

ฆ2.0	-1.8

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

Figure Vlll-I. Standard Precipitation Index (SPI) between 2000-2017
VIII.D. Critical Period Selection

The critical period of interest for the phosphorus load reduction analyses is intended to be
representative: of critical climatic conditions related to the water quality endpoints (chl-a and TP),
which may lead to excessive algal growth and cyanobacteria blooms in the Mystic River. Based on
discussions with the technical team, a minimum of a 10-year period was recommended in order to
capture the critical conditions that could lead to eutrophication. Initially, the period of 2008 to 2017
was identified to utilize the last 10-years of meteorological data and most recent water quality
conditions in the water bodies. However, 2017 SSO data from MassDEP was not available at the
time, so the critical period of 2007 to 2016 was selected for further review.

During the critical period, there was one very wet year (2008), two moderately wet years (2010,
2011), two moderately dry years (2012, 2015) and one severely dry year (2016). An average analysis
of the rainfall and potential evapotranspiration are 43.0 and 37.0 inches/year, respectively. In
comparison to the period of record (2000 to 2017), the average annual rainfall and potential
evapotranspiration are 43.1 and 37.0, respectively. Overall, 2007 to 2016 was selected as the
representative climatic period for further phosphorus load reduction analyses with the calibrated
BATHTUB model.

VIII.E. Extreme Rainfall Years

An additional evaluation of the data was also done to identify extreme rainfall years that may be used
to compare against average annual data during the selected critical period. This included looking at
both SPI values and 10th and 90th rainfall depth percentiles, which were used in the Upper/Middle
and Lower Charles River TMDLs to: evaluate the potential for exceedances for water quality and the
margin of safety. During the critical period, the 10th percentile rainfall depth is 34.1 inches, while the
90th percentile rainfall depth is 52.6 inches. Comparatively, over a longer term, from 1990-2016, the
10th percentile rainfall depth is 34.4 inches and the 90th percentile rainfall depth is 51.6 inches.
Comparing these values to Figure VIII-I above, 2008 would be identified as an extreme wet year,
while 2016 would be considered an extreme dry year.

102


-------
Mystic River Watershed TMDL Alternative Development — Final Report

IX. Evaluation of Watershed Phosphorus Load Reduction
Analysis

This section documents the BATHTUB modeling methodology, loading estimates and the scenarios
developed to evaluate the watershed phosphorus loading reductions necessary to address
eutrophication within the Mystic River Watershed and attain water quality target (chl-a) and
secondary indicatory (TP) identified in Section IV. The calibrated BATHTUB model discussed in
Section VII was used for this project; the critical period of interest was identified in Section VIII for
the phosphorus load reduction analysis, which is the 10-year period from 2007 to 2016. Our analysis,
as described below, suggests that CSO and SSO management Ent, while important, have far less
impact on annual phosphorus loads compared to stormwater.

IX. A. Phosphorus Loading Estimates for Critical Period of Interest

Five different types of load estimates were developed for each reach and corresponding tributary as
shown in Figure VII-II:

•	Stormwater loads.

•	Groundwater loads.

•	Sediment nutrient efflux loads.

•	Combined sewer overflow loads.

•	Sanitary sewer overflow loads.

IX.A. 1. Existing Conditions

The watershed phosphorus loading estimates from Section V were averaged for the 10-year period
from 2007 to 2016 and summarized by the five main segments (reaches) in the BATHTUB model:
Upper Lobe, Upper Lake, Lower Lake, Upper Mystic Basin and Lower Mystic Basin. This includes
both attenuated loads in the reaches and unattenuated loads in the tributaries. No adjustments were
made to attenuation factors or to the average loads for the "existing conditions" scenario.

The existing conditions loads for the critical period of interest were summarized to determine the
relative influence of each on the system. The analysis also included atmospheric loads, which are
included in the BATHTUB model (see Section VII).

IX.A.2. Future Conditions

In discussions with the TSC, further evaluation of phosphorus loadings under future conditions was
requested in order to account for the ongoing work being done in the watershed to address both
CSO and SSO overflows to downstream water bodies. The ERG team consulted with EPA,
MassDEP, and MWRA to determine what expected changes should be reasonably applied to the
loads under future conditions. The following are the assumptions used for the future condition
scenarios discussed in Section IX.B. Refer to Section VI for further detailed discussion on the
development of the CSO and SSO volumes and loads.

CSO Loads

In BATHTUB, the CSO loads and flows from the Cambridge combined sewer areas and parts of
Somerville combined sewer areas contribute to Alewife Brook tributary, the CSO loads and flows
from the rest of the Somerville's combined sewer area contributes to the Lower Basin segment. No

103


-------
Mystic River Watershed TMDL Alternative Development — Final Report

CSO flows go to the Upper Basin segment directly except via Alewife Brook. The annual CSO
Loads used for the calibrated BATHTUB model were developed using modeled CSO volumes from
MWRA's annual reports and a representative TP concentration of 3.1 mg/L, as discussed in Section
VI.A.4. Under future conditions, the CSO volumes are assumed to meet the long-term control plan
(LTCP) estimates for the typical rainfall year. As noted under MWRA's 2017 annual reporting of
CSO discharge estimates (MWRA, 2018), the annual LTCP CSO volumes for the Alewife Brook and
Upper Mystic River are 7.29 MG and 3.48 MG, respectively. To calculate the phosphorus loads,
volumes were multiplied by the representative TP concentration.

SSO Loads

Similar to CSO Loads, SSO loads were developed by using estimated volumes obtained from
MWRA and MassDEP and representative TP concentration of 5.23 mg/L, as discussed in Section
VI.A.4. Under future conditions, a 50 percent reduction in SSO volumes is assumed for ongoing
SSO mitigation work being done within the Mystic River Watershed. Volumes were multiplied by
the representative TP concentration to determine the phosphorus loads.

Stormwater Loads from Combined Sewer Separation

In addition to CSO reductions, EPA also reached out to the cities of Cambridge and Somerville to
determine potential for combined sewer separation (CSS) within areas still connected to CSOs at
Alewife Brook and the Upper Mystic River (above the Amelia Earhart Dam). As of November 2018,
the City of Cambridge has no immediate plans to evaluate or complete CSS and the City of
Somerville is in the process of completing an alternative analysis to determine if CSS is a viable
option. Upon discussion with the Technical Steering Committee, the ERG team developed three
future conditions loading estimates: 0 percent CSS, 25 percent CSS, and 100 percent CSS. The 100
percent CSS condition is intended to provide a bookend for the maximum phosphorus load
reductions that may be required and is not an expected outcome of ongoing or future efforts by the
cities of Cambridge and Somerville.

Modeling of CSS for BATHTUB translates to adding the combined sewer land area added to the
total acreage for each sub-basin HRU (discussed in Section V), which generate extra stormwater and
groundwater loads and annual flow volumes. The combined sewer drainage areas GIS data were
provided by the cities of Cambridge and Somerville (refer to Figure G-XI-1 through Figure G-XI-3
in Appendix G). The modeled land uses within the CSO drainage areas were extracted by overlaying
the drainage areas and the MassGIS land use and soils data.

It is the ERG team's understanding based on conversations with MassDEP the majority of the CSO
drainage areas draining to outfall SOM007A/MWR205A are treated at the Somerville Marginal
facility and discharging through outfall MWR205, which is located downstream of the Amelia
Earhart Dam. A volumetric comparison of the two outfalls during the critical period indicates that
approximately 90 percent of the volume discharges through outfall MWR205 to the Massachusetts
Bay, while 10 percent is discharging into the Upper Mystic River at SOM 007A/MWR205A.
Consequently, for the purposes of evaluating future conditions, the ERG team has assumed that 10
percent of the land uses that were assumed to be contributing to outfall SOM007A/MWR205A are
discharging to the Mystic River above the dam. Further, because the Somerville Marginal CSO
Facility provides screening and disinfection only, there are no reductions in TP concentrations
assumed for future conditions.

To evaluate the impact of the three CSS future conditions, the percent CSS (0, 25, or 100 percent)
was multiplied by the land uses contributing to either the Alewife Brook tributary or the Lower

104


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Basin reach. The land areas for each were added to the HRUs, which were then summarized for the
BATHTUB model. In addition, the CSO loads were reduced according to the CSS future condition
(25 or 100 percent), so that for the scenario with 100 percent CSS, the CSO loads were zeroed.

Other Stormwater Loads

No additional changes to the land uses or stormwater loads were assumed (e.g., new development or
redevelopment of land).

Groundwater Loads

Since groundwater loads are computed from stormwater loads, changes due to the CSS future
conditions results in changes to groundwater loads. The scale of the impact is dependent on the
percent CSS noted above and proportional to the change in stormwater loads.

When stormwater reductions were made in the various scenarios, the groundwater loads were not
changed.

Sediment Nutrient Efflux Loads

Under future conditions the ERG team assumed that there would be a decrease in sediment
nutrients as a result of management and land loads For modeling purposes, the ERG team
estimated that the sediment load is reduced by 50 percent of the estimated stormwater phosphorus
load reduction (e.g., Stormwater TP Load Reductions = 60 percent, Sediment Load Reductions = 30
percent). This methodology is slightly more conservative than the method used for the
Upper/Middle Charles Nutrient TMDL (EPA/DEP, 2011). Under that method, there was a 25
percent sediment nutrient efflux load reduction that could be assumed with the reduction of
phosphorus loads and sediment loads were approximately 50 percent of the total phosphorus load
reduction (including CSO/SSO loads, groundwater, etc.) required to meet the water quality target
(e.g., Total TP Load Reductions = 50 percent, Sediment Load Reductions = 25 percent).

IX.B. Scenarios for Evaluation of Phosphorous Reduction

The ERG team developed four scenarios to evaluate the watershed conditions necessary for the
water quality target to be met: existing conditions (1 scenario) and future conditions (3 scenarios)
(Table IX-1). As identified in Section IXA, future conditions include reductions in CSOs and SSOs,
in addition to evaluating conditions with CSS. Each of these four scenarios were run with a baseline
model (#1 ,2, 3, 4) and a water quality (WQ) target model (# 1A, 2A, 3A, 4A) to provide
comparisons between the results.

105


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table IX-1. Modeled Scenarios for Phosphorus Load Reduction Evaluations

Scenario #

Scenario Name

CSO Load
Reductions3

SSO Load
Reductions'3

CSSC

Stormwater

Load
Reductions

Sediment

Load
Reductions

1

Existing Conditions - Baseline











1A

Existing Conditions - WQ Target







X

X

2

Future Conditions 1 - Baseline

X

X







2A

Future Conditions 1 - WQ Target

X

X



X

X

3

Future Conditions 2 - Baseline

X

X

25%





3A

Future Conditions 2 - WQ Target

X

X

25%

X

X

4

Future Conditions 3 - Baseline

X

X

100%





4A

Future Conditions 3 - WQ Target

X

X

100%

X

X

a)	CSO volumes to the Alewife Brook and Mystic River (Lower Basin) reduced to meet the LTCP target.

b)	SSO volume reductions at 50 percent across all tributaries and sub basins.

c)	CSS percentage indicates percent of combined sewer area that is assumed to be separated. The separated

land uses are added to the sub-basin area and generate additional stormwater flow and loads.

For each of the WQ target scenarios (#1A, 2A, 3A, 4A), multiple model runs were completed to
evaluate the reductions in stormwater loads and sediment efflux necessary to meet the water quality
target in all model segments. In Section IV, chlorophyll-a (chl-a) was identified as the WQ target,
specifically the seasonal average chl-a (<10 (Jg/L). TP of 30 |ag/Lwas used as a secondary indicator.
This analysis focused on the seasonal average chl-a because the regressions of the TP and seasonal
average chl-a concentrations performed in Section IV indicate a linear relationship, such that
attainment of the seasonal average chl-a (10 (Jg/L) would provide attainment of the TP
concentration needed to meet water quality goals. Iterations were conducted for each WQ target
scenario until the seasonal average chl-a was met. Predicted TP concentrations are reported in the
results and was compared to the water quality secondary indicator (30 (Jg/L) for reference.

IX. C. Modeling Methodology
IX.C.1. Model Setup

The BATHTUB model used for the load reduction analysis was calibrated by Dr. Nigel Pickering in
November 2018.

IX.C.2. Model Inputs

The BATHTUB model allows the user to change a number of model inputs. The BATHTUB input
parameters that are adjusted for each segment in the model to achieve the various scenarios include
drainage area, flow, total nitrogen, inorganic nitrogen, total phosphorous, orthophosphate, and the
total phosphorus internal loading rate. The parameters are changed for the total segment, including
both unattenuated loads to the reach (from adjacent sub basin area) and the attenuated loads from
inflowing stream (tributaries). The flow and loads were average annual values calculated over the
critical period from 2007 to 2016. The input data for BATHTUB was converted to metric units (see
Section VII) from the units derived from the loading spreadsheet model (e.g., lb./year, ac-in/yr.,

106


-------
Mystic River Watershed TMDL Alternative Development — Final Report

mg/L). None of the calibrated model coefficients, model options, calibration factors, segment
morphometry, atmospheric loads, and global variables were changed for this analysis.

Once the input parameters were changed in the BATHTUB model to reflect the average annual
scenario conditions, the BATHTUB model was run. Model output was exported into an Excel
spreadsheet. The model outputs that were recorded for each scenario run include the average chl-a
and Total P concentrations for each model segment: Upper Lobe (1), Upper Lake (2), Lower Lake
(3), Upper Basin (4), and Lower Basin (5).

IX.C.3. Analysis with Wet and Dry Year Data

The model scenarios described above use the average annual flows and loads calculated over the
critical period (2007-2016). In order to evaluate whether the water quality target would be met
during extreme precipitation, additional model scenarios were developed to evaluate responses to
the water quality target during a wet year and for a dry year. The selection of the years was based on
precipitation and relative drought/wetness condition (as defined by the standard precipitation index
and the 10th and 90th percentile) noted in Section VIII as well as the annual total nutrient loads based
on data from Section V. While 2008 was considered to be the wettest year in the critical period, 2010
had the highest annual loads for total phosphorus and was considered to be moderately wet. 2016
was a severely dry year and had the lowest annual loads for total phosphorus. For extreme
precipitation analyses, 2010 and 2016 were selected for wet year and dry year analyses, respectively.

IX.D. Model Results with Average Annual Data

A total of 29 runs were performed for the water quality target scenarios (Table IX-2). A summary of
the key results is shown in Table IX-3; further detailed inputs and outputs are outlined in Appendix
J. The BATHTUB output spreadsheet files, which are not submitted as part of this memorandum,
can be made available upon request.

The starting point for the stormwater phosphorus load reduction runs was determined based on the
magnitude of reductions that is needed to meet the water quality target without sediment load
reductions, which is approximately in the 70 to 80 percent range. The modeling of the stormwater
phosphorus load reductions process started with reducing the percent reductions for stormwater
each run until the predicted water quality conditions transition from exceeding the target to meeting
it. Once the water quality target was met with only stormwater reductions, the sediment efflux
reductions were incorporated. The sediment efflux was reduced by half of the estimated stormwater
phosphorus load reduction. The iterative process then continued, lessening both the percent
reductions for stormwater loads and sediment efflux (half of the stormwater reductions) until the
water quality target was met. The ending point for the stormwater and sediment efflux reduction
runs was when the water quality target was exceeded for all segments. Text in red font in Table IX-2
indicates the run that met the chl-a water quality target.

Table IX-2. BATHTUB Model Runsawith Average Annual Data

Scenario

Run #

Stormwater
Reduction (%)

Sediment Efflux
Reduction (%)

1

1

0

0

la

2

80

0

3

78

0

107


-------
Mystic River Watershed TMDL Alternative Development — Final Report



4

78

39



5

70

35



6

68

34



7

67

33



8

66

33

2

9

0

0



10

80

0



11

73

0

2a

12

73

37

13

64

32



14

62

31



15

61

31

3

16

0

0



17

75

0



18

73

0

3a

19

73

37

20

63

32



21

61

31



22

60

30

4

23

0

0



24

75

0



25

71

0

4a

26

71

36

27

62

31



28

59

29



29

58

28

a) Red font indicates the run that met the water quality target for chl-a.

Table IX-3 is a summary of the key results for each scenario: the base run and the run that met the
water quality target with stormwater and sediment efflux reductions. This table shows the predicted
phosphorus concentration and chl-a concentration. In addition, the model input, TP Load, is shown
for comparison across scenarios. Cells highlighted in grey indicate segments that do not meet the
water quality target. Figure IX-I and Figure IX-II further show the key results by segment for
phosphorus and chl-a concentrations, respectively. More detailed scenario information is available in
Appendix H.

108


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table IX-3. Scenario Results3 with Average Annual Data

Scenario

Parameter

Upper Lobe

Upper

Lower

Upper

Lower

- Run #





Lake

Lake

Basin

Basin



Total P Load (Ib./yr.)

3,907

416

1,558

3,298

3,849

1-1

Predicted P Cone. (ng/L)

43.4

33.1

33.6

46.2

57.7



Predicted chl-a Cone. (ng/L)

13.9

7.6

7.7

16.4

19.4



SW P Load Reduction (%)

67



P Sediment Efflux Reduction (%)

33

la-7

Total P Load (Ib./yr.)

1,490

161

625

1,553

1,608



Predicted P Cone. (ng/L)

20.8

18.6

19.4

25.3

31.1



Predicted chl-a Cone. (ng/L)

5.9

4.1

4.3

8.1

10.0



Total P Load (Ib./yr.)

3,892

410

1,526

2,968

3,707

2-9

Predicted P Cone. (ng/L)

43.3

33.0

33.3

44.2

55.8



Predicted chl-a Cone. (ng/L)

13.9

7.6

7.7

15.6

18.8



SW P Load Reduction (%)

62



P Sediment Efflux Reduction (%)

31

2a -14

Total P Load (Ib./yr.)

1,655

175

663

1,354

1,632



Predicted P Cone. (ng/L)

22.4

19.8

20.4

24.9

31.1



Predicted chl-a Cone. (ng/L)

6.5

4.4

4.6

7.9

9.9



Total P Load (Ib./yr.)

3,892

410

1,526

3,033

3,704

3-16

Predicted P Cone. (ng/L)

43.3

33.0

33.3

44.2

55.6



Predicted chl-a Cone. (ng/L)

13.8

7.6

7.6

15.6

18.7



SW P Load Reduction (%)

61



P Sediment Efflux Reduction (%)

31

3a-21

Total P Load (Ib./yr.)

1,691

178

677

1,393

1,652



Predicted P Cone. (ng/L)

22.7

20.0

20.6

24.9

31.1



Predicted chl-a Cone. (ng/L)

6.6

4.5

4.6

7.9

9.9



Total P Load (Ib./yr.)

3,892

410

1,526

3,227

3,697

4-23

Predicted P Cone. (ng/L)

43.2

32.9

33.2

43.4

54.5



Predicted chl-a Cone. (ng/L)

13.8

7.5

7.6

15.3

18.3



SW P Load Reduction (%)

59



P Sediment Efflux Reduction (%)

29

4a-28

Total P Load (Ib./yr.)

1,763

186

705

1,490

1,680



Predicted P Cone. (ng/L)

23.4

20.4

21.1

25.0

31.1



Predicted chl-a Cone. (ng/L)

6.9

4.6

4.7

7.9

9.9

a) Grey highlighted cell indicates the predicted value does not meet the chl-a target.

109


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Predicted P Concentrations with Scenarios using Average Annual Data





•







ฆ







•



•



•

•
•









• 1-1

•

i



• la -7





•

• 2-9







• 2a-14





*

• 3-16

I





• 3a - 21

•

ซ



• 4-23

•





• 4a-28









Upper Lobe Upper Lake

Lower Lake

Upper Basin Lower Basin





Segment





Figure IX-i. Predicted Phosphorus Concentrations with Average Annual Data

20

Predicted Chl-a Concentrations with Scenarios using Average Annual Data





•

•
•



16



•







•
•





'J- 14

1

c

.2 12
ra

c

01

110

u
ra
j;

U 8









t





• 1-1







•	la - 7

•	2-9



• • 4



2a- 14
• 3-16

ฆV 8

0)

TJ

ill

•

9



•	3a-21

•	4-23

•	4a-28



•









! !



























Upper Lobe

Upper Lake Lower Lake Upper Basin
Segment

Lower Basin



Figure IX-II. Predicted Chlorophyll-a Concentrations with Average Annual Data
IX.E. Analysis with Wet and Dry Year Data

An additional eight runs were conducted to evaluate whether the chl-a water quality target would be
met during extreme wet and dry years. The starting point for these runs under each scenario was the

110


-------
Mystic River Watershed TMDL Alternative Development — Final Report

run that met the target (refer to Figure IX-II and Figure IX-III). The average annual flow and loads
were replaced by either the wet or dry year annual flow and load values, as appropriate. The same
stormwater reduction and sediment efflux reductions were applied as was used to meet the chl-a
water quality target in the scenario when using the average annual flow and load values. Table IX-4
shows the additional runs that were conducted for this analysis. Text in red font in indicate Table
IX-4 the run that met the chl-a water quality target.

Table IX-4. Model Runs3 with Wet and Dry Year Data

Scenario

Run #

Wet/Dry Year Data

Stormwater
Reduction (%)

Sediment Efflux
Reduction (%)

la

7

Wet (2010)

67

33

la

7

Dry (2016)

67

33

2a

14

Wet (2010)

62

31

2a

14

Dry (2016)

62

31

3a

21

Wet (2010)

61

31

3a

21

Dry (2016)

61

31

4a

28

Wet (2010)

59

29

4a

28

Dry (2016)

59

29

a) Red font indicates the run that met the water quality target for chl-a.

Table IX-5 is a summary of the key results for each run. Cells highlighted in grey indicate segments
that do not meet the water quality target. Figure IX-III and Figure IX-IV further show the key
results by segment for phosphorus and chl-a concentrations, respectively. More detailed scenario
information is available in Appendix H.

Table IX-5. Scenario Results3 with Wet and Dry Year Data

Scenario Parameter Upper Lobe Upper Lower Upper Lower
- Run # Lake Lake Basin Basin

la-7
(Wet)

SW P Load Reduction (%)

67

P Sediment Efflux Reduction (%)

33

Total P Load (Ib./yr.)

2,153

301

1,089

2,758

2,445

Predicted P Cone. (ng/L)

18.9

17.9

19.8

26.2

30.5

Predicted chl-a Cone. (ng/L)

5.2

3.8

4.3

8.3

9.5

la-7
(Dry)

SW P Load Reduction (%)

67

P Sediment Efflux Reduction (%)

33

Total P Load (Ib./yr.)

968

79

353

760

998

Predicted P Cone. (ng/L)

24.9

20.3

20.1

26.3

35

Predicted chl-a Cone. (ng/L)

7.5

4.7

4.6

8.6

11.6

2a -14
(Wet)

SW P Load Reduction (%)

62

P Sediment Efflux Reduction (%)

31

Total P Load (Ib./yr.)

2,328

304

1,053

1,884

2,124

Predicted P Cone. (ng/L)

20.0

18.7

20.2

23.2

27.4

Predicted chl-a Cone. (ng/L)

5.6

4.0

4.4

7.1

8.4

111


-------
Mystic River Watershed TMDL Alternative Development — Final Report

2a-14
(Dry)

SW P Load Reduction (%)

62

P Sediment Efflux Reduction (%)

31

Total P Load (Ib./yr.)

1,090

90

398

944

1,177

Predicted P Cone. (ng/L)

26.9

21.4

21.2

29.0

38.4

Predicted chl-a Cone. (ng/L)

8.3

5.0

4.9

9.7

12.9

3a-21
(Wet)

SW P Load Reduction (%)

61

P Sediment Efflux Reduction (%)

31

Total P Load (Ib./yr.)

2,377

310

1,073

1,939

2,155

Predicted P Cone. (ng/L)

20.4

19

20.4

23.3

27.5

Predicted chl-a Cone. (ng/L)

5.7

4.1

4.4

7.2

8.4

3a-21
(Dry)

SW P Load Reduction (%)

61

P Sediment Efflux Reduction (%)

31

Total P Load (Ib./yr.)

1,115

92

407

965

1,186

Predicted P Cone. (ng/L)

27.2

21.6

21.4

28.9

38.3

Predicted chl-a Cone. (ng/L)

8.4

5

5

9.7

12.9

4a-28
(Wet)

SW P Load Reduction (%)

59

P Sediment Efflux Reduction (%)

29

Total P Load (Ib./yr.)

2,474

322

1,112

2,079

2,206

Predicted P Cone. (ng/L)

21

19.5

21

23.4

27.7

Predicted chl-a Cone. (ng/L)

5.9

4.2

4.5

7.2

8.5

4a-28
(Dry)

SW P Load Reduction (%)

59

P Sediment Efflux Reduction (%)

29

Total P Load (Ib./yr.)

1,164

96

425

1,015

1,188

Predicted P Cone. (ng/L)

28.1

22

21.8

28.7

37.9

Predicted chl-a Cone. (ng/L)

8.7

5.1

5.1

9.6

12.7

a) Grey highlighted cell indicates the predicted value does not meet the chl-a target

112


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Predicted P Concentrations with Scenarios using Wet and Dry Year Data







1

~

•

•
f



•

•

•

ซ
t

ซ

•

•

I
ฆ



•

• •
ซ

•



















Upper Lobe

Upper Lake

Lower Lake
Segment

•	la - 7 (Wet)

•	la-7(Dry)

•	2a -14 (Wet)
2a -14 (Dry)

•	3a - 21 (Wet)

•	3a-21 (Dry)

•	4a - 28 (Wet)

•	4a - 28 (Dry)

Figure IX-III. Predicted Phosphorus Concentrations with Wet and Dry Year Data

III

Predicted Chl-a Concentrations with Scenarios using Wet and Dry Year Data



=2
a



•

• .



c
,0





* la - / (Wet)

|



:

i 1m

1 9

c





• 2d 14jWM>

U



•

• 2a - 14 (D'v'j

Z

D





• 5a - 71 (Wft)

1 6

•
•



• 5a 21 (Dry)

|

t

f

• -1a - 78 (Werl

a.

f

1

ซ la - JS (Dry)

6

Upper lobe Upper lake

tower take Upper Basin lower Rastrt







5ซgmซnt



Figure IX-IV. Predicted Chlorophyll-a Concentrations with Wet and Dry Year Data
IX.F. Discussion

In each base run for the four scenarios, the Upper Lake and Lower Lake segments meet the chl-a
water quality target without any stormwater and sediment load reductions. The Lower Basin

113


-------
Mystic River Watershed TMDL Alternative Development — Final Report

segment requires the most significant reductions to meet the water quality target (i.e., it is the critical
segment in this analysis). Overall, under existing conditions, a 67 percent reduction is required for
stormwater phosphorus loads with a 33 percent reduction in sediment efflux. The stormwater load
reductions required under future conditions (scenarios 2A, 3A and 4A) were between approximately
59 and 62 percent with a 30 to 31 percent required sediment efflux reduction. Table X provides a
total phosphorus load comparison between existing conditions and scenario 2A. Since the focus of
this table is the entire watershed, we report unattenuated land loads here. Under future conditions
scenarios 3A and 4A with CSS, the overall phosphorus loads, and flows increase, though not
proportionally, which appears to be resulting in lower overall phosphorus concentrations and lower
required reductions compared to scenario 2A.

Table IX-6. Total Phosphorus Load Reductions for Scenario 2A

Item

Stormwater

Groundwate
r

cso/sso

Internal

Atmospheri

c

Total

Existing













Conditions













Total P Load













(lb./yr.)

14,887

1,141

1,696

3,793

120

21,638

Scenario 2A













P Load













(lb./yr.

9,974

1,141

412

1,271

120

12,919

Reduction













(%)

67%

0%

24%

34%

0%

60%

For this analysis, stormwater and sediment efflux reductions were made consistently across all
tributaries and sub-basins. However, further analysis can determine what the minimum required
reductions are for the first three segments versus the last two segments (Upper Basin and Lower
Basin) in order to meet the water quality target. For example, in scenario 1 A, run #7, the water
quality target for chl-a is just met in the Lower Basin segment with a concentration of 9.9 (Jg/L.
However, in the Upper Lake and Lower Lake model segments, the concentrations are over half the
water quality target at 4.1 and 4.3 (Jg/L, respectively. It is evident that the Upper Lobe, Upper Lake,
and Lower Lake model segments do not require as significant reductions compared to the Upper
Basin and Lower Basin model segments. An evaluation of high intensity land uses (e.g.,
commercial/industrial uses, high density/multi-family residential, transportation, etc.) such was done
in the Upper/Middle and Lower Charles River TMDL evaluations, including analysis of relative
loads and required reductions by segment, could provide additional clarification in the future as the
TMDL implementation plan is established to meet the water quality targets.

Further analysis with wet and dry year data explores whether the chl-a water quality target can be
met during extreme conditions. During a wet year (such as 2010) the chl-a target appears to be met
without having to further adjust the stormwater reductions or sediment efflux reductions as
compared to the average annual loads. It appears that the wet year provided dilution of the loads,
and perhaps overall improved water quality, so much that less stormwater reductions would be
needed to achieve the target. In contrast, a dry year (such as 2016) would not meet the chl-a target
unless further stormwater reductions were applied. However, because 2016 was an extreme dry year
(below the 10th percentile threshold), the exceedance of water quality targets under these conditions
would be statistically infrequent. More detailed modeling with additional water quality data could

114


-------
Mystic River Watershed TMDL Alternative Development — Final Report

help to provide greater confidence in the recommended phosphorus load reductions and the
potential exceedances of chl-a.

115


-------
Mystic River Watershed TMDL Alternative Development — Final Report

X. Broad-Based Nutrient Stormwater Management
Strategies for the Mystic River Watershed using Opti-Tool

Highly urbanized areas often have limited opportunities for implementing large-scale Stormwater
Control Measures (SCMs) for treating stormwater runoff. Distributed green infrastructure (GI)
practices can provide cost-effective solutions that achieve load reduction numeric targets while
effectively integrating within urbanized landscapes. In New England, almost 50 percent of daily
rainfall events are less than 0.3 inches. The relatively small size of distributed GI facilities
substantially increases the feasibility to provide treatment to runoff from impervious surfaces in
constrained developed spaces and achieve meaningful water quality benefits in receiving waters.

Strategically optimizing the selection and placement of distributed SCMs within highly urbanized
settings can also help to develop management strategies that are more cost-effective than the
traditional approach of sizing BMPs at fixed locations to treat a design storm. The Opti-Tool, which
was developed for the United States Environmental Protection Agency (USEPA) Region 1, is a
continuous simulation model that can be used to optimize the selection and placement of distributed
GI practices at a watershed scale. This case study demonstrates how the Opti-tool can be used to
help with stormwater management planning in urban New England settings and highlights the value
of conducting strategic planning to address stormwater impacts for achieving water resource goals.
It presents an analytical framework that can be readily customized and applied in other settings to
inform the stormwater management planning effort. This section also provides examples of GI
implementation efforts in other locations where distributed GI practices were also found to be cost-
effective stormwater management strategies.

X.A. Study Objectives

Two stormwater management scenarios were formulated to evaluate two different stormwater
management approaches that meet the required annual TP load reduction target for the Mystic River
Watershed. These scenarios were configured and optimized using the Opti-Tool:

1.	Design-Storm Objective: Optimize distributed BMP locations by land use type with fixed
sizes to capture 1 inch of runoff for a design storm and develop cost-effectiveness curve
(CE-Curve).

2.	Mix-Storm Objective: Optimize distributed BMP locations and sizes by land use type to
reflect flexible sizing approach and develop CE-Curve.

This section presents a step-by-step, high-level technical approach to identify structural controls
associated with cost-effective stormwater management strategies. The Mystic River Watershed in
Massachusetts was selected as a watershed to test the sensitivities of the two stormwater
management scenarios and to identify the most cost-effective management approach that achieves
the phosphorus reduction objectives.

X.B. Pilot Sub-Watershed Selection

The Mystic River Watershed consists of 15 sub-watersheds, which drain into three primary river
segments, and seven large impoundments (ponds/lakes). This stormwater management portion of
the project uses one Mystic River sub-watershed to implement a high-level, generalized approach
and step-by-step guidance that is transferable to other sub-watersheds.

116


-------
Mystic River Watershed TMDL Alternative Development — Final Report

The pilot sub-watershed that was selected is representative of the overall land use distribution in the
Mystic River Watershed. This comparison was made by computing the root mean square error
(RMSE) of the percent area distribution of each land use category between individual sub-
watersheds and the entire Mystic River Watershed. The four watersheds with the minimum RMSE
(indicating a close match to the overall watershed distribution) were: Maiden. Tudkins. Lower Mystic.
and Mystic River. Each of the four were intersected with municipal boundaries. The Mystic River
sub-watershed was selected for the pilot study because of its central location within the watershed. It
also represents a large portion of one community, the City of Medford. The pilot sub-watershed is
highlighted in yellow in Figure X-I.

The pilot sub-watershed comprises 5,151 acres of land and 179 acres of water bodies. Combined
Sewer Overflow (CSO) basins within the pilot sub-watershed drain 1,010 acres of land. Half of the
sub-watershed area is on low-slope topography (i.e., less than 5%) and 12% of the sub-watershed
area has slope larger than 15% (Table X-l, Figure X-II). The dominant soil type in the pilot
watershed is hydrologic soil group (HSG) C, which makes up 70% of the total watershed area (Table
X-2, Figure X-III). Soil map units with no HSG attribute data were assumed to be C soils (Group C
soils have relatively low infiltration rates). The dominant land use type in the pilot watershed is high-
density residential (46%) followed by forest (24%) and commercial (15%) land uses (Figure X-IV).
The pilot watershed is 49% impervious and the impervious portion of the watershed is mostly high-
density residential (28%), followed by commercial (11%) and transportation (highways) (4%). The
pervious portion of the watershed is mostly forest (23%) followed by high density residential
pervious (18%) and commercial pervious (4%), as shown in Figure X-V and Table X-3.

One-fifth of the pilot sub-watershed area falls within a CSO basin. That area was excluded from the
analysis because the CSO drainage area does not include separate stormwater drainage systems that
discharge to the receiving water. Thus, no stormwater controls were explored in CSO areas.

117


-------
My stic River Watershed TMDL Alternative Development — Final Report

Streams
Lakes

AbeijonaRiverl
I I AbeijonaRiver2
Alewife
BlacksNook
Horn
Judkins

LowerMysticLake
Maiden

MysticRiver
UpperLobe
UpperMysticLake
Spy Pond
Wedge
fed! Winter

\//\ Cambridge CSO Basin 2017
, I//1 Somerville CSO Basins 2017

/ *	<.*> ฆ

Figure X-l. Location Map of Pilot Sub-Watershed, Mystic River (Highlighted in

Yellow)

Table X-1. Ground Slope Classification in Pilot Sub-Watershed, Mystic River

Ground Slope Classification

Percent Slope

Area (acres)

Low

0% - 5%

2,852.92

Moderate

5% -15%

1,861.32

High

>15%

615.65

Total area

5,329.89

118


-------
Mystic River Watershed TMDL Alternative Development — Final Report

I I Mystic River Subwatershed
VZA Somerville CSO Basin
Slope Classification

0% - 5%

4 5% -15%
f Wk >15%

JTErmh*... J
3 mile

Figure X-ll. Ground Slope Map of Pilot Sub-Watershed, Mystic River
Table X-2. Hydrologic Soil Group Classification in Pilot Sub-Watershed, Mystic River

Soil Classification

HSG

Area (acres)

No Data

C

3,601.45

HSG-B

B

560.90

HSG-C

C

85.04

HSG-C/D

C/D

283.45

HSG-D

D

799.05

Total area

5,329.89

119


-------
Mystic River Watershed TMDL Alternative Development — Final Report

I I Mystic River Subwatershed
Y/A Somerville CSO Basin
Hydrologic Soil Group
No Data

-IsmAbi..

3 mile

Figure X-lll. Soil Map (Hydrologic Soil Group) of Pilot Sub-Watershed, Mystic

River

120


-------
Mystic River Watershed TMDL Alternative Development — Final Report

High Oensity Residential, 46%

Highway, 6%

Industrial, 4%

Forest, 24%

Commercial, 15%

* I I Mystic River Subwatershed .
V7/1 Somerville CSO Basin
Landuse Classification
$ Agriculture
ฃ ฆ Commercial
ฆH Forest

H High Density Residential

Highway	Pปt

Industrial

Low Density Residential
4 ฆ Medium Density Residential
Open land

Figure X-IV. Land Use Map of Pilot Sub-Watershed, Mystic River

121


-------
Mystic River Watershed TMDL Alternative Development — Final Report

High Density Residential, 28%

I I Mystic River Subwatershed
V/A Somerville CSO Basin
Land Cover Type
>5 It Impervious
Pervious

Pervious Cover

Commercial, 4%
Forest, 23%

High Density Residential, 18%

Highway, 2% ,ndustrial> 1%
Open land, 3%

Impervious Cover

Commercial, 11%

3 mile

Figure X-V, Land Cover Map of Pilot Sub-Watershed, Mystic River

122


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table X-3. Land Use Classification in Pilot Sub-Watershed, Mystic River

Land Use Classification

Area (acres)

Percent Impervious

Percent Pervious

Commercial

769.79

11%

4%

Forest

1,225.94

0%

23%

High Density Residential

2,370.66

28%

18%

Highway

294.14

4%

2%

Industrial

235.22

4%

1%

Low Density Residential

0.55

0%

0%

Medium Density Residential

3.09

0%

0%

Open land

251.20

2%

3%

Water

179.30





Total

5,329.89

49%

51%

X.C. Technical Approach

The Opti-Tool provides the ability to evaluate options for determining the best mix of structural
BMPs to achieve water quality goals. Structural BMPs are permanent structures, provide stormwater
storage capacity, and rely upon vegetation and soil mechanisms in order to perform as intended. The
tool incorporates long-term runoff responses (HRU timeseries) for regional climate conditions that
are calibrated to regionally representative stormwater data and annual average pollutant load export
rates from nine land uses. The tool uses regionally representative BMP cost functions and regionally
calibrated BMP performance parameters for four pollutants, including total phosphorus, to calculate
long-term cumulative load reductions for a variety of structural controls. Structural controls
simulated by the tool include low impact development (LID) and green infrastructure (GI) practices,
such as infiltration systems, bio-filtration, and gravel wetlands.

The technical approach for applying the Opti-Tool is organized into three general steps:

1.	Develop stormwater management categories for SCMs known to be highly effective at
removing phosphorus (e.g., shallow filtration, infiltration, biofiltration) based on the site
suitability analysis of GIS layers;

2.	Estimate the available opportunity by BMP type (i.e., physical footprint area) within each
management category and summarize the upstream impervious drainage area that can be
managed for each management category, and

3.	Set up and run the Opti-Tool application to identify the most cost-effective combination of
BMP options that achieve the desired management objectives.

X.C.1. Stormwater Management Categories Development

Spatial data analyses were conducted during Phase 1 of the project to characterize watershed
features and identify the corresponding stormwater management categories that were suitable for
application with the Opti-Tool for the pilot study area. The GIS data used for the evaluation of
stormwater management categories for the Mystic River Watershed include: municipal boundaries,
watershed sub-basins, land use coverage, impervious cover, Hydrologic Soil Group (HSG),
wetlands, Digital Elevation Model (DEM) for ground slopes, Activity and Use Limitation (AUL)
and MGL Ch. 21E sites (for contaminated land), and property ownership. All data are from

123


-------
Mystic River Watershed TMDL Alternative Development — Final Report

MassGIS data layers except the watershed sub-basins, which were derived using FEMA catchments
and DEM data as described earlier in this report.

The following assumptions were made to develop the stormwater management categories and
estimate available BMP opportunity:

1.	Areas with no depth to groundwater data were assumed to have a depth to groundwater
greater than 2.5 feet.

2.	Areas with no HSG identified for soils were assumed to be classified as HSG C. This was
typically the predominant soil type in urban areas of the Mystic River Watershed.

3.	The extent of potential contamination from AUL and/or MGL Ch. 21E sites was
approximated using the parcel in which the site was located.

4.	Wetland areas were not included within the management categories because they were not
considered candidates for implementation of stormwater practices.

5.	Both public and private land areas were assessed for identifying the GI opportunity areas
based on the site suitability criteria.

Areas with impervious cover (IC) were also explored for certain management practices (i.e., porous
pavement) that not only replace IC but can also treat stormwater from adjacent IC (for example,
within parking lots), but at a higher unit cost. Table X-4 presents siting criteria for potential
stormwater management categories (SMC), which were derived from the GIS data analysis. Table
X-5 shows the maximum footprint areas of each SMC that is available in the pilot watershed and the
spatial locations of those SMCs are shown in Figure X-VI.

124


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table X-4. Potential Stormwater Management Categories and BMP Types in Opti-

Tool

Cover Type

Ground
Slope (%)

AUL/21E

HSG

Management
Category

BMP Type(s) in Opti-
Tool

Pervious
Area

LO
I

II

V

Is a AUL/21E
Site

A/B/C/D or No Data
(HSG C assumed)

Shallow
filtration

Biofiltration (e.g.,
Bioretention with
underdrain option)

Not a AUL/
21E Site

A/B/C or No Data
(HSG C assumed)

Infiltration

Surface Infiltration
Basin (e.g., Rain
Garden)

D

Biofiltration

Biofiltration (e.g.,
Gravel Wetland)

> 15

-

-

Less likely for
onsite BMP

-

Impervious
Area

<= 5

-

A/B/C/D or No Data
(HSG C assumed)

Shallow
filtration

Porous Pavement

>5

-

-

Less likely for
onsite BMP

-

Table X-5. Potential BMP Opportunity Areas (Maximum Footprints) in the Pilot

Sub-Watershed, Mystic River

Stormwater Management Category

Area (acres)

Percent Area (%)

Biofiltration

437.92

8%

Shallow filtration (pervious area)

513.13

10%

Infiltration

1,087.20

20%

Shallow filtration (impervious area)

1,477.06

28%

Less likely for BMP

1,814.58

34%

Total

5,329.89

100%

125


-------
Mystic River Watershed TMDL Alternative Development — Final Report

1 I Mystic River Subwatershed
V/A Somerville CSO Basin
Stormwater Management Categories

Biofiltration
H Infiltration

Less likely for BMP
PorousPavement
Shallow filtration

Oake/ Qckrtc/ f	V

Figure X-VI. Stormwater Management Categories Map of Pilot Sub-Watershed,

Mystic River

For this study, stormwater management siting was primarily evaluated for areas with pervious cover,
outside of wetland areas, and with ground slopes of less than 15%. Porous pavement on impervious
cover with ground slopes of less than 5% was also evaluated with the assumption of drainage
impervious area ratio of 2:1, meaning one acre of porous pavement was assumed to treat its own
footprint plus one additional acre of adjacent impervious land. Only 10% of suitable impervious
land (based on the siting criterion) was identified for this practice. While removal of some
impervious land cover and implementation of porous pavement are viable options within this
watershed, those practices are typically costlier than stormwater practices located on pervious land.
Therefore, a multiplier of 3X (unit cost) for installing porous pavement was used in the Opti-Tool to

126


-------
Mystic River Watershed TMDL Alternative Development — Final Report

account for installing underdrains and connections to the drainage system. In comparison, a
multiplier of 2x (unit cost) was used for other BMPs for installing new BMPs in developed areas.

X.C.2. Estimating BMP Footprints and Impervious Drainage Areas

The distribution of the BMP opportunity areas (i.e., BMP footprints) was estimated by land use
category group. This distribution represents the maximum available BMP footprint in the pilot
watershed, based on GIS spatial data analysis, and does not necessarily represent the actual
opportunity areas. The total impervious areas by land use group were proportionally distributed to
the BMP drainage areas based on the available percentage of opportunity area of that specific BMP
type by land use type as determined through the Management Category analysis (Table X-6 and
Figure X-VII). For example, if the opportunity area of Bio-filtration was 20% of the total available
opportunity area in commercial land, then 20% of the impervious area in the commercial land was
treated by Bio-filtration practices located on commercial land. For this case study, no field
verification was performed, and maximum opportunity areas were set to limit the BMP footprints
needed to capture up to 1 inch of runoff from the impervious drainage areas (Table X-7).

Table X-6. BMP-Treated Impervious Area (Drainage Area) Distribution by Land
Use Category Group in the Pilot Sub-Watershed, Mystic River

Land Use Type

Biofilt ration

Infiltration
HSG-B

Infiltration
HSG-C

Porous
Pavement

Total
(Land Use)

(acre)

(acre)

(acre)

(acre)

(acre)

Commercial

167.92

20.95

114.01

55.02

357.9

High Density Residential

9.74

31.78

714.35

116.01

871.88

Highway

86.81

13.67

46.76

-

147.24

Industrial

-

28.17

61.24

18.84

108.25

Open land

5.25

17.26

16.09

7.86

46.46

Total (BMP-treated)

269.72

111.83

952.45

197.73

1,531.73

127


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Treated Impervious Areas by BMP Type and Landuse Type

ฆ Commercial ฆ Industrial ฆ High Density Residential ฆ Open land ฆ Highway

Figure X-VII. Treated Impervious Areas by BMP Type and Land Use Type in the

Pilot Sub-Watershed, Mystic River

Table X-7. BMP Area (Footprints) Distribution by Land Use Category Group
Required to Treat 1 Inch of Runoff from the Impervious Surface in the Pilot Sub-

Watershed, Mystic River

Land Use Type

Biofilt ration

Infiltration
HSG-B

Infiltration
HSG-C

Porous
Pavement

Total
(Land Use)

(acre)

(acre)

(acre)

(acre)

(acre)

Commercial

10.00

0.87

4.75

27.51

43.13

High Density Residential

0.58

1.32

29.76

58.00

89.66

Highway

5.17

0.57

1.95

-

7.69

Industrial

-

1,17

2.55

9.42

13.14

Open land

0.31

0.72

0.67

3.93

5.63

Total (BMP-treated)

16,06

4,65

39.68

98.86

159.25

X.C.3. Opti-Tool Setup7

The following steps were performed to set up the Opti-Tool for the pilot sub-watershed.

1. Establish baseline condition: The climate data was extended to develop unit-area HRU
timeseries for the critical period of interest (Jan 2007 — Dec 2016), which was used as the

7 Opti-Tool User Guide provides the step-by-step instructions on how to setup the Opti-Tool project.

128


-------
Mystic River Watershed TMDL Alternative Development — Final Report

boundary condition to the BMP simulation model. The Opti-Tool provides a utility tool that
runs the SWMM models, calibrated to Region 1 specific land use average annual loading
export rates, and generates the HRU hourly timeseries in the format needed for the Opti-
Tool. These are the same loading HRUs being used in the watershed and water quality
modeling work done as part of the Mystic River Eutrophication analysis to determine
needed TP load reductions.

2.	Management objective: Identify the most cost-effective stormwater controls (types and
sizes) for achieving a wide range of TP load reductions at the watershed scale.

3.	Optimization target: Develop cost effectiveness curve (CE-curve) for TP average annual
load reduction.

4.	Land use information: Estimate the area distribution for the major land use groups within
the pilot watershed. Assign the corresponding unit-area HRU timeseries for each land use
group in the model.

5.	BMP information: Eighteen BMP types were selected on five major land use categories
based on the Management Category analysis and BMP specifications were set using the
default parameters and BMP cost function available in the Opti-Tool, see Appendix K).
Assign impervious drainage areas to be treated by each BMP type in the model.

6.	Run optimization scenario: Define the simulation period (2007 - 2016), the pollutant of
concern (TP), the objective function (minimize cost), and create an input file for the
optimization run. Run the optimization using the continuous simulation BMP model to
reflect actual long-term precipitation conditions that includes a wide range of actual storm
sizes to find the optimal BMP storage capacities that provide the most cost-effective
solution at the watershed scale. Each optimization run generates a CE-Curve showing the
optimal solutions frontier for a wide range of TP load reduction targets.

Figure X-VIII shows the main interface of the Opti-Tool, the left panel guides the user to follow
steps in the chronological order. The right panel allows the user to place BMPs on the map and
enter design specifications for each BMP type.

129


-------
Mystic River Watershed TMDL Alternative Development — Final Report

X.D. Management Scenarios

In this study, two management scenarios were created and optimized using the Opti-Tool. The most
cost-effective solution from each scenario was selected that met the TP average: annual load
reduction target for the pilot watershed. Two numeric targets; 67% and 62% (load reduction target
for existing condition [Section 1\. \. 11 and tuture condition 1 [Section IX.A.2], respectively) were
evaluated for each management scenario- The existing condition's target requires the load reduction
from stormwater only whereas the load reduction target for future condition 1 assumes additional
reductions to CSO volumes that meet the long-term control plan (LTCP) estimates for the typical
rainfall year and 50% sanitary sewer overflow (SSO) volume reductions for the ongoing SSO
mitigation work being done within the Mystic River Watershed. The load reduction targets for
tuture condition 2 and future condition 3 were not optimized in this study as it requires changing the
baseline by shifting land areas from the CSS areas. The details on the existing and future conditions
are discussed in Section IX of this report.

Scenario 1: Sizing the BMPs to capture one inch of surface runoff from the impervious drainage
areas and spatial optimization for the strategic locations (at the land use level) in the pilot
watersheds. The BMP sizes were fixed, and the optimization engine explored the best mix of BMP
types and strategic locations to identify the cost-effective solutions.

Scenario 2: Sizing the BMPs to capture from one-tenth to an inch of surface runoff from the
impervious drainage areas and spatial optimization for the strategic locations (at the land use level) in
the pilot watersheds. The BMP sizes were variable (increment of one-tenth of an inch to a maximum

130


-------
Mystic River Watershed TMDL Alternative Development — Final Report

of 1 inch) and the optimization engine explored the best mix of BMP types, sizes, and strategic
locations to identify the cost-effective solutions.

X.D.1. Results: Scenario 1

Figure X-IX shows the CE-Curve for Scenario 1 from the model simulation in the Opti-Tool. The

Scenario: The scenario term is used to describe the BMP siting criteria; scenario 1 captures a typical one-inch
storm si^e (i.e., one-inch of runoff depth from the IC) whereas scenario 2 captures a range of storm si ^es from
one-tenth to a one-inch storm.

Condition: The condition term represents the target sources; existing condition taigets the stormwater load
reduction only whereas future condition 1 also taigets the CSO and SSO had reductions in addition to the
stormwater had reduction.

Solution: The solution term represents the optimised mixture of different BMP types, si^es, and locations
that meet the given numeric had reduction target.

Cost: The coit estimates are intendedfor planning levelpurposes and are intended to highlight relative cost
differences among the scenarios.

curve is an interactive plot showing the target solution (red triangle for the existing condition and
orange triangle for the future condition 1) and all the iterations performed during the optimization
process. The grey-dots on the curve are the inferior solutions and the blue-diamonds form the cost-
effectiveness curve for a wide range of load reduction targets. Based on the given target reduction,
Opti-Tool searches for the closest solution and provides the information on the selected BMPs
under that target solution (BMP ID, BMP type, surface area, storage depth, treated the impervious
area, runoff depth, annual maintenance hours, and BMP cost).

The results of Scenario 1 (where BMPs are sized using a typical design criterion of capturing one inch
of runoff from the impervious drainage area) show that it would cost $106.37 million to meet a 67%
TP average annual load reduction target for the existing condition whereas it would cost $66.12
million to meet a 62% TP average annual load reduction target for the future condition 1 for the
pilot watershed (Figure X-9). The cost estimates are based on regional unit cost information for the
control types, a 35% add-on for engineering and contingencies and a site factor multiplier to account
for anticipated difficulties associated with installations. For this analysis, a multiplier of 2Xwas
assumed for all controls except for porous pavement for which a 3X multiplier was applied. These
cost estimates can be considered conservative because they do not reflect the potential for
significant cost offsets that could be achieved through the installation of SCMs as part of other
development/redevelopment, urban renewal and roadwork related projects.

Though the optimization engine did not optimize the BMP sizes, it still shows a significant co st
saving in optimizing the strategic locations (where to place a BMP and what BMP combination to
use). Table X-8 shows the selected BMP types in the optimal solution that meet the TP load
reduction target for the existing condition. Table X-9 shows the selected BMP types in the optimal
solution that meet the TP load reduction target for the future condition 1. The optimizer preferred
the infiltration BMPs because they provided the highest volume reduction and associated water
quality benefits compared to the more expensive practices such as biofiltration and porous
pavements.

131


-------
My stic River Watershed TMDL Alternative Development — Final Report



AllSolutions ~ Best Solutions A Target Solution 1

Target Solution 2

Target Solution 3

% TP Reduction
Annual Average Load

Figure X-IX. Scenario 1: Opti-Tool Cost-Effectiveness Curve (Optimize Locations
Only) of TP Annual Average Load Reduction for the Pilot Watershed

The shape of the Civ Curve itself provides valuable information for informing strategic stormwater
management planning. As indicated, the slope of the curve is relatively mild from 0" <> to 58% in TP
reduction and then increases sharply for higher TP reductions. The incremental cost of $ 40 million
to move from a TP reduction target of 62% to 67% is substantially higher than the $5 million
incremental cost increase associated with the same incremental increase in percent TP reduction of
53% to 58% on the flatter part of the curve. This curve highlights the potential high value of
investing in other measures including nonstructural control such as leaf litter management, high-
efficiency street cleaning, catch basin cleaning and fertilizer management that could achieve TP
reductions of 10-20% or higher and shift the target for the structural control retrofit program from
the steep part of the curve to the flatter portion. For example, assume that a 15% TP reduction
could be accomplished through nonstructural controls and that the target for structural control can
be reduced from 67% to 52% resulting in an estimated cost for structural controls of approximately
$40 million (less than half).. Not only is it far less costly to move from the steep portion of the curve
to the flatter portion but the total amount ot impervious cover area requiring treatment for a 52 %
reduction (909 acres) is substantially less than the 67% (1,379 acres) and 62% (1,211 acres) options
as indicated in Table X-10, Table X-9, and Table X-9 respectively.

132


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table X-8. Scenario 1: BMP Types and BMP Sizes for The Selected Target
Solution 1 (Existing Condition) in Opti-Tool

BMPID

BMP Type

Land Use

Treated
Impervious
Area (acres)

Runoff
Depth
(in.)

BMP
Storage
Capacity
(gallon)

BMP Cost ($)

BMP1

Infiltration-B

High Density Residential

31.78

1.00

863,013

$1,439,764

BMP2

Infiltration-B

Commercial

20.95

1.00

568,931

$949,148

BMP3

Infiltration-B

Industrial

28.17

1.00

764,973

$1,276,203

BMP4

Infiltration-B

Open land

17.26

1.00

468,813

$782,121

BMP5

Infiltration-B

Highway

13.67

1.00

371,150

$619,189

BMP6

Infiltration-C

High Density Residential

714.35

1.00

19,401,598

$32,367,663

BMP7

Infiltration-C

Commercial

114.01

1.00

3,096,513

$5,165,908

BMP8

Infiltration-C

Industrial

61.24

1.00

1,663,216

$2,774,741

BMP9

Infiltration-C

Open land

16.09

1.00

437,036

$729,107

BMP10

Infiltration-C

Highway

46.76

1.00

1,270,010

$2,118,756

BMP11

Biofiltration

High Density Residential

-

-

-

-

BMP12

Biofiltration

Commercial

167.92

1.00

4,559,847

$18,847,683

BMP13

Biofiltration

Open land

5.25

1.00

142,556

$589,242

BMP14

Biofiltration

Highway

86.81

1.00

2,357,192

$9,743,223

BMP15

Porous Pavement

High Density Residential

-

-

-

-

BMP16

Porous Pavement

Commercial

55.03

-

13,573,865

$28,963,024

BMP17

Porous Pavement

Industrial

-

-

-

-

BMP18

Porous Pavement

Open land

-

-

-

-

Total

1,379.27

1.00

49,538,712

$106,365,771

Table X-9. Scenario 1: BMP Types and BMP Sizes for the Selected Target Solution

2 (Future Condition 1) in Opti-Tool

BMPID

BMP Type

Land Use

Treated
Impervious
Area (acres)

Runoff
Depth
(in.)

BMP
Storage
Capacity
(gallon)

BMP Cost ($)

BMP1

Infiltration-B

High Density Residential

31.78

1.00

863,013

$1,439,764

BMP2

Infiltration-B

Commercial

-

-

-

-

BMP3

Infiltration-B

Industrial

28.17

1.00

764,973

$1,276,203

BMP4

Infiltration-B

Open land

17.26

1.00

468,813

$782,121

BMP5

Infiltration-B

Highway

13.67

1.00

371,150

$619,189

BMP6

Infiltration-C

High Density Residential

714.35

1.00

19,401,598

$32,367,663

BMP7

Infiltration-C

Commercial

114.01

1.00

3,096,513

$5,165,908

BMP8

Infiltration-C

Industrial

61.24

1.00

1,663,216

$2,774,741

BMP9

Infiltration-C

Open land

16.09

1.00

437,036

$729,107

BMP10

Infiltration-C

Highway

46.76

1.00

1,270,010

$2,118,756

133


-------
Mystic River Watershed TMDL Alternative Development — Final Report

BMPID

BMP Type

Land Use

Treated
Impervious
Area (acres)

Runoff
Depth
(in.)

BMP
Storage
Capacity
(gallon)

BMP Cost ($)

BMP11

Biofiltration

High Density Residential

-

-

-

-

BMP12

Biofiltration

Commercial

167.92

1.00

4,559,847

$18,847,683

BMP13

Biofiltration

Open land

-

-

-

-

BMP14

Biofiltration

Highway

-

-

-

-

BMP15

Porous Pavement

High Density Residential

-

-

-

-

BMP16

Porous Pavement

Commercial

-

-

-

-

BMP17

Porous Pavement

Industrial

-

-

-

-

BMP18

Porous Pavement

Open land

-

-

-

-

Total

1,211.24

1.00

32,896,167

$66,121,134

Table X-10. Scenario 1: BMP Types and BMP Sizes for the Selected Target

Solution 3 (52% Target) in Opti-Tool

BMPID

BMP Type

Land Use

Treated
Impervious
Area (acres)

Runoff
Depth
(in.)

BMP
Storage
Capacity
(gallon)

BMP Cost ($)

BMP1

Infiltration-B

High Density Residential

31.78

1.00

863,013

$1,439,764

BMP2

Infiltration-B

Commercial

20.95

1.00

568,931

$949,148

BMP3

Infiltration-B

Industrial

28.17

1.00

764,973

$1,276,203

BMP4

Infiltration-B

Open land

-

-

-

-

BMP5

Infiltration-B

Highway

-

-

-

-

BMP6

Infiltration-C

High Density Residential

714.35

1.00

19,401,598

$32,367,663

BMP7

Infiltration-C

Commercial

114.01

1.00

3,096,513

$5,165,908

BMP8

Infiltration-C

Industrial

-

-

-

-

BMP9

Infiltration-C

Open land

-

-

-

-

BMP10

Infiltration-C

Highway

-

-

-

-

BMP11

Biofiltration

High Density Residential

-

-

-

-

BMP12

Biofiltration

Commercial

-

-

-

-

BMP13

Biofiltration

Open land

-

-

-

-

BMP14

Biofiltration

Highway

-

-

-

-

BMP15

Porous Pavement

High Density Residential

-

-

-

-

BMP16

Porous Pavement

Commercial

-

-

-

-

BMP17

Porous Pavement

Industrial

-

-

-

-

BMP18

Porous Pavement

Open land

-

-

-

-

Total

909.25

1.00

24,695,027

$41,198,686

134


-------
Mystic River Watershed TMDL Alternative Development — Final Report

X.D.2. Results: Scenario 2

The results of Scenario 2 (where BMPs are sized to capture from one-tenth to an inch of runoff from
the impervious drainage area) show that it would cost $102.83 million to meet a 67% TP average
annual load reduction target for the existing condition whereas it would cost $51.13 million to meet
a 62% TP average annual load reduction target for the future condition 1 for the pilot watershed
(Figure X-X). For this scenario, the optimization engine optimized the BMP sizes and the strategic
locations (i.e., where to place a BMP, what BMP size to pick, and what BMP combination to use).
The results show more cost saving ($3.5 million for existing condition and $15 million for future
condition 1) for optimizing the BMP sizes as compared to only optimizing the strategic locations.
Table X-ll shows the selected BMP types in the optimal solution that meet the TP load reduction
target for the existing condition. Table X-12 shows the selected BMP types in the optimal solution
that meet the TP load reduction target for the future condition 1 for Scenario 2. However, similar to
scenario 1, the TP reduction targets of 67% and 62% are located on the portion of the curve with
the steeper slope indicating much higher incremental cost increases for increasing TP reduction
targets. Again, assuming that nonstructural controls could achieve a 15% TP reduction, the
estimated cost of achieving a 52% reduction using structural controls is $20 million and equal to 'A
of the estimated cost ($40 million) for achieving the same reduction using 1-inch design capacities.
Table X-13 summarizes the results of achieving 52% TP reduction for scenario 2. As indicate,
optimized sizing of the structural controls for 52% are notably smaller than was determined for the
62% and 67% reductions.

The optimizer mostly picks the infiltration BMPs because they provide the highest volume reduction
and water quality benefits compared to more expensive practices such as biofiltration and porous
pavement. For Scenario 2, the optimizer picks a combination of BMP sizes ranging from 0.1 to 1.0
inches, showing that BMPs designed to manage smaller-size storms in New England region may
provide more cost saving and increased feasibility for implementation in the highly urbanized
watershed such as the Mystic River.

135


-------
My stic River Watershed TMDL Alternative Development — Final Report

All Solutions ~Best Solutions A Target Solution 1 t Target Solution 2 Target Solution 3

% TP Reduction
Annual Average Load

Figure X-X. Scenario 2: Opti-Tool Cost-Effectiveness Curve (Optimize Locations
and BMP Sizes) of TP Annual Average Load Reduction for the Pilot Watershed.

Table X-11. Scenario 2: BMP Types and BMP Sizes for the Selected Target
Solution 1 (Existing Condition) in Opti-Tool

BMPID

BMP Type

Land Use

Treated
Impervious
Area (acres)

Runoff
Depth (in.)

BMP
Storage
Capacity
(gallon)

BMP Cost ($)

BMP1

Infiltration-B

High Density
Residential

31.78

0.70

604,107

$1,007,830

BMP2

Infiltration-B

Commercial

20.95

0.60

341,364

$569,497

BMP3

Infiltration-B

Industrial

28.17

0.80

611,981

$1,020,967

BMP4

Infiltration-B

Open land

17.26

0.40

187,524

$312,847

BMP5

Infiltration-B

Highway

13.67

0.80

296,922

$495,354

BMP6

Infiltration-C

High Density
Residential

714.35

1.00

19,401,581

$32,367,635

BMP7

Infiltration-C

Commercial

114.01

0.90

2,786,886

$4,649,358

BMP8

Infiltration-C

Industrial

61.24

0.80

1,330,580

$2,219,806

BMP9

Infiltration-C

Open land

16.09

0.90

393,340

$656,208

BMP10

Infiltration-C

Highway

46.76

0.70

889,019

$1,483,150

136


-------
Mystic River Watershed TMDL Alternative Development — Final Report

BMPID

BMP Type

Land Use

Treated
Impervious
Area (acres)

Runoff
Depth (in.)

BMP
Storage
Capacity
(gallon)

BMP Cost ($)

BMP11

Biofiltration

High Density
Residential

9.74

0.80

211,654

$874,853

BMP12

Biofiltration

Commercial

167.92

1.00

4,559,847

$18,847,683

BMP13

Biofiltration

Open land

5.25

1.00

142,555

$589,236

BMP14

Biofiltration

Highway

86.81

0.90

2,121,495

$8,768,993

BMP15

Porous Pavement

High Density
Residential

-

-

-

-

BMP16

Porous Pavement

Commercial

55.03

-

13,573,865

$28,963,024

BMP17

Porous Pavement

Industrial

-

-

-

-

BMP18

Porous Pavement

Open land

-

-

-

-

Total

1,389.01

o
1

O

47,452,719

$102,826,441

Table X-12. Scenario 2: BMP Types and BMP Sizes for the Selected Target



Solution 2 (Future Condition 1) in Opti-Tool



BMPID

BMP Type

Land Use

Treated
Impervious
Area (acres)

Runoff
Depth (in.)

BMP
Storage
Capacity
(gallon)

BMP Cost ($)

BMP1

Infiltration-B

High Density
Residential

31.78

0.60

517,806

$863,855

BMP2

Infiltration-B

Commercial

20.95

0.30

170,682

$284,749

BMP3

Infiltration-B

Industrial

28.17

0.70

535,483

$893,346

BMP4

Infiltration-B

Open land

17.26

0.40

187,524

$312,847

BMP5

Infiltration-B

Highway

13.67

0.80

296,922

$495,354

BMP6

Infiltration-C

High Density
Residential

714.35

0.70

13,581,106

$22,657,344

BMP7

Infiltration-C

Commercial

114.01

0.80

2,477,232

$4,132,763

BMP8

Infiltration-C

Industrial

61.24

0.50

831,613

$1,387,379

BMP9

Infiltration-C

Open land

16.09

0.40

174,818

$291,648

BMP10

Infiltration-C

Highway

46.76

0.70

889,019

$1,483,150

BMP11

Biofiltration

High Density
Residential

9.74

0.70

185,198

$765,496

BMP12

Biofiltration

Commercial

167.92

0.80

3,647,877

$15,078,146

BMP13

Biofiltration

Open land

5.25

0.90

128,299

$530,313

BMP14

Biofiltration

Highway

86.81

0.20

471,443

$1,948,665

BMP15

Porous Pavement

High Density
Residential

-

-

-

-

BMP16

Porous Pavement

Commercial

-

-

-

-

BMP17

Porous Pavement

Industrial

-

-

-

-

BMP18

Porous Pavement

Open land

-

-

-

-

Total

1,333.99

(0.2-0.9)

24,095,022

$51,125,054

137


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Table X-13. Scenario 2: BMP Types and BMP Sizes for the Selected Target

Solution 3 (52% Target) in Opti-Tool

BMPID

BMP Type

Land Use

Treated
Impervious
Area (acres)

Runoff
Depth (in.)

BMP
Storage
Capacity
(gallon)

BMP Cost ($)

BMP1

Infiltration-B

High Density
Residential

31.78

0.60

517,806

$863,855

BMP2

Infiltration-B

Commercial

20.95

0.30

170,682

$284,749

BMP3

Infiltration-B

Industrial

28.17

0.20

152,995

$255,242

BMP4

Infiltration-B

Open land

17.26

0.40

187,524

$312,847

BMP5

Infiltration-B

Highway

13.67

0.20

74,230

$123,838

BMP6

Infiltration-C

High Density
Residential

714.35

0.40

7,760,632

$12,947,054

BMP7

Infiltration-C

Commercial

114.01

0.30

928,962

$1,549,786

BMP8

Infiltration-C

Industrial

61.24

0.50

831,613

$1,387,379

BMP9

Infiltration-C

Open land

16.09

0.20

87,409

$145,824

BMP10

Infiltration-C

Highway

46.76

0.20

254,005

$423,757

BMP11

Biofiltration

High Density
Residential

9.74

0.10

26,457

$109,357

BMP12

Biofiltration

Commercial

167.92

0.10

455,985

$1,884,768

BMP13

Biofiltration

Open land

-

-

-

-

BMP14

Biofiltration

Highway

-

-

-

-

BMP15

Porous Pavement

High Density
Residential

-

-

-

-

BMP16

Porous Pavement

Commercial

-

-

-

-

BMP17

Porous Pavement

Industrial

-

-

-

-

BMP18

Porous Pavement

Open land

-

-

-

-

Total

1,241.93

(0.1-0.6)

11,448,300

$20,288,455

X.E. Summary

The results of this pilot study provide quantitative and qualitative technical guidance to support
watershed-based GI management planning. Opti-Tool analysis results help to identify optimal
stormwater controls (including categories of methods and sizing approaches) that could increase the
technical and economic feasibility of retrofitting needed stormwater management strategies into
developed watershed areas. This study highlights the computational power of optimization
algorithms in Opti-Tool for evaluating thousands of iterations for a combination of different BMP
types and BMP sizes at strategic locations. As demonstrated in Scenario 1 (BMPs sized for a typical
design storm), spatial optimization at the watershed scale can provide significant cost savings as
compared to picking locations by best professional judgment. Scenario 2 further demonstrates that
when location and size are optimized, there is potential for further cost savings for the same annual
average load reduction benefit. The results of both scenarios indicate that considerable cost savings
or avoidance may be accomplished through investing in nonstructural controls to reduce the TP

138


-------
Mystic River Watershed TMDL Alternative Development — Final Report

reduction target for structural controls from the steep portion of the CE-Curve to the flatter
portion. Table X-14 compares the two scenarios simulated in this case study.

Table X-14. BMP Scenarios Comparison in Opti-Tool

Scenario
ID

Scenario
Description

TP Load
Reduction
Target (%)

Impervious
Area Treated
(acre)

Runoff
Depth (in.)

BMP Storage
Capacity
(Million gallon)

BMP Cost
(Million $)

Scenario 1

BMP size (1 in.)
and optimize
the spatial
locations

67%

1,379

1.00

49.54

$106.37

62%

1,211

1.00

32.90

$66.12

52%

909

1.00

24.70

$41.20

Scenario 2

Optimize BMP

size (0.1 in.
increment and
max size 1 in.)
and the spatial
locations

67%

1,389

o
1

^|-
o.

47.45

$102.83

62%

1,334

(0.2-0.9)

24.10

$51.13

52%

1,242

(0.1-0.6)

11.45

$20.29

The CE-curve for both scenarios shows that the numeric targets for existing condition (67%) and
future condition 1 (62%) are above the knee-of-curve where optimal solutions tend to become
expensive due to the cheaper stormwater controls being exhausted. The performance curve provides
clear guidance on achieving the management objectives in a most-cost effective manner. The future
condition 1 load reduction target solution for Scenario 2 provides a cost saving of $51.7 million as
compared to the existing condition target solution for Scenario 2. By lowering 5% of load reduction
target (from 67% to 62%), it can provide almost 50% of cost saving (from $102.83 to $51.13) for
Scenario 2. The CE-curve provides optimal solutions for a range of load reduction targets, so it can
also be used to pick solutions for the intermediate milestones that show progress towards meeting
the final load reduction target. Additionally, the CE-Curve provides information to make a strong
case for investing in nonstructural and source reduction controls to achieve TP reductions and avoid
the need for installing the most expensive control in the most challenging locations.

This study provides planning level analysis with no site-specific project information but provides
guidance on which land use sources to target and what type of BMPs are suitable and how to size
those BMPs. For example, Scenario 2 of future condition 1 (Table X-12 and Table X-13) show
recipes for meeting the 62% and 52% TP load reduction targets by implementing 14 and 12
different BMP types, respectively (combination of land use, soil, and storage capacity) in the pilot

139


-------
Mystic River Watershed TMDL Alternative Development — Final Report

watershed. Next step would be performing the field investigation to identify the feasible sites in the
watershed and selecting the BMP types and sizes based on the guidance provided in Table X-13. For
example, identified suitable sites on high density residential land use should be designed as
infiltration practices to capture 0.6-inch runoff depth for underlying soil B type and to capture 0.4-
inch depth for underlying soil C type. For poor draining soil type D, biofiltration practices can be
designed to capture 0.1-inch of runoff depth. The required BMP storage capacity reflects the storage
volume of the control expressed in terms of runoff depth from the contributing impervious area.

X.F. Example Projects

This section provides two example projects of GI implementation efforts in other locations where
distributed GI practices were also found to be cost-effective stormwater management strategies.

X.F.1. Berry Brook Project, Dover, New Hampshire

Berry Brook Project in Dover, New Hampshire is a partnership between New Hampshire
Department of Environmental Services (NHDES), University of New Hampshire Stormwater
Center (UNHSC) and the City. This unique partnership between regulators, academics, and
committed city staff has reduced best management practice implementation costs, increased
effectiveness, and led to more maintainable stormwater management systems. The project goal is to
filter, infiltrate, and reduce stormwater runoff from Effective Impervious Cover (EIC) as a means
for managing pollutant loading and controlling runoff volumes to Berry Brook. The project has
become a prime example of how scientists and public works departments collaborated to improve
water quality in an urban watershed, using Low Impact Development (LID) and Green
Infrastructure (GI) retrofits, that reduced the effective impervious area in the 185-acre urban
watershed from 30% down to 10%. Below is the list of BMPs implemented for this project and
Figure X-ll shows their locations in the watershed. A detailed report on this project is available on
UNHSC website to download.

•	12 bioretention systems,

•	a tree filter,

•	a subsurface gravel wetland,

•	one acre of new wetland,

•	3 grass-lined swales

•	2 subsurface gravel filters

•	an infiltration trench system

•	3 innovative filtering catch basin designs

140


-------
Mystic River Watershed TMDL Alternative Development — Final Report

N Berry Brook BMPS

0 0.0450.09 0.18 0.27 0.36
am i iMiles

Legend
New BMPs

I | BB_Watershed

2015 1-foot Orthophotography

Figure X-XI. Green Infrastructure Retrofits for Berry Brook Project in Dover, New

Hampshire

X.F.2. The Advancing Green Infrastructure Program, New Haven, Connecticut

The Advancing Green Infrastructure Program in New Haven, Connecticut is a public-private
partnership that: promotes environmental protection and social justice through the construction of

141


-------
Mystic River Watershed TMDL Alternative Development — Final Report

hundreds of right-of-way bioswales to combat water quality pollution associated with stormwater
runoff and combined sewer overflows. Every year, approximately 260 million gallons of combined
sewage enters the waterways surrounding New Haven, contributing to pollution in local waterways
and the Long Island Sound, and negatively impacting ecosystem health and public recreation.

The bioswale program began as a pilot project in 2014, funded through the National Fish and
Wildlife Long Island Sound Futures Fund, with the goal of quantifying how effective green
infrastructure was at stormwater retention in the city. The project was spearheaded by the New
Haven Resources Initiative (URI), a non-profit in New Haven, and made possible by the
partnerships URI fostered in academia, the non-profit sector, and the city and water pollution
control authority. University of Rhode Island (URI), Yale School of Forestry & Environmental
Studies, the City of New Haven's Engineering Department and the Greater New Haven Water
Pollution Control Authority (GNHWPCA) sited eight bioswales in the Westville neighborhood of
New Haven. Considerations in siting were slope, obstacles (trees, telephone poles, driveways,
underground utilities), homeowner agreement and interest, and the desire to place bioswales as close
upstream of catch basins as possible.

Once bioswales were sited, outreach was conducted in the neighborhood to ask homeowners if they
would be willing to adopt the bioswales, ensuring that bioswales were taken care of early in the
process and maintained in the future. Each of the eight bioswales was adapted from the City of New
York standard 3.05 m x 1.5 m right of way bioswale optimized for high capacity stormwater
retention. These bioswales were generally smaller than typical NYC bioswales, and ranged from 8' x
5.2' (the smallest) to 16.5' x 6' (the largest). Four feet of soil was excavated, and a geo-textile was
installed in the base to prevent fine soil and sediment infiltration past the base of the swale. The
bioswale was backfilled with two feet of engineered soil (a mix of New Haven sandy loam and
compost/mulch organic material to facilitate plant growth) and one foot of river stone or gravel.
Soils in the New Haven area are highly permeable, consisting of sandy loams with infiltration rates
near 0.46 m/hr. Along one end of each bioswale, 0.6 — 0.7 m wide vertical wire fenced gabions were
placed in each bioswale and filled with river stones to encourage fast infiltration if ponding occurred.
New Haven high school students from Common Ground High School, alongside homeowners who
adopted the swales, planted native plants and shrubs, such as winterberry holly, black-eyed susans,
and grasses and chose their plant pallet (see Figure X-XII). The sidewalk curbs were cut to allow
water to infiltrate into the gardens.

This small pilot program turned into a citywide partnership program that recently won Harvard's
prestigious Roy Award for Environmental Partnership (City of New Haven, 2018). Funding for an
additional 275 bioswales has been secured, with 200 built within downtown New Haven to alleviate
flooding and 75 within the combined sewer area to mitigate combined sewer overflow pollution. In
small cities like New Haven that are considering sewer separation, bioswales have proven to be a
cost-effective alternative to this expensive and disruptive procedure, while also improving flooding
hazards and risks in urban areas. Bioretention systems are scalable to larger cities, especially those
that have aging infrastructure. If sited strategically and developed with adequate maintenance, green
infrastructure can serve as a long-term solution to aid in stormwater management.

142


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Figure X-XII. Homeowners and Common Ground High School students plant
perennials in a New Haven Bioswale

Photo taken by Kelsey Semrod, https://hixon.yale.edu/practice/bioswales.

143


-------
Mystic River Watershed TMDL Alternative Development — Final Report

XI. References

Arnold JG and PM Allen, 1999. Automated methods for estimating baseflow and groundwater
recharge from streamflow records. J Am Water Resource Assoc, 35(2), 411-424.

Arnold, JG, PM Allen, R Muttiah, and G Bernhardt, 2005. Automated base flow separation and
recession analysis techniques. Ground Water, 33(6): 1010-1018.

Berry Brook Project. (2017). Berry Brook Watershed Management Plan —Implementation Projects
Phase III. Prepared for: The New Hampshire Department of Environmental Services, NH.

Prepared by: The City of Dover and University of New Hampshire Stormwater Center, NH.
December 2017.

Breault, RF, Sorenson, JR and Weiskel, PK, 2002, Streamflow., Water Quality and Contaminant Loads in
the Lower Charles River Watershed, Massachusetts, 1999-2000. Water Resources Investigations Report 02-
4137, 139 p. U.S. Geological Survey, Westborough, Massachusetts.

City of New Haven. (2018). New Haven's "Advancing Green Infrastructure Program" Wins
Harvard's Roy Award for Environmental Partnership Promoting environmental protection, climate
resilience, and social justice in New Haven, CT. Accessed November 9:
https://www.newhavenct.gov/news/displaynews.htm?NewsID=516&TargetID=67

Clough JS, 2014. AQUATOX (Release 3.1 plus). Modeling Environmental Fate and Ecological
Effects in Aquatic Ecosystems. Volume 1: User's Manual. US Environmental Protection Agency,
Office of Science and Technology, Washington DC.

Demonstration of Opti-Tool: Buzzards Bay Watershed Case Study. Prepared for: U.S. EPA Region
1, Boston, MA. Prepared by: Tetra Tech, Inc. Fairfax, VA. December 30, 2016.

DEP, 2010. Draft Total Maximum Daily Loads of Phosphorus for Selected Boston Harbor (Mystic
Basin) Lakes. Department of Environmental Protection, Division of Watershed Management, MA.

DEP-EPA, 2011. Total Maximum Daily Load for Nutrients in the Upper/Middle Charles River,
Massachusetts. Control Number: CN 272.0. Massachusetts Department of Environmental
Protection, Worcester, MA and US Environmental Protection Agency, Boston, MA. Prepared by
Charles River Watershed Association, Weston, MA and Numeric Environmental Services, Beverly
Farms, MA.

DFW (no date). Various Bathymetry Maps. Department of Fish and Game, MA.

Dodds, WK, & Oakes, RM. 2004. A technique for establishing reference nutrient conditions across
watersheds affected by humans. Limnol. Oceanogr.: Methods 2, 2004, pgs. 333—341.

ENSR, 2000. Diagnostic/Feasibility Study of Winter Pond, Winchester, Massachusetts. Prepared for
The Town of Winchester and the Friends of Winter Pond Association.

EPA, 2016. Phosphorus TMDLs for Vermont Segments of Lake Champlain. U.S. Environmental
Protection Agency, Region 1, Boston, MA.

EPA, 2018. Bathymetric Depths and Soft Sediment Presence Measurements in the Mystic River.
Personal communication with Mark Voorhees and Monique Dulac of EPA Region 1.

Filstrup, CT and Downing, JA, 2017. Relationship of chlorophyll to phosphorus and nitrogen in
nutrient-rich lakes, Inland Waters, 7(4), p385-400, D01:10.1080/20442041.2017.1375176.

144


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Fischer, HB, List, EJ, Koh, RCY, Imberger, J and Brooks. NH, 1979. Mixing in inland and coastal
waters. Academic Press, New York, N.Y.

Flanagan, SA, DL Montgomery, and JD Ayotte, 2001. Shallow Ground-water Quality in Boston,
Massachusetts Metropolitan Area. USGS Water-Resources Investigations Report 01-4042.

HW, 2015, Stump Brook / Monponsett Pond Hydrologic and Water Quality Assessment. Prepared
for the Massachusetts Division of Ecological Restoration, MA by Horsley Witten Group, Sandwich,
MA.

IDEQ, 2015. Lower Boise River TMDL. Total Phosphorus Addendum. Idaho Department of
Environmental Quality, Boise, ID.

Knight, E. (2016). Aberjona River: Historical Background. Prepared by Ellen Knight, PhD for the
Town of Winchester's Flood Mitigation Program. Updated 6/25/2016.
https://www.winchester.us/DocumentCenter/View/1745 (Accessed June 28, 2017)

Koiv T, Noges T, and Laas A, 2011. Phosphorus retention as a function of external loading
hydraulic turnover time, area and relative depth in 54 lakes and reservoirs. Hydrobiologia 660, pl05—
115. DOI 10.1007/s 10750-010-0411-8.

Ludlam, S. D., & Duval, B. (2001). Natural and Management-Induced Reduction in Monimolimnetic
Volume and Stability in a Coastal, Meromictic Lake. Lake and Reservoir Management, 17(2), 71—81.

Massachusetts Department of Environmental Protection & U.S. EPA. 2007. Final — Total Maximum
Daily Load for Nutrients in the Lower Charles River Basin, Massachusetts. CN 301.0.

Massachusetts Department of Environmental Protection. 2016. Massachusetts Consolidated
Assessment and Listing Methodology Guidance Manual for the 2016 Reporting Cycle. CN 445.0.

McKee, T.B., Doesken N.J, and Kliest J. 1993. The relationship of drought frequency and duration
to time scales. In Proceedings of the 8th Conference of Applied Climatology, 17-22 January,
Anaheim, CA. American Meteorlogical Society, Boston, MA. 179-184.

MWRA, 2018. CSO Discharge Estimates and Rainfall Analyses for Calendar Year 2017. Technical
Memorandum. April 30, 2018.

MWRA, 2016. Industrial Waste Report. Number 32. October 2016, variously paged. Massachusetts
Water Resources Authority, Boston, Massachusetts.

NADP, 2016. Nitrogen from the Atmosphere. ISWS Miscellaneous Publication 207 and NADP
Brochure 2016-01. National Atmospheric Deposition Program, University of Wisconsin, Madison,
WI. http://nadp.slh.wisc.edu/lib/brochures/nitrogenAtmos.pdf.

NADP, 2018a. Map of Nitrate Ion Wet Deposition, 2015. National Atmospheric Deposition

Program, University of Wisconsin, Madison, WI.

http://nadp.slh.wisc.edu/maplib/pdf/2015/N03_dep_2015.pdf.

NADP, 2018b. Map of Ammonium Ion Wet Deposition, 2015. National Atmospheric Deposition

Program, University of Wisconsin, Madison, WI.

http://nadp.slh.wisc.edu/maplib/pdf/2015/NH4_dep_2015.pdf.

NADP, 2018c. Map of Total Nitrogen Deposition, 2015. National Atmospheric Deposition
Program, University of Wisconsin, Madison, WI.

http://nadp.slh.wisc.edu/committees/tdep/tdepmaps/preview.aspx#n_tw.

145


-------
Mystic River Watershed TMDL Alternative Development — Final Report

NCEI, 2018. National Centers for Environmental Information, formerly the National Climatic Data
Center, https://www.ncdc.noaa.gov/data-access.

NOAA, 2018. Hourly/Sub-Hourly Data Observation Map.

https://gis.ncdc.noaa.gov/maps/ncei/cdo/hourly. Accessed May 2018. National Oceanic and
Atmospheric Administration, Washington, DC.

Opti-Tool for Stormwater and Nutrient Management: User's Guide. Prepared for: U.S. EPA Region
1, Boston, MA. Prepared by: Tetra Tech, Inc. Fairfax, VA. June 30, 2016.

Park RA and Clough JS, 2014. AQUATOX (Release 3.1 plus). Modeling Environmental Fate and
Ecological Effects in Aquatic Ecosystems. Volume 1: Technical Documentation. US Environmental
Protection Agency, Office of Science and Technology, Washington DC.

Paul, M.J., J. Robbiani, L. Zheng, T. Rafi, S. Bai, and P. von Loewe. 2011. Development of Nutrient
Endpoints for the Northern Piedmont Eco-region of Pennsylvania: TMDL Application Follow-up
Analysis. Prepared by Tetra Tech, Inc., Owings Mills, MD for United States Environmental
Protection Agency, Region 3.

Powers SM, Tank JL, and Robertson DM, 2015. Control of nitrogen and phosphorus transport by
reservoirs in agricultural landscapes. Biogeochemistry 124, p417—439. DOI 10.1007/s 10533-015-
0106-3.

Sloto RA and MY Crouse, 1996. HYSEP: A Computer Program for Streamflow Hydrograph
Separation and Analysis. Water-Resources Investigations Report 96-4040. US Geological Survey,
Lemoyne, PA.

TetraTech, 2015. Lake Champlain BATHTUB Model Calibration Report. Prepared for the U.S.
Environmental Protection Agency, Region 1, Boston, MA.

Tipping, E, Benham, S, Boyle, J F, Crow, P, Davies, J, Fischer, U, Guyatt, H, Helliwell, R, Jackson-
Blake, L, Lawlor, AJ, Monteith, DT, Roweg, EC, and Tobermanac, H, 2014. Atmospheric
deposition of phosphorus to land and freshwater. Environ. Sci.: Processes Impacts, 2014 (16), 1608
—1617. https://doi.org/10.1039/c3em00641g.

U.S. EPA, 1986. Quality Criteria for Water.

U.S. EPA, 1998. Level III eco-regions of the continental United States (revision of Omernik, 1987).

U.S. EPA, 2001. Using Stressor-response Relationships to Derive Numeric Nutrient Criteria. EPA-
820-S-10-001.

Wagner K, 2009. LLRM — Lake Loading Response Model. Users Guide and Quality Assurance
Project Plan. Water Resource Services, Wilbraham, MA.

Walker, W, 1982. An empirical analysis of phosphorus, nitrogen, and turbidity effects on reservoir
chlorophyll-a levels, Canadian Water Resources Journal, 7(1), 88-107,

https://doi.org/10.4296/cwrj0701088. Published online 2013.

Walker, W, 1985. Empirical Methods for Predicting Eutrophication in Impoundments - Report 3:
Model Refinements", Technical Report E-81-9, U.S. Army Corps of Engineers, Waterways
Experiment Station, Vicksburg, MI.

146


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Walker W, 1996. Simplified Techniques for Eutrophication Assessment & Prediction: User Manual.
Instruction Report W-96-2. US Army Corps of Engineers, Waterways Experiment Station,
Vicksburg, MI. Updated September 1999.

Walker W, 2004. BATHTUB-Model Version 6.1. Simplified Techniques for Eutrophication
Assessment & Prediction. US Army Corps of Engineers, Waterways Experiment Station, Vicksburg,
MI.

WH, 1987. Clean Lakes Program: Diagnostic/Feasibility Studies of Black Nooks Pond. Wiitman
and Howard, MA.

WH, 1988. Clean Lakes Program: Diagnostic/Feasibility Studies of Wedge Pond. Wiitman and
Howard, MA.

147


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Appendix A: Water body and Monitoring Location
Summary Table

Table A-1. Number of Days Sampled from 2000 - 2016 and Maps of Monitoring
Locations.



Monitoring Program



/ Monitoring
Station

Baseline

Phos.
Loadin
g

Bosto
n

Harbo
r

cso

Event

Total
Sample
No.

Aberjona
River

558

39





597

ABR006

187

39

-

-

226

ABR028

J 90

-

-

—

190

ABR049

184

-

-

-

184

Horn Pond

-

2

-



2

HOPCTR

-

2

-

-

2

Horn Pond











Brook

-

1

-

-

1

HOB002

-

1

-

-

1

Winter Pond

-

5

-

-

5

WIPCTR

-

5

-

-

5

Wedge Pond

-

33

-

-

33

WEPCTR

-

33

-

-

33

Upper Mystic
Lake

183

60

_

_

243

UPL001

183

-

-



183

UPLCTR

-

33

-

-

33

UPLUPL

-

27

-

-

27

Lower Mystic
Lake



2





2

LOLCTR

-

2

-

-

2

~ABR049



~ABR028

waterbody

AHOPCTR

~ Aberjona River

\ \ / ' Stone Zoo'Q j

A Horn Pond



ฆ Horn Pond Brook

WIPCTR ฆHOBt)02

-|- Lower Mystic Lake

^ฆWEPCTR

0 Upper Mystic Lake



•)|(- Wedge Pond

~ABR006

Winter Pond

0UPLUPL



[gUPLCTR



0UPLOO1



-(-LOLCTR







Table A-2. Water Body/Monitoring Station and Program

Water body/
Monitoring
Station

Monitoring Program

Total
Monitoring
Events

Baseline

Phos.
Loading

Boston
Harbor

CSO
Event

Mill Brook

188

38

-

-

226

MIB001

188

-

-

-

188

MIB0045

-

38

-

-

38

Spy Pond

-

17

-

-

17

SPPCTR

-

17

-

-

17

Winns Brook

190

-

-

-

190

A-148


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Water body/

Monitoring Program

Total

Monitoring
Station

Baseline

Phos.
Loading

Boston
Harbor

cso

Event

Monitoring
Events

WIB001

190

-

-

-

190

Little River

-

-

-

442

442

MWRA174

-

-

-

442

442

Alewife Brook

192

39

-

1,393

1,624

ALB006

192

39

-



231

MWRA070

-

-

-

466

466

MWRA074

-

-

-

469

469

MWRA172

-

-

-

458

458

Mystic River
(Fresh)

190

109

995

2,0531

3,3291

MWRA056

-

-

-

365

365

MWRA057

-

-

-

373

373

MWRA059

-

-

-

367

367

MWRA066

-

-

548

-

548

MWRA067

-

-

-

70

370

MWRA083

-

-

261

85

646

MWRA177

-

-

186

194

3651

MYR071

190

38

-

-

228

MYR33

-

36

-

-

36

MYR43

-

35

-

-

35

Meetinghous
e Brook

191

13

_

_

204

MEB001

191

13

-

-

204

Maiden River

183

73

-

322

578

MAR003

-

36

-

-

36

Mystic River
(Salt)

1701



459

1,079

1,70s1

MWRA015

-

-

-

344

344

MWRA052

-

-

-

92

492

MWRA069

-

-

-

243

243

MWRA137

-

-

459

-

459

MYR275

82

-

-

-

82

MYRMMP

95

-

-

-

95

Mill Creek

93

-

-

-

93

MIC004

93

-

-

-

93

Chelsea River

95

-

-

370

465

CHR95S

95

-

-

-

95

MWRA027

-

-

-

370

370

Belle Isle Inlet

81

-

-

-

81

BEI001

13

-

-

-

13

BEI093

68

-

-

-

68

1.Sub-category values do not sum to this value due to sampling events at different
locations or programs occur same day.

A-149


-------
Mystic River Watershed TMDL Alternative Development — Final Report

4tri

A11
(7.,

-^SPPCTR
VVIB001 „



$Vr071

^ ^'wr4Reboi)1

I^MWRAOSe

lr< vsrsilv	\

^.L3C0€

^MWRA172

iMm*w

j \

Sarocfvitlc

Harvard
i ฆ •.••• Mity

O

Hnwarr-sij.mrr o Cambridge

.*$67
.1VVRA059



<••*ป, ft.ฎ

IcwMfftseE

-*ฆ U4!i.-ซE-:f	*

SfcWwl.	y

7D .Niwrfl
Mv'svafijSe.ltf ^
ฉ e * * :>arKral-4>&nibHv

O	- i'Jtrcrl' 3!JPe^

vvaterbody
• Ale wife Brook
A Little River
ฆ Meetinghouse Brook
—|- Mill Brook
ERI Mystic River (Fresh)
Spy Pond
Winns Brook

-sr

P>/ซ ;l

s,J

to 'Lu.

Ma: den *<-.
* •>

,r



".i v „

I.1AR036
ฆ

MWRA176
ฆ

MAR003

wr

TS>

MWl



Itomi

Sn,.

+

WC'004

;i*iwป f

M

131WYRI.WP 4k

UVVRAOl^MMW

Si

BEI093

BEI001

OU^'^OcnslrtL'.cr

j bo:
J 6nK
-------
Mystic River Watershed TMDL Alternative Development — Final Report

Appendix B: Water Quality Parameters Included in
Monitoring Programs

Table B-1. Water Quality Parameters Included in Monitoring Programs

Parameter Code

Parameter Name/Description

Speciation

ATTENUATION_COEFFICIENT

Light attenuation coefficient

-

CHLA

Chlorophyll-a

-

DO

Dissolved Oxygen

-

DO_SAT

Dissolved Oxygen, %
saturation

_

ECOLI

Escherichia coli

-

ENT

Enterococcus

-

FCOLI

Fecal Coliform

-

NH3

Ammonia

as N

N02

Nitrite

as N

N023

Nitrate + Nitrite

as N

N03

Nitrate

as N

PH

PH

-

PHAEOPHYTIN

Pheophytin a

-

P04

Orthophosphate

asP

SALINITY

Salinity

-

SECCHI

Secchi Disk Depth

-

SPCOND

Specific conductance

-

TDN

Nitrogen, total dissolved

as N

TDP

Dissolved Phosphorus

asP

TEMP_WATER

Water Temperature

-

TN

Total Nitrogen

as N

TP

Total Phosphorus

asP

TPC

Total Particulate Carbon

-

TPN

Particulate Nitrogen

as N

TPP

Phosphorus, Particulate
Organic

asP

TSS

Total suspended solids

-

TURB

Turbidity

-

B-151


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Appendix C: Descriptive Figures of Water Quality
Data by Water body

Note that values marked as "dot" denote observations that extend above or below the nearest hinge
(i.e., the 25th or 75th percentile) by a distance exceeding 1.5 times the inter-quartile range. Unusual
values are marked by a red circle and additional information on these values are provided in
Appendix D.

Figure C-1. Attenuation coefficient results by water body.

C-152


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Figure C-2. Chlorophyll-a results by water body.

Figure C-3. Dissolved oxygen results by water body.

C-153


-------
My stic River Watershed TMDL Alternative Development — Final Report

DO SAT

60D -

o

400-

a>

3
flj
>

0)

cn

200-





/

/



+ ^ + _



*



Figure C-4. Dissolved oxygen percent saturation results by water body

Figure C-5. Inorganic nitrogen (nitrite plus nitrate) results by water body.

C-154


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Figure C-6. pH results by water body.

PHAEOPHYTIN

20-

Figure C-7. Phaeophytin results by water body.

C-155


-------
Mystic River Watershed TMDL Alternative Development — Final Report

SECCHI

6-

Figure C-8. Secchi depth results by water body.

Figure C-9. Total nitrogen results by water body.

C-156


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Figure C-10. Total phosphorus results by water body.

Figure C-11. Total suspended solids results by water body.

C-157


-------
Mystic River Watershed TMDL Alternative Development — Final Report

Figure C-12. Turbidity results by water body

C-158


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Appendix D: Identified Extreme Values

Table D-1. Extreme values identified in Appendix C

Sample
Date &
Time

Monitori
ng

Program

Water

body

Name

Monitor
ing

Location

Paramet
er Name

Sample
Result

Uni
ts

Qu
al.

Fl

ag
ID

Flag

Descript
ion

Commen
t

Analytic
al

Method

Analytical

Method

Description

2/2/09
7:56

Baseline

Chelsea
River

CHR95S

Dissolve
d

Oxygen

26.66

mg

/I

-

E

instrume
nt error

YSI likely

not
calibrated
correctly

D888(B)

Dissolved
Oxygen by
Instrumental
Probe

6/25/03
7:36

CSO

Maiden
River

MWRA1
76

Dissolve
d

Oxygen
(%
Saturati
on)

616.7

%

-

-

-

-

D888(B)

Dissolved
Oxygen by
Instrumental
Probe

5/1/14
6:27

Baseline

Belle Isle
Inlet

BEI093

Total
Phos.

3.65

mg

/I

-

-

-

-

4500-P-E

Phosphorus in
Water by
Colorimetry-
Ascorbic Acid
Method

12/16/1
1 7:21

Baseline

Belle Isle
Inlet

BEI093

Total
Phos.

2.25

mg

/I

-

-

-

-

4500-P-E

Phosphorus in
Water by
Colorimetry-
Ascorbic Acid
Method

7/12/06
6:37

Baseline

Maiden
River

MAR036

Total
Phos.

1.9203
76

mg

/I

-

-

-

-

4500-P-J

Persulfate
Method for
Simultaneous
Determination
of Total
Nitrogen and
Total

Phosphorus

10/13/1
5 10:00

Phosphor

us
Loading

Horn
Pond
Brook

HOB002

Turbidity

4720

NT
U

-

-

-

-

180.1

Turbidity by
Nephelometry

D-159


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Appendix E: Modeled Stormwater TP Load and
Rainfall-Runoff Results

Table E-XI-1. Modeled Stormwater TP Load (Ibs./year; No Calibration)

Year

Critical Water Quality Segments

Impaired Ponds

Upper

Lobe

Basin

Upper
Mystic
Lake

Upper
Basin

Lower
Basin

Blacks
Nook Pond
(MA71005),
Cambridge

Horn Pond

(MA71019),

Woburn

Judkins
Pond

(MA71021),
Winchester

Mill Pond

(MA71031),

Winchester

Spy Pond

(MA71040),

Arlington

Wedge
Pond

(MA71045),
Winchester

Winter
Pond

(MA71047),
Winchester

1992

7,748

8,315

14,693

18,940

2.19

2,360

7,501

7,543

371

2,581

61

1993

5,428

5,737

10,408

13,718

0.73

1,490

5,260

5,291

288

1,646

36

1994

6,426

6,849

12,203

15,899

1.35

1,877

6,227

6,262

318

2,061

46

1995

5,868

6,276

11,211

14,556

1.44

1,726

5,684

5,716

294

1,894

43

1996

7,492

7,990

14,194

18,434

1.70

2,217

7,255

7,297

365

2,430

55

1997

3,749

3,939

7,327

9,771

0.28

951

3,635

3,657

218

1,060

22

1998

9,409

10,142

17,714

22,613

3.04

2,977

9,098

9,146

438

3,257

79

1999

6,852

7,367

13,067

16,801

2.02

2,087

6,628

6,664

341

2,292

55

2000

6,523

6,950

12,458

16,256

1.35

1,882

6,321

6,357

331

2,069

46

2001

4,004

4,218

7,697

10,193

0.40

1,073

3,880

3,904

218

1,190

25

2002

4,692

4,940

9,127

12,131

0.43

1,217

4,549

4,577

267

1,354

29

2003

5,449

5,758

10,419

13,739

0.68

1,500

5,282

5,313

288

1,659

35

2004

6,231

6,636

11,845

15,442

1.27

1,812

6,036

6,071

312

1,991

44

2005

5,451

5,763

10,519

13,878

0.76

1,478

5,283

5,314

295

1,635

36

2006

6,747

7,180

12,814

16,710

1.37

1,960

6,536

6,574

335

2,152

48

2007

7,376

7,926

13,968

17,976

2.16

2,264

7,141

7,181

350

2,475

58

2008

7,469

7,972

14,158

18,402

1.66

2,207

7,236

7,277

366

2,421

54

2009

5,235

5,521

10,146

13,454

0.55

1,381

5,076

5,106

293

1,534

32

2010

8,315

8,931

15,614

20,039

2.40

2,594

8,045

8,089

386

2,837

67

2011

6,559

6,950

12,543

16,483

1.02

1,832

6,356

6,394

342

2,022

44

E-160


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

2012

5,036

5,322

9,616

12,675

0.63

1,392

4,881

4,910

265

1,539

33

2013

5,491

5,844

10,498

13,707

1.09

1,574

5,318

5,349

283

1,734

39

2014

6,141

6,537

11,747

15,345

1.24

1,758

5,950

5,984

314

1,933

43

2015

4,077

4,288

7,947

10,586

0.32

1,045

3,953

3,977

235

1,165

24

2016

3,912

4,115

7,647

10,202

0.27

994

3,795

3,818

229

1,110

23

2017

5,236

5,544

10,061

13,246

0.78

1,442

5,075

5,105

278

1,593

35

Average
Annual

6,035

6,423

11,525

15,046

1

1,734

5,846

5,880

308

1,909

43

Table E-XI-2. Modeled Groundwater TP Load (Ibs./year; Concentration = 8 mg/L;

No Calibration)

Year

Critical Water Quality Segments

Impaired Ponds

Upper

Lobe

Basin

Upper
Mystic
Lake

Upper
Basin

Lower
Basin

Blacks
Nook Pond
(MA71005),
Cambridge1

Horn Pond

(MA71019),

Woburn

Judkins
Pond

(MA71021),
Winchester

Mill Pond

(MA71031),

Winchester

Spy Pond

(MA71040),

Arlington

Wedge
Pond

(MA71045),
Winchester

Winter
Pond

(MA71047),
Winchester

1992

562

597

1,102

1,432

0.15

168

544

547

32

184

4

1993

429

453

842

1,110

0.06

120

416

419

25

132

3

1994

522

552

1,021

1,339

0.10

151

506

509

30

165

4

1995

380

402

746

977

0.08

109

367

370

22

120

3

1996

634

673

1,238

1,612

0.15

189

613

617

36

207

5

1997

229

240

454

604

0.02

60

222

223

15

66

1

1998

744

790

1,450

1,881

0.19

226

719

723

42

247

6

1999

500

532

982

1,274

0.13

150

484

486

29

165

4

2000

490

519

962

1,259

0.10

142

475

477

29

155

3

2001

279

293

548

723

0.04

77

270

272

17

85

2

2002

337

354

666

883

0.04

90

327

329

21

99

2

2003

421

443

825

1,089

0.05

116

408

410

25

128

3

2004

524

555

1,024

1,339

0.11

153

507

510

30

168

4

2005

432

455

849

1,118

0.07

121

418

421

26

133

3

E-161


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

2006

648

686

1,264

1,651

0.14

191

627

630

37

209

5

2007

442

467

864

1,132

0.08

128

427

430

26

140

3

2008

641

678

1,252

1,638

0.13

187

620

624

37

205

4

2009

374

393

739

979

0.04

101

363

365

23

111

2

2010

685

727

1,333

1,733

0.16

207

662

666

38

226

5

2011

512

539

1,004

1,323

0.07

143

496

499

30

157

3

2012

336

354

662

875

0.05

93

326

328

20

102

2

2013

427

450

835

1,096

0.07

122

413

415

25

134

3

2014

492

520

965

1,266

0.09

141

477

479

29

155

3

2015

270

283

534

710

0.02

71

262

263

17

79

2

2016

251

263

497

662

0.02

65

243

245

16

73

1

2017

385

405

757

999

0.05

106

373

375

23

117

2

Average
Annual

459

486

901

1,181

0

132

445

447

27

145

3

1. All resu

ts rounc

ed to the nearest pound. Groundwater loads for Blacks Nook Pond are non-zero but

below the rounding level.

Table E-XI-3. Modeled Total Streamflow TP Load (Ibs./year) from Stormwater and
Groundwater (No Calibration and No Attenuation)

Year

Critical Water Quality Segments

Impaired Ponds

Upper

Lobe

Basin

Upper
Mystic
Lake

Upper
Basin

Lower
Basin

Blacks
Nook Pond
(MA71005),
Cambridge

Horn Pond

(MA71019),

Woburn

Judkins
Pond

(MA71021),
Winchester

Mill Pond

(MA71031),

Winchester

Spy Pond

(MA71040),

Arlington

Wedge
Pond

(MA71045),
Winchester

Winter
Pond

(MA71047),
Winchester

1992

8,310

8,913

15,795

20,372

2.34

2,528

8,045

8,090

403

2,764

65

1993

5,857

6,189

11,250

14,828

0.79

1,610

5,676

5,710

314

1,779

38

1994

6,948

7,401

13,224

17,238

1.45

2,028

6,732

6,771

348

2,226

49

1995

6,247

6,678

11,957

15,532

1.51

1,836

6,051

6,086

316

2,014

46

1996

8,126

8,663

15,432

20,046

1.86

2,407

7,869

7,914

401

2,637

60

1997

3,978

4,179

7,781

10,375

0.29

1,011

3,857

3,881

233

1,127

24

1998

10,152

10,932

19,165

24,494

3.24

3,203

9,816

9,869

480

3,503

84

E-162


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

1999

7,352

7,899

14,049

18,075

2.15

2,237

7,112

7,150

370

2,457

59

2000

7,014

7,469

13,420

17,516

1.45

2,023

6,795

6,834

360

2,224

50

2001

4,283

4,511

8,245

10,916

0.44

1,150

4,150

4,175

235

1,275

27

2002

5,029

5,294

9,793

13,015

0.47

1,307

4,876

4,906

288

1,453

31

2003

5,870

6,201

11,243

14,828

0.73

1,616

5,689

5,723

313

1,787

38

2004

6,755

7,190

12,869

16,781

1.37

1,965

6,543

6,581

342

2,159

48

2005

5,883

6,219

11,368

14,996

0.83

1,599

5,701

5,735

321

1,767

38

2006

7,394

7,866

14,079

18,361

1.51

2,151

7,163

7,204

372

2,361

53

2007

7,818

8,393

14,832

19,108

2.24

2,392

7,569

7,611

375

2,614

61

2008

8,110

8,650

15,411

20,040

1.80

2,394

7,856

7,900

403

2,626

59

2009

5,609

5,914

10,884

14,433

0.59

1,482

5,438

5,471

316

1,645

35

2010

9,000

9,657

16,947

21,771

2.56

2,801

8,707

8,755

424

3,063

72

2011

7,070

7,490

13,547

17,806

1.09

1,975

6,852

6,892

373

2,179

47

2012

5,373

5,677

10,278

13,550

0.67

1,484

5,207

5,238

285

1,641

35

2013

5,917

6,294

11,333

14,803

1.16

1,696

5,731

5,764

308

1,867

42

2014

6,633

7,057

12,712

16,611

1.34

1,900

6,426

6,464

343

2,088

47

2015

4,347

4,571

8,481

11,296

0.34

1,116

4,215

4,241

252

1,243

26

2016

4,163

4,379

8,143

10,863

0.29

1,059

4,039

4,063

245

1,182

24

2017

5,621

5,950

10,818

14,245

0.83

1,549

5,448

5,480

302

1,710

37

Average
Annual

6,495

6,909

12,425

16,227

1

1,866

6,291

6,327

335

2,054

46

Table E-XI-4. Modeled Stormwater Rainfall-Runoff (in-acre/year) Results
Attributable to Stormwater (No Calibration)



Critical Water Quality Segments

Impaired Ponds

Year

Upper

Lobe

Basin

Upper
Mystic
Lake

Upper
Basin

Lower
Basin

Blacks
Nook Pond
(MA71005),
Cambridge

Horn Pond

(MA71019),

Woburn

Judkins
Pond

(MA71021),
Winchester

Mill Pond

(MA71031),

Winchester

Spy Pond

(MA71040),

Arlington

Wedge
Pond

(MA71045),
Winchester

Winter
Pond

(MA71047),
Winchester

1992

163,532

172,857

299,273

386,996

34

48,984

158,337

159,243

7,458

53,541

1,201

E-163


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

1993

124,957

131,132

228,531

299,548

15

35,074

121,200

121,922

5,923

38,533

811

1994

151,887

159,854

277,229

361,607

23

43,887

147,257

148,122

7,020

48,090

1,026

1995

110,439

116,346

202,510

263,692

18

31,888

107,023

107,645

5,206

34,979

764

1996

184,500

194,666

336,333

435,720

35

55,105

178,670

179,708

8,282

60,185

1,334

1997

66,630

69,539

122,938

162,806

4

17,386

64,691

65,087

3,386

19,230

396

1998

216,286

228,638

394,132

508,494

45

65,734

209,281

210,472

9,710

71,879

1,618

1999

145,562

154,019

266,821

344,342

31

43,804

140,861

141,655

6,735

47,978

1,089

2000

142,610

150,205

261,067

340,099

23

41,244

138,230

139,038

6,647

45,187

980

2001

81,045

84,916

148,487

195,056

9

22,396

78,615

79,087

3,896

24,633

519

2002

98,038

102,569

180,500

238,187

8

26,240

95,165

95,740

4,882

28,956

599

2003

122,371

128,239

223,592

293,853

12

33,921

118,729

119,439

5,854

37,342

771

2004

152,491

160,591

278,158

361,786

25

44,629

147,742

148,606

6,987

48,869

1,064

2005

125,609

131,889

230,356

301,812

16

35,190

121,833

122,558

5,993

38,640

821

2006

188,424

198,535

343,313

445,967

33

55,633

182,546

183,616

8,494

60,793

1,329

2007

128,454

135,124

234,544

305,637

19

37,152

124,466

125,196

5,968

40,747

883

2008

186,341

196,318

340,070

442,459

30

54,472

180,579

181,632

8,558

59,652

1,291

2009

108,768

113,884

200,147

263,889

10

29,298

105,583

106,218

5,392

32,316

668

2010

199,181

210,309

362,159

468,354

38

60,157

192,826

193,935

8,871

65,750

1,455

2011

148,835

156,207

272,223

357,113

17

41,628

144,398

145,255

7,090

45,772

953

2012

97,844

102,650

179,543

235,965

10

26,972

94,958

95,523

4,748

29,693

616

2013

124,104

130,387

226,509

295,882

16

35,490

120,288

120,998

5,811

38,976

833

2014

143,166

150,681

262,035

341,763

22

41,170

138,785

139,599

6,695

45,131

973

2015

78,511

82,072

144,666

191,383

5

20,705

76,244

76,705

3,965

22,898

465

2016

72,989

76,278

134,516

178,297

4

19,059

70,913

71,341

3,724

21,118

420

2017

111,957

117,413

205,258

269,529

12

31,009

108,617

109,266

5,391

34,106

715

Average
Annual

133,636

140,589

244,420

318,855

20

38,393

129,532

130,292

6,257

42,115

908

E-164


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Table E-XI-5. Modeled Groundwater Flow (in-acre/year) Results

Year

Critical Water Quality Segments

Impaired Ponds

Upper

Lobe

Basin

Upper
Mystic
Lake

Upper
Basin

Lower
Basin

Blacks
Nook Pond
(MA71005),
Cambridge

Horn Pond

(MA71019),

Woburn

Judkins
Pond

(MA71021),
Winchester

Mill Pond

(MA71031),

Winchester

Spy Pond

(MA71040),

Arlington

Wedge
Pond

(MA71045),
Winchester

Winter
Pond

(MA71047),
Winchester

1992

310,282

329,668

608,065

790,428

80

92,834

300,072

301,790

17,708

101,464

2,269

1993

237,057

249,894

465,018

612,652

35

66,473

229,692

231,061

14,063

73,023

1,532

1994

288,158

304,720

563,690

739,100

54

83,174

279,075

280,713

16,668

91,133

1,939

1995

209,530

221,810

411,823

539,012

43

60,434

202,825

204,004

12,361

66,287

1,443

1996

350,059

371,192

683,395

890,007

83

104,434

338,607

340,574

19,664

114,054

2,520

1997

126,394

132,440

250,595

333,477

10

32,949

122,600

123,350

8,040

36,442

747

1998

410,390

436,066

800,564

1,038,307

106

124,578

396,619

398,876

23,055

136,215

3,056

1999

276,195

293,776

542,123

703,279

74

83,018

266,953

268,457

15,991

90,922

2,056

2000

270,564

286,351

530,850

695,146

55

78,165

261,968

263,499

15,783

85,633

1,851

2001

153,749

161,796

302,272

399,084

20-

42,445

148,987

149,883

9,250

46,680

979

2002

185,977

195,396

367,697

487,619

19

49,730

180,353

181,443

11,593

54,874

1,132

2003

232,145

244,344

455,090

601,152

28

64,287

225,010

226,356

13,899

70,766

1,457

2004

289,316

306,154

565,466

739,317

59

84,581

279,994

281,631

16,590

92,610

2,009

2005

238,295

251,350

468,759

617,306

38

66,691

230,892

232,267

14,230

73,226

1,552

2006

357,496

378,515

697,750

911,177

78

105,436

345,956

347,984

20,167

115,211

2,515

2007

243,707

257,573

476,959

624,767

46

70,411

235,882

237,266

14,171

77,219

1,668

2008

353,536

374,279

691,310

904,163

72

103,235

342,228

344,224

20,319

113,047

2,442

2009

206,333

216,968

407,645

540,156

23

55,525

200,096

201,300

12,802

61,242

1,262

2010

377,926

401,058

735,670

956,458

91

114,009

365,440

367,542

21,063

124,605

2,754

2011

282,354

297,679

553,937

730,413

40

78,893

273,658

275,283

16,834

86,742

1,802

2012

185,615

195,606

365,464

482,757

25

51,118

179,961

181,031

11,273

56,270

1,164

2013

235,449

248,509

460,735

604,953

39

67,260

227,965

229,310

13,797

73,861

1,573

2014

271,616

287,239

532,893

698,638

52

78,025

263,020

264,564

15,897

85,527

1,840

E-165


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

2015

148,931

156,333

294,781

391,900

12

39,241

144,494

145,368

9,415

43,394

879

2016

138,452

145,290

274,137

365,152

9

36,120

134,392

135,202

8,841

40,020

793

2017

212,390

223,734

417,794

551,403

29

58,768

205,847

207,077

12,800

64,634

1,351

Average
Annual

253,535

267,990

497,095

651,836

47

72,763

245,484

246,925

14,857

79,811

1,714

Table E-XI-6. Modeled Total Streamflow (in-acre/year; Stormflow + Groundwater)

Results

Year

Critical Water Quality Segments

Impaired Ponds

Upper

Lobe

Basin

Upper
Mystic
Lake

Upper
Basin

Lower
Basin

Blacks
Nook Pond
(MA71005),
Cambridge

Horn Pond

(MA71019),

Woburn

Judkins
Pond

(MA71021),
Winchester

Mill Pond

(MA71031),

Winchester

Spy Pond

(MA71040),

Arlington

Wedge
Pond

(MA71045),
Winchester

Winter
Pond

(MA71047),
Winchester

1992

473,814

502,525

907,338

1,177,424

114

141,818

458,409

461,033

25,166

155,005

3,470

1993

362,014

381,026

693,550

912,199

50

101,547

350,892

352,982

19,986

111,556

2,343

1994

440,045

464,573

840,920

1,100,707

76

127,061

426,332

428,835

23,688

139,223

2,965

1995

319,969

338,156

614,333

802,704

61

92,322

309,847

311,649

17,567

101,265

2,207

1996

534,559

565,859

1,019,728

1,325,728

118

159,539

517,277

520,282

27,947

174,239

3,855

1997

193,024

201,978

373,533

496,283

14

50,335

187,291

188,437

11,426

55,672

1,143

1998

626,677

664,704

1,194,696

1,546,801

151

190,312

605,899

609,347

32,765

208,095

4,674

1999

421,756

447,794

808,944

1,047,621

105

126,822

407,813

410,112

22,725

138,900

3,145

2000

413,174

436,556

791,917

1,035,245

78

119,408

400,198

402,537

22,430

130,820

2,830

2001

234,794

246,713

450,759

594,140

29

64,841

227,602

228,970

13,147

71,313

1,498

2002

284,014

297,965

548,197

725,806

28

75,969

275,518

277,183

16,475

83,830

1,731

2003

354,516

372,583

678,682

895,005

40

98,208

343,739

345,795

19,753

108,108

2,228

2004

441,808

466,744

843,624

1,101,102

84

129,210

427,736

430,237

23,577

141,479

3,073

2005

363,904

383,239

699,115

919,117

54

101,881

352,725

354,825

20,224

111,866

2,373

2006

545,920

577,050

1,041,063

1,357,145

110

161,069

528,502

531,600

28,660

176,004

3,844

2007

372,161

392,697

711,503

930,404

65

107,563

360,348

362,462

20,140

117,966

2,551

2008

539,877

570,597

1,031,380

1,346,621

103

157,707

522,808

525,856

28,876

172,699

3,733

E-166


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Year

Critical Water Quality Segments

Impaired Ponds

Upper

Lobe

Basin

Upper
Mystic
Lake

Upper
Basin

Lower
Basin

Blacks
Nook Pond
(MA71005),
Cambridge

Horn Pond

(MA71019),

Woburn

Judkins
Pond

(MA71021),
Winchester

Mill Pond

(MA71031),

Winchester

Spy Pond

(MA71040),

Arlington

Wedge
Pond

(MA71045),
Winchester

Winter
Pond

(MA71047),
Winchester

2009

315,101

330,853

607,792

804,045

33

84,823

305,679

307,517

18,195

93,558

1,930

2010

577,106

611,367

1,097,829

1,424,813

129

174,166

558,267

561,477

29,934

190,354

4,209

2011

431,189

453,885

826,160

1,087,526

57

120,521

418,056

420,539

23,924

132,514

2,755

2012

283,459

298,256

545,007

718,721

35

78,090

274,919

276,554

16,021

85,964

1,780

2013

359,552

378,896

687,244

900,835

55

102,750

348,254

350,308

19,609

112,837

2,406

2014

414,783

437,920

794,928

1,040,401

74

119,195

401,805

404,163

22,592

130,658

2,814

2015

227,442

238,405

439,447

583,284

18

59,946

220,738

222,073

13,381

66,293

1,344

2016

211,442

221,568

408,654

543,449

13

55,179

205,305

206,543

12,565

61,138

1,213

2017

324,347

341,148

623,052

820,932

42

89,777

314,464

316,343

18,191

98,740

2,067

Average
Annual

387,170

408,579

741,515

970,692

67

111,156

375,016

377,217

21,114

121,926

2,622

E-167


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Appendix F: Baseflow Estimates for Aberjona
River and Alewife Brook

Table F-XI-7. Baseflow Estimates for Aberjona River and Alewife Brook

Year

Aberjona River (1992 - 2016)

Alewife Brook (2006 - 2016)

Streamflow
(in-acre/year)

Baseflow (in-
acre/year)

Baseflow
Fraction

Streamflow
(in-acre/year)

Baseflow (in-
acre/year)

Baseflow
Fraction

1992

242,453

160,274

0.66

—

—

—

1993

287,207

193,927

0.68

—

—

—

1994

302,226

203,394

0.67

—

—

—

1995

199,736

145,786

0.73

—

—

—

1996

430,248

270,572

0.63

—

—

—

1997

223,833

162,732

0.73

—

—

—

1998

431,381

268,083

0.62

—

—

—

1999

233,642

155,263

0.66

—

—

—

2000

300,981

193,870

0.64

—

—

—

2001

316,739

195,302

0.62

—

—

—

2002

243,445

155,279

0.64

—

—

—

2003

360,401

238,359

0.66

—

—

—

2004

336,903

208,603

0.62

—

—

—

2005

386,938

266,063

0.69

—

—

—

2006

491,444

299,860

0.61

101,411

65,492

0.65

2007

293,250

184,530

0.63

70,514

47,048

0.67

2008

471,436

297,944

0.63

92,796

59,443

0.64

2009

394,291

266,786

0.68

76,214

54,396

0.71

2010

482,776

269,709

0.56

106,148

64,646

0.61

2011

444,306

290,130

0.65

102,179

71,934

0.70

2012

224,915

131,197

0.58

77,111

57,112

0.74

2013

255,365

189,513

0.74

70,244

53,077

0.76

2014

381,277

248,166

0.65

85,295

61,040

0.72

2015

261,463

179,430

0.69

73,946

54,779

0.74

2016

210,791

142,541

0.68

69,169

52,276

0.76

2017

312,960

210,840

0.67

95,648

72,209

0.75

Avg.

327,668

212,591

0.65

85,056

59,454

0.70

CV

0.28

0.25

0.065

0.16

0.12

0.069

F-168


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Appendix G: Maps Depicting CSO Drainage Basins

SOMOOi

SOM004-

MEDFORD

SOM002A/00:

T\ SOM001-

IOM007 U

CAM401
CAM400

-SOM007i

SOMOOf

/R205

CAM004
CAM401A-

SOMERVILLE

BOS 017-

Figure G-XI-1. MWRA CSO Map (updated in September 2015).

ARLINGTON

CAiViLitlDGE

^	Roadi

1,250

~ Feet
2.500

Legend

| Town Boundaries	V7A CSO (Cambridge)

Vteterbodies	KX>0i CSO (Somervilie)

Roads	CSO Outfalls (MWRA)

Interstate	• ACTIVE

US Highway	• CLOSED

State Route	SPECIAL (Treated CSO)

Mystic River Subbasins

Alewife Brook
Lower Mystic LaKe
Maiden River
Mystic River
! Spy Pond

MALDEN

G-169


-------
My stic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Alewife Sub-Watershed Eutrophication Study
City of Cambridge: 2000

	| Separated Area	ฉ CSO Outfall

j Combined Sewer Area	Watershed Boundary

[;"ฆ -y ; I Separated (With Loading From Town of Belmont)
] Separated (Wrth Active Common Manholes)

Figure G-XI-2. City of Cambridge CSO Drainage Basins (2000).

G-170


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

BOSTON

Alewife Sub-Watershed Eutrophication Study
City of Cambridge: 2017

| Separated Area	ฉ CSO Outfall

I I Combined Sewer Area	Watershed Boundary

X] Separated (With Loading From Town of Belmont)

Figure G-XI-3. City of Cambridge CSO Drainage Basins (2017).

G-171


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Legend

Area

MWRA Connection

Combined Sewer Overflow



Somerville Marginal Interceptor (SMI)

Mystic River



Alewife Brook Conduit (ABC)

Alewife Brook



Cambridge Branch Sewer (CBS)

Interactions with SMI and McGrath system



Primary to CBS with overflows to ABC

Alewife Brook, plus interactions with CBS

Figure G-XI-4. City of Somerville CSO Drainage Basins (2017).

G-172


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Appendix H: Water Quality Data Used in
Calibration and Validation of the Bathtub Model

Year

Segment

WQ Site ID

TP

Chl-a

2010

Upper Lobe

UPLUPL





2014

Upper Lobe

UPLUPL





2015

Upper Lobe

UPLUPL

0.052

16.790

2016

Upper Lobe

UPLUPL

0.060

17.103

2017

Upper Lobe

UPLUPL

0.053

13.351

2010

Upper Lake

UPLCTR





2010

Upper Lake

UPL001

0.029



2014

Upper Lake

UPLCTR





2014

Upper Lake

UPL001

0.038



2015

Upper Lake

UPLCTR

0.035

8.929

2015

Upper Lake

UPL001

0.029



2016

Upper Lake

UPLCTR

0.037

8.709

2016

Upper Lake

UPL001

0.022



2017

Upper Lake

UPLCTR

0.036

13.257

2017

Upper Lake

UPL001





2010

Lower Lake

MYR071

0.036



2014

Lower Lake

MYR071

0.034



2015

Lower Lake

MYR071

0.036



2016

Lower Lake

MYR071

0.036



2017

Lower Lake

MYR071

0.038



2010

Upper Basin

MWRA083

0.036

7.158

2010

Upper Basin

MWRA066

0.050

7.015

2014

Upper Basin

MWRA083

0.044

8.379

2014

Upper Basin

MWRA066

0.045

9.023

2015

Upper Basin

MYR43

0.056

17.944

2016

Upper Basin

MYR43

0.072

21.273

2017

Upper Basin

MYR43

0.062

22.476

2010

Lower Basin

MYR33





2010

Lower Basin

MAR003





2010

Lower Basin

MWRA167

0.054

14.582

2014

Lower Basin

MYR33





2014

Lower Basin

MAR003





2014

Lower Basin

MWRA167

0.047

15.574

2015

Lower Basin

MYR33

0.064

23.771

2015

Lower Basin

MAR003

0.059

22.881

2015

Lower Basin

MWRA167

0.054

23.951

2016

Lower Basin

MYR33

0.092

29.825

2016

Lower Basin

MAR003

0.094

26.699

2016

Lower Basin

MWRA167

0.080

35.384

J-173


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Year

Segment

WQ Site ID

TP

Chl-a

2017

Lower Basin

MYR33

0.072

26.674

2017

Lower Basin

MAR003

0.065

24.750

2017

Lower Basin

MWRA167

0.062

30.367

J-174


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Appendix I: Bathtub Model Inputs for Calibration

Calculation Input Worksheet (English units)

Type

Parameter

Upper
Lobe

Upper
Lake

Lower
Lake

Upper
Basin

Lower
Basin

Atmospheric

Precipitation

(m/yr.)

34.8

34.8

34.8

34.8

34.8

Atmospheric

Lake Evaporation
(m/yr.)

37.7

37.7

37.7

37.7

37.7

Atmospheric

Total P Load
(lb./ac/yr.)

0.27

0.27

0.27

0.27

0.27

Atmospheric

Inorg P Fraction (-)

50%

50%

50%

50%

50%

Atmospheric

Inorg P Load
(lb./ac/yr.)

0.13

0.13

0.13

0.13

0.13

Atten*
External

Area (mi2)

24.72





5.87

10.28

Atten*
External

Flow (ac-in/yr.)

226,408





95,529

123,300

Atten*
External

Flow (in/yr.)

14.31





25.43

18.74

Atten*
External

P Load (lb./yr.)

2,701





1,108

2,312

Atten*
External

P Load (lb./ac/yr.)

0.17





0.29

0.35

Sub-basin*

Area (mi2)

0.21

1.90

6.59

5.50

0.97

Sub-basin*

Flow (ac-in/yr.)

1,034

10,963

49,212

56,766

20,881

Sub-basin*

Flow (in/yr.)

7.69

9.02

11.67

16.13

33.64

Sub-basin*

P Load (lb./yr.)

21

225

996

1,150

623

Sub-basin*

P Load (lb./ac/yr.)

0.16

0.18

0.24

0.33

1.00

Ext* + Sub-
basin

Area (mi2)

24.93

1.90

6.59

11.37

11.25

Ext* + Sub-
basin

Flow (ac-in/yr.)

227,442

10,963

49,212

152,295

144,181

Ext* + Sub-
basin

Flow (in/yr.)

14.26

9.02

11.67

20.93

20.03

Receiving
Water

P Load (lb./yr.)

2,722

225

996

2,257

2,935

Ext* + Sub-
basin

P Load (lb./ac/yr.)

0.17

0.18

0.24

0.31

0.41

Ext* + Sub-
basin

P Cone (mg/L)

0.05

0.09

0.09

0.07

0.09

Ext* + Sub-
basin

Inorg P Fraction (-)

60%

60%

50%

50%

50%

Ext* + Sub-
basin

Inorg P Cone
(mg/L)

0.03

0.05

0.04

0.03

0.04

Ext* + Sub-
basin

Org P Cone (mg/L)

0.02

0.04

0.04

0.03

0.04

J-175


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Receiving
Water

Surface Area (ac)

35.6

140.9

92.8

56.9

124.2

Receiving
Water

Depth (ft)

10.0

42.5

30.8

3.8

4.5

Receiving
Water

Length (mi)

0.36

0.8

0.6

3.6

1.3

Receiving
Water

Epi Depth (ft)

10.0

10.0

10.0

3.8

4.5

Receiving
Water

Hypo Depth (ft)

0.0

32.5

20.8

0.0

0.0

Receiving
Water

Volume (ac-ft)

177.9

5987.5

2861.5

215.9

553.5

Receiving
Water

Cum Flow (ac-
in/yr.)

227,442

238,405

287,617

439,912

584,093

Receiving
Water

Retention Time (d)

3.4

110.0

43.6

2.1

4.2

Receiving
Water

Total P Cone
(mg/L)

0.05

0.03

0.04

0.06

0.06

Receiving
Water

Inorg P Fraction (-)

60%

60%

60%

60%

60%

Receiving
Water

Inorg P Cone
(mg/L)

0.03

0.02

0.02

0.03

0.04

Receiving
Water

Org P Cone (mg/L)

0.02

0.01

0.01

0.02

0.02

Receiving
Water

Chl-a Avg Cone

(mr/l)

16.8

8.9

4.7

17.9

23.5

Receiving
Water

Non-algal Turbidity

(1/ft)

0.17

0.17

0.17

0.17

0.17

Receiving
Water

Internal TP load
(lb./ac/yr.)

19.5

3.3

3.3

13.0

13.0

* Atten — attenuated, Ext — external, Sub-basin — unattenuated local sub-basin

J-176


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Model Input Worksheet (Metric units)

Case Title

Mystic River Alternative TMDL - Final Calibration

Number of Segments

5



Number of Tributaries

5



Number of Channels

0



Global Variables

Mean

CY

Averaging Period (yrs.)

1

0

Precipitation (m)

0.90

0

Evaporation (m)

1

0

Storage Increase (m)

0

0

Atmos. Loads (kg/km2-yr)	Mean	CY

Total P

30

0.5

Ortho P

15

0.5

Segment Data

Segment Number-
Segment Name
Outflow Segment Number-
Segment Group Number

1

2

3

4

5

Upper Lobe

2
1

Upper Lake

3
1

Lower Lake

4
1

Upper Basin

5
1

Lower Basin

0

1

Segment Morphometry

Surface Area (km2)

0.14

0.57

0.38

0.23

0.50

Mean Depth (m)

1.52

12.95

9.40

1.16

1.36

Length (km)

0.58

1.29

0.97

5.79

2.09

Mixed Depth (m)

1.52

3.05

3.05

1.16

1.36

Hypol. Depth (m)

0.00

9.91

6.35

0.00

0.00

Observed Water Quality

Non-Algal Turb (1/m)

0.55

0.55

0.55

0.55

0.55

Conservative Subst

0.00

0.00

0.00

0.00

0.00

Total P (ppb)

52.00

32.00

36.00

56.00

59.00

Chlorophyll-a (ppb)

16.80

8.90

4.70

17.90

23.50

Total P - Ortho P (ppb)

31

19

22

34

35

Segment Calibration Factors

Dispersion Rate

1

1

1

1

1

Total P

1

1

1

1

1

Chlorophyll-a

1

1

1

1

1

J-177


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Total P - Ortho P (ppb)

1

1

1

1

1

Internal Loading-Rates (me:/m2-
dav)











Total P

6

1

1

4

4

Tributary Data

Tributary Number-

1

2

3

4

5

Tributary Name
Segment Number-
Tributary Type Code

Upper Lobe

1
1

Upper Mystic

2
1

Lower Mystic
3
1

Upper Basin

4
1

Lower Basin
5
1

Drainage Area (km2)
Flow (hm3/yr.)
Total P (ppb)

Ortho P (ppb)

64.57
23.40
53
32

4.92
1.10
90
54

17.07
5.10
89
45

29.45
15.70
65
33

29.14
14.80
90
45

NonPoint Source Areas (km2)

Not used

0

0

0

0

0

Non-Point Source Export Coefficients

Not used

0

0

0

0

0

Transport Channels

Not used

0

0

0

0

0

Model Coefficients (Mean, CV)











Dispersion Rate
Total Phosphorus
Chl-a Model
TP-OP Model
HODv Model
MODv Model
Minimum Qs (m/yr.)

Chl-a Flushing Term
Chl-a Temporal CY
Availability Factor - Total P
Availability Factor - Ortho P

0.20
1

0.60
1
1
1

0.10
1

0.62
0.33
1.93

0.70
0.45
0.26
0.15
0.15
0.22
0
0
0
0
0







Model Options











Phosphorus Balance

Chlorophyll-a

Dispersion

1

2
1









J-178


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Phosphorus Calibration

1

Error Analysis

1

Availability Factors

0

Mass-Balance Tables

1

Output Destination

2

Appendix J: Bathtub Model Inputs and Outputs for
Scenarios

Table J-XI-8. Detailed BATHTUB Model Inputs and Outputs for Average Annual
Data

Model

Parameter

Scenario -

Upper

Upper

Lower

Upper

Lower

Input/



Run

Lobe

Lake

Lake

Basin

Basin

Output



















l-l

371,417





148,218

192,965





la - 7

371,417





148,218

192,965





2-9

371,386





147,525

192,963



Atten Trib Flow

2a -14

371,386





147,525

192,963



(ac-in/yr.)

3-16

371,386





157,040

192,963





3a - 21

371,386





157,040

192,963





4-23

371,386





185,594

192,963





4a - 28

371,386





185,594

192,963





1-1

1,856

20,243

84,185

90,239

30,380





la - 7

1,856

20,243

84,185

90,239

30,380





2-9

1,855

20,238

84,159

90,202

30,185



Sub-basin Flow

2a - 14

1,855

20,238

84,159

90,202

30,185



(ac-in/yr.)

3-16

1,855

20,238

84,159

90,331

31,315

Input



3a - 21

1,855

20,238

84,159

90,331

31,315





4-23

1,855

20,238

84,159

90,719

34,701





4a - 28

1,855

20,238

84,159

90,719

34,701





1-1

373,272

20,243

84,185

238,457

223,345





la - 7

373,272

20,243

84,185

238,457

223,345





2-9

373,241

20,238

84,159

237,727

223,148



Total Flow (ac-

2a - 14

373,241

20,238

84,159

237,727

223,148



in/yr.)

3-16

373,241

20,238

84,159

247,371

224,278





3a - 21

373,241

20,238

84,159

247,371

224,278





4-23

373,241

20,238

84,159

276,314

227,664





4a - 28

373,241

20,238

84,159

276,314

227,664



Atten Trib P Load
(Ib./yr.)

1-1

3,873





1,637

3,130



la - 7

1,476





872

1,186



2-9

3,858





1,352

3,128

J-179


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Model
Input/
Output

Parameter

Scenario -
Run

Upper
Lobe

Upper
Lake

Lower
Lake

Upper
Basin

Lower
Basin





2a -14

1,640





644

1,329





3-16

3,858





1,442

3,128





3a - 21

1,676





667

1,358





4-23

3,858





1,603

3,128





4a - 28

1,748





733

1,416





1-1

34

416

1,558

1,661

719





la - 7

14

161

625

681

421





2-9

34

410

1,526

1,616

579



Sub-basin P Load

2a - 14

15

175

663

710

303



(Ib./yr.)

3-16

34

410

1,526

1,618

577





3a - 21

15

178

677

725

294





4-23

34

410

1,526

1,624

570





4a - 28

16

186

705

757

264





1-1

3,907

416

1,558

3,298

3,849





la - 7

1,490

161

625

1,553

1,608





2-9

3,892

410

1,526

2,968

3,707



Total P Load

2a - 14

1,655

175

663

1,354

1,632



(Ib./yr.)

3-16

3,892

410

1,526

3,033

3,704





3a - 21

1,691

178

677

1,393

1,652





4-23

3,892

410

1,526

3,227

3,697





4a - 28

1,763

186

705

1,490

1,680





1-1

695

459

302

741

1,619





la - 7

466

307

202

497

1,084





2-9

695

459

302

741

1,619



P Sediment load

2a - 14

480

317

209

512

1,117



(Ib./yr.)

3-16

695

459

302

741

1,619





3a - 21

480

317

209

512

1,117





4-23

695

459

302

741

1,619





4a - 28

494

326

215

526

1,149





1-1

0.78

1.35

1.31

1.00

1.15





la - 7

0.78

1.35

1.31

1.00

1.15





2-9

0.78

1.35

1.31

1.00

1.15



N cone (mg/L)

2a - 14

0.78

1.35

1.31

1.00

1.15



3-16

0.78

1.35

1.31

0.96

1.15





3a - 21

0.78

1.35

1.31

0.96

1.15





4-23

0.78

1.35

1.31

0.85

1.14





4a - 28

0.78

1.35

1.31

0.85

1.14



Inorg. N cone

1-1

0.47

0.81

0.78

0.60

0.69



(mg/L)

la - 7

0.47

0.81

0.78

0.60

0.69

J-180


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Model
Input/
Output

Parameter

Scenario -
Run

Upper
Lobe

Upper
Lake

Lower
Lake

Upper
Basin

Lower
Basin





2-9

0.47

0.81

0.78

0.60

0.69





2a - 14

0.47

0.81

0.78

0.60

0.69





3-16

0.47

0.81

0.78

0.58

0.69





3a - 21

0.47

0.81

0.78

0.58

0.69





4-23

0.47

0.81

0.78

0.51

0.68





4a - 28

0.47

0.81

0.78

0.51

0.68





1-1

0.05

0.09

0.08

0.06

0.08





la - 7

0.02

0.04

0.03

0.03

0.03





2-9

0.05

0.09

0.08

0.06

0.07



P cone (mg/L)

2a - 14

0.02

0.04

0.04

0.03

0.03



3-16

0.05

0.09

0.08

0.05

0.07





3a - 21

0.02

0.04

0.04

0.03

0.03





4-23

0.05

0.09

0.08

0.05

0.07





4a - 28

0.02

0.04

0.04

0.02

0.03





1-1

0.03

0.05

0.04

0.03

0.04





la - 7

0.01

0.02

0.02

0.01

0.02





2-9

0.03

0.05

0.04

0.03

0.04



Inorg. P Cone

2a - 14

0.01

0.02

0.02

0.01

0.02



(mg/L)

3-16

0.03

0.05

0.04

0.03

0.04





3a - 21

0.01

0.02

0.02

0.01

0.02





4-23

0.03

0.05

0.04

0.03

0.04





4a - 28

0.01

0.02

0.02

0.01

0.02





1-1

0





la - 7

67





2-9

0



Stormwater load

2a - 14

62



reduction (%)

3-16

0





3a - 21

61





4-23

0





4a - 28

61





1-1

0





la - 7

33





2-9

0



P sediment load

2a - 14

31



reduction (%)

3-16

0





3a - 21

31





4-23

0





4a - 28

29

Output



1-1

43.4

33.1

33.6

46.2

57.7

J-181


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Model
Input/
Output

Parameter

Scenario -
Run

Upper
Lobe

Upper
Lake

Lower
Lake

Upper
Basin

Lower
Basin





la - 7

20.8

18.6

19.4

25.3

31.1





2-9

43.3

33.0

33.3

44.2

55.8



Predicted P cone
(|ig/L)

2a -14

22.4

19.8

20.4

24.9

31.1



3-16

43.3

33.0

33.3

44.2

55.6



3a - 21

22.7

20.0

20.6

24.9

31.1





4-23

43.2

32.9

33.2

43.4

54.5





4a - 28

23.4

20.4

21.1

25

31.1



Predicted chl-a

1-1

13.9

7.6

7.7

16.4

19.4



cone (ng/L)

la - 7

5.9

4.1

4.3

8.1

10.0





2-9

13.9

7.6

7.7

15.6

18.8





2a - 14

6.5

4.4

4.6

7.9

9.9





3-16

13.8

7.6

7.6

15.6

18.7





3a - 21

6.6

4.5

4.6

7.9

9.9





4-23

13.8

7.5

7.6

15.3

18.3





4a - 28

6.9

4.6

4.7

7.9

9.9

Table J-XI-9. Detailed BATHTUB Model Inputs and Outputs for Wet and Dry Year
Data

Model
Input/
Output

Parameter

Scenario -
Run

Upper
Lobe

Upper
Lake

Lower
Lake

Upper
Basin

Lower
Basin

Input



la - 7 (Wet)

574,239





218,660

285,460





la - 7 (Dry)

210,500





88,880

115,436





2a -14
(Wet)

574,105





216,564

285,449



Atten Trib Flow
(ac-in/yr.)

2a -14 (Dry)

210,499





89,101

115,436



3a - 21
(Wet)

574,105





230,094

285,449





3a - 21 (Dry)

210,499





95,041

115,436





4a - 28
(Wet)

574,105





270,695

285,449





4a - 28 (Dry)

210,499





112,870

115,436





la - 7 (Wet)

3,137

34,308

135,055

135,664

42,401





la - 7 (Dry)

942

10,127

45,401

52,853

19,407





2a -14
(Wet)

3,133

34,284

134,942

135,512

41,679



Sub-basin Flow

2a -14 (Dry)

942

10,127

45,401

52,853

19,488



(ac-in/yr.)

3a - 21
(Wet)

3,133

34,284

134,942

135,703

43,217





3a - 21 (Dry)

942

10,127

45,401

52,930

20,221





4a - 28
(Wet)

3,133

34,284

134,942

136,277

47,825

J-182


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Model
Input/
Output

Parameter

Scenario -
Run

Upper
Lobe

Upper
Lake

Lower
Lake

Upper
Basin

Lower
Basin





4a - 28 (Dry)

942

10,127

45,401

53,161

22,418

Total Flow (ac-
in/yr.)

la - 7 (Wet)

577,376

34,308

135,055

354,324

327,860

la - 7 (Dry)

211,442

10,127

45,401

141,733

134,842

2a -14
(Wet)

577,238

34,284

134,942

352,077

327,128

2a -14 (Dry)

211,442

10,127

45,401

141,953

134,923

3a - 21
(Wet)

577,238

34,284

134,942

365,797

328,666

3a - 21 (Dry)

211,442

10,127

45,401

147,970

135,657

4a - 28
(Wet)

577,238

34,284

134,942

406,972

333,274

4a - 28 (Dry)

211,442

10,127

45,401

166,031

137,854

Atten Trib P Load
(Ibs./yr.)

la - 7 (Wet)

2,127





1,607

1,579

la - 7 (Dry)

960





350

827

2a -14
(Wet)

2,303





818

1,752

2a -14 (Dry)

1,082





482

932

3a - 21
(Wet)

2,351





854

1,789

3a - 21 (Dry)

1,106





492

953

4a - 28
(Wet)

2,448





955

1,864

4a - 28 (Dry)

1,155





518

995

Sub-basin P Load
(Ibs./yr.)

la - 7 (Wet)

27

301

1,089

1,151

865

la - 7 (Dry)

7

79

353

410

170

2a -14
(Wet)

25

304

1,053

1,066

371

2a -14 (Dry)

8

90

398

462

244

3a - 21
(Wet)

26

310

1,073

1,085

365

3a - 21 (Dry)

9

92

407

473

232

4a - 28
(Wet)

27

322

1,112

1,124

343

4a - 28 (Dry)

9

96

425

496

193

Total P Load
(Ibs./yr.)

la - 7 (Wet)

2,153

301

1,089

2,758

2,445

la - 7 (Dry)

968

79

353

760

998

2a -14
(Wet)

2,328

304

1,053

1,884

2,124

2a -14 (Dry)

1,090

90

398

944

1,177

3a - 21
(Wet)

2,377

310

1,073

1,939

2,155

3a - 21 (Dry)

1,115

92

407

965

1,186

J-183


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Model
Input/
Output

Parameter

Scenario -
Run

Upper
Lobe

Upper
Lake

Lower
Lake

Upper
Basin

Lower
Basin





4a - 28
(Wet)

2,474

322

1,112

2,079

2,206

4a - 28 (Dry)

1,164

96

425

1,015

1,188

P Sediment load
(Ib./yr.)

la - 7 (Wet)

466

307

202

497

1,084

la - 7 (Dry)

466

307

202

497

1,084

2a -14
(Wet)

480

317

209

512

1,117

2a -14 (Dry)

480

317

209

512

1,117

3a - 21
(Wet)

480

317

209

512

1,117

3a - 21 (Dry)

480

317

209

512

1,117

4a - 28
(Wet)

494

326

215

526

1,149

4a - 28 (Dry)

494

326

215

526

1,149

N cone (mg/L)

la - 7 (Wet)

0.506

0.794

0.814

0.673

0.782

la - 7 (Dry)

1.383

2.691

2.422

1.683

1.900

2a -14
(Wet)

0.507

0.795

0.815

0.677

0.783

2a -14 (Dry)

1.383

2.691

2.422

1.680

1.899

3a - 21
(Wet)

0.507

0.795

0.815

0.650

0.782

3a - 21 (Dry)

1.383

2.691

2.422

1.607

1.894

4a - 28
(Wet)

0.507

0.795

0.815

0.579

0.777

4a - 28 (Dry)

1.383

2.691

2.422

1.420

1.878

Inorg N cone
(mg/L)

la - 7 (Wet)

0.304

0.477

0.489

0.404

0.469

la - 7 (Dry)

0.830

1.614

1.453

1.010

1.140

2a -14
(Wet)

0.304

0.477

0.489

0.406

0.470

2a -14 (Dry)

0.830

1.614

1.453

1.008

1.139

3a - 21
(Wet)

0.304

0.477

0.489

0.390

0.469

3a - 21 (Dry)

0.830

1.614

1.453

0.964

1.137

4a - 28
(Wet)

0.304

0.477

0.489

0.348

0.466

4a - 28 (Dry)

0.830

1.614

1.453

0.852

1.127

P cone (mg/L)

la - 7 (Wet)

0.016

0.039

0.036

0.034

0.033

la - 7 (Dry)

0.020

0.035

0.034

0.024

0.033

2a -14
(Wet)

0.018

0.039

0.034

0.024

0.029

2a -14 (Dry)

0.023

0.039

0.039

0.029

0.038

3a - 21
(Wet)

0.018

0.040

0.035

0.023

0.029

J-184


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Model
Input/
Output

Parameter

Scenario -
Run

Upper
Lobe

Upper
Lake

Lower
Lake

Upper
Basin

Lower
Basin





3a - 21 (Dry)

0.023

0.040

0.040

0.029

0.039

4a - 28
(Wet)

0.019

0.041

0.036

0.023

0.029

4a - 28 (Dry)

0.024

0.042

0.041

0.027

0.038

Inorg P Cone
(mg/L)

la - 7 (Wet)

0.010

0.023

0.018

0.017

0.016

la - 7 (Dry)

0.012

0.021

0.017

0.012

0.016

2a -14
(Wet)

0.011

0.023

0.017

0.012

0.014

2a -14 (Dry)

0.014

0.023

0.019

0.015

0.019

3a - 21
(Wet)

0.011

0.024

0.018

0.012

0.014

3a - 21 (Dry)

0.014

0.024

0.020

0.014

0.019

4a - 28
(Wet)

0.011

0.025

0.018

0.011

0.015

4a - 28 (Dry)

0.015

0.025

0.021

0.013

0.019

Stormwater load
reduction (%)

la - 7 (Wet)

67

la - 7 (Dry)

67

2a -14
(Wet)

62

2a -14 (Dry)

62

3a - 21
(Wet)

61

3a - 21 (Dry)

61

4a - 28
(Wet)

59

4a - 28 (Dry)

59

P sediment load
reduction (%)

la - 7 (Wet)

33

la - 7 (Dry)

33

2a -14
(Wet)

31

2a -14 (Dry)

31

3a - 21
(Wet)

31

3a - 21 (Dry)

31

4a - 28
(Wet)

29

4a - 28 (Dry)

29

Output

Predicted P cone
(Mg/L)

la - 7 (Wet)

18.9

17.9

19.8

26.2

30.5

la - 7 (Dry)

24.9

20.3

20.1

26.3

35.0

2a -14
(Wet)

20.0

18.7

20.2

23.2

27.4

2a -14 (Dry)

26.9

21.4

21.2

29.0

38.4

J-185


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Model
Input/
Output

Parameter

Scenario -
Run

Upper
Lobe

Upper
Lake

Lower
Lake

Upper
Basin

Lower
Basin





3a - 21
(Wet)

20.4

19.0

20.4

23.3

27.5

3a - 21 (Dry)

27.2

21.6

21.4

28.9

38.3

4a - 28
(Wet)

21.0

19.5

21.0

23.4

27.7

4a - 28 (Dry)

28.1

22.0

21.8

28.7

37.9

Predicted Chl-a
cone (ng/L)

la - 7 (Wet)

5.2

3.8

4.3

8.3

9.5

la - 7 (Dry)

7.5

4.7

4.6

8.6

11.6

2a -14
(Wet)

5.6

4.0

4.4

7.1

8.4

2a -14 (Dry)

8.3

5.0

4.9

9.7

12.9

3a - 21
(Wet)

5.7

4.1

4.4

7.2

8.4

3a - 21 (Dry)

8.4

5.0

5.0

9.7

12.9

4a - 28
(Wet)

5.9

4.2

4.5

7.2

8.5

4a - 28 (Dry)

8.7

5.1

5.1

9.6

12.7

J-186


-------
Mystic River Watershed TMDL Alternative Development — Phase 2 Technical Memoranda

Appendix K. BMP Design Parameters Used in the
Pilot Watershed

General
Information

BMP Parameters

Biofilt ration

Infiltration-B

Infiltration-C

Porous
Pavement

BMP Dimensions

Surface Area (ac)

Table X-7

Table X-7

Table X-7

Table X-7

Surface Storage
Configuration

Soil Properties

Orifice Height (ft)

0

0

0

0

Orifice Diameter
(in.)

0

0

0

0

Rectangular or
Triangular Weir

Rectangular

Rectangular

Rectangular

Rectangular

Weir Height
(ft)/Ponding Depth
(ft)

0.5

2

2

0.2

Crest Width (ft)

100

100

100

100

Depth of Soil (ft)

2.5

0

0

2.67

Soil Porosity (0-1)

0.2

0.4

0.4

0.23

Vegetative
Parameter A

0.9

0.9

0.9

0.1

Soil Infiltration
(in/hr.)

2.5

2.41

0.52

17.42

Underdrain
Properties

Consider

Underdrain

Structure?

Yes

No

No

Yes

Storage Depth (ft)

1

0

0

1.75

Media Void Fraction
(0-1)

0.4

0

0

0.4

Background
Infiltration (in/hr.)

0

2.41

0.52

0

Cost Parameters

Storage Volume
Cost ($/ft3)

$15.46

$6.24

$6.24

$5.32

Cost Function
Adjustment

BMP Development
Type

New BMP in
Developed
Area

New BMP in
Developed
Area

New BMP in
Developed
Area

Difficult
Installation in
Highly Urban
Settings

Cost Adjustment
Factor

2

2

2

3

Decay Rates

TP (1/hr.)

0.13

0.27

0.27

0.0051

Underdrain
Removal Rates

TP (%, 0-1)

0.43

0

0

0.1

K-187


-------