4>EPA
PA/600/R-14/179 I May 2015 I www.epa.gov/hfstudy
United States
Environmental Prot
Agency
Case Study Analysis of the Impacts of
Water Acquisition for Hydraulic
Fracturing on Local Water Availablility
Office of Research and Development
National Exposure Research Laboratory
-------
Water Acquisition for Hydraulic Fracturing May 2015
[This page intentionally left blank/
-------
Water Acquisition for Hydraulic Fracturing May 2015
Case Study Analysis of the
Impacts of Water Acquisition for Hydraulic Fracturing on
Local Water Availability
U.S. Environmental Protection Agency
Office of Research and Development
Washington, D.C.
May 2015
EPA/600/R-14/179
-------
Water Acquisition for Hydraulic Fracturing May 2015
Disclaimer
This document has been reviewed in accordance with U.S. Environmental Protection Agency policy and
approved for publication. Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
Preferred Citation: U.S. Environmental Protection Agency. 2015. Case study analysis of the impacts of
water acquisition for hydraulic fracturing on local water availability. EPA/600/R-14/179.
-------
Water Acquisition for Hydraulic Fracturing May 2015
Table of Contents
Disclaimer iv
Table of Contents v
List of Tables vi
List of Figures vii
Abbreviations xi
Units and Conversions xii
Preface xiii
Authors xiv
Acknowledgments xiv
Chapter 1. Executive Summary 1
Chapter 2. Introduction 5
Chapter 3. Research Objectives and Approach 13
Chapter 4. Marcellus Shale/Susquehanna River Basin 17
Marcellus Shale Geologic Setting 17
Susquehanna River Basin Background 19
Sources of Freshwater for Hydraulic Fracturing 23
Water Use Intensity Analysis at Self-supplied Sites 33
Analysis of Hydraulic Fracturing Water Acquisition on Groundwater 48
Vulnerability of Rivers and Streams to Depletion with Hydraulic Fracturing Withdrawals 57
Water Quality Impacts from Hydraulic Fracturing Withdrawals 60
Susquehanna River Basin Synopsis 62
Chapter 5. Piceance Basin/Upper Colorado River Basin 63
Uinta-Piceance Geologic Province 63
Upper Colorado River Basin Background 65
Sources of Freshwater for Hydraulic Fracturing 71
Water Used For Fracturing Wells 72
Water Use Intensity Analysis at Self-supplied Sites 82
Vulnerability of Streams to Depletions 84
UCRB Synopsis 88
Chapter 6. Synthesis and Summary 91
How much freshwater is used in hydraulic fracturing operations and what are its sources? 92
How might water withdrawals affect short- and long-term availability? 96
What are the potential impacts of hydraulic fracturing water withdrawals on water quality of
source waters? 98
Study Conclusions 98
References 101
Glossary 109
Appendix A. Acquired Data Source A-l
Appendix B. Surface Water Methods B-l
Appendix C. Groundwater Methods C-l
Appendix D. Water Use Estimation Methods D-l
Appendix E. Quality Assurance and Quality Control E-l
-------
Water Acquisition for Hydraulic Fracturing May 2015
List of Tables
Table 4-1. Statistics on injected and returned fluid volumes per well in the Susquehanna River Basin from
2008 to 2011 20
Table 4-2. Water use data sources and reports for the Susquehanna River Basin 23
Table 4-3. New permits for surface water withdrawal for the oil and gas industry 27
Table 4-4. Daily self-supplied water acquired for hydraulic fracturing in the Susquehanna River Basin 30
Table 4-5. Cross-validation of area-weighting method. Each pair represents two gaged watersheds on
which the area-weighting method was tested 33
Table 4-6. Count of daily SUI for all permit sites on days when water was withdrawn from 2009 to 2013 35
Table 4-7. Flow statistics at USGS gage in Towanda Creek for various time periods used in this analysis 41
Table 4-8. Description of hydraulic fracturing withdrawal scenarios applied to 26 years of streamflow
(1987-2012) in subbasins of the Towanda Creek watershed in Bradford County, Pennsylvania 43
Table 4-9. Count of days by surface water use intensity index category with withdrawal simulation of 1 million
gallons per day with and without passby flows applied to Towanda Creek (1987 to 2012) 46
Table 4-10. Groundwater aquifer volumes, annual recharge, and withdrawn volume for public and
domestic water supplies, agricultural uses, and by the O&G industry for hydraulic fracturing 52
Table 4-11. Estimated daily and annual groundwater use in the Towanda Creek basin 53
Table 4-12. Source of freshwater from private wellfields in the Susquehanna River Basin 56
Table 4-13. Minimum flow expressed in cubic feet per second needed to meet a surface water use intensity
index (SUI) value at various daily withdrawal volumes 58
Table 5-1. Undiscovered hydrocarbon reserves in the Piceance Basin 64
Table 5-2. Water use data sources for the Upper Colorado River Basin 71
Table 5-3. Hydraulic fracturing well starts by county 71
Table 5-4. Count of daily surface water use intensity index at active Parachute Creek withdrawal structures
from 2008 to 2013 83
Table 5-5. Water withdrawal volume for hydraulic fracturing withdrawal scenarios applied to 26 years of
streamflow (1987-2012) in subbasins of Parachute and Roan Creeks in Garfield County 85
Table 5-6. Approximate basin area threshold for median 95th percentile for surface water use intensity
indices (SUI) below 0.4 from 2008 to 2013 88
Table 6-1. Characteristics of well drilling and water use in the Susquehanna River Basin and Upper Colorado
River Basin during the peak year of drilling in each area 92
-------
Water Acquisition for Hydraulic Fracturing May 2015
List of Figures
Figure 2-1. The technologies of deep horizontal and directional drilling 5
Figure 2-2. Hydraulic fracturing well pad 6
Figure 2-3. Location of unconventional shale gas plays in North America 6
Figure 2-4. Photograph of mature wellfield in the Uinta Basin, Utah 7
Figure 2-5. Example of multiple subsurface directionally drilled wellbores radiating from each well pad 7
Figure 2-6. Estimated daily water used for hydraulic fracturing in natural gas shale plays 8
Figure 2-7. Intrinsic variability of water resources within the United States 9
Figure 2-8. Daily surface and groundwater usage by all user sectors in 15 states with unconventional oil
and gas reserves in the 2005 USGS water census 10
Figure 2-9. Daily water use (surface and groundwater combined) by major sectors in selected group of
states with unconventional gas reserves 10
Figure 2-10. Most water is trucked from source to well pad in Bradford County, Pennsylvania 11
Figure 3-1. Case study area location map 14
Figure 4-1. Location of the Marcellus Shale and the Devonian Shale 17
Figure 4-2. Natural gas production in the Marcellus Shale 17
Figure 4-3. Characteristics and extent of the Marcellus Shale 18
Figure 4-4. Map of theSusquehanna River Basin 19
Figure 4-5. Annual and cumulative well completions within theSusquehanna River Basin 19
Figure 4-6. West branch of the Susquehanna River near Renovo, Pennsylvania within the Appalachian
Plateau physiographic region 20
Figure 4-7. Major subbasins, counties, and towns in the Susquehanna River Basin 21
Figure 4-8. Aerial view of terrain, land use, and well pads in Bradford County 22
Figure 4-9. Susquehanna River Basin daily water use by sector 22
Figure 4-10. General location of public water suppliers in the Susquehanna River Basin, shown by their
dominant source of water supply (surface water or groundwater) 24
Figure 4-11. Daily sales by user sector for the 29 public water systems that supply water to oil and gas in
theSusquehanna River Basin 25
Figure 4-12. Proportion of public water supply facility capacity used to provide water to the oil and gas
industry at the 29 public water supply facilities registered to sell water for hydraulic fracturing 26
Figure 4-13. Characterization of oil and gas self-supplied water sources in theSusquehanna River Basin 27
Figure 4-14. Distribution of oil and gas self-supplied water sources for hydraulic fracturing in the
Susquehanna River Basin 27
Figure 4-15. Examples of self-supplied withdrawal locations permitted by the Susquehanna River Basin
Commission 28
Figure 4-16. Maximum daily withdrawal limits for oil and gas self-supplied water withdrawal sites 29
-------
Water Acquisition for Hydraulic Fracturing May 2015
Figure 4-17. Volume of water acquired from self-supplied sources annually by water source type 30
Figure 4-18. Total volume of water acquired by the oil and gas industry in the Susquehanna River Basin
from public water systems 31
Figure 4-19. Daily self-supplied water withdrawal for hydraulic fracturing in the Susquehanna River Basin
summed for all active sites 31
Figure 4-20. Acquisition of water for hydraulic fracturing by the oil and gas industry acquired from public
water systems and self-supplied, summed by county 32
Figure 4-21. Self-supplied permitted sites have a daily withdrawal limit and may have a minimum
maintenance flow (passby) 34
Figure 4-22. Count of daily volume withdrawals, all sites combined 35
Figure 4-23. Daily surface water use intensity index (SUI) for all withdrawal sites in the Susquehanna River
Basin used to source water for hydraulic fracturing from 2009 to 2013 35
Figure 4-24. Observed surface water use intensity index at permit sites on all days when water was
withdrawn for hydraulic fracturing from 2009 to 2013 37
Figure 4-25. Photograph of O&G freshwater impoundment to service hydraulic fracturing water needs 37
Figure 4-26. Towanda Creek basin in Bradford County, Pennsylvania 39
Figure 4-27. Visual display of land use, topography, and river size and form in the Towanda Creek basin 40
Figure 4-28. Long-term flow at the USGS gaging station at Monroeton 41
Figure 4-29. Comparison of contibuting watershed area of Towanda Creek subbasins used for scenario
analysis and Susquehanna River Basin Commission-permitted surface water sites in the
Susquehanna River Basin 42
Figure 4-30. SUI at peak drilling and median flow at subbasin streamflow prediction points in Towanda
Creek subbasins 44
Figure 4-31. SUI distribution statistics for withdrawal scenarios applied to subbasin streamflow prediction
points in Towanda Creek 44
Figure 4-32. U.S. Geological Survey Pennsylvania regional flow equations applied to the Towanda Creek
watershed with withdrawal scenarios 45
Figure 4-33. Simulated daily surface water use intensity index computed with the U.S. Geological Survey
gaging record at Towanda Creek from 1987 to 2012 46
Figure 4-34. Count of days each year that the flow fell below the passby flow assigned by the Susquehanna
River Basin Commission to U.S. Geological Survey Towanda Creek gaging station 47
Figure 4-35. SUI analysis of cumulative withdrawals from four permitted sites on Wyalusing Creek 47
Figure 4-36. Daily rate of groundwater supplied to the oil and gas industry from public water systems and
self-supplied sources 48
Figure 4-37. Groundwater wellfields currently available for water acquisition by the oil and gas industry
in the Susquehanna River Basin 48
Figure 4-38. Depth and pumping capacity of public and self-supplied groundwater sources registered to
supply water for hydraulic fracturing in the Susquehanna River Basin classified by geology type 49
Figure 4-39. Schematic of the hydrologic water balance with emphasis on groundwater interactions with
surface water flow and possible impacts from water withdrawals 49
-------
Water Acquisition for Hydraulic Fracturing May 2015
Figure 4-40. Daily hydrograph of streamflow in Towanda Creek at the U.S. Geological Survey (USGS) gage
at Monroeton and USGS observation well BR92 located in alluvium the same general area from
2008 to 2012 50
Figure 4-41. Annual baseflow the normalized to watershed area of 215 mi2 at the U.S. Geological Survey
Towanda gage at Monroeton from the period of record 1915 to 2012 using PART baseflow
separation 51
Figure 4-42. Conceptual graphic of groundwater volumes within the Towanda Creek watershed 52
Figure 4-43. Groundwater wells in the Towanda Creek watershed based on PADCNR (2014a) 53
Figure 4-44. Daily water use at PWS 2080003 municipal water supply in the Towanda Creek basin 54
Figure 4-45. Groundwater-surface water interactions at the PWS 2080003 municipal groundwater wellfield
under average recharge and baseflow conditions (2000-2001) 55
Figure 4-46. Groundwater cone of depression associated with cross section A-A' for the Canton Borough
municipal wellfield for four scenarios 56
Figure 4-47. Heat maps showing probabilities of experiencing various surface water use intensity index (SUI)
values at a range of watershed sizes under several daily withdrawal scenarios based on
streamflow records from 48 streams and rivers in Pennsylvania 59
Figure 4-48. General relationship between the concentration magnification and the percent of water
withdrawn (surface water use intensity index) at a withdrawal location 61
Figure 4-49. Potential magnification of water quality pollutant concentrations at observed hydraulic
fracturing withdrawals, assuming each site had an upstream point source discharge 61
Figure 5-1. Location of the Sevier orogenic belt in relation to the epeiric seaway in the Cretaceous period 63
Figure 5-2. Location of the Uinta-Piceance Province 63
Figure 5-3. Geologic formations in the southern part of the Uinta and Piceance basins 64
Figure 5-4. The hydrologic boundaries of the Colorado River Basin within the United States, plus the
adjacent areas of the basin states that receive Colorado River water 65
Figure 5-5. The Upper Colorado River Basin in Colorado below Grand Junction, where the Upper Colorado
River joins the Gunnison River 66
Figure 5-6. Location of the Upper Colorado River and its basin above Grand Junction, where this project
focuses assessment of hydraulic fracturing water acquisition 66
Figure 5-7. Geologic strata in Garfield County 67
Figure 5-8. Roan Creek Plateau 68
Figure 5-9. Colorado River alluvial valley bottom in the Garfield County along the Interstate 70 corridor
between Rifle and Glenwood Springs 69
Figure 5-10. Upper Colorado River (District 5) daily water use by sector in 2012 70
Figure 5-11. Annual and cumulative hydraulic fracturing well starts in Garfield County since 2000 72
Figure 5-12. Estimated use of water for hydraulic fracturing natural gas extraction in Garfield County 74
Figure 5-13. Estimated freshwater use in Upper Colorado River Division 5 74
Figure 5-14. Location of drinking water sources including public water supplies and private groundwater
wells in the Upper Colorado River Basin between Glenwood Springs and Cameo west of
DeBeque 75
-------
Water Acquisition for Hydraulic Fracturing May 2015
Figure 5-15. Annual water volume acquired for drinking water at three municipal public water systems
in the vicinity of Parachute and Roan Creeks, where current hydraulic fracturing drilling activity
is focused 76
Figure 5-16. Parachute and Roan Creek watersheds with completed gas wells, and O&G water withdrawal
locations 77
Figure 5-17. Sources of freshwater acquired for the oil and gas industry in the Upper Colorado River Basin 78
Figure 5-18. Parachute Creek water acquisition sites 79
Figure 5-19. Annual water withdrawn from surface waters in Parachute Creek for irrigation and for
hydraulic fracturing by the oil and gas industry 80
Figure 5-20. Daily streamflow and collective withdrawal volume summing 29 structures withdrawing water
from surface water structures in Parachute Creek 81
Figure 5-21. Surface water use intensity index for 29 primary water withdrawal sites in Parachute Creek 81
Figure 5-22. Surface water use intensity index observations at individual withdrawal sites (structures) in
Parachute Creek 82
Figure 5-23. Parachute Creek confluence self-supplied groundwater wellfield for oil and gas, showing
area of potential impact and surface water zone of capture for two scenarios based on GFLOW
simulations assuming averaged geohydrologic conditions (2007-2012) 84
Figure 5-24. Annual water pumping and groundwater use intensity index at the Union 76 wellfield 83
Figure 5-25. Distribution of surface water use intensity index at streamflow prediction locations in Parachute
and Roan Creek for the current drilling-directional wells scenario, modeled from 2008 to 2013 86
Figure 5-26. Surface water use intensity indices (SUI) for drilling scenarios using current and peak drilling
rates 87
Figure 5-27. Overview of current water use in Garfield County and existing or potential use to support oil
and gas development with hydraulic fracturing and oil shale extraction 89
Figure 5-28. Projected water supply and use in the Colorado River Basin with growing population in the
next decades 90
Figure 6-1. Total fluid and freshwater use in the the study basins in the peak year of drilling in each 94
Figure 6-2. Self-supplied water acquisition site on a river in the SRB 94
Figure 6-3. Trucking water to a hydraulic fracturing well site in Virginia 94
Figure 6-4. Volume of water sourced from municipal public water suppliers and self-supplied by the oil and
gas industry for hydraulic fracturing in the peak drilling year in each study
basin 95
Figure 6-5. Volume of water sourced from surface water and groundwater sources by the oil and gas industry
for hydraulic fracturing in the peak drilling year in each study basin 96
-------
Water Acquisition for Hydraulic Fracturing
May 2015
List of Abbreviations
ac-ft Acre-feet
ASL Above sea level
BLM Bureau of Land Management (United States Department of the Interior)
CDF Cumulative distribution function
CFS Cubic feet per second
CMS Cubic meters per second
CODWR Colorado Department of Water Resources
COGCC Colorado Oil and Gas Conservation Commission
CWCB Colorado Water Conservation Board
EIA United States Energy Information Administration
EPA United States Environmental Protection Agency
FEMA Federal Emergency Management Agency
ft Foot
ft3 Cubic foot
GIS Geographic information system
GUI Groundwater use intensity index
gal Gallon
HSPF Hydrologic Simulation Program—Fortran
HUC Hydrologic unit code
m3 Cubic meter
MG Million gallons
MGD Million gallons per day
MGY Million gallons per year
mi Mile
mi2 Square mile
NCDC National Climatic Data Center
NLCD National Land Cover Data Set
O&G Oil and gas
NS Nash-Sutcliffe efficiency score
PADEP Pennsylvania Department of Environmental Protection
PADCNR Pennsylvania Department of Conservation and Natural Resources
PAGWIS Pennsylvania Groundwater Information System
PEST Parameter estimation tool
PWS Public water system
Q?,2 The lowest seven-day average flow that occurs once every two years at a river gage
Qy.io The lowest seven-day average flow that occurs once every 10 years at a river gage
Qmean Mean annual flow
RMSE Root mean squared error
SRB Susquehanna River Basin
SRBC Susquehanna River Basin Commission
STATSGO State Soil Geographic Database
SUI Surface water use intensity index
UCRB Upper Colorado River Basin
USDA United States Department of Agriculture
USFWS United States Fish and Wildlife Service
USGS United States Geological Survey
WCD West Conservancy District
WNS Weighted Nash-Sutcliffe efficiency score
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Units and Conversions
Length
"~\TO
From ^"~\^
mi
km
ft
Mi
1
0.621
0.0001894
km
1.609
1
0.0003048
ft
5,280
3,281
1
Area
~\. To
From ^\^
mi2
km2
acre
ha
mi2
1
0.386
0.001563
0.003861
km2
2.590
1
0.004047
0.01
acre
640
247.1
1
2.471
ha
259.0
100
0.408
1
Volume
~\\ To
From^'\^
ft3
m3
gal
Mgal
ac-ft
ft3
1
35.315
0.134
133,681
43560
m3
0.02832
1
0.003785
3785
1,233
gal
7.481
264.2
1
1,000000
325,848
ac-ft
0.00002296
0.0008107
0.000003069
3.069
1
Flow Rate
"^\^^ To
From ^"~~~-\^
CFS
CMS
MGD
ac-ft/year
CFS
1
35.315
1.547
0.00138
CMS
0.02832
1
0.044
0.00003911
MGD
0.646
22.825
1
0.0008927
ac-ft/year
724.97
25,567
1,120
1
-------
Water Acquisition for Hydraulic Fracturing May 2015
Preface
The U.S. Environmental Protection Agency (EPA) is conducting a study of the potential impacts of
hydraulic fracturing for oil and gas on drinking water resources. This study was initiated in Fiscal Year 2010
when Congress urged the EPA to examine the relationship between hydraulic fracturing and drinking
water resources in the United States. In response, EPA developed a research plan (Plan to Study the
Potential Impacts of Hydraulic Fracturing on Drinking Water Resources) that was reviewed by the Agency's
Science Advisory Board (SAB) and issued in 2011. A progress report on the study (Study of the Potential
Impacts of Hydraulic Fracturing on Drinking Water Resources: Progress Report), detailing the EPA's
research approaches and next steps, was released in late 2012 and was followed by a consultation with
individual experts convened under the auspices of the SAB.
The EPA's study includes the development of several research projects, extensive review of the literature
and technical input from state, industry, and non-governmental organizations as well as the public and
other stakeholders. A series of technical roundtables and in-depth technical workshops were held to help
address specific research questions and to inform the work of the study. The study is designed to address
research questions posed for each stage of the hydraulic fracturing water cycle:
• Water Acquisition: What are the possible impacts of large volume water withdrawals from
ground and surface waters on drinking water resources?
• Chemical Mixing: What are the possible impacts of surface spills of hydraulic fracturing fluid on
or near well pads on drinking water resources?
• Well Injection: What are the possible impacts of the injection and fracturing process on
drinking water resources?
• Flowback and Produced Water: What are the possible impacts of surface spills of flowback and
produced water on or near well pads on drinking water resources?
• Wastewater Treatment and Waste Disposal: What are the possible impacts of inadequate
treatment of hydraulic fracturing wastewaters on drinking water resources?
This report, Case Study Analysis of the Impacts of Water Acquisition for Hydraulic Fracturing on Local
Water Availability, is the product of one of the research projects conducted as part of the EPA's study. It
has undergone independent, external peer review in accordance with Agency policy and all of the peer
review comments received were considered in the report's development.
The EPA's study will contribute to the understanding of the potential impacts of hydraulic fracturing
activities for oil and gas on drinking water resources and the factors that may influence those impacts.
The study will help facilitate and inform dialogue among interested stakeholders, including Congress,
other Federal agencies, states, tribal government, the international community, industry, non-
governmental organizations, academia, and the general public.
-------
Water Acquisition for Hydraulic Fracturing May 2015
Authors
Kate Sullivan, PhD1
Michael Cyterski, PhD1
Stephen R. Kraemer, PhD1
Chris Knightes, PhD1
Katie Price, PhD2
Keewook Kim, PhD2
Lourdes Prieto1
Mark Gabriel, PhD1
Roy C. Sidle, PhD1
1 U.S. Environmental Protection Agency, Office of Research and Development
2 Oak Ridge Institute for Science and Education
Acknowledgments
The EPA would like to acknowledge all of the organizations that provided data and information for this
report, including the Pennsylvania Department of Environmental Protection, the Pennsylvania
Department of Conservation and Natural Resources, the Susquehanna River Basin Commission, the
Colorado Department of Natural Resources, and the Colorado Oil and Gas Conservation Commission, as
well as U.S. EPA Regions 3 and 8 who assisted with project logistics.
This research was supported in part by two appointments to the Postdoctoral Research Program at the
Ecosystems Research Division administered by the Oak Ridge Institute for Science and Education through
Interagency Agreement Number DW8992298301 between the U.S. Department of Energy and the U.S.
Environmental Protection Agency.
-------
Water Acquisition for Hydraulic Fracturing
May 2015
1. EXECUTIVE SUMMARY
Hydraulic fracturing and its associated technologies have made vast reserves of oil and gas (O&G) economically
recoverable in the United States, which has resulted in a surge of natural gas and oil production. Approximately 0.8 to
5 million gallons of water are typically used to complete each O&G well on average. As the number of hydraulic
fractured wells increases across the United States, the need for water increases. Local water acquisition for hydraulic
fracturing use has the potential to impact the quantity and quality of drinking water resources. The local consequences
of water demand and acquisition on water quantity and quality are a function of water sources, users, permitting, and
hydraulic fracturing technology dictated by geologic characteristics.
At the O&G well site, water is used during
well construction and during the reservoir
stimulation process. The hydraulic fracturing
fluid is used to initiate and/or extend fractures
and carry proppant into the fractures to hold
them open. Water is the most commonly used
base fluid for hydraulic fracturing. Injection
volumes vary widely, with an average of 4
million gallons per well in the Marcellus Shale
and 2.3 to 2.9 million gallons in the Piceance
Basin, although how much of this is freshwater
versus reused hydraulic fracturing wastewater
depends on local conditions.
This study explored the impact of hydraulic
fracturing water demand in two U.S. river
basins where hydraulic fracturing operations
have been widely implemented: the
Susquehanna River Basin (SRB) in
Pennsylvania (in the humid east) and the
Upper Colorado River Basin of Colorado
(UCRB) (in the semi-arid west). These two
large river basins were chosen because they are
naturally, economically, and socially
important; and because they are focal points
for natural gas extraction using hydraulic
fracturing technology. The goal of this project
was to investigate the water needs and sources
to support hydraulic fracturing operations at
the river basin, county, and local spatial scales
and to place these demands in a watershed
context in terms of annual and daily water
availability. Through these analyses, the study
confirmed that, at larger drainage areas in these
two systems, hydraulic fracturing water
demand has minimal impact on water
availability and water quality. As spatial and
temporal scale decreased, the potential for
impact on water quality and quantity increased.
However, water management strategies at the
two study areas reduce the chance for
realization of impact.
• A combination of factors determine whether
hydraulic fracturing introduces imbalance in the
relationship between water supply and demand in a
region, including drinking water resources. These
factors include available water resources and their
capacity to yield water, industry needs influenced by
geologic characteristics of rocks in each play, other
user demands, and permitting or allocation controls.
• Minimal impacts to past or present drinking water
supplies or other water users resulting from hydraulic
fracturing water acquisition were found in either study
basin due to unique combinations of these factors in
each area.
• In the Susquehanna River Basin in Pennsylvania
(SRB), there is little use of public water supplies
(currently <8%) because water resources are well
distributed and available year round and hydraulic
fracturing operators have been able to develop
unallocated sources. In SRB, there are times or
locations when water sources can be stressed, but
water is managed to prevent overuse and minimize risk
at individual sources.
• Water in the Upper Colorado River Basin in Colorado
(UCRB) is strongly seasonal and over-allocated, but
unconventional gas production requires little
freshwater as the industry is able to reuse large
volumes of flowback and produced water instead. No
municipal drinking water supplies are used for
hydraulic fracturing in the areas studied within the
UCRB.
-------
Water Acquisition for Hydraulic Fracturing May 2015
Withdrawal impacts were quantified using a ratio between water use and available water volume: the water use
intensity index, ranging between 0 and 1. Higher values indicate higher vulnerability to water withdrawal. The water
use intensity indices were implemented for surface water (the SUI, or surface water use intensity index) and
groundwater (the GUI, or groundwater water use intensity index). Water use data were gathered from publicly
available databases from state, federal, and private sources. Surface water and groundwater volumes available for
consumptive use were calculated using observations at U.S. Geological Survey (USGS) gage stations where available,
or predicted from empirical areal-weighting techniques and Hydrologic Simulation Program Fortran Model (HSPF)
stream flow simulations. Groundwater availability was associated with pump ratings at well fields and geohydrologic
modeling. Analyses included use intensity calculations using actual observed hydraulic fracturing withdrawals, and
hydraulic fracturing withdrawals at current and projected drilling rates.
The Marcellus Shale covers 95,000 mi2 spanning Ohio, New York, West Virginia, Pennsylvania, and Maryland. It is
currently the largest producing shale gas play in the United States. Its most productive dry gas portion underlies the
Susquehanna River Basin (SRB), spanning Pennsylvania, New York and Maryland. As of 2014, water from rivers and
streams withdrawn from water sources by operators (self-supplied) is the dominant source of water used for hydraulic
fracturing activities, with public supplies currently providing approximately 8% of water. The Susquehanna River
Basin Commission (SRBC) regulates water acquisition and issues permits to operators for individual withdrawal sites.
Permits constrain the volume, rate, and timing of withdrawals. Permits assign daily withdrawal and pumping rate
limits, and set river passby flow thresholds that halt withdrawals during low flows.
The water use intensity quantification approach based on the SUI index demonstrated that the relative effects of
hydraulic fracturing withdrawals from streams were dependent on their size, defined by contributing basin area. Small
streams that supply water to hydraulic fracturing operators had the potential for SUI of 0.4 or higher (withdrawal of
40% or more of available water) during all or most of the year. Based on measured flow records throughout the SRB,
there was an increased probability of higher SUI at average daily withdrawal volumes in watersheds less than 10 mi2.
Watersheds up to 600 mi2 had some increased level of vulnerability at maximum withdrawal volumes during
infrequent droughts.
The water management system operated by SRBC is applied with the objective to provide water for all users while
protecting ecological values. The system is effective in maintaining low use intensity at virtually all sites across a
range of flow volumes. Water management in the SRB demonstrated that large withdrawals can be managed with
hydrological predictive measures and data made available by USGS. Hydraulic fracturing operations do not currently
provide a significant challenge to public water supplies at a regional, county or local scale in the SRB due in part to
use of different water sources and in part to oversight that controls the large-volume withdrawals and industry use
patterns that have distributed self-supplied water sources throughout a wide geographic area.
In the SRB, groundwater meets about 20% of the freshwater demand from the O&G industry, with a mixture of public
and private self-suppliers. Given the higher yields and higher permeability required, these groundwater wellfields are
located in the glacially deposited valley fill and alluvium of large rivers, like the Susquehanna River, and its smaller
tributaries. The potential for cumulative impact on regional aquifers was shown to be unlikely. However, we
examined the potential for local impact due to well drawdown and baseflow depletion at a representative public water
supply in Bradford County and a private wellfield in Wyoming County. We did not find any observed or reported
impact from hydraulic fracturing water acquisition on local domestic wells. Baseflow depletion was less than 10%
under average flow conditions. The SRBC manages drawdown and baseflow depletion through its permitting process.
Based on use intensity expressed as the proportion of water source volume withdrawn, a concentration magnification
approach was used to evaluate impacts on water quality in the SRB. This approach assumed that removing water for
hydraulic fracturing upstream of pollutant discharge locations would concentrate pollutants. Results showed that for
watersheds larger than 200 mi2, pollutant concentrations would increase 10% or less and usually 1% or less due to
reduced water volume. Water quality in watersheds smaller than 20 mi2 was more vulnerable to withdrawal, with
potential concentration magnification 2-10 times (although these smaller streams may be less likely to be permitted
for effluent discharges). The vast majority of observed SUI values in the SRB were less than 0.02 (2%), suggesting
that water withdrawal for hydraulic fracturing would have minimal impact on water quality.
The Piceance Basin (-7,250 mi2) in western Colorado has a large volume of recoverable natural gas in low-
permeability reservoirs within the Mesaverde formation. The Piceance contains one of the richest known oil shale
deposits (not to be confused with shale oil or shale gas) in the world and is the focus of most ongoing oil shale
-------
Water Acquisition for Hydraulic Fracturing May 2015
research and development extraction in the United States, with an estimated 64 trillion gallons of in-place oil shale
resources. According to the water use records maintained by the Colorado Department of Water Resources
(CODWR), freshwater obtained for hydraulic fracturing is serf-supplied with a mix of surface water and groundwater
resources. Irrigated agriculture is the largest user of water in the Upper Colorado River Basin (UCRB). Water
acquired for hydraulic fracturing in this region is highly concentrated within the tributary watershed of Parachute
Creek located in the vicinity of the Parachute gas field (198 mi2). Tributary streams and groundwater wells provide
about 50% of hydraulic fracturing water demand in the area.
CODWR manages water in Colorado using a system of water rights, where water is allocated based on prior
appropriations. The volume of water allocated exceeds available supplies in some locations at some times creating
localized shortages. In the 1970s, O&G companies collectively acquired significant water allocations in the Piceance
structural basin in anticipation of the need for very large volumes of water for extraction of hydrocarbons from oil
shale with current technologies. A very small portion of these allocations are currently used for hydraulic fracturing
for gas.
In the UCRB, hydraulic fracturing operators report a very high flowback volume of fluid returning to the surface
following the hydraulic fracturing process and during production. The Piceance tight sands have natural water content,
and produced water continues flowing from each well at a rate of approximately 140,000 gallons per well per year.
Hydraulic fracturing operators collect and treat the high volume of flowback and produced water and reuse it for
fracturing almost all new wells. Although the volume of fluid used per well is similar between the SRB and the
UCRB, the high water recycling rate observed in the tight sands of the Piceance results in use of much smaller
volumes of freshwater in this semi-arid region.
In Parachute Creek, a tributary to the Colorado River, hydraulic fracturing operators obtain water from small
reservoirs and a groundwater well field. The impact of cumulative daily withdrawals in Parachute Creek, primarily by
irrigators, generally had higher SUI than observed in the SRB. Many days (45%) had more than 10% of the available
water withdrawn, and 16% of the days had more than 40% removed. The percent of water removed daily from small
reservoirs within the watershed averaged 1 to 20%. The SUI values did not show a marked decrease with increasing
basin area as observed in the SRB. This is because withdrawal rights tend to allocate more water in the lower portions
of the watershed where irrigated agriculture occurs and water withdrawal volumes tend to increase faster moving
downstream than streamflow accumulates, resulting in higher SUI values in some larger basins.
Scenario analyses in the Parachute and Roan Creek watersheds were performed in a complementary manner to that
applied in the SRB, but with different hydraulic fracturing assumptions reflecting differences in freshwater use.
Without the influence of irrigation, there was a strong pattern of increasing potential impact of water removal (higher
SUI) with decreasing watershed size. The runoff rate in the semi-arid west is one-third that of the humid east, meaning
UCRB watersheds must be larger than in the SRB to meet the same SUI thresholds. SUI dropped below 0.4 under
current directional drilling rates when watersheds were larger than 30 mi2; with horizontal wells drilled at current
rates, the median 0.4 SUI threshold increased to a basin area of approximately 100 mi2.
Demand scenarios were simulated in Parachute Creek using precipitation and streamflow from a historically dry year,
2012, and a historically wet year, 2011. In the wet year, median historical demand including water used for hydraulic
fracturing, was generally met without issue. However, in the dry year, median historical demand throughout the basin
could not be satisfied, and large federal reservoirs were called onto augment supplies, as they were designed to do.
In the UCRB, groundwater meets a small percentage of the freshwater demand from the O&G industry. We
investigated one of the few private wellfields in Garfield County and explored the potential for local impact due to
drawdown and baseflow depletion at this site. There were no known domestic water wells within the maximum
predicted cone of depression of this wellfield. Baseflow depletion was less than 10% under average flow conditions.
Colorado state appropriation law manages alluvial wellfield acquisition as tributary flow.
In summary, the potential for higher intensity use locally due to water acquisition for hydraulic fracturing was related
to watershed size in both UCRB and SRB. Higher vulnerability (higher SUI values) was demonstrated in small
streams (<10 mi2) during most of the year and in larger watersheds (<600 mi2) during low flow periods. Higher use
intensity calculated from observed hydraulic fracturing withdrawals, however, occurred only at a few sites on smaller
streams in both basins, and localized high use intensity was found at only a few withdrawal locations. These results
were similar, not because the required water volumes for hydraulic fracturing were minimal, but rather because
3
-------
Water Acquisition for Hydraulic Fracturing May 2015
emergent local consumptive factors limited the need for freshwater while management and hydraulic fracturing
technology influenced withdrawal practices.
Water demands on limited water resources create potential for higher intensity use in the semi-arid Colorado River
basin. However, the geology of this play results in large return volumes of hydraulic fracturing fluids that can be
recycled and reused. Recycling rates of 100% are reported. A different story unfolds in the SRB, where surface water
and groundwater reserves have not approached depletion, with less demands among users and withdrawals managed
with daily and low flow restrictions.
A mosaic of factors that contributed to the outcome in each basin included water sources, users, permitting controls,
and hydraulic fracturing technology dictated by geologic characteristics. These factors should be thoroughly
understood to assess potential local impacts of hydraulic fracturing on water resources in other regions and for future
hydraulic fracturing development in these same areas—results from one area do not readily transfer to another.
However, the approach outlined in this report provides a methodology for assessing how water acquisition for
hydraulic fracturing might impact water quantity and quality in other regions. These findings provide information to
states, Tribes, communities, and industry to help understand the potential impacts of hydraulic fracturing on drinking
water supplies, and the protection of those resources for the future.
-------
Water Acquisition for Hydraulic Fracturing
May 2015
2. INTRODUCTION
Shale and tight sand sedimentary deposits contain significant quantities of natural gas and oil that until recently were
considered unrecoverable because of very low rock permeability, depth below surface, and thickness of geologic
formations (Aydemir 2012; Cueto-Felgueroso and Juanes 2013). Recent advances in technologies have allowed
economically viable gas and oil production from so-called "unconventional" deposits (Montgomery and Smith 2010;
Clark et al. 2012) (Fig.2-1). Horizontal and directional drilling techniques steer the drill head to the thin layers of
shale, coal bed, or tight sand deposits deep below surface, while hydraulic fracturing allows gas to flow by improving
rock permeability. Together, these technologies have enabled a tenfold increase in natural gas production from shale
and tight sand deposits in the United States since 2000 (U.S. EIA 2013a).
Conventional
non-associated
gas
Source US. Engigy Infcmttlon Mnirfetialion and U S Geological Surrey
Figure 2-1. The technologies of deep horizontal and directional drilling that target resource rich sedimentary strata
and hydraulic fracturing that increases permeability of the rock allow recovery of natural gas from unconventional
shale and tight sand formations. (Source: U.S. EIA 2011a.)
The Process of Hydraulic Fracturing. Wells are typically directionally bored, often thousands of feet, to tap the
targeted gas-bearing formation. Rather than strictly drilling in a vertical direction, directional drilling deviates the
wellbore along a planned path to a target located a given lateral distance and direction from vertical (Schlumberger
2014). Horizontal drilling is any wellbore that exceeds 80 degrees. During well construction, water is used in drilling
fluids to maintain downhole hydrostatic pressure, cool the drill head, and remove drill cuttings (Clark et al. 2013).
Once drilled, wells are cased with steel and cement and the producing intervals are then "stimulated" to release the gas
or oil tightly held within the fine-grained matrices of the rock. The major need for water comes during the typically
week-long stimulation phase, when large quantities of the base fluid (typically water) are mixed with a proppant and
various chemicals are injected under high pressure to induce and maintain fracture openings (Clark et al. 2012; Soeder
and Kappel 2009). The proppant is an inert material such as sand or ceramics that props the fissures and cracks open,
allowing the hydrocarbons to migrate into the wellbore. The chemical mixture introduces friction reducers, scale
inhibitors, and biocides into the well to maintain well functionality (Gregory et al. 2011). Some wells produce just gas
or liquids, but a significant portion produce a combination of gas and liquids (U.S. EIA2013a).
After the hydraulic fracturing procedure is completed and pressure is released, the direction of fluid flow reverses, and
a portion of the water and excess proppant flow back through the wellbore to the surface, referred to as "flowback."
Water may also continue to flow to the surface—along with the natural gas—for the life of the well. Some of this
"produced water" is returned fracturing fluid and some is natural formation water. The returned fluid is of poor quality
and is returned in volumes that can be substantial over time (Gregory et al. 2011; Maloney and Yoxtheimer 2012;
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Wilson and VanBriesen 2012). We refer to both flowback and produced water as "hydraulic fracturing wastewater" in
this report. How much hydraulic fracturing wastewater returns is highly variable among plays.
Well operations are centered at "well pads" where water is
stored, chemicals are mixed, and flowback and produced water is
collected and possibly treated for reuse (Fig. 2-2). Water is
delivered to each well pad where it is stored in temporary
reservoirs or tanks to be ready for hydraulic fracturing
stimulation. Flowback water is collected and trucked away or
reintroduced into other hydraulic fracturing wells. Operators can
typically drill numerous horizontal wells radiating outward from
one pad, reducing the construction footprint of pads and access
roads on the landscape.
A significant portion of the United States has potentially
recoverable dry natural gas or liquids within unconventional
shale and tight sand formations (Fig. 2-3). Natural gas and oil
production using hydraulic fracturing technologies is in the early
phases of a decades-long build-out of a network of wells that will
tap the 29 known commercially viable unconventional natural
gas and oil reserves. These reserves (also termed "plays")
underlie as many as 32 states in the United States (Entrekin et al.
2011; U.S. EIA 201 Ib) but do not follow state boundaries. The
annual pace of well drilling and production has increased
exponentially from 0.3 to 8.6 trillion cubic feet per year in the
past 10 years and is expected to continue to increase in the
coming decades (U.S. EIA 2013a).
Figure 2-2. Hydraulic fracturing well pad.
(Photo by SkyTruth: aerial overflight provided
by LightHawk.)
Muskwa-Otter Park,
Lowl, Evie-Klua
North American shale plays
(as of May 2011)
Niobrara* -
Cody
HHItert- Baxter-
Mancos-Nlobrara
^> Niobrara*
Montsrey-
TembJor
Monterey
Excello-
Mulky-—*
Woodford
eia
Current shale plays
Stacked plays
Shallowest/youngest
Intermediate depth / age
Deepest/oldest
1 Mixed shale & chalk play
"* Mixed shale & limestone play
'" Mixed shale & tight dolostone-
siltstone-sandstone play
Prospective shale plays
Basins
200 400 600
Source: U.S. Energy Infc
Updated: May 9. 2011
Figure 2-3. Location of unconventional shale gas plays in North America. There are also some gas fields in northern
Alaska not shown on the map. (Map obtained from U.S. EIA ZOllb.)
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Figure 2-4. Left is a photograph of mature wellfield in the Uinta Basin, Utah. Well pads are 2 acres in size with a
density (visible as small white squares) of 1 per 11 acres within this area. Large dark shapes are storage
reservoirs of hydraulic fracturing wastewater. Right is a close-up of image on the left. (Images from Google
Earth, Landsat.)
*
Parachute Creek Detail
HF Well Pads
HFWell Lines
0 0.5 1Mi
Gas production at individual wells declines with time as the area around the wellbore is slowly drained. An
individual well may yield gas for 20 or more years, though production lifetime of hydraulic fracturing wells and the
need for re-fracturing is not fully established (Clark et al. 2012; Cueto-Felgueroso and Juanes 2013). To develop a
play, O&G companies must drill the wells and link them to distant refining and distribution hubs through
transmission lines, a sort of natural gas highway typically buried below ground that maintains the flow of gas to
markets. To ensure a flow of gas from the play that meets customer demand, new wells must be drilled as older
wells decline. This means that an area will
experience episodic drilling and hydraulic fracturing
to keep gas flowing, often infilling between existing
wells. Oil and natural gas producers adjust their
formation targeting in response to changes in the
market value of natural gas and liquid hydrocarbons
in an attempt to focus on the more profitable
products in response to markets (U.S. EIA 2014a).
The final well density below ground is dependent on
the characteristics of the rock formation. Projected
below-ground well density in the Marcellus Shale in
Pennsylvania is one well per 132 acres (U.S. EIA
2012). The directionally drilled wellfield in the Uinta
Basin in Utah shown in Fig. 2-4 illustrates well pads
spaced approximately 700 feet apart, yielding a
surface and below ground density of one well per 11
acres, with apparently one pad supporting one well.
A dense network of wells far below ground and well
pads on the Earth's surface increasingly
characterizes the landscape in these regions as build-
out proceeds. Fig. 2-5 shows multiple directionally
drilled wellbores below ground within the formation
radiating outward from one well pad.
Figure 2-5. Example of multiple subsurface directionally
drilled wellbores radiating from each well pad. (Data
source: COGCC 2013.)
-------
Water Acquisition for Hydraulic Fracturing
May 2015
How Much Water Is Needed for Hydraulic Fracturing? The amount of water injected into wells for hydraulic
fracturing varies significantly between areas, depending on the permeability of the rock formation and well design
including length and depth of the wellbore (Gregory et al. 2011; GWPC 2009; Kargbo et al. 2010; Soeder and Kappel
2009). The amount of water consumed at individual wells also can vary widely between wells within each play.
Hydraulic fracturing injection volume in the Marcellus Shale has averaged 4 million gallons per well but varies from 2
to 13 million gallons (Clark et al. 2012; Gregory et al. 2011; Nicot et al. 2014; Vengosh et al. 2014). This average per
well volume is relatively similar among shale plays (Vengosh et al. 2014). It appears that relatively little additional
freshwater is required during the life of the producing well, although this is not well established.
Estimated Daily Water Use for HF in Shale Plays
(Million Gallons Per Day)
Bakken, 14.5
Others, 68.2
Eagle Ford,
47.6
Utica, 4.7
Niobrara,
12.9
Haynesville,
10.3
.Marcellus,
15.3
Figure 2-6. Estimated daily water used for hydraulic fracturing in
natural gas shale plays. Estimates are based on drilling rig counts.
(July 2014: U.S. EIA2014a.)
At the play level, water use can be roughly
estimated from drilling rig counts that drive the
pace of hydraulic fracturing along with
assumptions of drilling speed and injection
volume. Annual water use in seven major plays
and all others combined was computed
assuming that each hydraulic fracturing well
requires the average volume of water per well
determined from the FracFocus national well
registry (U.S. EPA 2015b), minus 8% reused
wastewater. Well numbers were estimated
from the distribution of drilling rigs in July
2014 (U.S. EIA 2014a) assuming that each rig
drills between 1 and 1.5 wells per month.
Water use estimates are very approximate, as
the time required to drill a well and the needed
water volume can vary significantly depending
on rock material and length of the wellbore.
Results expressed in million gallons per day
(MGD) are shown in Fig. 2-6.
According to these assumptions,
approximately 50 to 72 billion gallons of water
are needed each year nationally to supply about 27,000 wells developed with hydraulic fracturing technology, with
play-level demand ranging from 2 to 17 billion gallons. Nationwide, hydraulic fracturing uses 173 million gallons of
water per day. Daily use varies from 5 to 48 million gallons per day (MGD) among plays with the most water use in
the Eagle Ford play in Texas.
Water Sources. Hydraulic fracturing operators must obtain water from sources available in each area. Intrinsic water
availability is highly variable at national, regional, and local scales. The physiographic setting of landforms,
underlying geologic structure, climate, vegetation, and hydrography—coupled with societal investment in water
storage and delivery infrastructure—create a varied array of potential water sources (Padowski and Jawitz 2012).
These factors control the nature and extent of surface and groundwater resources available for use. Surface water
resources far exceed current allocations in the humid eastern states but constrained in the semi-arid/arid southwest,
which has less than a third of the annual runoff (Fig. 2-7 A) and lower groundwater recharge rates. Many of the drier
regions rely on large regional aquifers that have accumulated water over long time periods (Fig. 2-7B). Region-scale
aquifers, such as the High Plains aquifer that underlies 173,000 mi2 from Texas to South Dakota, are important water
sources for agricultural and drinking water supplies in these regions (Reilly et al. 2008). Overexploitation of surface
water and groundwater supplies is a concern for all users in water-scarce regions (GAO 2003), including areas with
hydraulic fracturing (Nicot and Scanlon 2012). If the pace of hydraulic fracturing development continues or increases,
there is potential for increased intensity of use of available water resources and competition for available supplies
among users. Just as intrinsic water availability is not uniform spatially, water imbalance issues are not generally
uniform within states or physiographic areas or over time (GAO 2003; Roy et al. 2005). Water imbalance is
particularly disruptive in dry regions where water supplies are insufficient to meet demand (Bureau of Reclamation
2005, 2012) and where populations are large or increasing (CWCB 2011; GAO 2003). Periodic droughts draw
attention to the limits of local and regional water supplies (Bureau of Reclamation 2005, 2012; GAO 2003, 2014;
Kenny et al. 2009).
-------
Water Acquisition for Hydraulic Fracturing
May 2015
A) Surface Water Runoff
2010
PR
cz?
^^ Explanation — Runoff in mm
0-49 50-99 100-199 200-299 300-399 500-749 750-999 >1000
B) Groundwater Aquifers
Principal Aquifers
Figure 2-7. Intrinsic variability of water resources within the United States. A) Estimated annual surface
water runoff for October 1, 2009, through September 30, 2010 computed for each of the eight-digit
hydrologic unit code cataloging units in the United States (Map source: USGS 2014c). B) Major
groundwater aquifers of the United States (Map source: USGS 2003). Annual runoff is three to four times
greater in the humid east relative to the semi-arid west. Regional aquifers are important water sources,
especially in more arid regions, and vary in location, extent, and water quality.
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Hydraulic fracturing operators are likely to
acquire water in the same places other users
do. Water use statistics for 15 states with
ongoing hydraulic fracturing development
activity are shown in Figs. 2-8 and 2-9, with
data taken from the U.S. Geological Survey
(USGS) 2010 water census (Maupin et al.
2014). This group of states includes most of
the hydraulic fracturing activity in shale gas
and tight sands plays and have varying
climate, water sources, and populations. Most
water in these states, as elsewhere throughout
the United States, is acquired for all uses from
surface water sources, although some states,
such as Texas and Arkansas, rely more heavily
on groundwater (Fig. 2-8).
Because water quality is a less significant
concern for hydraulic fracturing, lower
quality water than that used for human
consumption can be used (Vengosh et al.
2014). The need for freshwater can be
partially mitigated by use of non-potable,
brackish, or chemically contaminated
water. Hydraulic fracturing wastewater can
be used in other wells within the limits of
the hydraulic fracturing technology.
Hydraulic fracturing wastewater reuse is
increasing, but availability varies between
plays, reflecting the volume of flowback
and produced water that returns to the
surface, treatment and transportation costs,
and availability of less-expensive water
sources. This volume varies from as little as
5% of water injected into shales to as much
as 80% injected into tight sands (Clark et
al. 2012). Despite increased reuse,
freshwater remains the dominant source of
water for well stimulation in most plays
(Nicot et al. 2014; Vengosh et al. 2014).
Water Use
Every region has its own portfolio of water in
surface water and groundwater resources
shared among users that have evolved over
time, including drinking water supply,
agriculture, industries, and power generation
(Davis 2012; Kargbo et al. 2010; Kenny et al.
2009; Tidwell et al. 2012). Increasingly there
species and aquatic and terrestrial ecosystems
Total Water Use From Selected States
25,000
Q 20,000
13
15,000
10,000
5,000
Surface Water
I Groundwater
.
I II I
1
II
NM UT CO WY MT ND OK KS TX LA IN AR OH PA WV
Figure 2-8. Daily surface and groundwater usage by all user sectors
in 15 states with unconventional oil and gas reserves in the 2005
USGS water census. States were selected as examples of varying
climate, water sources, and population. (Data source: USGS data in
Maupin eta/. 2014).
25,000
20,000
'S
13
f 15,000
Daily Water Use-2005
ra 10,000
5,000
Hill..I
Thermoelectric
I Mining
I Industrial
Irrigation
I Public + Domestic
NM UT CO WY MT ND OK KS TX LA IN AR OH PA WV
Figure 2-9. Daily water use (surface and groundwater combined) by
major sectors in selected group of states with unconventional gas
reserves in 2005 water census. States are sorted from driest to
wettest from left to right. Hydraulic fracturing water use is included
in mining. (Data source: USGS water census data in Maupin et al.
2014.)
is demand to provide water to support the viability of endangered
(GAO 2003; Roy et al. 2005).
Fig. 2-9 shows daily water usage by major user sectors in the 15 exemplar states. The combined water use for public
and domestic water supplies is a relatively small portion of water consumption, but it is higher in states with larger
populations. More than 86% of the U.S. population relies on public water supplies for household use, with that
proportion increasing over time. As many as 43 million people, most living in rural areas, supply their own water from
groundwater wells (Maupin et al. 2014). The majority of water in these states and others is used to produce food and
10
-------
Water Acquisition for Hydraulic Fracturing
May 2015
electricity. Irrigation is the largest user in the drier western states, while thermoelectric power generation is the
dominant user in the more populous eastern states (Kenny et al. 2009). The USGS includes any hydraulic fracturing
water use in the mining user category. This category is such a small proportion of use in each state that it can barely
be seen in the bars in Fig. 2-9.
When comparing play-level hydraulic fracturing water demand (Fig. 2-6) and state-level water use (Figs. 2-8 or 2-9),
it is clear that hydraulic fracturing consumes a small fraction of water resources at this scale of assessment. Individual
state-level water consumption for mining in the 15 exemplar states ranges from 1,300 to 23,000 MOD, hundreds of
times greater than the hydraulic fracturing daily water use by play, which may span several states. Others have
independently compared hydraulic fracturing to total state water use and arrived at a similar conclusion: hydraulic
fracturing water use makes up less than 1% of state water use generally as documented in Texas (Nicot et al. 2014),
Oklahoma (Murray 2013), and Pennsylvania (Arthur et al. 2010). Even if hydraulic fracturing drilling rates continue
to increase, hydraulic fracturing water use is not likely to be a large component of any state's total water use.
Hydraulic Fracturing Water Acquisition Is Unique. While hydraulic fracturing water acquisition is not likely to
affect the balance between water supply and demand at the national, state, or large river basin scale, it can have local
impacts where the water is obtained (Nicot and Scanlon 2012; Nicot et al. 2014). Hydraulic fracturing has different
water use requirements than most other industries. Operators need large quantities of water episodically in many
different places spread over large areas. Water acquisition is spatially and temporally dynamic. The need for water is
short-lived at individual well pads and regionally dependent on drilling rig activity. Within a play, there are typically
multiple independent operators developing numerous wells at any time in widely dispersed areas with uncoordinated
demands (Entrekin et al. 2011; Nicot et al. 2014).
The hydraulic fracturing industry acquires water differently than most other users. Most water is trucked from source
to well, giving the industry high mobility and flexibility to acquire supplies close to where they will be used. Water is
tapped at locations that are accessible with proper permissions, and supplies need only be sufficient to complete as
few as one well at a time. The high cost of trucking water (as much as one-third to one-half of well drilling costs)
encourages the industry to obtain water as close to hydraulic fracturing wells as possible (Arthur et al. 2010; Kargbo
et al. 2010; Nicot et al. 2014).
Hydraulic fracturing operators are capable of withdrawing water from a variety of sources including rivers, streams,
farm ponds, lakes, municipal hydrants, groundwater wells, and wastewater treatment plants (Brantley et al. 2014;
Mitchell et al. 2013; Nicot et al. 2014). Stress on water supplies is more likely to arise locally than regionally because
of this flexibility (Davis 2012; Entrekin et al. 2011; Freyman 2014; Nicot et al. 2014).
Where and how water is acquired is
governed by state or regional authority
(GAO 2003; Richardson et al. 2013). The
doctrinary basis of many state laws (riparian
vs. prior appropriation) varies in approach,
regarding how water is allocated There are a
variety of water laws and regulations
overseeing water use, including allocation
rules, management and technology
requirements, reporting, and enforcement
(Entrekin et al. 2011; GWPC 2009; Murray
2013; Nicot and Scanlon 2012; Richardson
et al. 2013). This gives rise to variability in
how the states manage water (GAO 2003).
Figure 2-10. Most water is trucked from source to well pad in
Bradford County, Pennsylvania.
11
-------
Water Acquisition for Hydraulic Fracturing May 2015
Concerns About Increased Water Use to Supply Water for Hydraulic Fracturing. Water use imbalance arises
at the local level when the demand for water exceeds what is available in streams or groundwater aquifers within the
time span of natural replenishment. Water use imbalance is less likely when there is larger volume at the source, or
when the volume is withdrawn over a longer period. Water storage within the system alleviates short-term pressures
during high-use or lower-flow periods, and some areas of the county have water delivery systems in places for storage
and transfer of water to where it is needed. Use imbalance can be more consequential when water demands are higher,
as when irrigation is active and surface water flow is naturally lower.
Locally, hydraulic fracturing operators can rapidly withdraw the necessary volumes of water from available surface or
groundwater supplies. High use intensity resulting from these withdrawals is most likely in situations involving
extraction from smaller volume sources such as headwater streams, small ponds, or domestic water supplies (Clark et
al. 2012; Kargbo et al. 2010); cumulative withdrawals from proximal users (Dunlap 2011; Rahm and Riha 2012);
withdrawals during episodic shortages, such as low flows or droughts (Arthur et al. 2010; Brantley et al. 2014;
Mitchell et al. 2013); or withdrawal from sources that replenish slowly (GAO 2003).
Options for acquiring water depend upon volume and water quality requirements for each play, physical availability,
competing uses, and water management. Not all options are available to hydraulic fracturing operators in all situations
(API 2010), and there is likely to be considerable variability from state to state and play to play. Important factors that
influence the potential impacts of hydraulic fracturing on water resources are climate as it determines available water
sources, the portfolio of users, state and regional policies that oversee allocation and use, and the characteristics of the
geologic plays that determine the hydraulic fracturing engineering practices deployed by the industry. State spatial
scales do not provide sufficient granularity to detect effects given volume of use relative to state level use. County or
finer levels may be more informative (Freyman 2014). Ultimately, the most useful assessments will match the scale of
impact analyses to the water bodies where the water is taken.
Research Needs. While there is potential for local impacts with hydraulic fracturing water acquisition, there has
been little study to verify whether impacts occur (Brantley et al. 2014). Nicot and colleagues (Nicot and Scanlon
2012; Nicot et al. 2014) have reported on hydraulic fracturing water use in the Barnett, Haynesville and Eagle Ford
Shale plays in Texas, including water acquisition in the heavily populated Dallas-Fort Worth area. In Texas, water for
hydraulic fracturing has been obtained from the same surface and groundwater sources relied on by most other users,
including municipal supplies. The industry reduces some reliance on commonly shared freshwater sources by utilizing
some brackish groundwater and hydraulic fracturing wastewater. At the county scale, water supplies for a few rural
counties with small populations were considered most at risk, while depletion of important groundwater aquifers was
considered a more widespread potential problem (Nicot et al. 2014). Hydraulic fracturing operators draw water from
regional aquifers that are important sources of freshwater for municipal supplies and are considered depleted (Nicot et
al. 2014). Localized analysis is hindered by difficulties obtaining necessary data at fine spatial resolution or specific
to hydraulic fracturing activity relative to broader industrial use in many agency databases (Brantley et al. 2014;
Hansen et al. 2013; Nicot et al. 2014; Perrone et al. 2011).
12
-------
Water Acquisition for Hydraulic Fracturing May 2015
3. RESEARCH OBJECTIVES AND APPROACH
Goals and Objectives. The goal of this project was to investigate the water needs and sources to support hydraulic
fracturing operations at the river basin, county, and site spatial scales, and to place this demand in the context of
annual and daily water availability at water sources. Potential effects on water quality were estimated in the context of
changes to water volumes. Recognizing the unique geography and geology of unconventional oil and gas (O&G)
resources, the project adopted a case study approach within natural river basins.
Study Areas. The project was conducted in two study areas to explore and identify potential differences related to
water acquisition: the Susquehanna River Basin (SRB), located in the eastern United States (humid climate) and
overlying the Marcellus Shale gas reservoir, and the Upper Colorado River Basin (UCRB), located in the western
United States (semi-arid climate), overlying the Piceance structural basin and tight gas reservoir (Fig. 3-1). These two
river basins were among those studied for future climate change impacts on streamflow by the U.S. EPA (2013a).
The two study areas are similar in several ways. Both river basins are large (20,000+ mi2) and important to the natural,
economic, and social fabric of their regions. Each has considerable hydrocarbon reserves, an active oil and gas
industry using hydraulic fracturing technologies, and productive natural gas wells with long term development
prospects. Each has been targeted as a potential area of concern for hydraulic fracturing water acquisition in
nationwide analyses (e.g., Freyman 2014). There are also many differences in factors that could influence the extent of
hydraulic fracturing impact on water supplies. These include water management methods, existing water users,
inherent water availability, and hydraulic fracturing technology required by the geologic formations.
The objective of the project was to understand and quantify the effects of water acquisition, with emphasis on drinking
water supplies. Understanding drinking water supply impacts requires consideration of other water users as they can
have a strong influence on the availability of water when they compete. Ecological water use is both affected by
human water use and can limit supplies when ecological protection measures are invoked.
Overview of Approach. A basic premise of the project was that the water acquired for hydraulic fracturing would be
insignificant at either the state or large basin scale relative to other existing uses. The working hypothesis was that
impacts from hydraulic fracturing withdrawals are most likely to occur at the local sources where the water is
acquired. Considerable project effort was applied to quantify how the oil and gas industry obtains water in each study
area: How much water was withdrawn? Where and how often did this happen? How significant was the withdrawal
relative to the volume at the source?
Terminology. We use the term "water use" to refer to water that is withdrawn for a specific purpose, such as for
public supply, domestic use, irrigation, thermoelectric power cooling, mining, or industrial processing (Kenny et al.
2009). "Water withdrawal" or "acquisition" refers to water removed from the ground or diverted from a surface water
source for use. This is the total amount of water removed from the water source, regardless of how much of that total
is "consumptively" used, meaning not returned to the water resource system. For many uses, some fraction of the total
withdrawal will be returned to the same or a different water source after use or treatment, and is then available for
other withdrawals (Kenny et al. 2009). This project generally disregarded the consumptive use distinction by
assuming that all freshwater used for hydraulic fracturing was 100% consumed. We use the term "freshwater" to refer
to any water taken from water bodies, regardless of its water quality. Use of the term does not imply specific standards
or drinking water quality. Our primary distinction in water sources based on water quality is between freshwater and
hydraulic fracturing wastewater.
Data Acquisition. Two primary questions guided data collection:
• How much water was acquired from what sources for hydraulic fracturing?
• How much water was available at the sources where it was acquired?
The project gathered information on where and how much water was acquired in each study basin by querying
publicly available databases from state, regional government, and federal data sources, augmented by databases
maintained by private or nonprofit organizations. In characterizing water use, we emphasized direct monitoring of
usage and minimally relied on county-scale water census data provided by the U.S. Geological Survey (USGS) that is
13
-------
Water Acquisition for Hydraulic Fracturing
May 2015
not specific to hydraulic fracturing water use to understand other sector usage. In each area, we were able to develop
good-quality information on local sourcing of water by hydraulic fracturing operators from agency records that likely
included most if not all known withdrawal sites in the regulatory system.
Data acquired from state/regional/federal sources were used as obtained. We occasionally found what appeared to be
data errors. Obvious mistakes were repaired, or the agency was consulted to verify or correct. Members of the project
team visited each basin to familiarize researchers with the watersheds and to meet with representatives from
government and industry, who provided insight to how hydraulic fracturing water acquisition works at the field level
in each area. This included visits to acquisition sites, as well as some subsequent interaction to guide understanding of
the data systems.
SHALE AND TIGHT GAS BASINS
Major river basins
Study watersheds
Figure 3-1. Case study area location map. Major shale gas plays are shaded in blue. River basins include the
Susquehanna River Basin (SRB) in the Marcellus Shale region and the Upper Colorado River Basin in the
Piceance geologic basin in northwestern Colorado. Water acquisition in the SRB is studied in the Pennsylvania
portion of the basin. Subbasins shaded in yellow were used for local analyses within basins, as described in
each study area chapter. (Shale and tight gas basins data source: U.S. EIA 2014b.)
14
-------
Water Acquisition for Hydraulic Fracturing May 2015
Analytical Approach. The impacts of water withdrawals were characterized using the simple water balance approach
applied by Roy et al. (2005), Reeves 2010, Tidwell et al. (2012), and Freyman (2014). A ratio between water
withdrawn and the available volume of water quantifies the potential for water use imbalance, defined as:
Water Use
Water Use Intensity Index fWUIl = Equation 3-1. General water use intensity index
Available Volume
We adopt the terminology "water use intensity index" following Reeves 2010 and Weiskel et al. 2007. The water use
intensity index ranges between 0 and 1.0 and is dimensionless. The water use intensity index is simply the proportion
of water removed from the source. Water withdrawn is the volume removed from the source, and available volume is
the volume of water in the water body at the location of withdrawal in a relevant time-step. Daily was the minimum
practical time-step for calculating the water use intensity index. This analysis was performed at individual sites
withdrawing from surface water and groundwater sources. An index was computed for each:
Y. SW Water Withdrawn Volume or Flux
SUI = Equation 3-2. Surface water use intensity index
Available Surface Water Volume or Flux
Y. GW Water Withdrawn Volume or Flux
GUI = Equation 3-3. Groundwater use intensity index
Available Groundwater Volume or Flux
The water use intensity metric proved useful for characterizing local water acquisition. It allowed expression of the
relative proportion of water taken in short time intervals and in terms relevant to other water users, including
ecological resources. It could be readily quantified, consistently applied, and assessed on a daily basis and at every
level of flow. At the same time, it could be related to common flow frequency metrics. It has been used by researchers
understanding the nexus between energy and water (e.g., Tidwell et al. 2012) and others characterizing impacts of
water use (Reeves 2010, Freyman 2014). As will be shown, it provided a direct means to statistically quantify effects
on streamflow impact, and it could also be directly translated to water quality impacts.
The water use intensity metric provided a scalar of potential impact. Clearly, when consumption is high relative to
available volume (i.e., SUI or GUI approaches 1.0), there is greater risk of over-withdrawal, with potential impact on
other users. The index level where impact occurs may differ for each type of user; therefore we propose no particular
threshold value of concern. In comparing consumption to availability, some have referenced values as low as 0.4
(Freyman 2014; Hurd et al. 1999), 0.5 (Richter 2014), while others have referenced values as high as 0.7 (Tidwell et
al. 2012).
Water Availability. An estimate of the volume of water in a water body was needed to perform the water use
intensity calculation. USGS streamflow data were the primary source of information on water availability. We applied
various techniques to estimate daily streamflow volume at ungaged withdrawal locations based on gaged sites. These
included a combination of empirical and hydrological modeling that were used as needed given the availability of
weather and streamflow data in each area. The water use intensity index calculation did not require high precision in
flow estimates.
HSPF and SWAT are hydrologic models available for estimating daily streamflow. Each is a widely used
deterministic model that produces a watershed streamflow hydrograph on a daily or finer time-step while
accommodating some of the spatial and temporal variability in land use, vegetative cover, and soils that influence
streamflow response to precipitation. Calibration steps were important to ensure that a good match to known
streamflow records was achieved. We worked with both models, finding strengths in how each performed certain
tasks in the process of watershed hydrological modeling. Ultimately, we relied primarily on HSPF for streamflow
prediction, because it was computationally more conducive to the Monte Carlo approach we used for calibration and
uncertainty analysis. We were also interested in potential impacts of hydraulic fracturing withdrawals on groundwater
resources in the case study areas, so we used the groundwater model GFLOW™ to evaluate and visualize
groundwater volumes and flux. Streamflow at gaged locations was also directly extrapolated to ungaged sites using an
empirical approach based on area-weighting when appropriate. The empirical and modeled hydrologic data allowed
computation of SUI and GUI at individual withdrawal sites.
Water Quality. Water quality was not measured in this study. Possible impacts to water quality were assessed in the
context of the relationship between water volume and chemical concentration.
15
-------
Water Acquisition for Hydraulic Fracturing May 2015
Overview of Analyses. Each case study featured three types of analyses. First, the water use data were accessed,
compiled, and analyzed. From these data we developed facts, analyzed patterns, and summarized information at a
variety of scales from local to basin-wide. The facts of water acquisition were summarized and are presented as an
important product of this research. These data were then used to apply the water use intensity analytical approach to
systematically quantify the water balance effects of observed hydraulic fracturing withdrawals on water bodies.
Finally, scenario analyses were applied to reduce gaps related to water bodies, climate conditions, or levels of
hydraulic fracturing activity that were not well represented in observed hydraulic fracturing withdrawals in each study
area. The scenario analyses were typically focused in selected areas within the larger river basins, in a process that
involved applying hydraulic fracturing withdrawal assumptions to lengthy (26-yr) hydrological records. Hydraulic
fracturing activity has been ongoing in each study area sufficiently long to use observed usage patterns as a foundation
for future water acquisition scenarios. The scenario analyses were designed to improve the project's ability to provide
more generalized answers to its charge questions that could be applied beyond the study areas.
Deviations of Project from Outline in the EPA 2012 Progress Report. EPA's water acquisition project initially
took a top-down approach to assessing water acquisition impacts by emphasizing watershed modeling and scenario
analysis at the large basin scale, with some exploration of selected local areas within them as described in U.S. EPA
(2012a). In 2013, EPA received feedback on the approach from the Science Advisory Board peer review panel and
through technical workshops that encouraged greater focus on local effects. As a result, the project increased emphasis
and data gathering on local hydraulic fracturing water acquisition, including water management practices that guide
water allocation and industry practices that influence how much water is used and how it is sourced. This report
incorporates a bottom-up approach stimulated by public input while applying the overall strategy outlined in U.S.
EPA (2012a). The project has followed EPA's hydraulic fracturing Quality Management Plan (U.S. EPA 2012b,
updated in 2014) and the quality assurance project plan for Project 5b (U.S. EPA 2013b).
A main goal of the project was to generate information from the case study areas that would have general relevance
beyond the borders of the study areas and to properly account for climate fluctuations in representing effects on water
availability. Specific hydraulic fracturing consumption scenario analyses were proposed to ensure that a range of
potential hydraulic fracturing water demand with feasible drilling rates and potential reduction in demand through
reuse were considered. We conducted our analysis in the context of climate variability and examined a much wider
range of stream sizes and water body types than anticipated in U. S. EPA (2012a). With the greater emphasis on local
information gathering, we adapted the scenarios to the local use patterns and projections in each study area. We also
evaluated the potential for localized groundwater impact.
About the Project Report. Findings and analysis in each case study area are provided in chapters dedicated to each
basin. Chapter 4 reports analysis of the Marcellus/Susquehanna River and Chapter 5 reports analysis of the
Piceance/Upper Colorado River. These chapters describe data sources, analysis, empirical findings, and scenario
analysis results. Chapter 6 summarizes and synthesizes finding from the two areas, emphasizing general similarities
and differences and addressing the project charge questions. Data sources and quantitative methods used in each case
study area are described relatively briefly in each chapter. Additional detailed information is provided in four
appendices: a complete list of data sources and source information (Appendix A); a complete description of surface
water hydrology methods, including hydrological modeling calibration and uncertainty analysis (Appendix B); a
detailed description of groundwater hydrology methods, including GFLOW model calibration, uncertainty analysis,
and findings (Appendix C); and additional detailed data and assumptions on water use applied in scenario analyses
(Appendix D).
This report uses English units of measurement: water volumes are reported as gallons (gal), or millions of gallons
(MG) for large numbers. We found that the case study areas were oriented to different units of measure, so we also
provide volumes in acre-feet (ac-ft) in parentheses when discussing large volumes. Flow rate is expressed as cubic
feet per second (cfs) as reported by USGS gage records, often translated to million gallons per day (MOD) to
accommodate comparison with withdrawal volumes. (One cfs equals 0.646 MOD.) Area is expressed in square miles
(mi2). Large-volume water use data are rounded to three significant figures. All values are rounded independently, so
the sums of individual rounded numbers may not equal the totals. Percentage changes discussed in the text are
calculated from the unrounded data and are expressed as integers. All population data are rounded to two significant
figures. All statistics and graphing were performed with R Statistical Software, version 3.0.1 (R Core Team 2013) or
Microsoft Excel (2007).
16
-------
Water Acquisition for Hydraulic Fracturing
May 2015
4. MARCELLUS SHALE/SUSQUEHANNA RIVER BASIN
Marcellus Shale Geologic Setting
The Marcellus Shale underlies nearly 95,000 mi2 of the states of Ohio, New York, West Virginia, Pennsylvania, and
Maryland (Fig. 4-1). The potential gas production from the Marcellus Shale makes it an important play for the
nation's energy future, as it may contain as much as 363 trillion ft3 of recoverable gas—enough to supply the needs of
the entire nation for 15 years at the current rate of consumption (Soeder and Kappel 2009). The Marcellus is currently
the largest producing shale gas basin in the United States, accounting for almost 40% of U.S. shale gas production
after seven years of development (Fig. 4-2; U.S. EIA 2014c).
The Marcellus Shale is composed of fine-grained
sediments deposited as an organic-rich mud in a shallow
inland sea that covered most of the North American
continent over 350 million years ago during the
Devonian period (Soeder and Kappel 2009). Sediments
were transported westward from the Acadian orogeny,
located approximately where the Appalachian
Mountains now stand (Dott and Batten 1981). The basin
deposits are shown schematically in Fig. 4-3. The
Marcellus Shale was deposited across the Appalachian
Basin before burial by an influx of younger
continentally derived sediments (Fig. 4-3 A). The basin
floor subsided under the weight of sediment, resulting
in a wedge-shaped deposit where the organic layer thins
from east to west (Fig. 4-3B). Hydraulic fracturing gas
extraction targets the basal organic rich layer that ranges
in thickness from 150 to 300 feet and lies 4,000 to 7,000
feet below the earth's surface across the play.
Figure 4-1. Location of the Marcellus Shale and the
Devonian Shale. (Map modified after Milici and Swezey
2006).
Marcellus Region
Natural gas production
billion cubic feet per day
16
14
Marcellus Region
New-well gas production per rig
million cubic feet per day
2007
rig count
rigs
160
2009
2011
2013
Figure 4-2. Natural gas production in the Marcellus Shale. Natural gas production
has increased rapidly since 2009. (Source: U.S. EIA 2014c.)
17
-------
Water Acquisition for Hydraulic Fracturing
May 2015
A)
A -
IWestl
- Approximately 240 miles
i
Ohio ! Pennsylvania
TOP OF DEVONIAN
• A'
lEastl
7,000
B)
REGIONAL ORGANIC-THICKNESS
MAP OF THE MARCELLUS SHALE
with additional organic-rich shale beds in the Hamilton Group
included for \ew York, Pennsylvania, and West Virginia
Figure 4-3. Characteristics and extent of the Marcellus Shale, A) Thickness of the
vertical depth to 7,000 feet. (Source: Soeder and Kappel 2009.) B) Surface area and
thickness of organic-rich deposits targeted for hydraulic fracturing drilling. (Map
source: Erenpreisseto/. 2011.)
18
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Susquehanna River
Susquehanna River Basin Background
The most easterly and productive dry gas portion
of the Marcellus Shale formation underlies the
Susquehanna River Basin (SRB) that drains
much of central Pennsylvania and portions of
New York and Maryland into the Chesapeake
Bay (Fig. 4-4). About 85% of the 27,510 mi2
watershed is underlain by natural gas shale
(Arthur et al. 2010), and this area has been a
focal point for hydraulic fracturing in the
Marcellus Shale (Vengosh et al. 2014).
Hydraulic Fracturing and Drilling Activity.
Within the SRB, dry gas drilling activity has
been centered in north central Pennsylvania. The
earliest exploratory wells were drilled in the SRB
in 2005 and production began in 2008 (Fig. 4-5).
Drilling peaked in 2012 at 836 wells.
Approximately 3,000 wells have been completed
in the SRB since 2008 according to records
provided by the Susquehanna River Basin
Commission (SRBC) who tracks drilling activity
and manages water use in the SRB. Most wells
are located within a 17 county area that lies
wholly or partially in the SRB. The Pennsylvania
Department of Environmental Protection
(PADEP) manages water use elsewhere in the
state (2013b), as well as oil and gas extraction.
Annual drilling rates depend on economics. The
current annual drilling rates are low relative to
projections of future activity as 43,000 wells are
expected over the next two decades (Johnson
2010), although realized drilling activity will
depend on economics. Well density could reach 1 per 132 acres (U.S. EIA 2012b), and the annual drilling rate could
reach 2,800 wells if this projection is realized (Beauduy 2009).
Figure 4-4. Map of the Susquehanna River Basin, Marcellus
Shale (blue shading), and location of drilled wells. (Data
source: PADEP 2013a.)
3,000
2,500
tn
| 2,000
"a.
§ 1,500
U
IS
Susquehanna River Basin
••Annual
—(^Cumulative
Figure 4-5. Annual and cumulative well
completions within the Susquehanna River
Basin. (Data source: SRBC 2013a.)
2008 2009 2010 2011 2012 2013
19
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Table 4-1. Statistics on injected and returned fluid volumes per well in the Susquehanna River Basin from 2008 to
2011 (n = 748). (Data source: SRBC 2013a.)
Hydraulic Fracturing Fluid Characteristics
Total hydraulic fracturing water use (million gallons)
Total freshwater volume (million gallons)
% acquired water volume in injection fluid
% hydraulic fracturing wastewater in injection fluid
% flowback water returned
Mean
4.25
3.85
89.1%
13%
7.3%
Median
4.31
3.88
91.9%
8%
4.7%
SRBC tracked the volume of freshwater and hydraulic fracturing wastewater consumed at 748 individual hydraulic
fracturing wells from 2008 to 2011. Mean and median volumes are reported in Table 4-1. Average hydraulic
fracturing fluid volume injected into each well is 4.3 million gallons (MG), with about 13% composed of reused
hydraulic fracturing wastewater. Available hydraulic fracturing wastewater is largely reused in the Marcellus Shale
(Maloney and Yoxtheimer 2012), and return flow is 8% to 12% of what is injected. The portion of hydraulic
fracturing wastewater reuse in the SRB is similar to other shale gas formations as reported by Vengosh et al. (2014).
Geology, Hydrology and Climate. The SRB is a major tributary to the Chesapeake Bay, contributing 25,000 million
gallons per day (MOD) of freshwater to the bay on average. The climate in the SRB is classified as "cold humid" in
the revised Koppen-Geiger classification system (Kottek et al. 2006). Long-term average precipitation ranges from 37
to 43 inches per year (McGonigal 2005). Rainfall is evenly distributed throughout the year, with the lowest average
monthly rainfall in February (~2 inches) and the highest in July (~4 inches) (Arguez et al. 2012).
Surficial geology is made up of unconsolidated deposits of glacial and post-glacial origin and the nearly flat-lying
sediment bedrock of the Appalachian Basin. The glacial and post-glacial deposits consist of till, stratified drift, and
river alluvium. The bedrock consists primarily of interbedded shale, siltstone, and sandstone of Devonian to
Pennsylvanian age. To date, almost all hydraulic fracturing drilling activity has occurred in the west and middle
branches of the Susquehanna River and the Lower Chemung River subbasins (Fig. 4-7), which are geographically
coincident with the Appalachian Plateau physiographic province whose landform is illustrated by the photograph
shown Fig. 4-6. The uplands are broad and valleys narrow (Hunt 1974). Elevation ranges from 1,000 to 3,000 feet.
The main branches of the
Susquehanna River flow to the
south while the smaller tributaries
are constrained by the northeast-
southwest-trending orientation of
the Appalachian Mountains (Fig.
4-7).
Stratified drift aquifers and the
Loch Haven and Catskill bedrock
formations serve as primary
groundwater drinking sources.
Glacial till is also tapped as a
drinking water source in some
locations (Williams et al. 1998).
Figure 4-6. West branch of the Susquehanna River near Renovo, Pennsylvania
within the Appalachian Plateau physiographic region (Photo byJ. Marasco,
Source: Picasa. http://creativecommons.Org/licenses/by-nc-nd/3.0/)
20
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Susquehanna River Basin
SUBBASINS
I
squehanna
„ _^^ . :iawan
/ Wyoming
Sullivan /•
X v v^-.~z~
carneronV^tt^Renanna
\ , ^NCASTER
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Figure 4-8. Aerial view of terrain, land use, and well pads in Bradford County. Distance between the two well pads
is about 3500 feet. Multiple hydraulic fracturing wells can be drilled from one pad. The well pad in the upper left
corner has a fracturing-operator-constructed freshwater storage pond. (Image Google Earth, PADCNR-
PAMAP/USGS.)
Hydraulic fracturing activity in the SRB has been concentrated in rural areas, away from more populous urban and
industrial centers to the north in New York and south near Harrisburg. Land use is mixed agriculture and forests with
small to moderately populated rural communities. Hydraulic fracturing activity occurs in a 17-county area but is
centered in 14 within Pennsylvania including Bradford, Susquehanna, Tioga, Lycoming, Wyoming, Potter, Sullivan,
Lackawanna, Clinton, Centre, Clearfield, Cameron, Blair, and Columbia.
Both surface and groundwater sources are
generally widely available in the SRB
(Arthur et al. 2010; SRBC 2012). Mean
daily discharge of the Susquehanna River
at Wilkes-Barre, Pennsylvania, based on
100 years of record is 9,000 MOD (basin
area approximately 10,000 mi2). Data on
daily water use from the USGS 2005
water census is shown by user sector
within the 14-county area in Fig. 4-9.
Thermoelectric power generation and
public water supplies are the largest water
users. There is minor use by irrigation and
industrial activity in this area. Water
acquisition by the oil and gas (O&G)
industry is quantified in the following
section of this report based on data
acquired from agency sources listed in
Table 4-2. Aquaculture water was not
included. At the large river basin scale, O&G
water use is small relative to other sectors.
Daily Water Use in 14-County Area within SRB Including
Unconventional Gas Development (MGD)
PWS,68.0
I
Thermoelectric!
Cooling, 345.8
O&G PWS, 1.8
Domestic, 16.0
Industrial, 19.6
Mining, 9.2
Livestock, 10.4
Irrigation, 2.9
O&G Self, 7.3
Figure 4-9. Susquehanna River Basin daily water use by sector (in
million gallons per day (MGD). (Data source: USGS 2014a.)
22
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Water Allocation and Management. All aspects of hydraulic fracturing in the SRB are regulated by the
Pennsylvania Department of Environmental Protection (PADEP 2013 a), except water acquisition, which is managed
by the Susquehanna River Basin Commission (SPJ3C 2012). SPJ3C is a tri-state commission with broad authority to
control water allocations and withdrawals based on the principles of riparian rights, which entitles landholders and
others to whom acces has been granted to draw on the natural stream flow in reasonable amounts (SRBC 2012).
SRBC requires permits for any person or business seeking to withdraw water exceeding 100,000 gallons per day
(GPD) from surface or groundwater sources, although hydraulic fracturing operators must obtain a permit for any
water withdrawal. SRBC permits control daily withdrawals following policies to protect aquatic life. Daily pumping
rate restrictions and passby flow thresholds shut down withdrawals during low flows for non-municipal users,
depending on requested withdrawal amounts. O&G permittees must accurately monitor withdrawals and report daily
usage. Permits are reviewed every four years and permit conditions are modified to reflect evolving SRBC policies.
SRBC policies are available at SRBC (2013b). Elsewhere in Pennsylvania, PADEP requires a water management plan
that has some common elements to SRBC permits.
Water Use Data Sources. Water withdrawal data were obtained from sources listed in Table 4-2. All public water
suppliers must provide an annual report of water volumes delivered to customers to the Pennsylvania Department of
Environmental Protection (PADEP). Since 2010, each facility must itemize the volumes sold to O&G customers.
These data were accessed on PADEP's State Water Plan interactive website (PADEP 2014a). PWS records were
downloaded by county and searched for facilities reporting O&G sales.
SRBC maintains records of daily water withdrawn from self-supplied permitted sites monitored by O&G permittees.
SRBC (2013a) provided the daily data for 2009 to 2013 by bulk download in response to a data request. SRBC also
provided well consumption information on request (SRBC 2013d). SRBC policies, technical documents and maps are
readily accessed on the SRBC website portal (SRBC 2013c).
The volume of water allocated by agencies is a key limitation on potential water use. Site data and appropriation limits
were obtained for SRBC permits from the SRBC website portal (SRBC 2013c). Some PWS facility information was
available on the PADEP State Plan website accessed through the Chapter 110 query (PADEP 2014b). The data
sources used throughout the report are also compiled in Appendix A.
Table 4-2. Water use data sources and reports for the Susquehanna River Basin.
Source
PADEP
2014
PADEP
2013a
SRBC
2013
Data Type
a. Annual public water
system use report
b. PWS facility information
Well drilling reports
a. Water source acquisition
volume
b. Miscellaneous reports,
policies, maps
c. Docketed permits
d. Gas well volume
Location
http://www.pawaterplan.dep.state.pa.us/
StateWaterPlan/WaterDataExportTool/W
aterExportTool.aspx
http://www.portal.state.pa.us/portal/serv
er.pt/commun ity/oil_and_gas_reports/20
297
http://www.srbc.net/pubinfo/index.htm
http://www.srbc.net/publicinfo/index.ht
m
http://www.srbc.net/wrp/
http://www.srbc.net/pubinfo/index.htm
Query
Primary facility report, by year and
county
Chapter 110 (Act 220) registration
Wells drilled by county
Provided by the Susquehanna River
Basin Commission (SRBC) by written
request
Website search
Water Resource Portal/search for
projects
Provided by SRBC by written request
Sources of Freshwater for Hydraulic Fracturing
Water is supplied to the O&G industry for hydraulic fracturing from a combination of surface water and groundwater
sources at public and serf-supplied withdrawal sites. Some public water suppliers sell water to the O&G industry,
while the majority of water is self-supplied from SRBC permitted sites. Water use data were available for 2010 to
23
-------
Water Acquisition for Hydraulic Fracturing
May 2015
2013 for the public water supplies and from 2009 to 2013 for the self-supplied withdrawal locations. During this time,
2,800 wells were drilled and hydraulically fractured in the SRB.
Public Water Suppliers and Use Statistics
Public water suppliers include municipal suppliers that typically provide water to a variety of customers including
domestic, commercial, industrial, and institutional, and for municipal functions such as watering parks or public pools.
Public water suppliers also include many private entities with a collective water supply such as mobile home parks
and homeowner associations. Many large institutions such as universities or prisons are also classified as public water
suppliers. In all of Pennsylvania, there are about 24,000 public water suppliers in the PADEP database. Statewide, 105
of them sold 2.8 MOD of water to hydraulic fracturing operators in 2012—an average of 27,000 GPD per
participating facility.
PADEP records for all PWSs within the SRBC and surrounding counties were searched for records of sales to O&G in
each year since 2010, when separate reporting for O&G began. Of hundreds of PWS facilities within the counties
where hydraulic fracturing drilling has been active (Fig. 4-10), 25 municipal suppliers distributed within 12 counties
in the SRB have provided water to hydraulic fracturing operators at some time between 2010 and 2013. Four
additional facilities are registered for O&G sales with SRBC but have not provided water to date. It appears that no
type of PWS other than municipals have sold water to O&G and no PWS outside of Pennsylvania sold water for use in
in the SRBC.
Legend
• Public Water Systems
• PWS that Supply Water to O&G
Counties
Susquehanna River Basin
New York
ot v. • • t ?. -.•••
•,«:. •••>••.•.-'„ Ivv;":"v•/• - ***''
ix J'x2^^^'w5a
^•C^>a^>rS^
.- tj r:^?.v?^v..rj
Pennsylvania
Figure 4-10. General location of public water suppliers in the Susquehanna River Basin. Small circles
shown all public water systems; large circles are public water systems that have sold water to oil and
gas. (Data source: U.S. EPA 2012c for PWS location, PADEP 2014a for PWS that sold water to O&G.)
24
-------
Water Acquisition for Hydraulic Fracturing
May 2015
PWS Water Sold to O&G
|D e
i 4
i 24-1
AllSusquehanna River Basin
•12010 a 2011 a 2012 2013
JIL
Figure 4-11. Daily sales by user sector summed for the 25 public water systems that supply water to oil
and gas in theSusquehanna River Basin. On average PWS provide about 5% of their water to the oil and
gas industry for hydraulic fracturing. The category "Unaccounted for" includes leakage and other
undocumented losses within the infrastructure. (Data source: PADEP 2014a.)
Each facility annually reports the total number of connections and average daily water use by sector. User sectors
include domestic, commercial, industrial, institutional, other PWSs, miscellaneous uses, and unaccounted for volumes
that represent undocumented losses within the infrastructure. Daily sales summed for the 25 public systems that have
sold water to O&G in the SRB are shown in Fig. 4-11. Collectively within this group, domestic households used 10
MOD, or about 30% to 35% of the facility sales. Commercial and industrial usage was 16% of daily sales. In 2010,
the O&G portion of water sales by PWS was 7% of water sold (1.8 MOD). By 2013, water demand for hydraulic
fracturing increased but the portion taken from PWS declined to 1%, indicating the industry was acquiring water from
other sources. Because most PWSs acquire all or some of their water from groundwater sources, about 85% of the
O&G water obtained from public suppliers came from groundwater.
The relative distribution of the sales by sector evident in Fig. 4-11 is consistent among most municipal PWSs, except
for the O&G component. Many PWS facilities report that 30% or more of their daily production is routinely
unaccounted for due to leakage and other undocumented losses within the infrastructure, about as much as supplied to
households. O&G acquisition of water from public suppliers has declined by 65% since the 2011 peak. During this
time, hydraulic fracturing operators have increased their ability to self-supply by obtaining permits, as described in the
next section.
Each PWS facility has a daily operating capacity defined by permit or by pumping capacity. Sufficient facility
operating information was obtained from the PADEP Chapter 110 query (PADEP 2014b) to establish an estimate of
facility capacity, although our estimate may underestimate potential supply, especially for several of the facilities with
multiple supply sources. The daily volume delivered as a portion of facility capacity is shown for the 25 PWSs in the
2011 peak use year in Fig. 4-12. Note that many of these facilities had no sales to O&G in this year, as was the case in
all years. The participating PWS facilities routinely delivered about 30% to 70% of their operating capacity (average
49%) to all of their customers, although some appeared to operate near capacity. Some of these may have additional
water sources that are not accounted for in our total. Public suppliers generally did not use much of their facility
capacity to supply water to the O&G industry—only 4.6% on average, with a median less than 0.1% (Fig. 4-12).
However, there was a wide range among individual PWS facilities. Two provided as much as 40% of their total
production capacity for O&G use in 2011 (2080003 and 4410175), but neither operated near maximum capacity.
25
-------
Water Acquisition for Hydraulic Fracturing
May 2015
PWS Daily Water Distribution Relative to Capacity-2011
100%
U
ro
o.
TO
U
U
TO
o
o
'•E
o
Q.
i_
O
o.
90% -
80% -
70% -
60% -
50% -
40% -
30% -
20% -
10% -
0%
Other Sectors Portion of Capacity
O&G Portion of Capacity
oo m
CM o
o o
o o
00 00
01 O
(N (N
o o
o o
00 00
o o
(N (N
i/i m i/i m
CM m m i/i
o o
00 (N
o
-------
Water Acquisition for Hydraulic Fracturing
May 2015
A)
180 i
160 -
140
120 -
100
80 -
60
40 -
20 -
0
-Number Permitted
Number Active
B) Active Self-Supplied Sites by Source Type
6.
4. 4
River or Stream
iGroundwater
Lake/Pond
I Mine Impaired
Treated
2009 2010 2011 2012 2013
Figure 4-13. Characterization of oil and gas self-supplied water sources in the Susquehanna River Basin. A)
Number of sources by year. B) Number by water body type. (Data source: SRBC2013a.) Most permitted sites
withdraw water from rivers and streams.
Permitted withdrawals are now widely distributed within
,s^ 17 counties within the SRB and concentrated within five
7JjyfY-^"v''~~ "~^\ (Fig. 4-14). Thirty-six withdrawal sites are located on the
mainstems of the west and middle branches of the
Susquehanna River (watershed area > 1,000 mi2). Note
that there are a few SRBC withdrawal sites outside the
basin boundary in adjacent counties, indicating that there
may be some inter-basin transfer of water. A check of
their records showed that a very minor amount of water
was brought into the SRB. Any water transported out of
the basin is included in the analyzed records.
The 31 O&G companies operating in the SRB register
public and self-supply sites where they may acquire water
with SRBC (2014). Just one small operator obtains all
water from public suppliers. Although most variously list
the PWSs shown in Fig. 4-12 as potential sources, most
are known to self-supply from permitted sites that they or
water service providers operate. It appears that companies
share use of some sites.
Table 4-3. New permits for surface water withdrawal
for the oil and gas industry. (Location data source: SRBC
2013a. Basin areas were determined in this project.)
Figure 4-14. Distribution of oil and gas self-supplied
water sources for hydraulic fracturing in the
Susquehanna River Basin. Sites are now, or have been,
permitted by the Susquehanna River Basin Commission
from 2008 to 2013. (Data source: SRBC 2013a.)
Basin Area (mi2)
Year
2008
2009
2010
2011
2012
2013
Number
25
23
28
35
25
10
Median
300
150
75
109
150
75
Range
8.9-8,884
5.2-8,261
1.7-8,489
3.4-10,527
13.6-8,248
1.5-8,720
27
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Middle
Branch
Susquehanna
River
withdrawal
site near
Towanda, PA
(6,500 mi2)
Middle Susquehanna River water acquisition depot
where water is first pumped into holding tanks
Mine-drainage impaired stream—Fall Brook (7 mi2)
Tunkhannock Creek (380 mi2)
Figure 4-15. Examples of self-supplied withdrawal locations permitted by the Susquehanna River Basin
Commission. (Approximate basin watershed area in parenthesis.)
Nearly all water is trucked to its point of use, although some operators are building water delivery infrastructure,
including pumping facilities and pipelines, to increase efficiency and to reduce truck traffic and costs. The widely
distributed network of sites (Fig. 4-14) provides operators the flexibility to withdraw from sources nearer to gas wells
and avoid concentrating withdrawals in any one location. Fig. 4-15 shows photographs of several active self-supplied
withdrawal sites on a range of river sizes within the SRB.
SRBC withdrawal permits constrain the volume and timing of withdrawals with specifications developed for each site.
Streamflow and groundwater are protected by limits on daily withdrawal volume and pumping rate. We considered
the daily withdrawal limit as the "capacity" of the permit site to supply water to the O&G industry. Maximum daily
permit limits are shown in Fig. 4-16. Permitted daily volume ranges between 0.04 and 3.0 MOD, shown categorically
by withdrawal volume in Fig. 4-16A. Withdrawal volume varies between sites reflecting, but not solely determined
by, river size (Fig. 4-16B). Withdrawals up to 3.0 MOD are allowed on the Susquehanna River. Some permittees have
requested relatively high daily withdrawal limits (1 MOD) on very small streams, with the expectation of using them
only during higher flows to fill storage reservoirs. Current permits would collectively allow 140 MOD to be
withdrawn within the SRB to support hydraulic fracturing activity (Fig. 4-16C).
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Fig. 4-16C organizes the information in 4-16A and B to illustrate the withdrawal capacity in relation to river size. The
daily maximum permit capacities are cumulatively summed, ordering sites from largest to smallest basin area. The
large river sites (>1,000 mi2) make up a substantial portion of O&G permitted water withdrawal capacity (60 MOD).
Sites with watershed area between 400 and 1,000 mi2 add about 40 MOD of capacity; small streams with watershed
area less than 40 mi2 add another 20 MOD. There are 12 permitted sites on small streams (<10 mi2) that can provide a
small amount of water. In any year, all of these sites could be tapped for some portion of the annual water supply.
Permit capacity is already far greater than currently or likely needed in the future. Freshwater needs for hydraulic
fracturing can be roughly approximated by well drilling rates. At the 2012 rate of 836 drilled wells (Fig. 4-5), each
requiring 3.85 MG of freshwater for hydraulic fracturing (Table 4-1), about 9 MOD would be consumed (Fig. 4-16D).
Drilling activity is projected to increase and could reach as high as 2,800 wells per year (Johnson 2010), which would
require 30 MOD of freshwater (Fig. 4-16D) (Beauduy 2009; SRBC 2013e). The increased demand for freshwater
under these projections could be readily accommodated at existing permit sites (compare Fig. 4-16D to 4-16C).
Nevertheless, new sites are added each year to the portfolio of permitted sites which has had the effect of minimizing
travel distance. In researching transportation characteristics of the hydraulic fracturing industry in the SRB, distance
Gilmore et al. (2014) found that most truck trips to haul water were less than 10 miles.
A)
45 -|
40 \
35 \
30 J
25
20-
15-
' ' '
ID O ID C
rsi ID r- C
D L/I o L/I o
D rN ID P- O
.. 1
LT) O LT) O
rsi ID r- O
B)
C)
E
D
E
Maximum Daily Permit Limit (MGD)
Cumulative Permit Capacity
10.00 g
1.00
0.10 =
0.01
D)
Daily Permit Volume Limits
O
13
Daily Maximum Withdrawa
40 -
20 •
n H
• •-•"• '*
.«.«»«•
(,t«^
• *
.-«•-
•••-«"
• •-»
.#*
i«*
.'v
10,000 1,000 100 10
Basin AreaAbove Permitted Withdrawal Site (mi2]
10
100 1,000 10,000 100,000
Watershed Area (mi2)
HF Water Use
516°
S- 140
I 120
Q_
3 100 -
o
u 80
"oj
5 60
cu
"I 40
S 20
LL.
0
Figure 4-16. Maximum daily withdrawal limits for oil and gas self-supplied water withdrawal sites. A) Count of permits
by daily limit category. B) Daily permit limit in relation to contributing watershed area. C) Cumulative sum of site permit
capacity, with sites ordered from largest to smallest basin area; for example, 60 million gallons per day can be acquired
from sites with watershed areas greater than 1,000 mi2. (Data source: SRBC2013c.) D) Estimated hydraulic fracturing
well freshwater consumption based on current annual drilling rate (840 wells) and future projected maximum rate of
2,800 (Beauduy 2009; Johnson 2010). O&G wells use an average of 3.85 million gallons per well of freshwater for
hydraulic fracturing (Table 4-1). (Data source: SRBC 2013a.)
29
-------
Water Acquisition for Hydraulic Fracturing
May 2015
SRBC has a low flow policy to protect aquatic biota and municipal supplies, requiring shutoff of hydraulic fracturing
withdrawals when streamflow drops below a threshold, termed the "passby" flow. A passby is assigned only if the
requested daily withdrawal volume exceeds a percentage of the lowest flows likely to be observed at a site (SRBC
2012). Low flow is defined by the Qv.io flow statistic, a common benchmark of extreme minimum flows at a stream
gaging site. The Qv.io is the average minimum streamflow observed over seven consecutive days once every 10 years
(USGS 1982). Note that the passby threshold is generally set at 20% to 25% of the mean annual flow if a low flow
threshold is required and is likely to trigger in most years. The statistic is computed from measured flow records at
USGS gages and applied to ungaged sites based on a basin area weighting methodology. Permit sites are referenced to
real-time USGS gages, so that operators can be readily informed via internet when the passby is invoked. From 2009
to 2011, SRBC applied a single annual low flow value. Since 2012, SRBC has assigned biennial and then monthly
passby thresholds. The passby flows can halt withdrawals at all sites except eight on Susquehanna River segments.
These sites have no passby restriction given small maximum daily withdrawals relative to large river streamflow.
The annual volume of water that is serf-supplied by the O&G industry was determined by summing the daily
withdrawal volumes obtained from SRBC sites reported by operators (2013a). Results summed annually by source
type are shown in Fig. 4-17 and daily in Table 4-4. In recent years, the O&G industry has serf-supplied 2 to 3 billion
gallons of water each year from 77 sites found mostly on rivers and streams distributed widely in the SRB and ranging
in size from very small to very large. Less than 5% of the water is acquired from waters impaired by mine drainage.
Virtually none comes from lakes, ponds, or impoundments other than those constructed by the O&G industry for
temporary storage after initial withdrawal from permitted sites; sources of this type have only been used
intermittently. Several commercial (but not public) groundwater wells supply water to O&G. Serf-supplied daily
withdrawals range from 0 to 3 MOD at the site level and sum to about 7 MOD for the SRB as a whole.
Self-Supplied Sources of Waterto O&G
3,500
(J 3,000
:E
tu
E 2,500
| 2,000
-a
« 1,500
1 1,000 -
c
500 -
Impaired 6)
Impoundments (4)
IGroundwater (5)
Streams/Rivers (97)
10,000
Figure 4-17. Volume of water acquired from
self-supplied sources annually by water source
type. Number in legend refers to total number
of active permitted sites in each category
between 2009 and 2013. (Data source: SRBC
2013a.)
2009
2010
2011
2012
2013
Table 4-4. Daily self-supplied water acquired for hydraulic fracturing in the Susquehanna River Basin (in million
gallons per day). (Data source: SRBC 2013a.)
Water Body Type 2009 2010 2011 2012 2013
Quality surface water
Groundwater
Impaired water
Total
Streams and rivers
Lakes/Ponds
Groundwater
Impaired
0.77
0.00
0.00
0.01
0.78
5.07
0.00
0.20
0.43
5.54
6.50
0.00
1.49
0.50
7.35
6.00
0.00
1.62
0.38
6.90
6.90
0.00
0.57
0.14
7.39
30
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Summary of Water Acquisition Volumes at the Susquehanna River Basin and County Scale
The combined public and self-supplied water volume supplied to the O&G industry for drilling and development of
hydraulically fracturing O&G wells in the SRB is shown in Fig. 4-18. Annually, the O&G industry and service
providers acquired about 3 billion gallons of freshwater for hydraulic fracturing, 82% of which was self-supplied (Fig.
4-18A) and most of which came from surface water (Fig. 4-18B). Almost 18% was acquired from PWSs in the peak
year of 2011 (Fig. 4-18A), but this source has declined in significance. The cumulative total daily use at self-supplied
sites has ranged widely during this period, peaking at 14 MOD in winter 2012 (Fig. 4-19). The daily summation
reveals that water use occurs every day of the year.
The water volumes accounted for at sources matched reasonably well with water consumed at well pads
independently tracked by SRBC and PADEP (Fig. 4-18A). The difference between the two is hydraulic fracturing
wastewater reuse (about 12% on average), consistent with the individual well water consumption statistics at
hydraulic fracturing sites (Table 4-1).
Self-supplied
I Public
B)
10 -i
Q
(3
HF Water Sources--SRB--2011
Self-supply
• Public supply
72%
2009 2010 2011 2012 2013
9.5%
5.5% 12.4%
Surface Water Groundwater
Figure 4-18. Total volume of water acquired by the oil and gas industry in the Susquehanna River Basin
from public water systems (data source: PADEP 2014) and self-supplied from permitted sites. A) Annual
acquired water volume and volume consumed at well pads tracked independently by the Susquehanna
River Basin Commission. B) Daily source of water. (Data sources: PADEP 2014a; SRBC 2013a.)
Daily Self-Supplied HF Withdrawal-SRB
Figure 4-19. Daily self-supplied water withdrawal for hydraulic fracturing in the Susquehanna River Basin
summed for all active sites. (Data source: SRBC 2013a.) Withdrawals occur every day. Withdrawals peaked in
2012 at 14 MGD and have dipped with slowed drilling reflecting lower gas prices in 2013.
31
-------
Water Acquisition for Hydraulic Fracturing
May 2015
A)
B)
i
Water Volume Supplied by PWS to O&G summed by County-from
participating PWS Facilities-2011
Self-Supplied Withdrawals by County
1,000
600 -
200 -
n -
• 2010 2011 2012 2013
•
I Ib L _ 1
.
| |
k
1
1
ll 1
III
3,500
3,000
2 500
OJ
2,000 £
i
0)
1,500 £
1,000
- 500
- n
<»*
<*"
Figure 4-20. Acquisition of water for hydraulic fracturing by the oil and gas industry acquired from public water
systems and self-supplied, summed by county. A) Daily sales to oil and gas from 25 public water systems in 2011,
summed by county. (Data source: PADEP 2014a). B) Annual volume of freshwater self-supplied by the oil and gas
industry, summed by county. (Data source: SRBC 2013a.)
The volume of freshwater supplied by PWSs and self-supplied sources is summed annually by county in Fig. 4-20.
Purchase of water from public water systems by the oil and gas industry was distributed among counties (Fig. 4-20 A).
Most of the PWS water was acquired in Bradford and Lycoming Counties, but the O&G proportion of sales was not
large relative to other users in any county. Annual self-supplied volume for hydraulic fracturing summed by county is
shown for individual years in Fig. 4-20B. Early in the history of hydraulic fracturing development in 2010, much of
the freshwater was acquired in Bradford County. As the hydraulic fracturing activity center expanded, water
acquisition increased in Lycoming, Susquehanna, and Wyoming Counties and decreased in Bradford County. Most
water was acquired in these four counties, where drilling is most active, and very little has been acquired elsewhere to
date.
32
-------
Water Acquisition for Hydraulic Fracturing May 2015
Water Use Intensity Analysis at Self-supplied Sites
Most water used for hydraulic fracturing is self-supplied from rivers and streams (72%), so water balance analysis
focuses on this group. Oil and gas operators must monitor all water withdrawals at each site every day with a good-
quality meter, and submit records to SRBC quarterly. Withdrawals are recorded in GPD. These data were used as
provided from SRBC (2013a). There were 87,581 daily observations in the combined site data record for surface
water sites from 2009 to 2013, including days with no withdrawal. Withdrawals occurred on 29,907 days (termed
"site-days"). The impact of withdrawals on site water balance is evaluated with the surface water use intensity index
(SUI, eq. 3-2) on observed hydraulic fracturing withdrawals at permitted sites. Groundwater withdrawals were
assessed with the groundwater use intensity index (GUI, eq. 3-3) at the end of this chapter.
Daily streamflow volume was required to calculate SUI for each location. However, nearly all of the withdrawal sites
were ungaged. Therefore, daily streamflow was estimated for each withdrawal site from a nearby USGS gaged site
using an area-weighting factor. (SRBC uses a similar method for estimating streamflow characteristics for permit
applications.) There were 56 USGS gages ranging in watershed area from 5.2 to 11,220 mi2 available to estimate
ungaged sites, matching the range of watershed area in the populations of withdrawal sites. Withdrawal sites were
paired with USGS gages based on nearby location and contributing watershed size.
Some error was introduced in SUI calculations using Tab|e 4.5 Cross.va|idation of area-weighting method.
estimated streamflow. The area-weighting method was Each r nts two d watersheds. Flow for
cross-validated using five pairs of USGS gages. In each ,. . . * . • ,.-,..,,.
. ,, . . . r. i j i j the second member of each pair was estimated by
pair, flow was estimated for one gage based on observed . . . , . ,. . , . .
flow at its paired gage, replicating the procedure used to extrapolating values from the first member of the pair,
estimate flow at ungaged sites. A Nash-Sutcliffe (NS) with the Nash-Sutcliffe (NS) and In-transformed NS,n
model efficiency coefficient that compares daily estimated scores indicating the fit between this estimation and
and observed flow as a ratio (Nash and Sutcliffe 1970) was the actual observed flows. "Distance" is the straight-
calculated from daily comparisons of estimated and line distance between the two gaging stations.
measured flow. The closer the model efficiency is to 1, the
more accurately the estimated data represents the observed — ^^ Area (mi2) Distance (mi) NS Nsln
data. Comparisons are provided in Table 4-5. USGS 01541500 371
1 3 0.83 0.91
USGS 01541303 474
The In-transformed NSt score exceeded 0.90 for four of the USGS 01532000 215
five pairs, indicating an excellent match between the paired 2 uscsoissigos 112 4 °'96 °'95
sites despite large differences in watershed area between the uses 01541000 315
pairs in some cases. The NS in score was 0.70 for the smallest 3 uses 01541200 367 9 °'76 °'97
sites. Although this was still a quite satisfactory score, 4 uses 01553700 si 22 o 76 o 7
results suggest that extrapolated streamflow records for uses 01552500 24
smaller streams were likely to have more error. See 5 USGS oisosooo 2232 3g Q gg Q gg
Appendix B for more discussion of hydrologic streamflow USGS oisisooo 4773
methods and USGS data used for SUI analysis.
Observed Hydraulic Fracturing Withdrawals. Examples of the daily withdrawal records at six sites illustrate
some general patterns of hydraulic fracturing operations (Fig. 4-21). The sites range in contributing watershed area
from 7.4 mi2 (Fall Brook) to approximately 10,000 mi2 (Susquehanna River). Streamflow is shown as a daily flow
volume expressed in MOD to facilitate calculation of SUI. One MOD equals a daily instantaneous flow rate of 1.55
cfs as typically reported at USGS stream gages. Note that the vertical axis for daily streamflow and withdrawal
volume is log-scaled.
Use varied by site, but sites on larger rivers tended to be used more often (~40%-65% of the days), while those on
smaller streams tended to be used more episodically. Smaller sites were collectively used about 20% of days. Whether
maximum daily limits were withdrawn varied considerably between sites and between days at each site. Operators
withdrew the maximum allowed at times, but often withdrew less. Most sites were used at about 10% to 12% of their
total capacity combining days of use and permit allowance to estimate total volume available. Hydraulic fracturing
operators withdrew year-round except when passby flows were invoked and adhered to their permit restrictions
regarding maximum daily withdrawals and passby periods, as evident in Fig. 4-21.
33
-------
Water Acquisition for Hydraulic Fracturing
May 2015
1,000,000.000
100,000.000
10,000.000
— 1,000.000
5. 100.000
2 10.000
o
J5 1.000
ra
0.100
0.010 -
0.001
SusquehannaRiver8,748 mi2 SID 3043
Streamflow
+ Withdrawal
Permit Capacity
1,000,000.000
100,000.000
10,000.000
„ 1,000.000
Q
~ 100.000
E
.| 10.000
j-j 1.000
0.100
0.010
0.001
WestSusquehanna RiverS,229 mi2 SID 3311
Streamflow
-I- Withdrawal
Passby Threshold
Permit Capacity
\\\\\\\\\\\\
1,000.000
E. Branch WyalusingCreek 17.3 mi2 SID 3595
Streamflow
+ Withdrawal
Passby Threshold
Permit Capacity
Hr \ \ \
v •<><><>
Fall Brook7.4 miz SID3538
Streamflow
+ Withdrawal
passby Threshold
Permit Capacity
Figure 4-21. Self-supplied permitted sites have a daily withdrawal limit and may have a minimum maintenance
flow (passby). Permittees must accurately monitor and report daily withdrawals. Daily Streamflow, withdrawals,
permit limits, and passby flows (if required) are shown for six oil and gas withdrawal sites representing a range of
contributing watershed sizes as examples. Note that the Y-axis is log-scaled. Fall Brook (7.4 mi2) is the smallest
stream shown and is on a mine-drainage impaired stream. Three sites do not have passby thresholds. (Data
source:SRBC2013a.)
34
-------
Water Acquisition for Hydraulic Fracturing
May 2015
A count of daily withdrawal volume for all sites and
all days is shown in Fig. 4-22. Median daily
withdrawal of all observations was 0.2 MOD.
Streamflow ranged over six orders of magnitude from
the largest to smallest streams and generally increased
exponentially as a function of basin area. Flow in the
Susquehanna River fluctuated around 10,000 MOD,
while flow in Fall Brook fluctuated around 10 MOD.
Streamflow ranged at least two orders of magnitude
between storms and intervening dry periods over the
course of the year at each site. Daily withdrawals
could only vary from 0 to 3 MOD reflecting permit
limits. Withdrawals were nearly four orders of
magnitude lower than flow at the large river sites but
just one order of magnitude lower than flow in the
smallest streams (Fig. 4-21). Thus, higher SUI was
more likely in smaller streams.
100,000
Observed Withdrawal Volume-SRB
, 10,000
ro
T3
i
.td
LO
1,000
o
u
100
10
median = 0.2 MGD
n=29,907
O
O
Daily Withdrawal Volume (MGD)
Figure 4-22. Count of daily volume withdrawals, all sites
combined. The majority of daily water withdrawals were
less than 0.5 MGD. (Data source: SRBC 2013a.)
Surface Water SUI Analysis. The 97 surface water
sites had a wide variety of Streamflow and daily
withdrawals as illustrated in Fig. 4-21. Water balance
represented by SUI divides withdrawal volume by Streamflow and ranges from 0 to 1.0, with 1 meaning all water was
taken. SUI should be strongly dependent on Streamflow, which in turn strongly depends on watershed area. Daily SUI
computed for all combined site-days from 2009 to 2013 is shown in relation to Streamflow in Fig. 4-23 A. The highest
6% of observations are shown in relation to basin area in Fig. 4-23B. Count by SUI category is provided in Table 4-6.
The categorized SUI distribution is shown for individual sites in Fig.4-24.
Overall, SUI due to hydraulic fracturing withdrawals was very low; 98% of the SUI values were less than 0.1 and
94% were less than 0.02. SUI greater than 0.1 occurred only when Streamflow was less than 20 MGD (30 cfs) (Fig. 4-
23 A). Flow this low is rare in larger rivers (>1,000 mi2), but occurs with increasing probability in smaller rivers and
streams. SUI did not exceed 0.01 when flow exceeded 200 MGD (309 cfs). Higher values of SUI had a strong
association with watershed area (Fig. 4-23B) and did occasionally occur in the smallest streams. SUI greater than 0.4
and 0.7 was observed 83 and 14 times, respectively, during the five-year period from 2009 to 2013 (Table 4-6).
A)
SRB-AII Sites Combined
1.000
0.100
0.010
0.001
B)
10 100 1,000
Streamflow (MGD)
10,000
1.0
13
00
IT 0.8
0.02
o
0 0
° o n = 1,774
0 o 5. 5% of observations
o o
o
o
0
0
8
1
L
0
IQ
I
10 100 1,000
Basin Area (mi2)
Figure 4-23. Daily surface water use intensity index (SUI) for all withdrawal sites in the Susquehanna River Basin
used to source water for hydraulic fracturing from 2009 to 2013. Only values for days when water was withdrawn
are included. A) SUI in relation to daily Streamflow volume. Note that the axes are logio scaled. B) SUI values greater
than 0.02 plotted by watershed area. (Data source: SRBC 2013a.)
35
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Table 4-6. Count of daily SUI for all permit sites on days when water was withdrawn from 2009 to 2013 (n =
29,907). (Data source: SRBC 2013a.) The no passby scenario has more observations since there are no restrictions
on withdrawals.
SUI Category
0-0-0.10
0.11-0.20
0.21-0.30
0.31-0.40
0.41-0.50
0.51-0.60
0.61-0.70
0.71-0.80
0.81-0.90
0.91-1.0
Total
Actual Observed
Number of
Observations
29,313
344
107
60
34
25
10
3
6
5
29,907
%of
Observations
98.0%
1.15%
0.36%
0.20%
0.11%
0.08%
0.03%
0.01%
0.02%
0.02%
Scenario Assuming
No Passby Shutdown
Number of
Observations
32,585
1,587
610
329
247
154
103
104
78
1,589
37,386
Fig. 4-24 A uses boxplots to show five years of SUI observations at individual sites, identified by their basin area.
Higher SUI was rare at river and stream locations under standard conditions, e.g. when permitting requirements for
daily withdrawal limits and passby flows were in place. SUI never exceeded 0.1 when watershed area was greater than
17 to 27 mi2, regardless of daily permit limits. SUI values greater than 0.2 occurred only in the smallest streams with
watershed areas less than 7.8 mi2, although we note that the bulk of SUI observations were less than 0.2 at all sites.
Two of the very small streams were mine-drainage-impaired and had special permitting considerations allowing
larger-volume withdrawals (1 MOD). Two were required to draw from small offstream ponding structures that may
prevent impacts implied by the high SUI but maybe not realized due to the onsite water storage. Nevertheless, they
also illustrate that small streams defined by contributing basin area less than 10 mi2 are generally vulnerable to high
water use intensity from hydraulic fracturing withdrawals. All SUI values exceeding 0.4 occurred at the three sites
with basin area less than 7 mi2 (Fig. 4-24A).
The general patterns of SUI in relation to observed rate of withdrawals and streamflow indexed by basin area shown
in Fig. 4-24A are influenced by SRBC's low flow policies. The passby flow threshold designated in permits is
designed to reduce impacts of water withdrawals on the daily water balance by ceasing withdrawals during lower
flows. We illustrate the influence of the passby restriction by recalculating SUI on days when a passby shutdown
would have been in effect, assuming the permit limit was withdrawn when normally no water would be pumped. This
assumption will overestimate withdrawal effects, because hydraulic fracturing operators do not always withdraw their
full quota, nor would they necessarily have used the sites on these days as evident in Fig. 4-21. The distribution of
SUI assuming no passby flows were in place is shown by site in Fig. 4-24B and SUI count is included in Table 4-6.
The scenario analysis suggests that the frequency of higher SUI would increase significantly and could involve sites
with basin area up to about 700 mi2 without the passby shutdown.
The passby policy protects aquatic life and supplies used by other users including PWS, as public supplies are not
subject to low flow restrictions. There are strategies that allow withdrawals during passby shutdown periods.
Operators can divert to public suppliers. Increased O&G use of PWSs during shutdown periods could not be explored,
as PWS use is reported as an annually averaged daily value (PADEP 2014a).
O&G companies have constructed numerous small impoundments distributed throughout the SRB that store
freshwater. One can be seen in the aerial photograph in Fig. 4-8, and one is shown at ground level in Fig. 4-25.
Slonecker et al. (2012) counted 560 larger and 121 smaller temporary impoundments in Bradford County alone while
performing a landscape analysis of hydraulic fracturing in the region. Some—like the one pictured in Fig. 4-25—are
supplied by pipeline rather than trucks.
36
-------
Water Acquisition for Hydraulic Fracturing
May 2015
=)
co
1.0 -
0.8 -
0.6 -
0.4 -
0.2 -
o.o -
0 8 °
8S
8 ° 0 A) Observed
O ^
0 0
o
O T °
o
I I I I I I I I I I I I I I I I I I I
r--^rr~incncM^TCMcnocncocoT-inooco^r^;
I
en
oo
1.0
0.8
0.6
0.4
0.2
0.0
CO XT HJ
t- CM CO
in
en
CD
CM
(£>
CM
CO
CN
in
in
CM
CO
Basin Area (mi )
B) Assumed no passby
r- 10
CM CO
Basin Area ( mi;}
Figure 4-24. Observed surface water stress index at permit sites on all days when water was withdrawn for
hydraulic fracturing from 2009 to 2013 (n = 29,907). Sites are identified and displayed in order of their contributing
watershed area. A) Observed record. B) Simulated effect of passby by withdrawing each site's maximum daily limit
on days when passby would have been in effect. Upper and lower lines show the 5th and 95th percentiles; circles
show the remainder of the population between the 95th and 100th percentiles. (Data source: SRBC 2013a.)
Using Google Earth measuring tools to assess
dimensions, we estimated that most of these
impoundments have a 2 to 5 acre storage area
and can hold 5 to 10 MG of water, or enough to
fracture 2.5 wells. This total was confirmed in
interviews with Southwestern Energy
(November 14, 2014). In Bradford County,
O&G impoundments collectively can store
more than 5,500 MG of water to augment
supplies during periods of passby shutdowns,
enough to hydraulically fracture almost 1,400
wells. The increasing amount of privately
owned storage capacity could have contributed
to the decreasing use of municipal water
supplies evident in Fig. 4-18A.
The observed withdrawals in SRB from 2009
to 2013 may not represent the full potential
impact of hydraulic fracturing withdrawals. The
time frame of available data was short and
Figure 4-25. Photograph of O&G freshwater impoundment to
service hydraulic fracturing water needs.
37
-------
Water Acquisition for Hydraulic Fracturing May 2015
extreme low flows may not be represented. Very small streams can provide water for hydraulic fracturing but are not
well represented in the permitted sites at present—their use could be more prevalent in the future here or in other
areas. The growing number and increased density of permitted sites could have a "cumulative" effect not addressed in
site-by-site analyses. Hydraulic fracturing drilling rates have been relatively static for three of the five years of
available data and are projected to be higher in the future. Groundwater wells provide water to hydraulic fracturing,
but availability of water is difficult to quantify. We addressed these potential effects of these factors on SUI and GUI
by extending what we learned from observed hydraulic fracturing water use patterns with scenario analysis and
modeling.
Scenario Analysis of Hydraulic Fracturing Withdrawal on Water Use Intensity
Climate, watershed size, cumulative withdrawals, increased water demand, and groundwater impacts were further
explored by hindcasting SUI for hydraulic fracturing withdrawal scenarios at local stream sites using historical
streamflow records. Analyses were conducted in selected subbasins within the SRB focusing on smaller basin areas
that appear most likely to experience high use intensity from withdrawals (less than 500 mi2). Small towns and more
than 60 of the permitted withdrawal sites are located in 26 named subbasins of smaller rivers and creeks within this
basin size (HUC 8-12). We present four scenario analyses related to water balance associated with hydraulic
fracturing water acquisition designed to:
• Improve understanding of small stream, climate, and potential hydraulic fracturing water demands on
local water use intensity by dividing a watershed into first- to fifth-order streams and applying
withdrawal rates to mimic current and projected maximum drilling rates over 26 years;
• Demonstrate passby effects on SUI at a USGS gaged site over a lengthy flow record;
• Consider cumulative permit effects based on observed withdrawals in a watershed with multiple active
hydraulic fracturing withdrawal permits;
• Explore groundwater volumes and pumping impacts at several groundwater wells currently providing
water for hydraulic fracturing.
Surface water availability was determined from long-term USGS gages extrapolated to subwatersheds using various
methods. Observed records were directly used for some analysis or extrapolated to small stream sites using the
hydrologic simulation model HSPF. Groundwater modeling techniques were used for aquifer analysis.
We focused our scenario analysis on Towanda Creek, a small river in Bradford County with watershed area of 215
mi2 (Fig. 4-26). The watershed is representative of hydraulic fracturing activity as well as natural physiography and
land use in the area.
We chose this watershed, because it has two USGS streamflow gaging sites and two nearby rain gages with a lengthy
record to allow model parameterization and calibration. There has been hydraulic fracturing well drilling in the basin,
and there are two permitted self-supplied withdrawal sites. Towanda Creek joins the middle branch of the
Susquehanna River at the town of Towanda, Pennsylvania. There are two municipal water suppliers in the watershed
that have provided water for hydraulic fracturing from groundwater wells. See Figs. 4-4 and 4-26 A for location of the
Towanda Creek watershed within the SRB, Fig. 4-26B for hydrologic use and measurement landmarks within the
watershed, and Fig. 4-26C for location of existing hydraulically fractured wells.
Bradford County and the Towanda Creek watershed are located in the Appalachian Plateau physiographic province
within the Pleistocene glacial margins. Surficial geology is composed of horizontal inter-bedded sandstones,
siltstones, and claystones and unconsolidated glacial and alluvial deposits in the valleys. Towanda Creek consists of
two main branches (HUC-10) that join above the gage at Monroeton. Two sedimentary formations characterize the
two branches, with the Chemung formation in the northern branch and the Lock Haven formation in the southern. This
gives rise to a dichotomy in topography and land use that occurs more generally throughout much of the Appalachian
Plateau region in the SRB (Fig. 4-27A).
Land use in Towanda Creek in the northern branch is mixed agriculture with interspersed small towns (Fig. 4-27A).
The southern branch, with greater relief, is predominantly forested and is protected as a state game reserve. There is
moderate topographic relief in the Towanda Creek with elevations ranging from 770 feet at the outlet to 2436 feet
above sea level along the northwestern divide. Topography is subdued with broad valleys and moderate relief in the
northern branch subbasin and steeper with incised valleys in the southern branch. Soils are primarily well-drained
38
-------
Water Acquisition for Hydraulic Fracturing
May 2015
loamy inceptisols and entisols formed in glacial and alluvial parent material (Grubb 1986). Rainfall differed somewhat
between the two weather stations, and land use and topography varied with the geology. The variable physiography,
land use, and weather within the Towanda Creek basin allowed it to represent the wider range of variability in these
characteristics within the region.
A)
B)
Susquehanna River Basin
C)
o o
°0
A O&G water withdrawals
• HF wells
10Mi
H
Figure 4-26. Towanda Creek basin in Bradford County, Pennsylvania. A) Basin location map within the
Susquehanna River Basin, with permitted self-supplied hydraulic fracturing withdrawal sites (red dots). B)
Shaded relief map of the Towanda Creek watershed, showing meteology sites as black triangles labeled "TW"
and "CM" (source: NOAA 2013), USGS gages (source: USGS 2014b), and permitted withdrawal sites (red dots). C)
Map with existing hydraulic fracturing wells added as orange circles, showing withdrawal sites as blue triangles.
(Data sources: PADEP 2013a for hydraulic fracturing wells, SRBC 2013a for permitted withdrawal locations.)
39
-------
Water Acquisition for Hydraulic Fracturing
May 2015
A)
B)
C)
Figure 4-27. Visual display of land use, topography, and river size and form in the Towanda Creek basin. A) Aerial
photograph at 15,000 feet, showing a portion of the southern branch of Towanda Creek with forest and steeper
topography at the bottom, and the northern branch with subdued topography and agricultural land use at the
top, divided by a ridgeline at center. (Image from Google Earth, USDA Farm Service Agency, 2013). B) Landscape
perspective looking into the northern branch of Towanda Creek from the ridgeline between the north and south
branches. A gas pipeline right-of-way is in the foreground. C) Towanda Creek mainstem at about river mile 15.
(Source: Wikipedia.org; image by Labenedict.)
40
-------
Water Acquisition for Hydraulic Fracturing
May 2015
A)
Towanda Creek at Monroeton (01532000)
133DD
B)
Annual deviation from long-term mean flow, Towanda Creek (01532000): 1914-2014
u
E
E
E
•2 o
ro
*>
wet years
Mean flow = 292 cfs
11 . . i i .. i .. ii,. i i i.i , . ml, II . 1 1
1 "i i HI ii ii 1 1 i i|ii 'I ' ' IN ir i1
dry years
1
n
n
2011
i III
||T II
CTi O O
CTi O O
III
a
Figure 4-28. Long-term flow at the USGS gaging station at Monroeton (USGS 01532000). A) Mean daily flow,
in cfs, with the average of daily means for the total period shown as the red line. B) Annual deviation from
the mean, in cfs. (Data source: USGS 2014b.)
Table 4-7. Flow statistics at USGS gage in Towanda Creek (01532000; 215 mi2) for various time periods used in this
analysis. The scenario analysis used 1987 to 2012; 2009 to 2013 is the period of water withdrawal for hydraulic
fracturing. Data in cfs; 1 cfs = 0.65 MGD. (Data source: USGS 2014c.)
Time Period
1914-2013
1987-2012
2009-2013
Mean of Daily
Flow
(cfs)
293
306
322
Deviation from
Long-Term Mean
(cfs)
0
+13.6
+29.6
Median of
Daily Flow (cfs)
120
139
155
Average Seven-
Day Low Flow (cfs)
10.9
10.6
9.7
Minimum Flow
(cfs)
0.7
1.7
4.1
41
-------
Water Acquisition for Hydraulic Fracturing
May 2015
The USGS flow gaging station in Towanda Creek at Monroeton (USGS 01532000) has operated since 1914. The 100-
year streamflow record is shown in Fig. 4-28, with mean daily (A) and deviation from the annual mean (B). Flow
statistics for time periods pertinent to this study are provided in Table 4-7, including the long-term record (1914 to
2013), hydraulic fracturing activity (2009 to 2013), and scenario analysis (1987 to 2012). SUI will vary commensurate
with deviation of streamflow from the norm. The recent period of hydraulic fracturing activity was generally wetter
than average, with higher mean and minimum flows.
Scenario: Assess Hydraulic Fracturing Withdrawal Over Longer Climate Record and Stream Size. In this
scenario analysis, we extended analysis of hydraulic fracturing withdrawals from five years to 26 years, thus
increasing coverage of variable climatic conditions. The scenario also considered much smaller subbasins, including
first-order headwaters as small as 0.3 mi2. This scenario simulated hydraulic fracturing withdrawals to meet current
and projected water needs with future drilling rates.
10
Ol
4^
Lo
M—
O
L_
Ol
-Q
E
nn -
80 :
60 :
40 :
20 :
0;
• Modeled Scenario
• Actual Permitted
T-H rsi LO r-*
ill II
00000000
00000000
T-HT-HrsioOLnOLno
0 0
0 0
0 0
Basin Area (mi2) ^
Figure 4-29. Comparison of contibuting watershed area of
Towanda Creek subbasins used for scenario analysis and
Susquehanna River Basin Commission-permitted surface water
sites in the Susquehanna River Basin. (Data source for permitted
sites: SRBC 2013a.)
First, the watershed was divided into a nested
set of subbasins from first to fifth order using
the Strahler stream ordering method to establish
streamflow prediction points below each
change in stream order. The process was
automated by the Arc SWAT extension for
ArcGIS 10.0, which divides the watershed into
units associated with individual stream
segments that are defined based on a user-
specified accumulation threshold for perennial
streamflow. The perennial streamflow location
was estimated as the "blue line" on USGS
1:24,000 maps, and a portion were field-
verified during annual low flow in November
2013. Subbasins were aggregated into
watersheds encompassing the entire contiguous
area upstream of each subbasin outlet where
streamflow was predicted. In all, 168 subbasins
were delineated: 134 were located on first-order
streams, 26 on second-order streams, five on
third-order streams, two on fourth-order streams
(the north and south branches of Towanda Creek),
and one on the fifth-order basin corresponding to
Towanda Creek at the Monroeton USGS gaging
station. The areas of subbasins established in the Towanda watershed overlap SRB permitted sites as shown in Fig. 4-
29. The scenario analysis contains many smaller streams that are not permitted in SRBC at present, but could feasibly
provide water for the O&G industry in the SRB and elsewhere.The subbasins are varied in land use and vegetative
cover between the north and south branches reflecting the variability illustrated in Fig. 4-27.
Scenario Water Availability Estimates. There are no USGS gages of comparable size to the smallest subbasins that
could be used for empirical extrapolation of daily streamflow. Thus we opted to use a streamflow simulation model
(HSPF) to generate a daily flow record at each stream prediction location. HSPF is a widely used, freely available,
FEMA-endorsed deterministic model for streamflow and water quality simulation (Bicknell et al. 1997; FEMA 2014).
Based on user-supplied input data including spatially and temporally distributed weather, soils, topography, and land
cover, the model partitions water to hydrologic processes including evapotranspiration, infiltration, surface runoff, and
exfiltration of water from subsurface as streamflow (Bicknell et al. 1997). Land use was acquired from the Multi-
Resolution Land Characteristics Consortium. Climate data including temperature and precipitation were obtained from
the NOAA National Climate Data Center (NOAA 2013) for the two meteorological sites. Soils data were obtained
from the Natural Resources Conservation Service. (See Appendix A for a compendium of data sources and Appendix
B for additional discussion of the hydrologic model, data sources, calibration procedures, and uncertainty analysis.)
Streamflow measured at the USGS gaging station was used for calibration and validation. HSPF was calibrated using
a Monte Carlo method to optimize the parameter set that best matched simulated with observed streamflow at the
42
-------
Water Acquisition for Hydraulic Fracturing
May 2015
USGS-gaged Towanda Creek outlet. The objectives of the calibration were to (1) favor good prediction of low flows
over high flows and (2) minimize over-fitting the model by retaining as few parameters as possible. In general, HSPF
tended to overestimate low flows and underestimate high flows, but the calibration approach identified an optimized
parameter set that agreed well with observed flows across a range of magnitudes. Originally, the project intended to
use both HSPF and SWAT hydrologic models to estimate streamflow. The choice of Monte Carlo calibration methods
precluded use of SWAT due to computational inefficiencies introduced with this approach. (See Appendix B for more
discussion of model application.) Calibration resulted in a weighted Nash-Sutcliffe (WNS) score of 0.74, which
exceeds common performance thresholds used for raw WNS (Moriasi et al. 2007; Narasimhan et al. 2005).
Once calibrated parameter sets were established, streamflow was simulated at all 168 streamflow prediction points in
Towanda Creek for the period 1987 to 2012, chosen due to availability of necessary weather data. This period
increased representation of low flows relative to the 2009 to 2013 period of hydraulic fracturing activity (Table 4-7),
but still did not encompass the lowest flow of 0.7 cfs observed in Towanda Creek in 1932.
Scenario Consumption Estimates. Scenarios were designed to test all flows in the 26-year period in proportion to
their occurrence at each site. Each day, a volume of water was withdrawn that included a "background" volume
(accounting for unmeasured withdrawals for uses such as irrigation, livestock watering, and residential), and a
hydraulic fracturing volume assumed to be consumed in hydraulic fracturing wells drilled in the Towanda Creek
basin. Background consumption was estimated from county water use rates determined from data from the USGS
water census, U.S. Census Bureau, and the U.S. Department of Agriculture. Background use was on the order of 0.1
MOD per mi2, but varied with land use in each subbasin. See Appendix D for more information on derivation of water
use assumptions in scenario analysis.
Three hydraulic fracturing scenarios were applied as summarized in Table 4-8. Two reflected current withdrawals in
the SRB, including the median (0.19 MOD), and mean (0.31 MOD) daily withdrawal observed at all permitted sites.
The third represented a future peak drilling rate. U.S. EIA (2014c) projects increased production in the Marcellus
Shale in future decades, with well density eventually reaching 4.9 wells per mi2 (U.S. EIA 2014c). Johnson (2010)
and SRBC (Beauduy 2009) estimate that as many as 2,800 hydraulic fracturing wells could be drilled annually. Based
on area, the total prorates to about 52 hydraulic fracturing wells per year in the Towanda basin, or about one well per
week (4 MG), requiring 0.57 MOD of freshwater. In 2011, 33 hydraulic fracturing wells were drilled and stimulated
in the Towanda basin (PADCNR 2014a), so the selected maximum drilling rate appears reasonable. The daily water
withdrawals for hydraulic fracturing used in all three scenarios have been routinely observed in streams with basin
area as small as 1.7 mi2. The scenario analysis assumed that all hydraulic fracturing wells consume water from one
source every day for 26 years, rather than distributing withdrawals among the subbasins. This analysis ensures that the
entire flow history is sampled in proportion to the frequency that each flow occurs. SUI calculations were performed
at each stream prediction location. No passby was used.
Table 4-8. Description of hydraulic fracturing withdrawal scenarios applied to 26 years of streamflow (1987-2012)
in subbasins of the Towanda Creek watershed in Bradford County, Pennsylvania.
Scenario ID
Median observed
Mean observed
Peak drilling
Background
Towanda Creek Withdrawal Scenarios for Surface Water Use Intensity Index Analysis
Water Demands
Hydraulic fracturing
current demand
Hydraulic fracturing
current demand
Hydraulicfracturing peak
demand
Unmeasured background
water use by other users
Hydraulic Fracturing Demand Assumptions
Median daily of all permitted withdrawal
sites
Mean observed of all permitted withdrawal
sites
One well per well @ 4.0 MG/week
Irrigation, livestock, and residential
assuming rates based on county-scale
assessment
(on the order of 1,000 GPD/mi2)
MGD
0.19
0.31
0.57
Variable depending
on land use in each
subbasin
cfs/day
0.3
0.5
0.9
-
43
-------
Water Acquisition for Hydraulic Fracturing
May 2015
50th Percentile
• Peak Drilling
O Mean Observed
A Median Observed
V , , JB.B •
10 100
Basin Area (mi2)
95th Percentile
1,000
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
A
A
• Peak Drilling
• Mean Observed
A Median Observed
10 100
Basin Area (mi2)
1,000
4-30.Surface water use intensity (SUI) distribution statistics for
withdrawal scenarios applied to subbasin streamflow
prediction points in Towanda Creek. Withdrawal scenarios
include peak drilling, mean, and median withdrawals (see Table
4-8). A) 50th percentile of the distribution simulated for 26
years. B) 95th percentile of the observations.
SUI: 0.571 Consumption Scenario
Median Flow
SUI
<0.2
0.2 - 0.4
0.4 - 0.7
>0.7
1 st order stream
2nd order stream
- 3rd order stream
• 4th order stream
• 5th order stream
o O OO
1st 2nd 3rd 4th 5th
Streamflow Prediction Points
Long-term SUI. SUI values from the 26-year
simulated flows are summarized by population
distribution statistics of median and 95th percentile in
Fig. 4-30. The small streams emphasized in this
analysis have relatively low streamflow much of the
year. SUI follows basin area closely as an index of
available flow and varies with withdrawal volume.
The scatter within each withdrawal scenario reflects
the differences in rainfall and land use between the
subbasins. In small streams less than 10 mi2,
withdrawals can approach or equal available
streamflow frequently during most years evident in
SUI exceeding 0.4 for 50 to 95 percent of the time.
For example, the median SUI under the mean
observed withdrawal scenario exceeds 0.5 for
watersheds less than approximately 2 mi2 (Fig. 4-
30A). The 95th percentile of SUI values for all three
scenarios equaled 1.0 in watersheds less than 30 mi2
(Fig. 4-30B).
The 95th percentile of SUI declined to less than 0.2 in
watersheds greater than about 200 to 300 mi2 for the
peak drilling scenario and at about 70 mi2 for the
median scenario. The same threshold was observed at
about 5 mi2 at actual hydraulic fracturing withdrawals
sites (Fig. 4-24A), reflecting the higher flows
observed during the limited period of active hydraulic
fracturing operations (2009 to 2013). The scenario
analyses demonstrate the vulnerability to over-
withdrawal in smaller streams, especially in
watersheds smaller than 10 mi2 and extending to
larger watersheds infrequently. It is clear that SUI is
sensitive to basin area and volume withdrawn.
Fig. 4-31 shows SUI at streamflow prediction points
in Towanda Creek for the peak drilling withdrawal at
median streamflow. The larger subbasins could
maintain lower SUI for the withdrawal volumes at
commonly occurring flows, but most small streams
that are potential withdrawal sites for hydraulic
fracturing could not.
Figure 4-31. SUI at peak drilling and median flow
at subbasin streamflow prediction points in
Towanda Creek subbasins. The Towanda Creek
basin is approximately 23 miles from mouth to
headwaters and 11 miles in width.
44
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Regional Flow Equation Estimates of SUI Thresholds. Basin area thresholds to support minimum streamflow
necessary to support hydraulic fracturing withdrawals could also be estimated from regional flow equations. These
equations apply multiple linear regressions to long-term USGS flow records, including geographic factors that
influence streamflow, to develop estimates of flow rates for flow frequency statistics. Equations predict streamflow
(cfs) as a function of basin area and often include parameters such as precipitation or land use factors that improve
prediction precision. Regional regression equations are used as a basis for SRBC permit specifications (SRBC 2003,
2012). Stuckey (2006) provided regionalized streamflow regression equations for Pennsylvania that include basin
area, mean annual precipitation, mean
PA Regional Flow Equations (Stuckey 2006) , . , r •, i . •
10,000.000 3— elevation, and percent of the watershed in
forest cover, glaciated, and in urban land
use as predictive variables. These methods
are more readily available for most users
than spatially explicit hydrologic modeling
as used in our scenario analyses, but they
do not provide daily flow estimates.
1,000.000
100.000
Qmean
10.000 - A-r,«T — — T
1.000
0.100
0.010
0.001
Qmean
BaselOyr
A 07,2
A Q7,10
_-- 1MGD
— — PeakDrilling
^^™ • Mean Observed
_•• Median Observed
I-H 1—i i i 111
The regression predictions for several flow
statistics, including mean annual and the
average annual seven-day low flow with 2-
year (Q?,2) and 10-year return intervals
(Qv.io) for the Towanda Creek basin, are
shown in Fig. 4-32. A few individual data
points are included to assist the reader
locate the diagonal line for the four flow
frequency statistics.
10
100
1,000
Basin Area (mi2)
Figure 4-32. Pennsylvania regional flow equations applied to the
Towanda Creek watershed with withdrawal scenarios. (Model source:
Stuckey 2006.) SUI exceeds 0.4 where flow statistics (diagonal)
intersect withdrawal volumes (horizontal lines).
In this example, we seek the basin areas for
each of the withdrawal scenarios (Table 4-
8) that will produce a SUI no greater than
0.4. We included a fourth withdrawal of 1
MOD for this analysis. A minimum flow of
25% was maintained to mimic the passby
flow. The flow threshold can be determined
with the regional equation by converting
the total withdrawal volume to streamflow rate (cfs) and dividing by the SUI threshold of interest. The streamflow
regressions remain in the same position on the chart with each withdrawal scenario, but the withdrawal lines move up
and down depending on the selected SUI threshold. Fig. 4-32 shows results for SUI 0.4 or lower.
For SUI to be less than 0.4, flow must be above the horizontal withdrawal line. The basin area thresholds can be
identified by tracing horizontally along the withdrawal volume line to where each intersects the flow statistic line. For
example, the median withdrawal of 0.19 MOD (lowest horizontal line) can be supported while maintaining the SUI at
less than 0.4 at mean annual flow when basin area exceeds 3 mi2, at the 10-year baseflow when basin area is about 10
mi2, at Qv,2 when basin area exceeds 100 mi2, and at the 10-year low flow Qv.io when basin area exceeds 300 mi2. The
regressions are primarily determined by basin area, but they are also sufficiently sensitive to the land use and climate
input parameters that results will vary somewhat from basin to basin. The regional equations confirm that the
necessary watershed size to support typical hydraulic fracturing withdrawals in the SRB varies over orders of
magnitude from low flow to high flow dependent on withdrawal volume.
We compared regional regression with HSPF simulation for low flow (Qv.io) and annual mean flow (Qmean). Mean
flows across a wide range of drainage areas were very close using these two methods. The best-fit HSPF Q7,io
estimates agreed closely with the regional equation-derived estimates, particularly in smaller watersheds, which were
the focus of much of this analysis. HSPF tends to estimate lower Q7,io flow than the regional equations, particularly in
larger watersheds. (See graphical comparisons of the regional equation and HSPF comparison in Appendix B.)
Scenario Illustrating the Effect of Passby on SUI in a Small River. This scenario analysis demonstrates the
effect of the passby flow at a site in a small river. For this computation the same volume of water was withdrawn daily
from Towanda Creek at the location of USGS gage (01532000, B.A. 215 mi2) from 1987 to 2012. The passby
45
-------
Water Acquisition for Hydraulic Fracturing
May 2015
threshold (12 cfs) and withdrawal limit of 1 MOD were specified in the permit of a nearby O&G withdrawal site. The
computations were performed with and without the passby flow assigned to the site. Results are shown in Fig. 4-33.
No background consumption was included.
The minimum flow restrictions assigned by the passby threshold would have been invoked on 770 days during the 26-
year period. SUI in the more recent years of 2009 to 2013 was generally lower than in the decade of the 1990s. The
passby was significant in preventing higher SUI. Without the passby, SUI greater than 0.4 would have occurred on
103 days, and SUI exceeding 0.7 would have occurred on 25 days (Fig. 4-33 and Table 4-9). Even in this small river
(see Fig. 4-27C), SUI would have reached as high as 0.9. With the passby flow restrictions, there would have been no
days exceeding 0.1 (Fig. 4-33B). The count of days of SUI with and without passby is provided in Table 4-9.
Towanda Creekat Monroeton 01532000
06-
1 1
liLJ
xLJ
uuJiJ
A) Without Passby
!|
^JLllJIji
*^O *^O "^O ^3 ^/O ^/O ^/*J
\ \ \ % % °AJIL.iul MJUUuJu^MUULAi, IL^,^ Jl -Ji JIL JL-Ulu J Jk
Figure 4-33. Simulated daily surface water use intensity index computed with the U.S. Geological Survey gaging
record at Towanda Creek (USGS 01532000) from 1987 to 2012, assuming 1.0 MGD withdrawal for hydraulic
fracturing and background consumption of 0.3 million gallons per day. Scenarios calculated with passby flow (A)
and without passby flow (B). (Data source: USGS 2014b.)
Table 4-9. Count of days by surface water stress
index category with withdrawal simulation of 1
million gallons per day with and without passby
flows applied to Towanda Creek (1987 to 2012,
n = 9,497). (Data source: USGS 2014b.)
SUI Category
Count of Days
Less than or equal to With Passby Without Passby
0-0-0.10
0.11-0.20
0.21-0.30
0.31-0.40
0.41-0.50
0.51-0.60
0.61-0.70
0.71-0.80
0.81-0.90
0.91-1
0
0
0
0
0
0
0
0
0
549
188
64
42
20
16
20
3
~
46
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Passby threshold flows can shut down hydraulic fracturing withdrawals a number of days each year, since they
generally are set at 20% to 25% of the mean annual flow. The full 100-year flow record at the Towanda gage was
analyzed for frequency of occurrence of the passby on an annual basis. SRBC applied an annual low flow passby until
permit renewal in 2013 when monthly values were assigned. The number of days on which a passby would have
occurred using both approaches is shown in Fig. 4-34. Over 100 years, the annual passby would have been invoked
7.4% of total days (2,726), and would have involved as many as 160 in the 100-year drought (1931). The monthly
passby tends to add 10 to 20 days of closure each year. Although it may appear in Fig. 4-34 that passby occurrence
was less frequent during the recent 26-year modeling period, the annual proportion of days in passby was about the
same as in the 71 preceding years.
Towanda Creek at Monroeton (01532000)
Passby Invoked
250
200
150 -
100
50
• Annual (12 cfs) • Monthly (12-105 cfs)
Mil .1
nil
i
1
1
1.1,11 ll
1
1
J
I
1,1.
1,1
I
1,
.1
Figure 4-34. Count of days each year that the flow fell below the passby flow assigned bytheSusquehanna River
Basin Commission to U.S. Geological Survey Towanda Creek gaging station. SRBC first assigned an annual passby
flow level. A monthly passby is now assigned during renewal of permits.
(Data source: USGS 2014b.)
Scenario Evaluating Cumulative Permit Effects. As the number of
permitted sites grows, they begin to accumulate in 30 HUC 8 to 10
tributaries to the Susquehanna River. Some are in relatively close
proximity, probably reflecting locations where streams can be legally and
safely accessed near major roads. In this scenario, we briefly examine the
cumulative effects of individual permits. We selected one location on
Wyalusing Creek where four withdrawal sites are located within a basin
area of approximately 150 to 190 mi2. We summed the daily withdrawals
and the permit capacity at the four sites and withdrew the combined
volume from the streamflow at one of the sites. Total permit capacity for
the group was 5.43 MOD (nearly twice the permitted volume at any single
site in the SRB).
0.10 -
Fig. 4-35 shows SUI associated with the summed observed withdrawals
and maximum capacity, assuming all volume was taken at each site
every day. Observed SUI were low for actual withdrawals (< 0.01), and
increased to a maximum of 0.16 had the full capacity been used. This
example does not indicate significant cumulative effect from multiple
permits, which are limited by the passby shutdowns, as shown in Fig. 4-
3 3. SUI would be significantly higher at this level of withdrawal if
passby flows were not in effect as illustrated in Fig. 4-33.
Figure 4-35. SUI analysis of cumulative
withdrawals from four permitted sites on
Wyalusing Creek. Box plots are the
distribution of the SUI for the combined
daily withdrawals. Boxes are 75%
percentiles; bars are 95th percentiles. (Data
source: SRBC 2013a.)
47
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Analysis of Hydraulic Fracturing Water
Acquisition on Groundwater
Groundwater supplies about 20% of the water
used for hydraulic fracturing in the SRB, or about
2 MOD (Fig. 4-36). Groundwater-based
providers are distributed throughout the SRB and
include seven self-supplied sites and 15 active
PWSs (Fig. 4-37). Of this group, two commercial
(but not public) wells have provided most of the
self-supplied groundwater, and two municipal
PWSs have provided 40% of all of the publically
supplied water (Fig. 4-36). The majority of
groundwater is obtained in Bradford, Lycoming,
and Wyoming Counties. Figure 4-37 includes all
PWSs that are available as identified to SRBC.
While available, they do not all provide water to
O&G.
o
(D
3.0
Groundwater Sources For O&G in SRB
I Self-supplied
PWS
2010
2011
2012
2013
Figure 4-36. Daily rate of groundwater supplied to the oil
and gas industry from public water systems and self-
supplied sources. (Data sources: PADEP 2014a; SRBC 2013a.)
Self-supplied
O PWS
60
120 Miles
Figure 4-37. Groundwaterwellfields currently available for water acquisition by the oil and gas industry
in the Susquehanna River Basin. Circles represent public water systems with documented supply to the
oil and gas industry. (Data source: U.S. EPA 2012c.) Squares represent self-supplied (private) sources
permitted by the Susquehanna River Basin Commission. (Data source: SRBC 2013c.)
48
-------
Water Acquisition for Hydraulic Fracturing
May 2015
In this rural area within the SRB where
hydraulic fracturing is active, water resource
use is low as there is relatively low population
or industrial use. Usage is centered near towns
that utilize public supply wells, springs, or
domestic wells. Many towns within the
Susquehanna River valley are situated on or
adjacent to alluvial fans of large tributary
streams that overlie saturated sand and gravel
(outwash or ice-contact deposits) and
commonly tap thin permeable sand and gravel
zones just above bedrock or a few feet into
fractured bedrock for water supply (Heisig
2012). The alluvial floodplains are about 50 to
150 feet thick, and store a significant volume
of water that is in exchange with the river and
continually recharged by rainfall. Confined and
unconfined aquifers within the valley-fill are
the only potential groundwater source of large
municipal, commercial, or industrial supplies in
this area (Heisig 2012). Usage in the upland
area consists of widely spaced domestic wells
that tap the bedrock aquifer.
o.o
1.0
Rated Pump (MGD)
2.0 3.0
4.0
5.0
~ u ';
S
IT -100
(J
= -200 •
o
ID -300 ;
CO
-C
S--400 -
a
| -500
cnn -
A A
* 1
AA
A
O O O
OAlluvial Deposits
• Outwash Deposits
A Bedrock
A
Figure 4-38. Depth and pumping capacity of public and self-
supplied groundwater sources registered to supply water for
hydraulic fracturing in the Susquehanna River Basin classified
by geology type. (Data sources: PADCNR 2014a; PADEP 2013a;
SRBC 2013c.)
The relationship between well depth, pumping capacity, and geologic substrate at the public and private groundwater
supply sites that are registered to provide water to hydraulic fracturing operators is demonstrated in Fig. 4-38. The
highest yielding wells are found in the river alluvium and glacial outwash. The majority of wells are relatively low-
yielding and mixed between tapping alluvium, glacial outwash, and bedrock aquifers. Domestic wells in the uplands
tend to be deeper.
We were interested in assessing the potential impacts of hydraulic fracturing acquisition on groundwater resources,
although determining available water volume was not as straightforward as it was for surface water. In this section we
explore groundwater withdrawal effects at two spatial scales: the Towanda Creek watershed scale (215 mi2) and the
local site scale. The questions guiding the analysis were: 1) What is an effective extent of the groundwater reservoir?
and 2) How does groundwater use at the watershed and local scale affect other users and streamflow?
Dry Well
i i i
^ Recharge^ 2
Figure 4-39. Schematic of the hydrologic water cycle with emphasis on groundwater interactions
with surface water flow and possible impacts from water withdrawals.
49
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Groundwater Basics. Precipitation transfers water from the atmosphere to the land surface and initiates the water
cycle (Fig. 4-39). Some water is lost back to the atmosphere through evaporation and transpiration, and some travels
to streams relatively quickly during storms via overland flow. Precipitation infiltrates into the soil, where some is
tightly held by surface tension in the unsaturated soil matrix and some moves slowly downgradient through porous
soil and rock to topographic lows carved by streams. Groundwater saturates the valley low points and exfiltrates to
become streamflow. A small fraction percolates into the deep bedrock strata. Streamflow and subsurface water are
intimately associated and are in a continuous process of exchange (Dunne and Leopold 1978.) Groundwater supplies
streamflow between rain events and streams also feed water into the saturated alluvium. The top of the saturated zone,
called the water table, is largely coincident with the water surface in streambeds (Freeze and Cherry 1979).
The dynamic coupling between groundwater and surface water is evident when the USGS streamflow gage is
compared with a nearby USGS groundwater observation well in the alluvial valley in the Towanda Creek basin for a
4-year period (Fig 4-40). The base of the well is in the bedrock of the Lock Haven formation. The stream and water
table in the adjacent alluvium are tightly coupled, as the elevation of the water table rises during rainfall and falls as
the aquifer drains to the stream during intervening dry periods. The portion of precipitation that makes it way to the
water table and eventually to streams through subsurface flow paths is termed "recharge." We use the annual recharge
as the approximation for the dynamic and replenishable storage volume in the watershed's groundwater reservoir.
Watershed Groundwater Volume. Recharge to streams through groundwater flow paths can be estimated from
streamflow records by partitioning the hydrograph into stormflow and baseflow (periods between rainfall events).
This can be a subjective interpretation if only the flow record is available, as there are no clear indicators of a change
from stormflow to groundwater in the streamflow trace. Stream chemistry can be used in field studies to more
accurately identify stormwater. There are a variety of baseflow separation methods and different methods give
different results reflecting stormflow partitioning choices (Eckhardt 2008). USGS has developed several tools that
apply systematic rules to automate baseflow separation of USGS daily streamflow data, including RORA and PART.
PART designates groundwater discharge to equal streamflow on days that fit a requirement of antecedent recession,
and linearly interpolates groundwater discharge for other days; the method has been shown to be effective in
characterizing baseflow over long periods of record (Rutledge 1998).
Towanda Creek Streamflow and WaterTable Rise
100,000
40
10,000
WaterTable Rise
Streamflow
Month, Year
Figure 4-40. Daily hydrograph of streamflow in Towanda Creek at the U.S. Geological Survey (USGS) gage at
Monroeton (01532000) and USGS observation well BR92 (414330076280501) located in alluvium in the same
general area from 2008 to 2012. Water table rise is referenced from a datum set at 12 feet below land surface (the
maximum depth observed) to reverse the data provided as depth to water table for easier visualization of the
tight coupling of water table rise and streamflow. (Data source: USGS 2014b.)
50
-------
Water Acquisition for Hydraulic Fracturing
May 2015
A)
Baseflow-TowandaCr at Monroeton (01532000)
20
18 -
16 -
14 -
12 -
10
8
6
4
B)
20
18
is
14
10
Baseflow = 0.305ppt - 0.52
R2 = 0.52
0 10 20 30 40 50 60 70
Annual Rainfall (in)
Figure 4-41. Annual baseflow normalized to watershed area of 215 mi2 at the U.S. Geological Survey Towanda
gage at Monroeton from the period of record 1915 to 2012 using PART baseflow separation. A) Annual baseflow,
and B) Relationship between annual rainfall and annual baseflow. (Data source: USGS 2014b.)
PART was used to determine annual baseflow for the Towanda Creek basin from 1915 to 2012 using the flow record
at the USGS Towanda gage (Fig. 4-41 A). Annual baseflow is expressed as inches per year by normalizing the annual
baseflow water volume by watershed area. Annual streamflow and baseflow ranged widely from year to year based
largely on precipitation, with annual baseflow averaging 31% of annual rainfall (Fig. 4-4 IB). Average baseflow for
the 100-year record was 9.8 inches/year, ranging from 4.8 to 17.9 inches. To estimate the basin's annual recharge, we
used a baseflow of 9.6 inches per year (the 50th percentile) as the high estimate and 6.0 inches per year (5th percentile)
as a low estimate. For managing groundwater withdrawals, SRBC defines the annual baseflow (recharge) during a 1 in
10 year average annual drought to be the sustainable limit for groundwater withdrawals, equal to 5.7 inches per year in
this location (SRBC 2005). We assumed that the annual groundwater recharge was equal to the annual baseflow at the
USGS Towanda gage. The baseflow volumes normalized by watershed area were translated to water volume and
listed in Table 4-10 as assumptions for groundwater reservoir estimates.
The storage of groundwater beneath a watershed is filled by groundwater recharge. We estimated the groundwater
reservoir volume based on water table depths observed in drilled well logs assisted by the groundwater model
GFLOW. GFLOW solves for regional and steady groundwater flow in single layer aquifers (Haitjema 1995). The
model is well documented and accepted within the groundwater modeling community (Hunt 2006; Yager and Neville
2002), with particular application to shallow groundwater flow systems involving groundwater/surface water
interactions (Johnson and Mifflin 2006; Juckem 2009) and for recharge estimation (Dripps et al. 2006).
We used GFLOW™ to distribute areal recharge volume and to generate the spatial distribution of the elevation of the
water table surface within the watershed. The water table was set by anchoring it at the location of the upper extent of
the perennial stream channel network. The perennial streamflow location was estimated as the "blue line" on USGS
1:24,000 maps, and a portion were field-verified during annual low flow in November 2013 with good agreement.
GFLOW created a gridded digital model of the water table elevation from the stream elevation points considering the
hydraulic conductivity and saturated thickness of the substrate. The water table at annual average baseflow is shown
as the upper surface in Fig. 4-42. The water table generally follows the topography (Freeze and Cherry 1979). See
Appendix C for a detailed discussion of how the groundwater model was applied and calibrated.
51
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Aquifer depth below the water table
surface was determined from well drilling
logs. Domestic well logs were obtained
from the Pennsylvania Department of
Conservation and Natural Resources
Groundwater Information Systems
(PADCNR 2014a). When drilling
hydraulic fracturing wells in this area,
O&G operators record the depth of the
freshwater aquifer based on permeability
as they drill through to deeper gas
reserves. Hydraulic fracturing well logs
were obtained from PADCNR (2014b).
Hydraulic fracturing well logs indicate
average fresh groundwater thickness of
about 420 feet depth on average in this
watershed. Risser et al. (2005) measured a
freshwater aquifer thickness of 442 feet at
the Gleason Test Hole located just to the
west of the Towanda Creek watershed.
Domestic water wells are drilled to an
average depth of 160 feet. Domestic well
pump placement is often shallower than
maximum aquifer depth to minimize the
depth necessary to ensure a reliable supply.
The freshwater aquifer with the two
thicknesses is shown in Fig.4-42.
Figure 4-42. Conceptual depiction of groundwater volumes within
the Towanda Creek watershed. The vertical scale is exaggerated
relative to the other two scales. The top surface is the water table
generated by the groundwater model GFLOW. The light blue aquifer
depicts the average depth of domestic water wells. The darker blue
aquifer is the additional depth to the base of the freshwater aquifer
based on hydraulic fracturing well logs. (Data sources: Well drilling
logs for the watershed were obtained from the Pennsylvania
Groundwater Information Systems database, PADCNR 2014a.
Hydraulic fracturing drilling logs were obtained from PADCNR
2014b.)
The volume of water within the freshwater
aquifer depends on the drainable porosity,
or specific yield, of the rock. Two estimates
of groundwater aquifer volume were calculated based on the aquifer thickness of the Towanda Creek watershed area,
and a high and low estimate for porosity of 10% and 5% respectively. Results are provided in Table 4-10. With these
assumptions, the calculations suggested a large volume of water within the groundwater reservoir in Towanda Creek.
The minimum estimate is 359,000 million gallons of water (1 million acre-feet). The annual recharge is about 3% to
6% of the total reservoir volume.
Table 4-10. Groundwater aquifer volumes, annual recharge, and withdrawn volume for public and domestic water
supplies, agricultural uses, and by the O&G industry for hydraulic fracturing. High and low estimates envelop the
range of aquifer parameters used to estimate the volume of water in the groundwater reservoir. See Table 4-11
for withdrawal estimates.
Groundwater
Reservoir
Scenario
High storage
volume
Low storage
volume
Assumptions
Porosity: 10%
Recharge: 9.6 inches per year
Total freshwater aquifer thickness: 400 feet
Oil and gas municipal withdrawal: 75 MGY
Porosity: 5%
Recharge: 6.0 inches per year
Domestic well aquifer thickness: 160 feet
Oil and gas groundwater withdrawal: 75 MGY
Total Aquifer
Volume
1,800,000
MG
359,000
MG
Average
Recharge
35,800 MGY
(130 MGD)
22,400 MGY
(61 MGD)
Withdrawn from
Aquifer
378 MGY
(1.04 MGD)
378 MGY
(1.04 MGD)
52
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Groundwater Withdrawal. There are at least 126 active water wells in the Towanda Creek watershed drawing from
both bedrock and alluvial aquifers that are used for a variety of purposes, including domestic supplies primarily, some
industrial use including O&G, and farming (Fig. 4-43). About 8,800 people live in the basin in several small towns
(U.S. Census Bureau 2010). Average per capita use rate is assumed to be 60 GPD. Groundwater use in the Towanda
Creek watershed, on a daily and annual
basis, is shown in Table 4-11. Municipal
supplied water use by specific user sectors
was obtained from the PWS data from
PADEP (2014a), as discussed in an earlier
section. As done for SUI analysis,
irrigation and livestock use for the basin
were estimated using USDA and USGS
county-level estimates of water use.
At the watershed scale, the groundwater
reservoir and annual recharge are large
compared to annual withdrawals by all
users (Table 4-10). Annual average
groundwater use in the Towanda Creek
watershed including pumping withdrawals
supporting hydraulic fracturing ranges
between 1.1% to 1.7% of annual recharge
for the high and low cases, respectively.
The potential for hydraulic fracturing
impact on groundwater resources at the
watershed scale of 215 mi2 appears to be
small.
Municipalwells
Domesticwells
Other wells
16 Miles
Figure 4-43. Groundwater wells in the Towanda Creek watershed
based on PaDCNR (2014a). (The unpopulated southern part of the
watershed is the forested portion, protected as a state game
reserve.)
This does not mean that site-scale impacts
cannot occur. Zhou (2009) observed that groundwater systems are in dynamic equilibrium in which long-term average
recharge equals long-term average discharge under natural conditions. However, pumping groundwater locally
disturbs this equilibrium, and will cause a decrease in groundwater levels and induce new recharge patterns. A "safe"
or sustainable yield cannot be defined by natural recharge alone (Bredehoeft 2002; Zhou 2009).
Pumping groundwater resources at a well can affect the
sustainable production of other wells or streams within
its vicinity. When a well is pumped, water flows
through the permeable material toward the pump,
creating a so-called "cone of depression" radiating
outward from the pump's location and depth.
Drawdown of the water table is directly proportional to
the pumping rate and inversely proportional to aquifer
properties (Freeze and Cherry 1979). If pumping is
too strong relative to the aquifer's ability to move
water, nearby wells could go dry or the well could
capture the natural streamflow from adjacent
channels. The concept of pumping drawdown and
potential impacts to other wells or streamflow is
illustrated in Fig. 4-39. In the next section, we
examine local drawdown impacts using the
groundwater flow model GFLOW™.
Table 4-11. Estimated daily and annual groundwater use in
the Towanda Creek basin. Municipal withdrawals are
based on reported volumes in 2011 (PADEP 2014a). Self-
supplied domestic water is based on daily use per person.
Agricultural uses are estimated from U.S. Department of
Agriculture data.
Use
Domestic Water Self Supplied
Municipal WaterSupply
Daily Use
MGD
.411
.093
Annual
Total
MG
150
34
Municipal Other .146 53
O&G from PWS ( Pea k Yea r) .205 75
Agricultural Irrigation .021 8
Agricultural Livestock
TOTAL
.160
0.943
58
378
53
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Local Groundwater Use and
Impact. One small municipal water
supply located in the Towanda Creek
basin (Fig. 4-43) is one of several in the
SRB that have had larger sales to O&G
(Fig. 4-12). PWS 2080003 supplies
water from two wells drilled into the
stratified drift upstream of the
confluence and between Towanda Creek
and a smaller tributary. Drillers' logs
report total well depth of 110 and 120
feet, and pumps are rated at 300 and 350
gallons per minute. The capacity of the
two wells combined is 0.942 MOD.
There is no other permit limit.
PWS ID 2080003
Domestic and Other
O&G
GWSI
2010
2011
2012
2013
Figure 4-44. Daily water use at PWS 2080003 municipal water supply in
the Towanda Creek basin. Facility groundwater stress index was
calculated as proportion of total water delivered from the facility in
relation to its pumping capacity. (Data source: PADEP 2014b.)
PWS 2080003 typically provides 0.24
MOD of water for domestic, commercial,
industrial, and municipal use (Fig. 4-44),
using about 25% of facility water
production capacity. PWS 2080003 has
sold from 0.05 to 0.2 MOD to O&G,
depending on year, using an additional 6
to 22% of facility capacity. O&G water
sales peaked in 2011. This contributed to
facility-level sales of 0.48 MOD that consumed 75% of available capacity (GUI=0.75) in the peak use year. Sales to
O&G have declined since 2011. Because pumping rates are known at this facility, it serves as an example where
groundwater drawdown effects on the water table can be demonstrated with modeling.
The municipal wellfield introduces changes to the dynamic equilibrium of the subsurface geohydrologic system, as
does any pumping well. The groundwater system response to rainfall is quick, and from the modeling perspective,
instantaneous. The GFLOW™ groundwater model was shown to be effective at characterizing steady state flow in the
unconfined glacial outwash aquifer, both at the scale of the Towanda Creek watershed (215 mi2) and the local
pumping wellfield at the PWS, as shown next. The model was calibrated to USGS observed baseflow and USGS
observed static water table observations. The groundwater modeling (GFLOW™) and mapping (SUPJTiR™)
produced maps of the continuous water table surface (spatial grid), and the flow into, or out of, stream segments. The
details of groundwater methods are described in Appendix C.
Potential for high groundwater use intensity has two aspects. In addition to the localized drawdown of the water table
about the pumping center, the wellfield can capture induced recharge and water from nearby surface water features.
Both influences are shown for this wellfield in Fig. 4-45 for three pumping scenarios: (1) steady annual averaged
pumping to satisfy base-level drinking water and other municipal demands; (2) steady annual averaged pumping,
reflecting base-level pumping plus historical annual sales to O&G; and (3) maximum rated pumping with drawdown
to the base of the aquifer, assuming that the PWS sold 100% of maximum daily production. The following analysis
was for average geohydrologic conditions (11.0 inches/year recharge associated with 2000-2011 observations).
The maximum pumping scenario sets the facility GUI at 1.0 and produces the largest area of drawdown, as shown in
Fig. 4-45E. In comparison, the GUIs associated with scenario 1 (shown in Fig. 4-45A) and scenario 2 (Fig. 4-45C) are
0.26 and 0.40, respectively, which is a typical use level compared to other PWSs (Fig. 4-12). A vertical cross-section
of the drawdown associated with the pumping scenarios is shown in Fig. 4-46. Any other wells in the cone of
depression could be affected by drawdown in the municipal wellfield, but there are no known private drinking water
wells within the potential impact area for any of the pumping rates. Source water assessment and wellhead protection
programs for the PWS are in place to manage any potential risks.
54
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Scenario 1: Standard Pumping, No O&G A)
B)
Scenario 2: Standard + Average O&G C)
Scenario 3: Maximum Pumping
E)
well
pumping
consumes
1.6% of
available
baseflow
F)
max rated pumping
well
pumping
consumes
6.1% of
available
baseflow
(ft)
Figure 4-45. Groundwater-surface water interactions at the PWS 2080003 municipal groundwater wellfield under
average recharge and baseflow conditions (2000-2011). Towanda Creek is the lower stream. Scenario 1 involved no
sales to O&G and historical well supply to domestic and other uses; (A) shows the cone of depression associated
with a GUI = 0.016 and (B) shows the zone of capture and consumption of creek baseflow (1.6%). Scenario 2 involved
average historical sales to O&G in addition to average supply to domestic and other uses and GUI = 0.40; (C) shows
the cone of depression associated with GUI = 0.40, and (D) shows the zone of capture and consumption of creek
baseflow (2.4%). Scenario 3 involved the maximum rated pumping; (E) shows the cone of depression associated with
GUI = 1.00, and (F) shows the zone of capture and consumption of creek baseflow (6.1%).
55
-------
Water Acquisition for Hydraulic Fracturing
May 2015
V
.<:::
2000
(ft)
Figure 4-46. Groundwater cone of
depression associated with cross section A-
A' for the wellfield for three scenarios
shown in Figure 4-45. From the top down:
pre-pumping condition, scenario 1 (no sales
to oil and gas), scenario 2 (average sales to
oil and gas), and scenario 3 (maximum
drawdown). The vertical axis is greatly
exaggerated relative to the horizontal axis.
The PWS wellfield receives groundwater that originated as nearby surface water. As shown in Fig. 4-45F, under the
maximum pumping scenario, the wellfield captures water from the tributary and Towanda Creek adjacent to PWS
2080003, in addition to water from the outwash valley to the west, and a small amount from recharging waters in the
shaded source water zone. In the maximum pumping scenario, the volume of water captured from the baseflow of the
streams, in comparison to the total baseflow at the confluence, is 6.1%, which represents SUI of 0.06. Under scenario
1 with GUI equal to 0.26, the capture of baseflow from the tributary is about 1.6% of baseflow. Under scenario 2, the
capture of baseflow from the tributary is about 2.4% when the facility GUI is 0.40. Under dry conditions (6
inches/year recharge, one in 10 year recurrence), the stream capture use intensity for maximum rated pumping
increased to 10% at maximum facility pumping capacity. These findings support use of the background consumption
in surface water analysis reported earlier in that these withdrawals can influence surface water availability. In
summary, the PWS sales to date to O&G have been 10 to 40% volume of the municipal capacity, but the additional
well discharge was shown to have localized impact on drawdown and streamflow capture was shown to be 2 to 6%
under average and low flow conditions.
Self-Supplied/Private Water Use and Impact (O&G). There are six serf-supplied groundwater permitted sites in
the SRB, of which two have not been used to date. Permit and use information on the four active sites is provided in
Table 4-12. These wells supply nearly 1 MOD of water in the SRB, or about 15% of self-supplied freshwater for
hydraulic fracturing, varying by year. Streamflow capture and pumping effects such as demonstrated for the public
wellfield example can occur at these well-used sites, as previewed in Appendix C for SID 3711/3712.
Groundwater withdrawals are also permitted by SRBC. As a condition of permitting, permittees must conduct
constant rate pumping tests and monitor the drawdown at nearby wells to evaluate potential impacts that may require
restrictions or mitigation, all of which are contained in docket reports (SRBC 2013c). SRBC has assigned a surface
water passby flow to protect the stream from groundwater withdrawal during low flow, when tests demonstrate
interaction with adjacent streams. Because of the requirements for pump tests and site evaluation, it is unlikely that
there are impacts on other users from the operation of these wells to supply water for O&G.
Table 4-12. Source of freshwater from private wellfields in the Susquehanna River Basin (water use in MGD; pump
ratings data from SRBC 2013a; total depth of wells data from PADCNR 2014a). Average daily volume was
computed as total volume taken by number of days used.
Name
SID 3823 and 3837
(two wells)
SID 3792
SID 3594
SID 3711/3712
(3 wells)
County
Tioga
Bradford
Wyoming
Wyoming
Pump
Rated
MGD
0.54
0.364
0.54
0.864
Total Depth
(ft)
126
225
122
60
2010
MGD
0
0
0.04
0.08
2011
MGD
0.05
0
0.13
0.57
2012
MGD
0.18
0
0.16
0.55
2013
MGD
0.10
0.26
0.17
0.36
Peak Pumping
Groundwater Use
Intensity Index
0.46
0.72
0.77
0.66
56
-------
Water Acquisition for Hydraulic Fracturing May 2015
Vulnerability of Rivers and Streams to Depletion with Hydraulic Fracturing Withdrawals
Local effects of hydraulic fracturing water withdrawals can be assessed on a daily basis by relating the amount of
water acquired for hydraulic fracturing to the water available at the source (SUI). This study has shown the SUI of
actual water withdrawals subject to controls and modeled streams under scenarios of hydraulic fracturing demand.
Here we provide the general relationships between basin size, flow likelihood, withdrawal volume, and resulting SUI
values that has broad application beyond this study area. Only with an examination of these factors working together
can the full picture of the potential impacts from withdrawals be demonstrated.
SUI is a straightforward volumetric calculation with a simple mathematical structure that has the same meaning
wherever it is applied. A given withdrawal volume compared to a given flow volume will always produce a specific
value of SUI. The streamflow rate (expressed in cfs as done in USGS gaging records) necessary to meet a SUI target
for specific withdrawal volumes (expressed in MOD) is provided in Table 4-13. The table gives Qcritical values across a
range of SUI thresholds and withdrawal volumes up to 13 MOD. To compute the table, the streamflow rate was
converted to a daily volume, expressed as MOD.
Qcriticai (cfs) = Withdrawal Volume/SUI Equation 4-1
Flow must exceed the critical cfs value for each withdrawal volume to remain lower than the selected SUI value. For
example, if 1 MOD was withdrawn and the SUI threshold of interest was 0.5, the Qcriticai would equal 3.1 cfs. If
streamflow were less than this flow rate, SUI would exceed 0.5.
What varies between rivers is the probability of observing various flow rates. At a site, streamflow fluctuates from
seasonal low flow to floods in response to precipitation, and the range of possible flows is dependent on watershed
size, which determines the collection area for rainfall. It often requires decades to experience the full range of
streamflow at a site with mesoscale climate fluctuations. This is characterized at long-term streamflow gages with
flow frequency metrics that express the probability of seeing certain flow volumes given the observed record.
The extensive history of daily flow monitoring data available from USGS gaged sites can be harnessed to determine
the probability of observing Qcriticai values at specific locations. We acquired data for 48 gaged streams (ranging from
5 to 11,000 mi2) throughout the Susquehanna River Basin having records of at least 25 years, and added eight
randomly selected streamflow records from the small subbasins (0.5-4 mi2) in Towanda Creek generated by HSPF
modeling. (See Appendix B for a list of USGS gages used in SRB analyses.) Examining only sites with a sufficiently
long record ensured that climatic variability would be captured. We computed the frequency of flow observed and its
probability of occurrence. With the streamflow probabilities established, we computed the likelihood of observing the
SUI Qcnticai thresholds forthe range of withdrawal shown in Table 4-13. The water use intensity "heat maps" are
shown for six withdrawal volumes in Fig. 4-47. The probability of observing the SUI is on the right axis, and the SUI
range is represented by the spectrum of color. The gray area in each display indicates SUI less than 0.1. The six
withdrawal scenarios include the median (0.2 MOD) and mean (0.5 MOD) of actual observed withdrawals observed
in the SRB.
At these common daily withdrawals, very small streams have higher SUI most of the time for all withdrawal volumes.
Water is available a limited portion of the time. For example, if withdrawing 0.2 MOD, SUI is greater than 0.4 about
30% of the time (indicated by the red shades in Fig. 4-47). As withdrawal rate increases, higher values of SUI occur,
though infrequently, in surprisingly large rivers. For example, at withdrawal of 3.0 MOD, SUI could exceed 0.4 in
rivers up to about 500 mi2 (red or yellow on the figure). We include the very high withdrawal rate of 10 MOD,
because it is the sum of permitted withdrawals on several sites clustered very close together on the large Susquehanna
River.
The flow in rivers differs due to factors other than basin area. However, for the same reason that regional flow
equations can provide reasonable predictions of streamflow duration characteristics over wide areas, albeit with
variation, Fig. 4-47 has broad applicability in this part of Pennsylvania and in general within the humid eastern
climatic region. The withdrawal examples show that even larger rivers can occasionally have high water use intensity
indicated by high SUI from moderate and feasible hydraulic fracturing withdrawals. SRBC prevents this by shutting
off withdrawals at most sites when streamflow declines below thresholds established relative to the annual mean flow.
57
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Table 4-13. Minimum flow expressed in cubic feet per second needed to meet a surface water use intensity index
(SUI) value at various daily withdrawal volumes, expressed in million gallons per day. The flow threshold is
referred to as Qmticai in this report. Each combination of withdrawal volume and SUI target has a Qmticai value
expressed in the table. For example, for a withdrawal of 1 MGD and SUI<0.3, flow must be equal or exceed 5.16
cfs.
Surface Water Use Intensity Index (SUI)
<0.1 <0.2 <0.3 <0.4 <0.5 <0.6 <0.7 <0.8 <0.9 < 1.0
Withdrawal
(MGD) Minimum Flow in cfs to Meet SUI at Withdrawal Rate (Q^ai)
0.1
0.2
0.3
0.4
0.5
0.6
0.7
OS
0.9
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
IS
1.9
2
2.1
2.2
2.3
2.4
2.5
2.6
2.7
23
2.9
3
3.5
4
4.5
5
6
7
8
9
10
11
12
13
1.55
3.09
4.64
6.19
7.74
9.28
10.83
12.38
13.92
15.47
17.02
18.56
20.11
21.66
23.21
24.75
26.30
27.85
29.39
30.94
32.49
34.03
35.58
37.13
38.68
40.22
41.77
43.32
44.86
46.41
54.95
62.80
70.65
78.50
94.20
109.90
125.60
141.30
157.00
172.70
188.40
204.10
0.77
1.55
2.32
3.09
3.87
4.64
5.41
6.19
6.96
7.74
8.51
9.28
10.06
10.83
11.60
12.38
13.15
13.92
14.70
15.47
16.24
17.02
17.79
18.56
19.34
20.11
20.88
21.66
22.43
23.21
27.48
31.40
35.33
39.25
47.10
54.95
62.80
70.65
78.50
86.35
94.20
102.05
0.52
1.03
1.55
2.06
2.58
3.09
3.61
4.13
4.64
5.16
5.67
6.19
6.70
7.22
7.74
8.25
8.77
9.28
9.80
10.31
10.83
11.34
11.86
12.38
12.89
13.41
13.92
14.44
14.95
15.47
18.32
20.93
23.55
26.17
31.40
36.63
41.87
47.10
52.33
57.57
62.80
68.03
0.39
0.77
1.16
1.55
1.93
2.32
2.71
3.09
3.48
3.87
4.25
4.64
5.03
5.41
5.80
6.19
6.57
6.96
7.35
7.74
8.12
8.51
8.90
9.28
9.67
10.06
10.44
10.83
11.22
11.60
13.74
15.70
17.66
19.63
23.55
27.48
31.40
35.33
39.25
43.18
47.10
51.03
0.31
0.62
0.93
1.24
1.55
1.86
2.17
2.48
2.78
3.09
3.40
3.71
4.02
4.33
4.64
4.95
5.26
5.57
5.88
6.19
6.50
6.81
7.12
7.43
7.74
8.04
8.35
8.66
8.97
9.28
10.99
12.56
14.13
15.70
18.84
21.98
25.12
28.26
31.40
34.54
37.68
40.82
0.26
0.52
0.77
1.03
1.29
1.55
1.80
2.06
2.32
2.58
2.84
3.09
3.35
3.61
3.87
4.13
4.38
4.64
4.90
5.16
5.41
5.67
5.93
6.19
6.45
6.70
6.96
7.22
7.48
7.74
9.16
10.47
11.78
13.08
15.70
18.32
20.93
23.55
26.17
28.78
31.40
0.22
0.44
0.66
0.88
1.11
1.33
1.55
1.77
1.99
2.21
2.43
2.65
2.87
3.09
3.32
3.54
3.76
3.98
4.20
4.42
4.64
4.86
5.08
5.30
5.53
5.75
5.97
6.19
6.41
6.63
7.85
8.97
10.09
11.21
13.46
15.70
17.94
20.19
22.43
24.67
26.91
0.19
0.39
0.58
0.77
0.97
1.16
1.35
1.55
1.74
1.93
2.13
2.32
2.51
2.71
2.90
3.09
3.29
3.48
3.67
3.87
4.06
4.25
4.45
4.64
4.83
5.03
5.22
5.41
5.61
5.80
6.87
7.85
8.83
9.81
11.78
13.74
15.70
17.66
19.63
21.59
23.55
34.02 | 29.16 | 25.51
0.17
0.34
0.52
0.69
0.86
1.03
1.20
1.38
1.55
1.72
1.89
2.06
2.23
2.41
2.58
2.75
2.92
3.09
3.27
3.44
3.61
3.78
3.95
4.13
4.30
4.47
4.64
4.81
4.98
5.16
6.11
6.98
7.85
8.72
10.47
12.21
13.96
15.70
17.44
19.19
20.93
22.68
0.15
0.31
0.46
0.62
0.77
0.93
1.08
1.24
1.39
1.55
1.70
1.86
2.01
2.17
2.32
2.48
2.63
2.78
2.94
3.09
3.25
3.40
3.56
3.71
3.87
4.02
4.18
4.33
4.49
4.64
5.50
6.28
7.07
7.85
9.42
10.99
12.56
14.13
15.70
17.27
18.84
20.41
58
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Withdrawal = 0.2 MGD
Withdrawal = 0.5 MGD
0.2 0.4 0.6 0.8 1
SUI
II1!, LI
in in in mom o o tr»o o o
,Q ,_" CN ^ N- O O O
Basin Area (mi2)
• o
8
1
09
03
07
06
05
04
03
02
01
0.01
0001
tn in in tr> om o o in
o
O -
Basin Area (mi )
1
0.9
08
0.7
06
05
0.4
03
0.2
0.1
0.01
0001
Withdrawal = 1.0 MGD
Withdrawal = 2.0 MGD
Basin Area (mi2)
Withdrawal = 3.0 MGD
1
0.9
08
07
06
0.5
04
0.3
02
01
0.01
0001
o
O
v
.a
o
18
Basin Area (mi2
1
0.9
03
07
06
05
04
03
02
01
0.01
0.001
Withdrawal = 10.0 MGD
Basin Area (mi )
o
O
v
E
Q_
Basin Area (mi )
Figure 4-47. Heat maps showing probabilities of experiencing various surface water use intensity index (SUI)
values at a range of watershed sizes under several daily withdrawal scenarios based on streamflow records from
48 streams and rivers in Pennsylvania. Gray areas in the figures indicate SUI below 0.1. The SUI color scale used
for all withdrawal volumes is shown as an inset in the upper left box. (Data source: USGS 2014a.)
59
-------
Water Acquisition for Hydraulic Fracturing May 2015
Fig. 4-47 shows that SUI is a function of basin area as a surrogate for streamflow volume and illustrates the
probability of flow being less than critical flow dependent on withdrawal volume. We combined the 7,700
observations displayed in these figures and fit a multiple linear regression model to characterize the general observed
patterns in the six panels. The regression equation (4-2) has an adjusted R2 of 0.80:
Log(SUI) = 0.057 + 0.07*Withdrawal-0.525*Log(Area)-0.954*FlowProb Equation 4-2
where withdrawal is in MOD and basin area is in mi2 and the FlowProb is the probability of experiencing flows below
the Qcritical threshold over a lengthy period of record. A probability of 0.9 means that 90% of flows in the basin are less
than the observed flow. This equation will somewhat over-predict small values of SUI and under-predict large values
of SUI.
A practical formulation of the relationship between these variables is to ask, "What is the probability of exceeding a
threshold SUI at a site given its basin area with a certain withdrawal volume?" Another multiple linear regression
model was fit to the data to answer this question:
Prob(SUI > threshold) = 0.67 + 0.432*Log(Withdrawal) -0.46*Log(Area) -0.46*Log(SUI) Equation 4-3
The adjusted R2 of the model increased greatly when very low SUI values (O.01) and SUI equal to 1 were excluded.
Adjusted R2 for the clipped dataset was 0.85.
Water Quality Impacts from Hydraulic Fracturing Withdrawals
The SUI approach also has direct implications for interpreting the potential impact of withdrawals on water quality.
Water quality is determined by the concentration of pollutants within a volume of water; if the same amount of
pollutant is added to a smaller volume of water, its concentration will be higher. If a pollutant such as sediment or
nutrients is already in the stream, withdrawal of a volume of water for hydraulic fracturing will not impact water
quality, as the pollutant is removed from the water equal to its concentration. If a discharge point is in the immediate
area of a withdrawal location, the withdrawal will reduce the water volume and increase the concentration of the
discharged pollutant. This effect is "concentration magnification" (CM)—the opposite of dilution. Without knowing
the pollutant or its concentration, we can say it will be more concentrated in proportion to the withdrawal volume, and
CM will be more significant where flow is lower.
The SUI metric we have used throughout this report is quite conducive to assessing the potential for pollutant
concentration magnification, since it is a straightforward volumetric calculation. For those hydraulic fracturing
withdrawal locations upstream of point source discharges, the CM can be calculated using the SUI (the percent of
water removed from the stream at the hydraulic fracturing withdrawal site):
CM = 1/(1-SUI) Equation 4-4
The relationship between CM and SUI resulting from the equation is shown in Fig. 4-48. The CM approaches infinity
(no water for dilution) as the SUI approaches 1.0. When SUI is 0.5, pollutant concentration magnifies by 2 times
(CM=2).
The project quantified SUI for nearly 30,000 observed daily withdrawals of freshwater for hydraulic fracturing in the
SRB. Most sites included in this analysis are not directly influenced by a point source discharge. However, this
analysis uses these data assuming that each site could potentially be in this situation as a worst case scenario. Using
these data, we examined potential real-world CM values that would have resulted from those withdrawals. SUI
calculated for the 29,907 daily withdrawals from 2008 to 2013 was inserted into the CM equation (4-4). The
calculation was performed as observed with a passby low flow withdrawal restriction in place. The calculation was
also performed assuming no low flow restriction was in place, as was done for Fig. 4-24. Results are shown in Fig. 4-
49, with the CM expressed as a percentage above the baseline constituent concentration, assuming no withdrawal
occurred.
-------
Water Acquisition for Hydraulic Fracturing
May 2015
SUI was shown to be generally less than 0.1 in the SRB, even with no low flow passby restrictions in place (Figs. 4-
23, 4-24; Table 4-6). This translates to CM nearly always less than 1% above the baseline case. However, CM above
30% would have been observed in almost 10% of the observations, reaching up to 100% on occasion. The CM would
show the same relationship to basin area as an index of streamflow as SUI. CM values above 2 times in Fig. 4-49
were—not surprisingly—all in the very small streams. The passby cutoff limits also reduced potential impacts to
water quality by limiting CM to less than 20% above baseline.
100
OJ
CD
CO
c
o
4—'
CD
«! 10
c
QD
CD
O
Figure 4-48. General relationship
between the concentration
magnification and surface water
use intensity index, SUI) which
indicates the proportion of water
taken from a withdrawal
location.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Surface Water Use Intensity Index, SUI
Water Quality Concentration Magnification
30,000
27,116
o
OJ
J3
o
c
o
u
D Low Flow Passby Restriction
• No Low Flow Passby Restriction
Count Labels are for No Passby
3,655
1,800 i 350
540 300 180 130 90 85 60 70 420
V V
Water Quality Concentration Magnification
Figure 4-49. Potential magnification of water quality pollutant concentrations at observed hydraulic fracturing
withdrawals, assuming each site had an upstream point source discharge. The potential concentration
magnification was determined daily from the surface water use intensity index (SUI) determined by
streamflow and withdrawal volume at each site. Withdrawals ranged from 0.05 to 3.0 million gallons per day;
basin area of sites ranged from 1.7 to 10,547 mi2 (n = 29,907).
61
-------
Water Acquisition for Hydraulic Fracturing May 2015
Susquehanna River Basin Synopsis
Public water systems provided about 20% of the total freshwater used for hydraulic fracturing in the SRB as a whole
in 2011 and that volume has been declining significantly as serf-supplied water has become available. Use of serf-
supplied water taken from rivers and streams in the 17-county area of the SRB where hydraulic fracturing has been
active is essentially a new sector within the regional user portfolio as there is relatively little current use of this
resource for drinking water or other municipal or agricultural uses. Other users rely primarily on groundwater. As
these sources are developed through permitting, recent trends show reduced reliance on public water systems as
planned by SRBC (Beauduy 2009). Serf-supplied withdrawal sites are now widely distributed throughout the portion
of SRB active in O&G extraction. Hydraulic fracturing wastewater is reused to the extent that it is available,
contributing 13% of injection fluids used by O&G operators in the SRB.
Water acquisition by the O&G industry in the Susquehanna River Basin is managed by SRBC, which issues permits
to operators for individual withdrawal sites. PADEP performs a similar function outside the SRB, and regulates and
monitors gas well drilling procedures and activity. Permits include a variety of constraints related to how much, how
fast, and when water withdrawals can take place. SRBC permits assign daily withdrawal and pumping rate limits, and
set passby flow thresholds that cut off withdrawals during lower flow. The water management system operated by
SRBC relies on minimum passby flow calculations, referenced to real-time flow monitoring stations that provide
operators with timely information (via the Internet) to adjust operations.
The SUI approach used to evaluate hydraulic fracturing water withdrawals demonstrated that streams can be
vulnerable from typical withdrawals, dependent on their size as indexed by contributing basin area. Small streams that
could (and do) supply water to oil and gas operators have potential for high SUI for all or most of the year. Based on
measured flow records throughout the region, and dependent on the withdrawal volume, there is an increased
probability of higher SUI at average daily withdrawal volumes in watersheds less than 25 mi2. In the absence of a
passby, watersheds up to 600 mi2 have some probability of higher SUI during infrequent droughts and at higher
withdrawal volumes.
Groundwater resources are a major source of serf-supplied and community drinking water supply. They have been a
small component of O&G freshwater use. SRBC regulates groundwater sources requiring pump tests to ensure that
neighboring wells and potentially connected streams are not affected by requested pumping rates. Several self-
supplied groundwater permitted sites pump at fairly high rates, but well tests support these rates and there is no
indication of problems.
SRBC water management is designed to ensure water availability for all uses including municipal water supplies and
ecological communities. The system maintained very low surface water use intensity values at virtually all sites across
a range of flow volumes using simple hydrological predictive measures and data made available by USGS. Hydraulic
fracturing operations do not currently provide a challenge to public water supplies at a county or local scale in the
SRB, due to controls on the large-volume withdrawals and industry use patterns that have distributed self-supplied
water sources throughout a wide geographic area.
62
-------
Water Acquisition for Hydraulic Fracturing
May 2015
5. PICEANCE BASIN/UPPER COLORADO RIVER BASIN
Uinta-Piceance Geologic Province
The Uinta-Piceance geologic province covers about 40,000 mi2 in
northwestern Colorado and Utah. The hydrocarbon sedimentary basin
was deposited during the Cretaceous period 75 million years ago.
Water and eroded sediment flowed from the mountains west of the
present-day Rockies into a large epeiric sea that extended eastward
across what is the Great Plains today (Hettinger and Kirschbaum
2002) (Fig. 5-1). Fine-grained sediments laid down in this sea are
now the source of much of the shale gas and other hydrocarbon
resources found in the interior basin from North Dakota to Texas.
The Uinta-Piceance basin is made of sediments derived from
mountains and carried eastward by meandering rivers that deposited
coarser sediments in nearshore and coastal-plain environments. The
organic-rich sediments accrued in thick deposits in alluvial fans,
floodplains, beaches, and swamps as the coastal zones of the epeiric
seaway were gradually filled in with fine-grained muds (Johnson
1989) (Fig. 5-1). Overtime, the sandy shoreline migrated repeatedly
back and forth across the region, creating a thick
transgressive/progressive sequence of mudstones and sandstones (Fig.
5-3). Originally one large sedimentary basin, the unit divided into the
Uinta and Piceance Basins with the Laramide uplift at the Douglas
Creek Arch 30 million years later, shown in Fig. 5-2 (Pranter and
Sommer 2011). The area! extent of the Piceance Basin (locally
pronounced "PEE-awnce") is about 6,000 mi2.
Figure 5-1. Location of the Sevier orogenic
belt in relation to the epeiric seaway in the
Cretaceous period. Patterned areas
represent land masses. (From Johnson
1989.)
"""'(TWasatch
Plateau
nd Junction ^
M«» \ |MU G.mnlsol>
Delta
[a'-;
\Montrpse
COLORADO
Figure 5-2. Location of the Uinta-Piceance Province. (Map from USGS Uinta-Piceance Assessment Team 2003.)
63
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Southern Part of
Uinta Basin ut h ] Colorado
Southern Part of
Piceance Basin
Green
Format g
Wasatch Formation
Williams Fork
Castlegaie
<1>
V
5 243
s^l
lies
With burial and heating, the organic-rich sediments
formed a variety of recoverable hydrocarbons trapped
within the tight sandstones and mudstone organic-rich
sediments (Johnson 1989). Formations of various
sedimentary and hydrocarbon characteristics make up
the total petroleum system (USGS Uinta-Piceance
Assessment Team 2003), including the Mesaverde
sandstone group and the Mancos Shale (Fig. 5-3),
which both produce gas from unconventional
reservoirs. The lower part of the Mesaverde is
composed of blanket-like and near blanket-like
sandstone reservoirs, whereas mainly discontinuous
lenticular sandstone reservoirs deposited in fluvial
coastal plains make up the upper part of the
Mesaverde (Johnson 1989; Pranter and Sommer 2011;
Dietrich and Johnson 2013). The Mesaverde
sandstones thin to the east and interweave with the
Mancos Shale mudrock deposits that accumulated in
the offshore and open-marine environments of the
interior seaway. The Mancos Shale deposits occur at
shallower depths below the Piceance than in the Uinta
Basin. The Green River formation near the top of the
sedimentary sequence has kerogen-rich oil shale.
The clastic-rich reservoirs of the Piceance contain
enormous reserves of natural gas, oil, and gas liquids
in conventional and unconventional (continuous)
deposits (Johnson and Roberts 2003; USGS Uinta-
Piceance Assessment Team 2003). These include dry
gas and wet gas in unconventional tight sand and
shale deposits, coal methane, conventional oil, and oil
shale. The USGS Uinta-Piceance Assessment Team
(2003) assessed undiscovered conventional and
continuous (unconventional) oil and gas in the
Piceance (Table 5-1). The Piceance Basin contains one of the thickest and richest known oil shale deposits in the
world and has been the focus of most ongoing oil shale research and development extraction projects in the United
States (Johnson 1989). The U.S.EIA rated the Uinta-Piceance province as eighth in the United States in proven wet
natural gas and 10th for proven crude oil plus condensate reserves (U.S.EIA 2009). There are an estimated 64 trillion
gallons of in-place oil shale resources (USGS Uinta-Piceance Assessment Team 2003; Johnson 1989) of which 1.3
trillion may be recoverable (BLM 2006).
Blackhawk
Mancos
Shale
Utah j Colorado
Figure 5-3. Geologic formations in the southern part of the
Uinta and Piceance basins. (After Hettinger and Kirschbaum
2002.)
Table 5-1. Estimated hydrocarbon reserves in the Piceance Basin. (Data source: USGS Uinta-Piceance Assessment
Team 2003.)
Type
Undiscovered Reserves
Natural gas
21 trillion cubic feet
Gas liquids
Oil plus lease condensate
(oil shale)
Oil
43 million barrels
1.3 trillion barrels
598 million barrels
64
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Upper Colorado River Basin Background
The waterways of the Colorado River basin drain nearly 246,000 mi2 of largely semi-arid to arid lands before entering
the Gulf of California in Mexico (Fig. 5-4) (Bureau of Reclamation 2012). The Colorado River provides for the
drinking water needs of 40 million people and industrial and agricultural water use in seven states and Mexico,
including residents of Los Angeles, Phoenix, Tucson, Las Vegas, Denver, and Albuquerque. At the same time, the
river system irrigates nearly 5.5 million acres of crops and pasture. The region that relies on the Colorado River for its
water supply is one of the fastest-growing areas in the nation (Bureau of Reclamation 2005, 2012; GAO 2003), with
projected water deficits in coming decades (Bureau of Reclamation 2012; CWCB 2014). Water shortages may be
exacerbated, as climate change predictions suggest drying trends for this already mostly arid region (U.S. EPA 2013a;
Bureau of Reclamation 2012; CWCB 2014).
In 1922, the states of the Colorado River Basin signed the Colorado River Compact. The pact defined the upper and
lower basins of the river and apportioned 7.5 million acre-feet of water per year to each. Most of the water originates
in the upper basin states of Colorado, Utah, Wyoming, and New Mexico. The Colorado River is managed and
operated under numerous compacts, federal laws, court decisions and decrees, contracts, and regulatory guidelines,
collectively known as the "Law of the River" (Bureau of Reclamation 2012). This collection of documents apportions
the water and manages use of the Colorado River among the seven basin states and Mexico. There are increasing
concerns that the pact cannot be met, as
the agreed-on water volumes were set at
what history would prove to be a high
point in streamflow and low point in
population (Bureau of Reclamation 2012).
An important source of water in the
Colorado River system is the Upper
Colorado River Basin (UCRB) (Fig. 5-5).
The river originates in the rugged
mountains of the Southern Rockies in
Colorado that reach elevations of 10,000+
feet. The mountains capture precipitation
as snow in what is otherwise a semi-arid
climatic region (Spahr et al. 2000). The
snowpack fuels a robust recreation-based
economy in the headwaters. More
importantly, it produces most of the water
used by Colorado's population and
satisfies a significant portion of the needs
of the lower basin states. With
dependence on snowmelt, natural
streamflow is strongly seasonal. Some of
the snowmelt flows directly downstream,
while some is captured in large federal
Bureau of Reclamation and smaller
reservoirs in the headwaters for later
release. Water naturally flows westward
but some water is also piped through the
mountains for use by more than one
million people living in the cities of the
eastern slopes of the Rockies (Bureau of
Reclamation 2012; CWCB 2011; Spahr et
al. 2000).
Figure 5-4. The hydrologic boundaries of the Colorado River Basin
within the United States, plus the adjacent areas of the basin states
that receive Colorado River water. (Map source: Bureau of
Reclamation 2012.)
65
-------
Water Acquisition for Hydraulic Fracturing
May 2015
SOUTHERN ROCKY
MOUNTAINS
COLORADO PLATEAU
PHYSIOGRAPHIC
PROVINCE
r—i
' Utah M* i Colorado
Upper Colorado River Basin
Figure 5-5. The Upper Colorado River Basin in Colorado below Grand Junction, where the Upper
Colorado River joins the Gunnison River. Counties, major towns, and physiographic provinces are
shown. (Map modified from Spahr et al. 2000.)
Upper Colorado River
UINTA-PICEANCE
BASIN
Figure 5-6. Location of the Upper Colorado River and its basin above Grand Junction, where this project
focused assessment of hydraulic fracturing water acquisition. The Uinta-Piceance structural basin is shown
as blue/gray shading, with lighter shading where it intersects the Colorado River Basin. Hydraulic fracturing
wells are shown as circles. Yellow designates Parachute and Roan Creek tributaries where the surface
water use intensity is analyzed (Data source for well locations: FracTracker Alliance 2014.)
66
-------
Water Acquisition for Hydraulic Fracturing
May 2015
The Upper Colorado River flows westward, descending from its alpine headwaters, and enters the Colorado Plateau
physiographic province after it is joined by the Roaring Fork River near the town of Glenwood Springs (Fig. 5-5).
This is the northeastern portion of the Colorado Plateau, which extends southward to Arizona and New Mexico and
westward through Utah. The Piceance structural basin intersects the Upper Colorado River Basin in Colorado
westward from approximately the town of Rifle to DeBeque (Fig.5-2 and Fig. 5-6).
The Colorado Plateau is characterized by structural geology consisting of the nearly horizontal sedimentary
formations that have been uplifted thousands of feet since they were deposited in the Cretaceous period, as well as
occasional igneous intrusions. The general surface of the plateau at the modern-day river valley is 5,000+ feet above
sea level, and some of the uplifted plateaus reach nearly 10,000 feet (Hunt 1974). Within the UCRB, the Roan Plateau
found on the north side of the Colorado River and west of the town of Rifle is one of those features. The top of the
plateau is at 9,900 feet while the river base is at 6,500 feet (Fig. 5-7A). The drainage system is deeply incised and
forms steep-walled canyons exposing the sedimentary strata (BLM 2006). The land surface on top of the plateau is
relatively flat. The general position of the Roan Plateau is shown in Fig. 5-7.
Hydraulic Fracturing Drilling
Activity.
Hydraulic fracturing drilling activity
within the river basin has mostly
occurred in the area bounded to the
east by the town of Rifle and to the
west where the river passes the
Douglas Creek Arch (Fig 5-2). The
gas-bearing Williams Fork formation
illustrated in Fig. 5-3 is composed of
horizontally discontinuous sand
lenses within fluvial deposits that
have been the primary target of
directional drilling and hydraulic
fracturing (Fig. 5-7 upper panel).
Increasingly, O&G companies have
been horizontally drilling into the
Mancos Shale formation below the
Williams Fork. Hydraulic fracturing
wells have been drilled in the alluvial
valleys of the Colorado River and its
tributaries and on top of the Roan
Plateau.
The kerogen-rich oil shale deposits are
found in the Green River formation in
the uppermost strata exposed in the
Roan Plateau cliffs and forms steep,
500-to-l,000 ft cliffs and slopes (BLM
2006) (Figs. 5-7 lower panel and 5-8A,
B). The oil shale deposit extends
northward into the White River basin
in Rio Blanco County (Fig. 5-2).
10,000
Roan Plateau
7500
Colorado River
* Alluvium
0>
03
5000-
-5000
Miles
Figure 5-7. Geologic strata in Garfield County. Upper Panel: Schematic of
Garfield County's surficial geology and the recoverable gas formations of
the Williams Fork formation below the current valley surface, as well as
the exposed Green River formation, which bears kerogen-rich
sedimentary deposits. Lower Panel: Photograph of the Green River
formation. (Schematic after Dennison 2005; photo from Google Earth.)
67
-------
Water Acquisition for Hydraulic Fracturing
May 2015
A)
B)
Figure 5-8. Roan Creek Plateau. A) Aerial view of the Roan Plateau looking south across the top of the plateau
to the Colorado River and alluvial plain at the top of the photograph. The photograph looks into the Parachute
Creek subbasin, which joins the Colorado River in the upper right. The Grand Hogback monocline, a prominent
northwest trending feature that separates the Colorado Plateau from the White River Plateau, is in the upper
left. (Image from Google Earth; Landsat; ©Digital Globe.) B) Exposed bedrock outcrops on the southeast rim
include the upper portion of the Piceance Basin sequence, including the Green River formation. (Image from
Miamifittv.com.)
68
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Hydrology, Climate, and Land Use. The climate of the Colorado Plateau is semi-arid to arid and there is a general
shortage of water (Bureau of Reclamation 2012; CWCB 2007; GAO 2003; Hunt 1974). This area has annual
precipitation that ranges from 10 inches at lower elevations in the valley (which has little or no snow accumulation) to
approximately 25 inches atop the plateau, where enough snowpack develops to sustain a spring snowmelt season in
the tributaries (BLM 2006). Most of the population lives in towns in the alluvial valley along the UCRB and relies on
the river and its upstream reservoirs for its water supply (Fig. 5-9). The upper elevations of the plateau are managed
primarily by the Bureau of Land Management. Water from the river is withdrawn for municipal supplies and is also
shunted through natural and engineered structures, such as ditches and pipelines that route it to irrigated farmlands on
the mainstem and tributary valleys. River flow is highly seasonal, depending on snowmelt with occasional summer
storms
The shallow rock aquifers of the area are capable of yielding sufficient supplies for agricultural or domestic use, but
the water quality is variable (Robson and Banta 1995) and withdrawals within the Piceance Basin appear to be
minimal (Colorado Geological Survey 2003), although several thousand households in Garfield County have private
drinking water systems. Most municipal groundwater wells are located in the valleys and alluvium of the Colorado
River and its tributaries. The primary water use is irrigated agriculture within Garfield County, where most hydraulic
fracturing occurs.
•
Google eartff
Figure 5-9. Colorado River alluvial valley bottom in the Garfield County along the Interstate 70 corridor
between Rifle and Glenwood Springs. Irrigated agriculture is the dominant water use throughout the region.
Many of the semi-regularly distributed white dots are hydraulic fracturing well pads. (Image from Google
Earth 2014.)
The Upper Colorado River flows 230 miles from its headwaters to the Utah border, encompassing a land area of
17,800 mi2. Interstate 70 traces its path along most of the river's length in this area. The counties most dependent on
the river—and at the same time experiencing hydraulic fracturing resource extraction—are Garfield County, with a
population of about 56,000, and Mesa County, with a population of 146,000 (including Grand Junction, the largest
city in western Colorado). Water must be passed downriver through Garfield County to other users, including Grand
Junction, for municipal and agricultural use, low-head hydropower and endangered species habitat (CWCB 2007;
Bureau of Reclamation 2012). Total daily water use for the UCRB from headwaters to Grand Junction (Division 5) is
shown in Fig. 5-10. By far the largest users are irrigation, power generation from a low-head dam that does not
consume water, and transfer of water to the east slope cities. Most of the O&G resource extraction occurs in the
Colorado River valley and on the Roan Plateau on the north side of the river; some occurs on the south side of the
river (Fig. 5-6).
69
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Daily Water Use in Division 5--Upper Colorado River (MGD)
Power
Generation,
1384
Evaporation, 36
rrigation, 1505
Miscellaneous
267
Livgs^ock, 23
Fisheries and _
Wildlife, 133 Domestic, 4
Municipal, 62
Figure 5-10. Upper Colorado River (District 5) daily water use by
sector in 2012. A number of categories were grouped in the
"Miscellaneous" category. (Data source: CODWR 2012.)
Water Allocation and Regulation. The
Colorado Water Courts allocate water based
on a prior appropriations doctrine and water
is managed by the State Engineer in the
Department of Water Resources (CODWR)
in the Colorado Department of Natural
Resources (Grantham 2011; CWCB 2014).
The Colorado Water Conservation Board
(CWCB) is an agency with responsibilities
for water conservation, flood mitigation,
watershed protection, stream restoration, and
water supply planning. Courts decree a water
right to an applicant for specified beneficial
use(s) and volume. Priority is awarded based
on date of appropriation. Later rights are
junior to earlier rights and may not receive
their appropriation in times of shortage until
senior needs are filled. A water right can be
transferred to another owner maintaining the
original priority. Users include irrigation,
municipal supplies, hydropower, livestock,
industrial, commercial and endangered
species. Municipalities hold rights like other
users and may also augment supplies from
the Bureau of Reclamation reservoirs by
reserving volumes through contract.
Fresh surface water used by the hydraulic fracturing industry can be obtained from allocations on the mainstem
Colorado River or its tributaries. Water can be purchased from large federal reservoirs and delivered via the Colorado
River to a collection depot, from where it is transported to its final destination. This transaction may occur through a
third party such as state-sanctioned water conservancy districts or private owners, as long as industrial use is defined
as a use at the withdrawal point (a "diversion" structure). Since the 1980s, O&G companies have collectively acquired
access to large allocated volumes of water and contracts held throughout the Piceance play (URS 2008). In this
engineered system, water can be reallocated within the infrastructure among local and distant sources, allocated back
and forth between tributaries, the river, and small private reservoirs distributed throughout the region, or transferred
from structure to structure.
Irrigation has historically been the largest water user in this basin and throughout Colorado (Fig. 5-10). Withdrawal
reporting at most structures occurs only during the irrigation period (mid-April to end of October) when the water is
used. Some structures such as municipal structures report use year-round. Water used for hydraulic fracturing is not
separated from other industrial uses.
Data Sources. CODWR tracks water use at the structures where water is taken. There are more than 18,000 active
acquisition locations (structures) in the upper Colorado River basin. Every use location is identified, no matter how
small. CODWR generates water use records on a daily, monthly, or annual basis (dependent on source) (CODWR
2014a). The CODWR online database was accessed for information on water allocation and use throughout Divisions
4, 5, and 6, with a focus on subbasins and the Upper Colorado River (Division 5) within Garfield County where
hydraulic fracturing has been most active (CODWR 2014b,c,d). Individual database queries are listed in Table 5-2.
The water acquisition system and the CODWR database that tracks it are very complex, and there are challenges to
quantifying O&G industry water acquisition. Some of the structures where water is acquired are owned by O&G
companies and can be identified in the data system, while others sourced from contract purchases cannot be readily
identified. The O&G industry is designated as industrial use the same as other industries. Primary data sources for
water use and hydraulic fracturing wells are listed in Table 5-2.
70
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Well drilling practices are regulated by the Colorado Oil and Gas Conservation Commission (COGCC). The agency
also monitors O&G resource production and well status. Since 2010, COGCC has required the O&G industry to
report injection volume and chemicals used at individual wells in the FracFocus chemical disclosure database
(COGCC 2012; FracFocus 2014). Well counts and produced water volumes were obtained from COGCC. Well fluid
volume use was obtained from the FracFocus database (FracFocus 2014).
Table 5-2. Water use data sources for the Upper Colorado River Basin.
Agency/Organization
Colorado Division of
Water Resources
(CODWR 2014)
Colorado Oil and Gas
Conservation
Commission
(COGCC 2014)
FracFocus (2014)
Description
b. Water rights
information
c. Structure
information
d. Structure water
use
e. StateMod water
planning program
a. Well starts and
completions
b. Produced Water
Well fluid volumes
Source/Query
http://cdss.state.co.us/onlineTools/Pages/WaterRig
hts.aspx
http://cdss.state.co.us/onlineTools/Pages/Structure
sDiversions.aspx
http://cdss.state.co.us/onlineTools/Pages/Structure
sDiversions.aspx
(query structures for diversion reports)
http://cdss.state.co.us/Modeling/Pages/SurfaceWat
erStateMod.aspx
http://cogcc.state.co.us
Query: staff report
http://cogcc.state.co.us/COGCCReports/production.
aspx?id=MonthlvWaterProdBvCountv
http://www.fracfocusdata.org/DisclosureSearch/
(query by county, look at individual well reports)
Data Use
Priority, decreed use
Locations, history,
ownership
Daily, monthly, annual
volumes used
Scenario analysis of use
and structure priority
Well counts
Estimates of hydraulic
fracturing wastewater
reuse
Total well consumption,
counts, timing
Sources of Freshwater for Hydraulic Fracturing
Development of unconventional gas reserves in the Piceance has been ongoing for almost two decades, but the pace
has increased rapidly since 2000 (Hill 2013). Gas extraction using hydraulic fracturing has occurred in CODWR
Division 5 (the Upper Colorado River from headwaters to Grand Junction), Division 4 (the Gunnison River basin),
and Division 6 (the Yampa/White Rivers primarily in Rio Blanco County). Eighty-five percent of the wells have been
drilled in Garfield County and Mesa County within about 30 miles north and south of the Upper Colorado River. Well
starts by year are shown in Table 5-3 based on data provided by COGCC (2014a). Hydraulic fracturing has also been
active in the While River Basin in Rio Blanco County in CODWR Division 6.
Annual hydraulic fracturing activity increased steadily in this area after 2000, peaking in 2008 when 1,688 wells were
started in Garfield County (Fig. 5-11). The drilling rate has recently declined to only 390 in 2013. This decrease
coincided with a shift in drilling from dry gas to liquid-rich reservoirs in central and eastern Colorado and a relative
increase in the price of oil compared to the price of natural gas during this period (U.S. EIA2013b).
Table 5-3. Hydraulic fracturing well starts by county. (Data source: COGCC 2014a.)
CODWR
County Division 2000
Garfield
Mesa
Routt
Gunnison
Delta
Montrose
Rio Blanco
5
5
4
5
4
4
e
190
2
5
0
0
0
51
2001
251
12
12
0
0
3
82
2002 2003 2004 2005
245
12
1
1
0
0
47
417
13
0
0
5
2
83
585
25
1
1
4
1
92
799
89
6
1
6
0
95
2006
1,005
156
3
9
5
0
107
2007
1,304
209
2
5
2
1
95
2008
1,688
222
0
1
0
2
203
2009 2010
768
14
2
4
0
0
116
904
1
1
5
4
1
107
2011
879
39
2
2
1
0
72
2012
495
4
4
4
6
0
53
2013
392
6
2
1
0
0
36
71
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Garfield County
12,000
, 10,000 -
8,000 -
X
"5
i_
CD
.a
£
6,000
4,000
2,000
o
o
o
rxl
rxl
O
O
rxl
O
O
rxl
O
O
rxl
00
O
O
rxl
O
rxl
O
rxl
This project focused analysis where hydraulic fracturing
activity is centered in the UCRB (Division 5) and Garfield
County within the division, where almost 10,000 wells have
been drilled; see Figs. 5-6 and 5-11. However, data were
examined from other water divisions in the Piceance in
Colorado and are discussed as relevant.
Water Used For Fracturing Wells
There is no direct reporting of how much freshwater is used
for hydraulic fracturing gas wells in either oil and gas or
water use databases managed by the state of Colorado
(Table 5-2), although these data sources could be used to
estimate the freshwater consumption for hydraulic fracturing
activity. These estimates were supplemented by information
provided by oil and gas operators to federal agencies in this
area (USFWS 2008), including interviews conducted for this
project. We use the term "freshwater" to distinguish from
hydraulic fracturing wastewater. The term "freshwater" does
not infer that water obtained from local supplies is of
drinking water quality and may include untreated, brackish,
saline, or contaminated water.
Figure 5-11. Annual and cumulative hydraulic
fracturing well starts in Garfield County since 2000.
(Data source: COGCC 2014a.)
O&G companies have reported that 100% of the water used
in the hydraulic fracturing process to stimulate gas wells in this area is hydraulic fracturing wastewater (BLM 2006;
USFWS 2008). Freshwater is only used for drilling and associated activities. High reuse rates are possible because
nearly all of the hydraulic fracturing fluid (80% to 100%) injected into directionally drilled tight gas wells in the
Williams Fork formation returns to the surface within the first few months after fracturing, according to local
operators and agencies (BLM 2006; USFWS 2008; WPX Energy, onsite interview, January 8, 2014). Returned
hydraulic fracturing wastewater is of relatively good quality for industrial use and the industry captures, treats, and
reuses it in other wells. In addition, the Piceance tight sands have naturally high water content (Johnson 1989) and
formation water continues to flow from each producing well over time. COGCC tracks volumes of produced water;
we queried COGCC (2014b) by county to quantify this volume. Each producing well in Garfield County returns
approximately 140,000 gallons per year (0.43 ac-ft).
The freshwater volume needed for hydraulic fracturing in Garfield County was estimated with well counts and a set of
assumptions on required injection volumes and water available for reuse as reported by the industry. Given the
reported high reuse rate, we start from the assumption that all wells are fractured with reused hydraulic fracturing
wastewater if it is available. Any shortfall in water needed for hydraulic fracturing must be supplied from freshwater
sources. We quantify water use for hydraulic fracturing based on the following data and assumptions.
Injection Fluid Volume Needed for Hydraulic Fracturing. The volume of water needed for hydraulic fracturing
depends on the number of directional and horizontal wells and the average volume used per well. For the latter, there
is variability depending on each company's drilling strategies (targeted formation, depth, and so on.)
• Injection volume: Data obtained from FracFocus 1.0 disclosures (U.S. EPA 2015) was used to determine
fluid injection volume per well for wells drilled from 2011 to 2013 in Garfield County. FracFocus 2.0 was
directly accessed by this project for wells drilled in 2010. We assumed that numbers reported from 2010 to
2013 were representative of prioryears. The injection volume generally ranged from2 to 3 MG (6.1 to 9.2
ac-ft), averaging 2.4 MG (7.37 ac-ft).
• Number of wells: For the number of wells hydraulically fractured each year, we used well starts provided by
COGCC (2014a), as listed in Table 5-3.
Available Hydraulic Fracturing Wastewater. The available hydraulic fracturing wastewater for Garfield County as
a whole was computed as the sum of flowback and produced water. Any surplus was carried over to the next year.
72
-------
Water Acquisition for Hydraulic Fracturing May 2015
Operators report that drilling directional wells and associated development activities use 0.25 MG (0.77 ac-ft) (BLM
2006; USFWS 2008; URS 2008). WPX Energy has started to drill horizontal wells into the Mancos Shale in recent
years—they report that 1.05 MG (3.2 ac-ft) are needed for drilling these deeper and longer wells (WPX Energy, onsite
interview, January 8, 2014).
• Flowback water: We computed the pool of flowback water available each year by assuming a proportion of
flowback per well (80% to 100%) multiplied by the number of wells drilled that year. In the calculation, we
offset this volume by six months to allow time for treatment. We did not know the exact length of time
wastewater spends in storage, but this assumption somewhat improved our fit to observed data. Assumed
percentage was 70% until 2002, 80% in 2003, 90% in 2004, and 100% thereafter to reflect improving
technology and infrastructure development.
• Produced water: Produced water volume was reported by COGCC for Garfield County (COGCC 2014b).
This number represents the cumulative number of producing wells in the county and grows each year. The
volume available in 2000 was 26 MG (80ac-ft), climbing to 1,700 MG (5,200 ac-ft) by 2013.
Freshwater Volume Needed for Hydraulic Fracturing. The necessary volume of freshwater was the sum of water
used for hydraulic fracturing and the water used for drilling and associated activities.
• Hydraulic fracturing injected freshwater: Each year, the required volume of injection fluid needed was
computed from well count and average total injection volume. The volume of available hydraulic fracturing
wastewater was computed based on well count and produced and flowback water volume. In the calculation,
the wastewater was put into the wells first. Any deficit between required volume and wastewater was
fulfilled with freshwater.
• Drilling and associated activities: All drilling and associated activities required freshwater of 0.25 MG (0.77
ac-ft) for each directional well and 1.05 MG (3.2 acre-feet) for each horizontal well as reported by O&G
companies.
We combined these factors to estimate the total volume of hydraulic fracturing wastewater and freshwater used in
Garfield County annually; results are shown in Fig. 5-12. Total annual injection volume at the peak of drilling in 2008
was almost 4,500 MG (14,000 ac-ft); freshwater was a small part of the total. Freshwater was needed for hydraulic
fracturing until 2005, when a surplus of hydraulic fracturing wastewater developed (resulting from the cumulative
increase in wells producing hydraulic fracturing formation water, combined with the large volume of flowback). There
seems to have been enough hydraulic fracturing wastewater to accommodate the high drilling rates in 2008. Surpluses
have continued to grow as drilling rates have declined. Freshwater has made up less than 10% of the annual volume of
fluids used to drill and fracture primarily directional wells and has been used only for drilling since 2006. This project
did not determine whether more freshwater will be needed for horizontal drilling into shales due to any technological
differences other than longer wellbores. These estimates indicated how much freshwater must be acquired from local
sources.
Water Acquired for Hydraulic Fracturing
We determined the volume of water acquired by the O&G industry for hydraulic fracturing by querying the CODWR
structure database (CODWR 2012b, 2012c). Each structure has a water rights allocation, a history of activity pertinent
to the allocation, and data on water use at the structure. Each structure has an owner and one or more rights allocation
with one or more designated uses and a daily or annual volume limit. This system is highly complex; only the most
salient elements of use designation and allocation were used and discussed in this report. CODWR maintains a
database of water taken from each structure that can be accessed using the "structure" query. Beginning there, we
iteratively queried the database looking for water supplied to industrial users (the category to which use by the O&G
industry is assigned) and/or owners that could be identified as an O&G company or provider by name. Eventually the
search widened to other areas and structure ownership; water was not transported between major river basins at
present as trans-basin transfers are also tracked. Note that water acquired at an O&G-owned structure is not
necessarily used for hydraulic fracturing. Many of these structures are still used for irrigation.
73
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Garfield County
01
E
000
500 :
000 :
500 :
000 :
500 :
000 :
500 \
000 :
500 :
0
I Recycled Water for HF
I Freshwater for HF
Freshwater for Drilling
...ll
h
16,000
14,000
12,000
10,000
8,000
01
6,000 <
4,000
2,000
0
Figure 5-12. Estimated use of water for hydraulic fracturing natural gas extraction in Garfield County.
The total includes reused hydraulic fracturing fluid and freshwater used for hydraulic fracturing and
for drilling. (Data sources: COGCC 2014a, 2014b; FracFocus 2014 for individual well injection volume.)
0)
E
TO
3
C
500
450 -
400 -
350 -
300 -
250 -
200 -
150 -
100 -
50 -
0
Freshwater Acquisition for HF-Division 5
Water Supplied
Freshwater NEEDED
1,400
1,200
1,000
- 800
OJ
OJ
600 <
400
- 200
0
Figure 5-13. Estimated freshwater use in Upper Colorado River Division 5. Water supplied was
the freshwater volume accounted for in the Colorado Division of Water Resources structure use
monitoring database. Water needed was the total volume of freshwater estimated in Fig. 5-12 as
blue portions of the bars. (Data source: CODWR 2014d.)
74
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Later sections of this report analyze hydraulic fracturing water withdrawals in more detail; here, we characterize the
total volume of freshwater acquired for hydraulic fracturing within the search area. Fig. 5-13 shows volumes of
freshwater accounted for at withdrawal location and volumes needed for hydraulic fracturing activities (blue areas at
the top of the bar chart in Fig. 5-12). We were able to account for a volume of water each year that was close to
hypothesized based on the assumptions of hydraulic fracturing wastewater reuse as reported by hydraulic fracturing
operators (Fig. 5-13). The accounting was better in some years than others, disagreeing by more than 50% in 2000 and
2002. Water can be obtained from third party contracts such as with the Conservancy Districts that acquire water
reserved in the BLM reservoirs. This water is not tracked by individual users in the Colorado Water Resources
database (CODWR 2014d) and any water supplied from these sources would contributed to differences between
estimated water supplied and needed in Figure 5-13. Some water from this source was accounted for based on
documents available at http://www.wdwcd.org/content/library. . For the 14-year period as a whole, 86% of the
freshwater demand for hydraulic fracturing wells was accounted for at water withdrawal locations. These results
support industry reports that 100% of the hydraulic fracturing injection fluid is reused hydraulic fracturing
wastewater.
Note, however, that we generally have less confidence in the O&G water need and acquisition estimates in this case
study area. Uncertainty is greater due to lack of direct data on freshwater use in hydraulic fracturing wells, as well as
inclusion of hydraulic fracturing water with all other industrial uses in the CODWR water use data. Estimates appear
to be the right order of magnitude, but may be biased low. Average annual freshwater acquisition by the O&G
industry from 2000 to 2013 was 200 MG per year (600 ac-ft). In the peak year 330 MG (1000 ac-ft) were acquired in
Division 5. This value is low relative to shale plays and compared to irrigated agriculture in the area where most of the
O&G water is acquired.
Freshwater Sources for Hydraulic Fracturing
The Piceance gas fields within the UCRB are found mostly within 20 miles north and south of the Colorado River
(Fig 1, upper panel and Fig. 5-14) as it flows westward between Glenwood Springs and Grand Junction. Much of the
hydraulic fracturing activity in Garfield
County occurs within the northern
tributaries including Parachute and Roan
Creeks that together comprise an area of
about 800 mi2. These subbasins also
contain rich oil shale deposits in the
Piceance formations overlying the deeper
unconventional gas reservoirs currently
under development.
Land use in the alluvial valleys along the
Colorado River and its tributaries has
traditionally been irrigated agriculture, as
continues today in Roan Creek. Parachute
Creek has become a hub for natural gas
extraction with numerous well pads and
wastewater treatment facilities now
intermixed with agricultural lands. Water
sources in Garfield County, where hydraulic
fracturing is most active, are the Colorado
River and its major tributaries—mainly on
the north side of the river draining the Roan
Plateau (Fig. 5-14). Groundwater resources
are primarily found in the alluvial valleys
associated with the rivers and streams.
Public water supplies are taken from the
Colorado River for the most part, where
they may also pick up reservoir water. The
20
20
Miles
EXPLANATION
Household wells in DWR
database 1972-2012
Catchment betw. USGS
gages Glenwood Springs
& Cameo
Quaternary alluvium
Municipal Water
System
Figure 5-14. Location of drinking water sources including public water
supplies and private groundwater wells in the Upper Colorado River
Basin between Glenwood Springs and Cameo west of DeBeque (CODWR
2014f) and municipal wells (U.S. EPA 2012c) in the area of primary
hydraulic fracturing drilling activity. (Base map from CGS 2003).
75
-------
Water Acquisition for Hydraulic Fracturing
May 2015
O&G industry primarily self-supplies freshwater from a mixture of surface water and groundwater sources.
The alluvium of the Colorado River between Glenwood Springs and Cameo, where most of the population in Garfield
County lives, is a primary groundwater source. There are many private and a few municipal groundwater wells that
are not used for drinking water supply (Fig. 5-14). Just one municipal system pumps 0.9 MOD of groundwater from
the alluvium of the Colorado River at 64% of facility pumping capacity (Hill 2013), but this water is not used for
drinking water supplies. Some of the private wells are industrial gravel pits that fill with water seepage between the
river and alluvium. The water allocation system treats a gravel pit as a groundwater tributary to the Colorado River, so
water diverted for the O&G industry require plans for replacement of any depletion to river flow through
augmentation plans (COGCC 2012). Gravel pits are designated for industrial use and any sales to O&G from these are
not separately documented. We received anecdotal information from operators that some water has been obtained
from one or more of these gravel pits, but we were unable to discern hydraulic fracturing use from CODWR use
records.
Public Water Supplies. It appears from the CODWR database that hydraulic fracturing operators do not obtain
water from public water suppliers. Consequently, the discussion of public water supplies and drinking water sources
in this area is brief. Municipalities hold rights for water withdrawals from structures having designated beneficial
uses, priority, and volume limits like all other users. They often have allocations at more than one location to supply
water for a variety of municipal, commercial, and industrial uses. Many have water reserved in the Bureau of
Reclamation reservoirs to augment supplies during shortages. We focus on the small municipalities of Parachute (pop.
1,095) and Battlement Mesa (pop. 4,500), which are co-located at the confluence of the Colorado River and Parachute
Creek, while DeBeque (pop. 500) is located at the confluence with Roan Creek (Fig. 5-14). Each of these towns has
one structure that appears dedicated primarily to drinking water supply: each draws from the Colorado River. The
town of Parachute also augments its total supply from springs (Hill 2013).
The combined annual water intake for the three municipal suppliers is shown in Fig. 5-15. Water supply to these three
municipal structures was sufficient to provide for domestic water use (100 to 200 gallons per person per day);
augmentation from the reservoir was needed during the dry period in 2012-2013 when flow in the Colorado River in
2012 was the third lowest since records began in 1936. The reservoirs were generally called on to augment water
supplies for many users throughout the region, including 25% to 44% of the annual use for these municipalities. These
observations demonstrate the importance of the reservoir system, which was designed to sustain water supplies,
especially during drier periods. According to records, none of the three municipal drinking supply structures provided
Combined Municipal Facilities
Vicinity of Roan/Parachute Creeks
700
600
n? 500
- 400
300
200
100
Colorado River • Reservoir
I'
2000
- 1500
1000
- 500
2009
2010
2011
2012
2013
Figure 5-15. Annual water volume acquired for drinking water
at three municipal public water systems in the vicinity of
Parachute and Roan Creeks, where current hydraulic
fracturing drilling activity is focused. Water was obtained
from the Upper Colorado River or augmented from Bureau of
Reclamation reservoirs during low flow in 2012 and 2013 to
satisfy per capita demands (shown as dotted horizontal lines
set at 100 and 200 gallons per day). There was no acquisition
of water bvO&G at these municipal water suoolv structures.
76
water for hydraulic fracturing industrial operations
during the study period.
Self-Supplied Sources. CODWR structure records
in Divisions 4, 5, and 6 were searched to identify
locations with industrial use (use 4 in the CODWR
system). In doing so, we found that a number of
O&G companies collectively have rights to a sizable
volume of water at 449 structures within the
Piceance structural basin that includes Divisions 5
(Colorado River basin) and 6 (Yampa River Basin).
Many are found in Parachute and Roan Creeks
tributaries coincident with the Parachute gas field
where much of the hydraulic fracturing is ongoing
(Figs. 5-6 and 5-16). O&G water allocations were
acquired in the late 1970s and 1980s to supply the
large volumes of water necessary to extract the oil
from kerogen in the oil shale deposits (AMEC
2011; URS 2008), the richest of which are in the
Green River formation within the Roan Plateau
(BLM 2006). These water allocations were acquired
to supply projected water needs for oil shale
extraction that could reach high end projections of
-------
Water Acquisition for Hydraulic Fracturing
May 2015
40,000 to 123,000 MG (120,000 to 399,000 ac-ft) per year, depending on scale of operations and technology advances
(AMEC 2011; URS 2008).
Hydraulic fracturing operators obtain water
at various locations along the Colorado
River. Acquisition sites include structures
with O&G owners,as well as collection sites
shared among multiple users and where
obtained by contract from organizations such
as the West Conservancy District (WCD).
The conservancy districts are Colorado
governmental entities whose primary service
is to provide water right augmentation within
their service areas from supplies reserved in
the upstream reservoirs for this purpose.
They help those without access to enough
water in the water allocation system to
obtain supplies. Some of this water is picked
up at five water depots that draw from the
Colorado River, distributed along Interstate
70 interchanges in Garfield and Mesa
Counties. Hydraulic fracturing operators and
trucking companies have obtained water
through this source. We included estimates
of water supplied through contracts based on
archived materials from the WCD website
(http://www.wdwcd.org/), although we were
not able to obtain this data directly. In
general, water withdrawn from the Colorado
River was more difficult to track, because it
generally is acquired through various third
party sources. As planned within the system,
the volume of water from the Colorado River
would have come from the reservoirs and was
insignificant relative to the volume of water
in the river.
Municipal withdrawals
A O&G withdrawals
- HF wells
Figure 5-16. Parachute and Roan Creek watersheds. Completed
gas wells and oil and gas water withdrawal locations are
shown. The location of the municipal public water systems
discussed relative to Fig. 5-15 are also shown but note that
they are not used as a water source for hydraulic fracturing.
(Data source: gas wells from FracTracker Alliance 2014; water
acquisition locations from CODWR 2014b, 2014c.)
The CODWR water use data reports the
volume of water taken for the designated uses
at each structure separately. Each structure with assigned industrial use was examined for records of withdrawals. It
appeared that irrigators and municipal suppliers provide little water, if any, to hydraulic fracturing operators, as their
monitoring records showed no industrial use.
Although hydraulic fracturing activity was distributed throughout Garfield County, 50% or more of the freshwater
was obtained from tributary streams and groundwater wells within the Parachute Creek watershed (198 mi2) (Fig. 5-
17). The location of Parachute and Roan Creeks within the UCRB is shown in Figs 5-6 and 5-14, and the distribution
of hydraulic fracturing wells and withdrawal locations is shown in Fig. 5-16. Water acquired in Parachute Creek is
taken from O&G owned structures. However, a very small fraction of O&G allocated water is withdrawn relative to
rights, and a handful of structures are used by hydraulic fracturing operators. It appeared that Parachute Creek was
used steadily, while the Colorado River was tapped more heavily when drilling rates were higher. There has been no
industrial use of water in the Roan Creek watershed to date, although structures there are also part of the oil shale
water delivery system (Stantec 2013).
77
-------
Water Acquisition for Hydraulic Fracturing
May 2015
CODWR Division 5
450
1400
Parachute Creek Colorado River
Figure 5-17. Sources of freshwater acquired for the oil and gas industry in the Upper Colorado River Basin. All
water was acquired in Garfield and Mesa Counties from miscellaneous distributed sites on the Colorado River
mainstem or from structures within the Parachute Creek watershed. (Data source: CODWR 2014d.)
Parachute Creek Water Use. Land use in the alluvial valleys along the Colorado River and its tributaries has
traditionally been irrigated agriculture; this continues today in Roan Creek. While irrigated agriculture is still active in
the Parachute Creek valley, this area has also become a hub for hydraulic fracturing drilling, industrial operations, and
water acquisition within Garfield County and the Piceance Basin (Fig. 5-18). The O&G industry has built water
treatment facilities to clean hydraulic fracturing wastewater for reuse in hydraulic fracturing wells. There are also
groundwater well complexes, especially in the middle and upper reaches of the valley. At the same time, irrigation
farming remains active in the lower valley.
Water acquisition sites, including irrigation ditches and reservoirs, are pictured in Fig. 5-18. The O&G industry
primarily takes water from several small reservoirs that intercept the main tributaries in the upper valley with about
850 MG (2,600 ac-ft) of storage capacity, and from a groundwater well midway up the valley within the industrial
complex shown in Fig. 5-18C, located above most irrigation. There are nearly 100 more O&G-owned small instream
locations in the Parachute Creek headwaters; it is unknown if any hydraulic fracturing operators have obtained water
from them, as use is not tracked in the CODWR database.
Most of the water in Parachute Creek is diverted from the stream into ditches and used for irrigation (Fig. 5-18A).
Only a small fraction of the water that could be taken from O&G-owned diversion structures in Parachute Creek has
been used for hydraulic fracturing (Fig. 5-19).
78
-------
Water Acquisition for Hydraulic Fracturing
May 2015
A)
B)
C)
v
Figure 5-18. Parachute Creek water acquisition sites. A) Irrigators divert water from the mainstem of Parachute
Creek through headgates into ditches that carry it along the sides of the valley on each site of the stream. B) Small
reservoirs in the upper valley store water from the main tributaries of Parachute Creek. C) A groundwater
wellfield and reservoir complex in the upper valley at the junction of three main tributaries provides most of the
water for hydraulic fracturing. (Images from Google Earth.)
79
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Parachute Creek Annual Water Diversion
to
c
_o
ro
O
c
O
il
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Parachute Creek
Figure 5-20. Daily streamflow and collective withdrawal volume summing 29 structures withdrawing water
from surface water structures in Parachute Creek. Streamflow was estimated using HSPF model, with
empirical fitting to observed streamflow. The vertical axis is truncated to emphasize low flows. (Data source
for water withdrawals: CODWR 2014d.)
Parachute Creek
Figure 5-21. Surface water use intensity index, SUI for combined withdrawals at 29 primary ditch sites in
Parachute Creek. Most water is used for irrigation. (Data source: CODWR 2014d.)
81
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Water Use Intensity Analysis at Self-supplied Sites
During the spring snowmelt, streamflow peak ranged from 3 50 MOD (540 cfs) in the very wet year of 2011 to a low
of 35 to 37 MOD (77 cfs) during the very dry year of 2012. Streamflow descended to just 0.7 MOD (1 cfs) by October
(Fig. 5-20). Total basin withdrawals accounting for structure efficiencies peaked at about 6 MOD and exceeded
streamflow at times. While there was abundant water during the winter and spring months, shortages develop during
most summers, at times taking much of the streamflow as shown by the SUI for this period (Fig. 5-21). In 2012,
withdrawals exceeded streamflow, and records show that water supplies were imported into the subbasin from Ruedi
Reservoir via the Colorado River.
SUI values computed from daily records at each surface water withdrawal structure are shown in Fig. 5-22. Three
structures supply water to hydraulic fracturing, including two sites at the left of the figure (orange shaded boxes) and
one in the upper valley at about 20 mi2 (small range bar between sites at 14 and 34 in Fig. 5-22).These three hydraulic
fracturing withdrawal sites are small reservoirs located in the Parachute Creek valley upstream of irrigators (Fig. 5-
ISBandC). SUI at irrigation withdrawal sites ranged from about 0.01 to 1.0, with a median of 0.05. SUI did not
decrease with basin area (as observed in the Susquehanna River Basin), because withdrawal tends to be greater in the
lower portions of the watershed and withdrawals increase faster than streamflow accumulates downstream. Median
SUI in the reservoirs in the upper part of the basin used for hydraulic fracturing water supply ranged from 0.01 to
0.25, calculated as withdrawal relative to inflow to the reservoir and not accounting for the storage that mitigates
effects. Counts by SUI category on all withdrawal days for all structures combined from 2008 to 2013 are provided in
Table 5-4: 55% were less than 0.1 while 16% of sites exceeded 0.4 and 3.1% were near 1.
LO
x"
OJ
_c
^
'in
c
OJ
OJ
l/l
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Table 5-4. Count of daily surface water use intensity index at active Parachute Creek withdrawal
structures from 2008 to 2013 (n = 16,012). Most withdrawals are for irrigation.
Surface Water
Use Intensity
Index (SUI)
0-0-0.10
0.11-0.20
0.21-0.30
0.31-0.40
0.41-0.50
0.50-0.6
0.61-0.70
0.71-0.80
0.81-0.90
0.91-1.0
N
8,825
3,381
1,263
639
536
300
263
155
161
489
%
Observations
55.1%
21.1%
7.9%
4.0%
3.3%
1.9%
1.6%
1.0%
1.0%
3.1%
Groundwater Well Analysis. About 40% of the freshwater obtained in Parachute Creek is taken from a groundwater
wellfield located at the confluence of the west, middle, and east forks of Parachute Creek (Fig. 5-18C). Well Number
1 (structure 395298) was active for 2007-2010, and Well Number 1A (structure 395301) was pumped from 2011-
2012. The driller's logs report that the pumps were rated at 236 gallons per minute (124 MG per year), and the wells
were drilled to a depth of 57 feet. The structure permit capacity (decreed) is 80 MG per year (246 ac-ft).
The pumping wellfield receives its water from the outwash aquifer and the nearby creeks. The state of Colorado
considers this a tributary wellfield of Parachute Creek and ultimately of the Colorado River, and thus under prior
appropriation (Grantham 2011). The groundwater use intensity index (GUI) based on pumping capacity for this well
varied from 0.5 to 0.7 in most years (Fig. 5-23). The groundwater modeling (GFLOW) and mapping (Surfer™)
technology were used to calculate and map the cone of depression and the streamflow capture about the pumping
center (Fig. 5-24), as performed in the SRB. The building of the GFLOW model for the Parachute Creek confluence is
described in detail in Appendix C. The GFLOW model was parameterized to represent water balance using an
estimation of baseflow and regional
O&G GroundwaterWell-Parachute Creek recharge. GFLOW evaluated the
0.20 T— —r i.o transmissivity of alluvium and rock needed
to support the maximum rated pumping at
the O&G private wellfield.
Q
ID
0)
in
13
01
4->
ro
0.15
0.10
0.05 -
0.00 -I
0.2
0.1
0.0
2008
2009
2010
2011
2012
2013
Figure 5-23. Annual water pumping and groundwater use intensity
index (GUI) at the groundwater wellfield, including wells 1 (structure
39-5298) and la (structure 39-5301). Structure capacity (water
allocation) is 124 million gallons per year (380 ac-ft). (Data sources:
CODWR 2014b, 2014d.)
The zone of capture is illustrated using
GFLOW and mapping of streamlines, as
shown in Fig. 5-24B and D. Under average
observed pumping, the wellfield captures
about 6% of available baseflow, for a SUI
of 0.06. The wellfield captures about 9.4%
of available baseflow under maximum
rated pumping, for a SUI of 0.09. The
drawdown for the maximum pumping
scenario was about 13 feet to the base of
the aquifer. There were no domestic
groundwater wells in the area of potential
impact of the O&G well.
83
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Scenario 1: Average Pumping
(A)
(B
well pumping
consumes 6.2% of available
^ I baseflow
Scenario 2: Maximum Rated Pumping
(C)
(D)
well pumping
consumes 9.4% of available
baseflow
Figure 5-24. Parachute Creek confluence self-supplied groundwater wellfield for oil and gas, showing area of
potential impact and surface water zone of capture for two scenarios based on GFLOW simulations assuming
averaged geohydrologic conditions (2007-2012). Scenario 1 was based on the average annual observed pumping
rate; A) shows the cone of depression (drawdown in feet) associated with a groundwater stress index of 0.34, and
B) shows the associated baseflow capture of 6.2% from the nearby creeks and surface water use intensity index of
0.06. Scenario 2 is based on maximum rated pumping; C) shows the drawdown to the base of the aquifer and a
groundwater use intensity index of 1.0, and D) shows the baseflow capture of 9.4% of available and a surface
water use intensity index of 0.09. (Water use data source: CODWR 2014d.)
Vulnerability of Streams to Depletions
Water from Parachute Creek is heavily used for 6 to 7 months of the year for irrigation. The O&G industry currently
uses a small amount of water throughout the year to support hydraulic fracturing drilling. Hydraulic fracturing drilling
peaked in 2008 at a relatively high rate of drilling compared to current activity. Therefore, we assume that peak well
drilling rates are represented in the observed records and base scenario water requirements on 2008. Not well
represented in observations to date are withdrawals from very small streams, withdrawals from flows not influenced
heavily by irrigation, and volumes needed for horizontal wells that require more freshwater. Scenarios were applied to
address these considerations.
-------
Water Acquisition for Hydraulic Fracturing
May 2015
The same techniques for defining subbasins and applying scenarios used in the SRB were used in Parachute (198 mi2)
and Roan Creeks (509 mi2), but with different hydraulic fracturing use assumptions. Although most withdrawals
currently occur in Parachute Creek, Roan Creek was included to increase the sample size for small streams and
because it could be used for O&G water needs in the future. The watersheds were divided into a number of smaller
subbasins by stream ordering marked by a stream prediction point. There were 47 subbasins, ranging in area from 4.4
to 509 mi2. Streamflow was estimated from the simulated daily flow record at Parachute Creek based on watershed
area. The scenario withdrawals were performed daily at each site. Acknowledging the highly managed nature of water
withdrawals via water rights during the irrigation period (mid-April through October), and the abundant information
already provided on large volumes of water use during this period, SUI were only computed from November through
mid-April (the "free river" period). The Streamflow determined for Roan and Parachute Creeks from 1986 to 2012 as
described earlier was used for this analysis.
Hydraulic fracturing water use scenarios represented current and maximum observed drilling rate, and either all
directional or all horizontal wells. They assumed that freshwater was used only for drilling, and that all hydraulic
fracturing fluid was reused hydraulic fracturing wastewater. These scenarios enveloped the likely hydraulic fracturing
water use in this subbasin, assuming that enough hydraulic fracturing wastewater was available to supply 100% of
hydraulic fracturing injection fluid as currently occurs. To obtain daily water demand from Parachute Creek, we
applied the following assumptions:
• The current rate of drilling is 600 hydraulic fracturing wells per year in Garfield County (average from
2011 to 2013). At the peak rate of drilling in 2008, 1,700 wells were drilled.
• Parachute Creek has typically provided about 50% of the hydraulic fracturing freshwater needs in Garfield
County, and therefore would continue to supply 50% of the water for the number of wells for the current
and peak scenarios.
• The groundwater wellfield routinely supplies 41 MG of water per year, which was subtracted from the total
required of Parachute Creek before computing the required amount from surface waters.
• Directional and horizontal wells require 0.25 MG and 1.05 MG respectively. The groundwater wellfield
could therefore supply 164 directional or 40 horizontal wells. The remainder is supplied from each of the
Streamflow prediction locations.
The daily water demand for the four scenarios is listed in Table 5-5.
Table 5-5. Water withdrawal volume for hydraulic fracturing withdrawal scenarios applied to 26 years of
Streamflow (1987-2012) in subbasins of Parachute and Roan Creeks in Garfield County. Background use was
obtained from the USGS water use census (Ivahnenko and Flynn 2010).
Scenario ID
Current drilling —
directional
Current drilling-
horizontal
Peak drilling —
directional
Peak drilling —
horizontal
Background
Parachute and Roan Creek Withdrawal Scenarios for Surface Water Use Intensity Index Analysis
Water Demands
34.0 MG per year
(104ac-ft)
273 MG per year
(838 ac-ft)
172 MG per year
(528 ac-ft)
850 MG per year
(2,609 ac-ft)
Unmeasured background
water use by other users
Hydraulic Fracturing Demand Assumptions
136 directional wells per year
0.25 MG (0.77 ac-ft) per well
261 horizontal wells per year
1.05 MG (3.22 ac-ft) per well
686 directional wells per year
0.25 MG (0.77 ac-ft) per well
811 horizontal wells per year
1.05 MG (3. 22 ac-ft) per well
Domestic residential, livestock, assuming
rates based on county-scale water use
assessment
MGD
0.10
0.75
0.47
2.33
0.38
cfs
0.14
1.16
0.73
3.01
0.54
85
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Daily withdrawals ranged from 0.01 to 2.3 MOD, well within the range currently allowed at the 29 main structures.
As in the SRB, background consumption was added to account for domestic and livestock use based on USGS water
use census data for Garfield County (Ivahnenko and Flynn 2010). See Appendix D for additional information on
scenario assumptions.
The SUI distribution of withdrawal locations for the current-drilling directional wells scenario is shown in Fig. 5-25.
Water demand in this scenario was most similar to current use. The scenarios withdrew more from small streams than
probably actually occurs, and withdrew less from the primary structures discussed previously than actually occurs.
SUI declined in the scenario analysis with increasing basin area and streamflow. Median SUI was less than 0.1 and the
95th percentile was less than 0.6 in subbasins of approximately 20 mi2 for the current drilling directional scenario. All
subbasins less than 6 mi2 routinely had simulated SUI greater than 0.5. Simulated SUI in subbasins larger than 100
mi2 were similar to observed data.
The small to moderate-sized streams in the UCRB are vulnerable to withdrawals of relatively moderate volume. The
median (50th percentile) and 95th percentiles of the SUI for all four scenarios are shown in Fig. 5-26. The approximate
basin area where SUI declined below 0.4, as one example reference point, is provided in Table 5-6 for each of the
scenario withdrawals. The basin area threshold where median SUI was less than 0.4 increased from 6 mi2 to 150 mi2
with increasing withdrawal rates. The 95th percentile figures indicate that higher SUI can occur with moderate
withdrawals up to several hundred square miles.
Current Drilling—Directional Wells Scenario (0.1 MGD)
oo
1.0 -
0.8 -
0.6 -
0.4 -
0.2 -
0.0 -
ea
S9*e
I
to
I
CM
CO
CT
O
Ml
(mi )
Figure 5-25. Distribution of surface water use intensity index (SUI) at streamflow prediction locations in Parachute
and Roan Creek for the current drilling-directional wells scenario, modeled from 2008 to 2013. Hydraulic fracturing
withdrawal was 0.1 MGD. This scenario best represents current water volume used in Parachute Creek. Boxes
envelop the 25th and 75th percentiles of the distribution, dotted lines extend to the 5th and 95th percentiles, and dots
are the remainder of the observations (n = 4,325).
-------
Water Acquisition for Hydraulic Fracturing
May 2015
A) Directional withdrawal = 0.10 MGD
Horizontal withdrawal = 0.75 MGD
Current Drilling Scenarios —50th Percentile
1.0 -
0.8 -
.
0.6 -
0.4 -
0.2 -
n n -
CXEK3
^
\
3330 1 1 1 1 1|
°0
O
v
4*
A Directional Wells
O Horizontal Wells
0
AA 22
1 10 100
Basin Area (mi2)
C) Directional withdrawal = 0.47 MGD
Horizontal withdrawal = 2.33 MGD
Peak Drilling Scenarios - 50th Percentile
1000
0.8 -
0.6 -
0.4 -
0.2 -
n n -
i ii»
ADirectiona
O Horizonta
,UJl_/V2r-UAS_AI
A
4
4
WellL ^L
Wells
3
O
O
n
O
10 100
Basin Area (mi2)
1000
B) Directional withdrawal =0.10 MGD
Horizontal withdrawal = 0.75 MGD
Current Drilling Scenarios — 95th Percentile
0.8 •
0.6 •
0.4 •
0.2 •
0.0 •
U_J_U1_
A Directions
O Horizonta
4
A
A
Wells
Weld
u u
O
O
O
•^
L 10 100 10
Basin Area (mi2)
D) Directional withdrawal = 0.47 MGD
Horizontal withdrawal = 2.33 MGD
Peak Drilling -- 95th Percentile
1.0 -I
0.8 •
0.6 •
0.4 •
0.2 •
0 0
1 1 cniir
A Directional W
O Horizontal W<
His
Ms
OOO OO 1
A
A
A
A
L 10 100 10(
Basin Area (mi2)
Figure 5-26. Surface water use intensity indices (SUI) for drilling scenarios using current and peak drilling rates.
Scenario SUI analysis covered subbasins in Parachute and Roan Creeks ranging from 4.4 to 509 mi2; both
directional and horizontal wells were investigated in each scenario. See Table 5-5 for withdrawal rates and
assumptions. Computations were for November through April (n = 4,325). A) 50th percentile of SUI
observations for current drilling scenarios. B) 95th percentile of SUI observations for current drilling scenarios.
C) 50th percentile of SUI observations for peak drilling scenarios. D) 95th percentile of SUI observations for
peak drilling scenario.
87
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Table 5-6. Approximate basin area threshold for median and 95th percentiles of the surface water use intensity
index (SUI) below 0.4 from 2008 to 2013, including wet and dry years at each streamflow prediction point. The
full range of SUI for each scenario is shown in Fig. 5-26.
Scenario
Current — directional wells
Current — horizontal wells
Peak drilling — directional wells
Peak drilling — horizontal wells
Hydraulic
Fracturing
Withdraw Rate
(MGD)
0.10
0.75
1.97
2.33
Basin Area (mi2)
Median
SUI <0.4
7
50
35
150
95th percentile
SUI <0.4
50
500
150
>1,000
UCRB Synopsis
The UCRB is located in a semi-arid climate and water shortages can occur at various spatial scales. Most of the water
within the basin where hydraulic fracturing is ongoing is used for irrigation. Regional infrastructure is able to
supplement water needs for users including municipal water suppliers, who hold reservoir contracts, but that capacity
is diminishing as population grows and reservoir water is increasingly committed (CWCB 2011, 2007, 2014).
The region is rich in a variety of hydrocarbon reserves, and several of these can require larger quantities of water to
extract. Water supplies accessed to meet energy extraction demands are concentrated in Garfield County. Existing
water use by various user sectors in Garfield County and existing water use (Ivahnenko and Flynn 2010) and potential
water demand from energy extraction in the UCRB (AMEC 2011) are summarized in Fig. 5-27. Freshwater use for
hydraulic fracturing is small, because the industry is able to reuse hydraulic fracturing wastewater for nearly all needs.
Most of the freshwater used for hydraulic fracturing is taken from the Upper Colorado River and from surface water
and groundwater in Parachute Creek, a 198 mi2 tributary to the Upper Colorado River that is centrally located at the
Parachute gas field. The O&G industry assembled an extensive water supply system in the 1970s, anticipating the
need for large water volumes to support oil shale resource extraction. Gas extraction currently taps a small portion of
those allocations. Hydraulic fracturing operators have flexibility to purchase water from conservancy districts or other
water rights holders with access to Colorado River water. We did not find any records that indicated that
municipalities supplied water for hydraulic fracturing from their primary drinking water sources or other municipal
supplies.
The volume of freshwater used for hydraulic fracturing is small (427 MG per year) despite high drilling rates during
the past decade (Fig. 5-27). By accounting for how water is generally used in this basin for hydraulic fracturing, this
study was able to indirectly confirm what is commonly reported that the water used for injection into wells is 100%
reused wastewater from prior operations. Freshwater is only used for drilling and associated activities. The large
quantities of water returned as flowback and produced water that are treated by the industry for reuse minimize the
need for freshwater and impact on water supplies.
To date, most wells have been directionally drilled (s-shaped) into the tight sand formations. O&G companies have
started to drill into the Mancos Shale underlying the Williams Fork formation and to shift to horizontal drilling, which
will increase annual water use accordingly. Drilling into the Mancos Shale will increase water use per well by about 4
times (up to 1,600 MG per year) (URS 2008, WPX Energy, onsite interview, Jan 8, 2014). However, even with this
increase, total hydraulic fracturing water use in Garfield County will remain relatively small at 1,800 MG year and
relative to other uses, and is not likely to impact drinking water supplies.
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Due to the high rate of hydraulic fracturing wastewater reuse by the O&G industry, we did not find any locations in
the Piceance play where unconventional gas development contributed to locally high water use intensity at current
levels of freshwater use. We also found no evidence that the O&G industry acquires water from municipal or domestic
supplies. We are not aware of any reports of impact on drinking water availability in groundwater aquifers due to
groundwater acquisition for hydraulic fracturing tight gas in the UCRB.
GarfieId County
140,000
— 120,000
100,000
^ 80,000
60,000
40,000
20,000
230 b
Figure 5-27. Overview of current water use in Garfield County and
existing or potential use to support oil and gas development with
hydraulic fracturing and oil shale extraction. Water use by irrigators,
municipalities, and other non-energy sectors was obtained from U.S.
Geological Survey water census data for Garfield County (Ivahnenko
and Flynn 2010). Hydraulic fracturing water use was determined in
this study. Horizontal wells were assumed to require about 4 times
more water than vertical wells (URS 2008) and the projection assumed
all wells drilled in the county would be horizontal. Projections of
potential future water use for oil shale development were taken from
URS (2008) and AMEC (2011). Values shown for oil shale are 50% of
total Piceance estimates, applying volumes equally to Garfield and Rio
Blanco Counties.
Greater water demand for energy extraction
could occur should the far more water-
intensive oil shale extraction ramp up in
coming decades in the Colorado River and
White/Yampa River basins (CWCB 2014).
The O&G industry may likely exercise its
water rights to a greater extent to meet the
high water demands of petroleum
extraction from oil shales (URS 2008;
AMEC 2011). Although pilot oil shale
projects have been undertaken in the area,
current technology for obtaining oil from
kerogen is costly, and extraction is not
currently pursued commercially (U.S. EIA
2014d).
Two projected oil shale high-use water
demand scenarios provided by URS (2008)
and AMEC (2011) are shown in Fig. 5-27,
varying with assumptions of the scale of
operations and energy extraction
technology eventually deployed. Values
shown represent 50% of the total water
estimate, assigning 50% to Garfield County
and 50% to Rio Blanco County. Depending
on projections, oil shale development could
increase water use for energy extraction in
Garfield and Rio Blanco Counties.
The Upper Colorado River—in terms of the
population it serves throughout the
Colorado River Basin and east of the Rocky
Mountains—is one the fastest-growing
regions in the nation (GAO 2003). Gaps
between water supply and demand are
expected in the UCRB and elsewhere
throughout the southwestern states by 2050
with population growth and potential
climate change (Bureau of Reclamation
2012; CWCB 2014; CWCB 2011; CWCB
2007; GAO 2003). Existing trends in warming were confirmed by Lukas et al. (2014). Fig. 5-28 shows Bureau of
Reclamation (2012) estimates of past and projected use relative to water supply in the entire Colorado River Basin
through 2050. Water use and supply have already come to the crossroads of supply and demand, and the gap will
likely widen in the future (Bureau of Reclamation 2012). Water use in this region has been and will continue to be an
important topic for public resource planning at the local, state, and federal level CWCB (2014).
89
-------
Water Acquisition for Hydraulic Fracturing
May 2015
25
20
I 15
m
Historical Supply and Use
3
10
Water Supply
(10-year Running Average)
Water Use
(10-year Running Average)
Projected future Supply and Demand
Projected Demand
Projected Water Supply
(10-year Running Average)
Pre iminary Results
Year
Figure 5-28. Projected water supply and use in the Colorado River Basin with growing
population in the next decades. (From Bureau of Reclamation 2012.)
90
-------
Water Acquisition for Hydraulic Fracturing May 2015
6. SYNTHESIS AND SUMMARY
Whether hydraulic fracturing water demand affects water resources and other users depends on how much freshwater
is used and where water is acquired, especially in areas with water shortages due to population, climate, or drought.
Hydraulic fracturing has different water use requirements than most other industries—it needs large quantities of
water episodically and for short duration in many different places spread over wide areas. The volumes of water
needed for hydraulic fracturing and the fast-paced expansion of hydraulic fracturing development in many areas of the
country have raised concerns about the potential impact on local water supplies and users.
Hydraulic fracturing energy development in the United States operates within a variable overlay of water resources,
climate, population centers, energy networks, ancient geologic formations, and water management, creating a mosaic
of natural and operational situations that define water use and supplies. The options for acquiring water depend on
volume and water quality requirements for a given hydraulic fracturing play, water availability, competing uses, and
permitting constraints. These large-scale factors influence how much water is needed, who has access to it, and how it
is shared. Not all water sourcing options are available to the oil and gas industry in all situations (API 2010).
At the state scale, assessments of hydraulic fracturing water use lack the granularity necessary to predict local effects
on water resources, drinking water supplies, or other users. This project studied how the O&G industry acquires water
for hydraulic fracturing in two large river basins that reflect the wide variability in contributing factors that exist
nationally. The two study areas included the Susquehanna River Basin (SRB), located in the eastern United States
(humid climate) and overlying the Marcellus Shale gas reservoir; and the Upper Colorado River Basin in Colorado
(UCRB), located in the western United States (semi-arid climate) and overlying the Piceance structural basin with a
total petroleum system that includes coal bed methane, shale and tight gas reservoirs, and oil shale.
The study areas were similar in some ways. Each has large hydrocarbon reserves, an active hydraulic fracturing
industry, and productive natural gas wells with an anticipated long-term future of energy extraction. National
assessments of the potential impacts of water use for hydraulic fracturing identified counties within each basin as
areas of concern, based largely on their rate of drilling activity (Freyman 2014). In each basin, hydraulic fracturing
activities are concentrated in largely rural areas, but within large river basins that generate water to support large
populations that are remote from drilling and water use. There are also many differences between the watersheds in
the factors that were ultimately important to whether hydraulic fracturing impacted availability of water resources.
These included: (1) the dominant water users, 2) inherent availability of water; (3) water allocation and management
oversight; and (4) geologic characteristics that influenced the gas development technologies and the amount of water
needed for hydraulic fracturing.
The project gathered detailed information on where and how water is acquired in each study basin by querying
publicly available databases from state, regional government, and federal data sources. The two basins each had
state/regional agency databases that track water use, and our research could not have examined water acquired for
hydraulic fracturing at the local scale without them. The Susquehanna River Basin Commission (SRBC) and
Pennsylvania Department of Environmental Protection (PADEP) data systems specifically tracked water used for
hydraulic fracturing in both public water supplier and self-supplied permitting systems. The complexity of the water
allocation system in Colorado was reflected in their water use data system; freshwater used for hydraulic fracturing
was more difficult to identify because O&G use is not separated from other industries, although considerable useful
information was available that allowed the picture of water use for hydraulic fracturing to emerge. The most useful
data included daily withdrawal volumes from individual water sources along with any allocation limits. We did not
find the additional complexity of tracing water from source to specific gas well to be necessary for this project. We
were able to examine water use at every local site where water was acquired in each area with these data.
This chapter summarizes and synthesizes the results described in earlier chapters dedicated to each river basin. It is
organized to address the project framing questions on the potential impacts to drinking water resources of water
acquisition to support hydraulic fracturing.
91
-------
Water Acquisition for Hydraulic Fracturing
May 2015
How much freshwater is used in hydraulic fracturing operations?
The volume of hydraulic fracturing fluid that must be acquired each year depends on the drilling rate in the basin and
the volume needed to drill and fracture each well. How much freshwater must be acquired depends on how much
hydraulic fracturing wastewater is available and has been prepared for reuse. The characteristics of the rock
formations targeted for resource extraction determine the volumes of flowback and produced water returned to surface
from the formations, and whether it is economical to treat and reuse. Characteristics of well drilling and water use in
the peak drilling year in each basin are summarized in Table 6-1. The two study areas varied significantly in geologic
formation characteristics and total freshwater use.
In the SRB, O&G wells are horizontally drilled into the Marcellus Shale at a depth of about 7,000 ft. The drilling and
development of each well requires 4.3 MG of fluid on average, most of which is freshwater. The Marcellus Shale
returns a low percentage of hydraulic fracturing fluid of relatively poor quality as flowback, constraining the industry
to substitute reused water up to 13% of the total water needed for hydraulic fracturing. This reuse rate is similar to
those reported from other shale formations (Vengosh et al. 2014). About 3,350 million gallons of freshwater were
needed in the SRB during the peak drilling year of 2012 (Fig. 6-1). Drilling rates in the Marcellus are expected to
increase in the next decades, depending on economics, and could reach as high as 2,000 wells per year (Johnson 2010)
raising potential water needs to 11,000 MG per year (Beauduy 2009).
In the Upper Colorado River Basin (UCRB), wells are primarily directionally drilled (S-shaped) into the Mesaverde
sandstones in the Piceance basin to an average depth of about 8,000 ft. The average well uses 2.5 million gallons of
fluid. The Williams Fork tight sand formation—the current target of exploration—returns most of the hydraulic
fracturing fluid of relatively good quality. In addition, gas wells continue to bring formation water to the surface
during their producing life. The industry captures, treats, and reuses the hydraulic fracturing wastewater, reducing the
need for freshwater to 10% of total water per well and within the region as a whole.
Table 6-1. Characteristics of well drilling and water use in the Susquehanna River Basin in (SRB, 2012) and Upper
Colorado River Basin (UCRB, 2008) during the peak year of drilling in each area.
Factor
Geologic
Characteristics
Water Use
Characteristics
State water
management
Parameter
Formation
Material
Drilling Type
Depth Below Surface
Maximum Drilling Rate (wells
per year)
Likely future drilling
Flowback Returned
Average Fluid Volume Injected
per Well
Total Annual Water Use
Freshwater Portion of
Hydraulic Fracturing Fluid
Wastewater Portion of
Hydraulic Fracturing Fluid
Total Annual Freshwater Use
Allocation controls
Susquehanna River
Basin (SRB)
Marcellus
Shale
Horizontal
~7,000 ft
836(2012)
t
7.3%
4.3 Million Gallons
3,594 MG
~87%
~13%
3,350 MG
Permitting
Upper Colorado River
Basin (UCRB)
Piceance
Total Petroleum System
Tight Sands predominantly
(Williams Fork formation)
Directional
~8,000 ft
1,688 (2008)
•V
100% +
2.5 Million Gallons
4,220 MG
~0%
~100%
430 MG (drilling only)
Appropriation
92
-------
Water Acquisition for Hydraulic Fracturing
May 2015
In the UCRB, freshwater was needed only for well
drilling and dust abatement at the site. The ability to
recycle hydraulic fracturing fluids substantially
reduced the hydraulic fracturing impact in this
generally water-stressed area. Compared to 3,350
MG of freshwater used in SRB, 430 million gallons
of freshwater were needed in the UCRB in the peak
drilling year 2008 (Fig. 6-1). Drilling rates in
UCRB may not return to this high of a level in
future years. However, should the O&G industry
increase emphasis on drilling horizontal wells into
the Mancos Shale beneath the tight sands, the
longer wellbores and perhaps different technology
requirements could translate to increased freshwater
use. It is difficult to project future use with these
uncertainties, but freshwater needs could increase
up to 2,000 MG per year, based on company
estimates (WPX Energy interview Jan 8, 2014).
Injected Fluid in Peak Drilling Year
l/l
_g
TO
(3
o
i=
~
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Figure 6-2. Self-supplied water acquisition site on a river in
theSRB.
None of the water used for hydraulic fracturing in the UCRB came from municipal supplies. We note that sales to
O&G were not specifically tracked in water use records, but we are reasonably confident that the available water use
records supplemented by discussions with district engineers from the Colorado Division of Water Resources
(CODWR) support this conclusion.
Access to Freshwater for Self-Supply
In both study areas, hydraulic fracturing operators
as relatively "new" industrial water users have
access to water on a site-by-site basis and are
allowed to withdraw from water sources within
constraints imposed by state and regional
authorities. In the SRB, constraints are assigned in
site permitting by the Susquehanna River Basin
Commission (SRBC) (SRBC 2012) and by the
water rights allocation system managed by the
State Engineer and CODWR in the UCRB
(Grantham2011).
In the SRB, companies and operators gain access
to water by requesting a permit at specific water
sources from the SRBC. Most permitted sites draw
water from rivers and streams. There were not
many previous allocations to municipal or
industrial users from surface water sources in the
areas where hydraulic fracturing is most active.
Groundwater resources are significant in the SRB,
and they are the primary supply to many
municipalities, other industrial users, and most of the private domestic users in much of the basin. Overall, SRBC
permitting limits competition for available water among user sectors and largely separates domestic supplies from
other serf-supplied sites, including those used by the
O&G industry.
In the SRB, the O&G industry can self-supply water at
140 permitted locations distributed throughout a 17
county area in Pennsylvania. One of these acquisition
^| sites is shown in Fig. 6-2. Water is typically
transported to gas wells via trucks (Fig. 6-3). Some
permitted sites are used regularly while others are used
infrequently or not at all.
SRBC has designed a system for hydraulic fracturing
water acquisition by applying a number of specific
measures that limit daily and seasonal withdrawals,
including shutdown during low flows (Beauduy 2009).
O&G operators could elect to use municipal water
supplies during the shutdown periods.
In UCRB, the O&G industry acquires water within the
Colorado water allocation system based on prior
appropriations. Most available water is already
allocated in the water rights system. During shortages,
water is obtained in order of priority where the most senior (oldest) rights obtain water first. The O&G industry must
obtain rights or purchase water from sources with existingappropriations. Some water is available for purchase from
state-sanctioned conservancy districts or the industry can buy water from rights holders as long as industrial use is
assigned to their right.
Figure 6-3. Trucking water to a hydraulic fracturing well
site in Virginia (Photo from Virginia Dept. of Mines,
Minerals and Energy.)
94
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Decades ago, individual O&G companies acquired many water right allocations in surface waters coincident with the
Piceance structural basin anticipating oil shale development that requires large volumes of water to extract
hydrocarbons with current known technologies (AMEC 2011; URS 2008). Many of these water sources serving the
Parachute gas field (U.S. EIA. 2009) are spatially coincident with the oil shale deposits that are found in the same
area. Ownership of water rights by the oil and gas industry in Colorado is currently unique to the Piceance structural
basin. Acquisition of water for hydraulic fracturing in the UCRB is concentrated at a few of the O&G owned
structures in Parachute Creek where the gas field is centered, including small private reservoirs and a groundwater
well. The rest of the needed freshwater is obtained from the Colorado River through contracts.
Large regional high-yielding shallow groundwater aquifers are not available in either river basin. Shallow
groundwater aquifers are ubiquitous in both study areas and many public and private water wells are strategically
located in the highly permeable alluvial or valley-fill deposits of rivers and streams where the aquifers are in tight
communication with rainfall recharge and surface water. Given the humid climate in the SRB, with over 40 inches
per year of rainfall, the water replenishment rate tracks rainfall all year. Many of these river-adjacent aquifers are the
source of municipal and domestic water supplies in the SRB.
Even though the O&G activity is concentrated in a semi-arid climate in the UCRB, the annual pulse of Colorado
Rocky Mountain snowmelt supplies the Colorado River and its alluvium with significant annual recharge of
freshwater. Due in part to poorer water quality, there are far fewer community drinking water wells in the Piceance
overlapof the UCRB. In the UCRB, the wells are usually classified as tributary diversion structures and managed by
the state appropriation laws, and the well pumping must be augmented through purchase and release of upstream
reservoir storage water (Grantham 2011).
Volume of self-supplied freshwater used for hydraulic fracturing
In the SRB, most freshwater is serf-supplied from rivers and streams with basin areas ranging from 2 to 11,000 mi2. A
portion of the surface water comes from smaller rivers of sizes shown to be vulnerable to water use imbalance from
over-withdrawal during low flow periods. The total annual volume that was self-supplied by hydraulic fracturing
operators was 2,925 million gallons during the peak
use year (2012), or 7.3 MOD and averaging 0.3
MOD at individual sites (Fig. 6-4). Site volumes ___
are controlled by permit and vary significantly £ 4'000
Source of Freshwater
between sites. The largest daily withdrawals of 3.0
MOD are permitted in the Susquehanna River
mainstem.
In the UCRB, 1.2 MOD of freshwater was obtained
during the peak use year of 2008. Parachute Creek
acquisition sites annually supply enough surface
and low quality groundwater to supply about 50%
of the freshwater used for hydraulic fracturing in
Garfield County amounting to 430 million gallons
(Fig. 6-5). Withdrawal of water from three small
reservoirs and a tributary averaged 0.03 MOD.
Hydraulic fracturing water use does not appear to
interfere with downstream irrigation or municipal
users, nor is it taken from drinking water supplies.
Shallow alluvial groundwater wells provided 20%
to 25% of water used for hydraulic fracturing in both
study areas. In the SRB, most of this water was self-
supplied from a few commercial wells. Some water
I Self-supply
ro
(3
3,500
= 3,000
ro
g 1,500
_c
-------
Water Acquisition for Hydraulic Fracturing
May 2015
was taken from municipal supplies that relied on
groundwater sources. Groundwater was also self-
supplied in the UCRB from a single high-volume
groundwater wellfield that produced 0.14 MOD. No
direct impacts due to pumping drawdown at these
groundwater wells were discovered in this study
area.
How might water withdrawals affect short-
and long-term availability?
Intrinsic Vulnerability
Regional or local water use imbalance from
hydraulic fracturing water use may arise depending
on the overall gas field development rate and
individual well need, intrinsic water availability,
and competition for water from other user sectors.
Short-term and/or local water use imbalance depends
on how much is withdrawn from a water source, how
much is available at the source, and how fast water
replenishes. Surface water flows and storage in
human-engineered reservoirs are the dominant source
of freshwater for hydraulic fracturing in the study
areas, and these could be readily quantified.
Source of Freshwater
12,000
TO
ID
T3
OJ
0)
4->
TO
TO
C
Surface Water 10,000
Susquehanna River Upper Colorado
Basin River Basin
Figure 6-5. Volume of water sourced from surface water
and groundwater sources by the oil and gas industry for
hydraulic fracturing in the peak drilling year in each basin
(SRB 2012, UCRB 2008). (Data sources: SRBC 2013a,
CODWR 2014d).
With annual average precipitation over 40 inches distributed evenly through the year, the SRB has significant surface
and groundwater resources. A relatively low population and rural economy where hydraulic fracturing is most active
further minimizes competing demands. Shortages may develop during droughts, and low flow periods in the smaller
streams, but intrinsic vulnerability to water shortages is generally relatively low. The population is not projected to
increase substantially in this portion of the SRB (U.S. EPA 2013a).
The UCRB receives less than 20 inches of annual precipitation on average, and much of it occurs as snowfall on the
western slope of the Colorado Rocky Mountains. The spring through summer snowmelt supplies streamflow in the
Colorado River and tributaries. Some of the surplus streamflow is diverted into reservoirs for storage or transfer for
use by communities and irrigators on both the west and the east sides of the Rocky Mountains. Surface water supplies
are limited and in high demand in the UCRB. Under historic agricultural use, water shortages relative to demand
evolve in many tributaries and the mainstem Colorado during the dry summer and early fall months. Regional
infrastructure is able to supplement water needs with reservoir releases, but that capacity is diminishing as the
population grows and reservoir water is increasingly committed (CWCB 2011). Water availability in the entire
southwestern United States is dependent wholly or partially on water from the Upper Colorado River (Bureau of
Reclamation 2012) and population growth is expected to significantly increase water demands at the same time that
climate change is projected to decrease annual precipitation (CWCB 2011, 2014; GAO 2003, 2014). The region is
thus intrinsically vulnerable to water shortages.
The volume of water needed for hydraulically fracturing at individual wells is large (2 to 4 million gallons) and the
general rate of drilling can create a relatively high demand in an area. Nevertheless, the hydraulic fracturing demand
will always be small compared to other users at the state and play spatial scales (e.g. Murray 2013, Nicot et al. 2014).
Any imbalances that may arise between water supply and demand and competition among users are more likely to
evolve at county or finer scales, especially at sources where water is actually acquired.
At the local level, whether withdrawals impose high use intensity on water resources depends on the withdrawal
volume relative to the volume in the waterbody. This project quantified potential water imbalance using a simple
96
-------
Water Acquisition for Hydraulic Fracturing May 2015
water use intensity index (SUI and GUI) calculated as water volume withdrawn relative to available source volume on
a daily basis. The index is a straightforward volumetric calculation that represents the proportion of water removed
from the source in an immediate timeframe. The water use intensity index provided a quantitative scalar of potential
imbalance and was used to represent vulnerability of local water sources to withdrawals for any use. No threshold
values of concern were used in this study; others have used thresholds of 0.4 (Freyman2014; Hurde/a/. 1999) or 0.7
(TidweUe/a/. 2012).
The study evaluated the effects of typical water withdrawal amounts for hydraulic fracturing on the water use intensity
over a range of stream sizes and streamflow conditions in each basin, largely using hydrologic models and/or scenario
modeling. A given withdrawal volume compared to a given flow volume produces a specific value of the use intensity
index (SUI). Rivers and streams vary in the probability of observing low volume flow, dependent largely on
contributing watershed area, and reflecting the climate in each location. At the same rate of withdrawal, a smaller
stream is more likely to have a higher use intensity index than a larger river, as the former experiences smaller flow
volumes during a greater part of the year.
Analysis in both study areas showed that smaller streams (watershed areas <10 mi2) are vulnerable to typical hydraulic
fracturing withdrawals 0.1 to 1.0 million gallons per day (higher SUI or proportion of water removed) most of the
time. In general, three times more watershed area was needed to support typical hydraulic fracturing withdrawals in
the semi-arid UCRB. However, the analysis also revealed that surprisingly large rivers in humid areas are vulnerable
to typical hydraulic fracturing withdrawals during infrequent low flows that occur during infrequent regional droughts.
Rivers draining watersheds up to 600 square miles were shown to be potentially vulnerable to hydraulic fracturing
withdrawals (experience high SUI) in the SRB despite the abundance of water resources and humid climate.
Current and Potential Future Water Use Intensity
The potential for cumulative impact associated with pumping groundwater sources for O&G water needs on drinking
water supply is low in the SRB. However, there is the potential for localized effects on household or community
drinking water wells in the drawdown of the hydraulic fracturing water-supplying wells. If the induced drawdown
lowers the water table below the screen interval of a household or community well, the well could go dry. However,
no effects were discovered at the few public and private groundwater sources that supply relatively larger volumes of
water for hydraulic fracturing in the SRB. Groundwater wells located in alluvium and valley-fill can draw a part of
their source water from nearby streams. In a limited investigation of SRB and UCRB study areas, this use was
estimated to be less than 10% of the available annual baseflow. The SRBC manages streamflow depletion from
groundwater pumping through the passby flow program.
At the local level, and reflecting the water management systems in place, water use for hydraulic fracturing is
currently low relative to water availability at all but a few local sites in both basins. Despite the potential for higher
water use intensity, we found no impacts of hydraulic fracturing water acquisition in either case study area, expressed
in terms of the proportion of water withdrawn from the water body or PWS facility relative to the volume of
freshwater available at each location. We identified no shortages or hardships imposed on other users or at public
water suppliers by hydraulic fracturing in the two study areas. Many natural climatic and hydrologic differences
between the Susquehanna River and Upper Colorado River basins influence water availability, but other factors
related to water management influenced consumption patterns and contributed substantially to outcomes. Differences
in water demands and water management factors in these two regions resulted in the same outcome of no adverse
effect on water supply.
In the SRB, the industry trend is to widely distribute acquisition sites within the region and to locate them primarily in
rivers and streams that are not extensively used by others. The rate and timing of withdrawals are managed by permit
to protect other users and aquatic life. Although the rate of hydraulic fracturing activity is likely to increase in the
future, there is currently enough capacity in the serf-supplied permitting system to accommodate future hydraulic
fracturing water needs from rivers and streams without the need for significant contributions from public water
suppliers. Furthermore, O&G companies and hydraulic fracturing service companies in both areas are increasingly
building piping infrastructure to move water, and have built numerous small impoundments distributed around the
landscape to store water. There is currently sufficient storage in the SRB to supply freshwater for hydraulic fracturing
for a year at current drilling rates, further reducing the need for public water supplies during low flow shutdowns that
typically occur seven or more days each year.
97
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Water use intensity is already high in the semi-arid Colorado River basin, primarily due to large irrigation
withdrawals, but the contribution to this use intensity from hydraulic fracturing withdrawals is very low. The
favorable geologic characteristics that return large volumes of reusable hydraulic fracturing fluids currently minimize
the use of freshwater by the O&G industry. Freshwater usage for hydraulic fracturing will increase with development
of the Mancos Shale below the tight sands with horizontal drilling, due in part to longer wellbores. These additional
demands should be within historical scope and should not significantly impact water availability in the area given
available water sources that do not conflict with other users. Extraction of energy from oil shales may require the
O&G industry to use as much as 50 times more water than hydraulic fracturing demands (CWCB 2011; AMEC 2011;
URS 2008). Thus, significantly greater water demand would emerge should the far more water-intensive oil shale
extraction ramp up in future decades when the economics of resource extraction make extraction more feasible.
What are the potential impacts of hydraulic fracturing water withdrawals on water quality of source
waters?
A preliminary exploration was conducted on the potential impact of hydraulic fracturing withdrawals on water quality
in streams. Water quality is determined by the concentration of chemical stressors within a volume of water.
Hydraulic fracturing withdrawals reduce water volume and thus have the potential to increase chemical concentration
if taken upstream of discharge points or a chemical spill site. In this case, the chemical concentration would be
magnified in proportion to the water withdrawal volume. The SUI index used for this study could be mathematically
reformulated to determine the concentration
magnification of given withdrawal rates, and was
applied to the observed withdrawal data in the
SRB assuming all sites were affected by point
source discharges. In the SRB, concentration
magnification was mostly less than 10% but
could range as high as 30%. Because SUI and
concentration magnification are mathematically
related, the same factors that influenced SUI
influence the magnitude of water quality
response.
Study Conclusions
We studied the potential impact on drinking
water resources of water acquisition for hydraulic
fracturing by the unconventional oil and gas
industry, with particular focus on the implications
of hydraulic fracturing for natural gas in
tight/shale formations. We focused our study on
two large river basins where significant O&G
activity has occurred and is predicted to continue
in the future: (1) the Susquehanna River basin,
which overlies the Marcellus shale; and (2) the
Upper Colorado River basin of Colorado, which
overlies the Piceance structural basin. No
significant negative impacts to past or current
drinking water supplies or other water users
resulting from hydraulic fracturing water
acquisition were found in either of the study
basins. However, this underscores the importance
of careful planning and management to protect
water resources. This conclusion is based on data
on both available water and water volumes
obtained at local sources on fine temporal scales.
A combination of geology, industry practices,
state allocation and management, and other
• A combination of factors determine whether hydraulic
fracturing introduces imbalance in the relationship
between water supply and demand in a region, including
drinking water resources. These factors include available
water resources and their capacity to yield water,
industry needs influenced by geologic characteristics of
rocks in each play, other user demands, and permitting
or allocation controls.
• No significant negative impacts to past or present
drinking water supplies or other water users resulting
from hydraulic fracturing water acquisition were found in
either study basin due to unique combinations of these
factors in each area.
• In the Susquehanna River Basin in Pennsylvania (SRB),
there is little use of public water supplies (currently <8%)
because water resources are well distributed and
available year round and hydraulic fracturing operators
have been able to develop unallocated sources. In SRB,
there are times or locations when water sources can be
stressed, but water is managed to prevent overuse and
minimize risk at individual sources.
• Water in the Upper Colorado River Basin in Colorado
(UCRB) is strongly seasonal and over-allocated, but
unconventional gas production requires little freshwater
as the industry is able to reuse large volumes of flowback
and produced water instead. No municipal drinking
water supplies are used for hydraulic fracturing in the
areas studied within the UCRB.
98
-------
Water Acquisition for Hydraulic Fracturing May 2015
factors, in each study basin reduced the actualization of impact. The potential for future impact was explored by
distributing hypothetical point withdrawals within representative sub-watersheds. Results showed that unmanaged
withdrawals in small streams (watershed area <10 mi2) have the potential for frequently large impact. Since multiple
local factors must be understood to anticipate the effects of water acquisition on drinking water supplies,
generalization to other plays and regions is not possible. The study only involved two large river basins. Other areas
with active hydraulic fracturing will likely have a unique combination of the important driving factors that determine
water balance that could result in adverse impacts that were not observed in the two study basins.
Several recent national scale studies have used downscaling techniques based on water census and hydraulic
fracturing activity and other data to predict areas of potential water use imbalance or high use intensity at the county
level throughout the United States (e.g. Freyman 2014). Where our study basins overlapped these national studies,
data at the local scale did not identify any impacts on water availability for other users in several counties highlighted
in the national assessments. This underscores a need for integrating the array of key factors that influence water use
and water availability when assessing the likely effects of water acquisition for hydraulic fracturing on water
resources and drinking water supplies in a local area.
99
-------
Water Acquisition for Hydraulic Fracturing May 2015
[This page intentionally left blank/
100
-------
Water Acquisition for Hydraulic Fracturing May 2015
REFERENCES
AMEC. 2011. Energy development water needs assessment Phase II. Prepared for the Colorado River Basin
Roundtable and Yampa/White Basin Roundtable.
API (American Petroleum Institute). 2010. Water management associated with hydraulic fracturing. API Guidance
Document HF2. Available at: http://www.api.org/policv-and-issues/policv-
items/hf/apihf2 water management.aspx
Arguez, A., I. Durre, S. Applequist, R. S. Vose, M. F. Squires, X. Yin, R. R. Heim, and T. W. Owen. 2012. NOAA's
1981-2010 U.S. climate normals: An overview. Bulletin of the American Meteorological Society 93(11):
1687-1697.
Arthur, J. D., M. Uretsky, and P. Wilson. 2010. Water resources and use for hydraulic fracturing in the Marcellus
Shale region. ALL Consulting. Available at: http://www.all-
llc.com/publicdownloads/WaterResourcePaperALLConsulting.pdf
Aydemir, A. 2012. Comparison of Mississippian Barnett Shale, northern-central Texas, USA and Silurian Dadas
formation in southeast Turkey. Journal of Petroleum Science and Engineering 80(1): 81-93.
Beauduy, T. W. 2009. Accommodating a new straw in the water: Extracting natural gas from the Marcellus Shale in
the Susquehanna River Basin. Susquehanna River Basin Commission, Harrisburg, PA.
Bicknell, B. R., J. C. Imhoff, J. L. Kittle Jr., A. S. Donigian Jr., andR. C. Johanson. 1997. Hydrological simulation
program—FORTRAN: User's manual forversion 11. EPA/600/SR-97/080. U.S. Environmental Protection
Agency, National Exposure Research Laboratory, Athens, GA. 755 pp.
BLM (Bureau of Land Management). 2006. Roan Plateau planning area proposed resource management plan
amendment and environmental impact statement. U.S. Department of the Interior, Bureau of Land
Management, Glenwood Springs, CO.
Brantley, S. L., D. Yoxtheimer, S. Arjmand, P. Grieve, R. Vidic, J. Pollak, G. T. Llewellyn, J. Abad, and C. Simon.
2014. Water resource impacts during unconventional shale gas development: The Pennsylvania experience.
International Journal of Coal Geology 126: 140-156.
Bredehoeft, J. D. 2002. The water budget myth revisited: Why hydrogeologists model. Ground Water 40(4): 340-345.
Bureau of Reclamation. 2005. Water 2025: Preventing crises and conflict in the West. U.S. Department of the Interior,
Washington, D.C.
Bureau of Reclamation. 2012. Colorado River Basin water supply and demand study. U.S. Department of the Interior,
Bureau of Reclamation. Available at:
http://www.usbr.gov/lc/region/programs/crbstudy/fmalreport/Executive%20Summary/Executive_Summary_
FINAL_Dec2012.pdf).
COS (Colorado Geological Survey). 2003. Ground water atlas of Colorado. Available
at: http://coloradogeologicalsurvev.org/water/groundwater-atlas/
Clark, C. E., A. J. Burnham, C. B. Harto, and R. M. Horner. 2012. The technology and policy of hydraulic fracturing
and potential impacts of shale gas development. Environmental Practice 14: 251-261.
Clark, C. E., R. M. Horner, and C. B. Harto. 2013. Life cycle water consumption for shale gas and conventional
natural gas. Environmental Science and Technology 47: 11829-11836.
CODWR (Colorado Division of Water Resources). 2012. Cumulative statistics. Available
at: http://water.state.co.us/DWRDocs/Reports/Pages/CumStats.aspx
CODWR (Colorado Division of Water Resources). 2014a. Data search. Available at:
http://water.state.co.us/DataMaps/DataSearch/Pages/DataSearch. aspx#adminrpts
CODWR (Colorado Division of Water Resources). 2014b. Water rights. Available at:
http://cdss.state.co.us/onlineTools/Pages/WaterRights.aspx
CODWR (Colorado Division of Water Resources). 2014c. Structures (diversions). Available at:
http://cdss.state.co.us/onlineTools/Pages/StructuresDiversions.aspx
101
-------
Water Acquisition for Hydraulic Fracturing May 2015
CODWR (Colorado Division of Water Resources). 2014d. Structures (diversions). Available at:
http://cdss.state.co.us/onlineTools/Pages/StructuresDiversions.aspx. Query: individual structures for
diversion reports.
CODWR (Colorado Division of Water Resources). 2014e. Surface water (StateMod). Available at:
http://cdss.state.co.us/Modeling/Pages/SurfaceWaterStateMod.aspx
CODWR (Colorado Division of Water Resources). 2014f. Colorado's well permit search. Available at:
http://www.dwr.state.co.us/WellPermitSearch/default.aspx
COGCC (Colorado Oil and Gas Conservation Commission). 2012. COGCC hydraulic fracturing rules: Rule 205A
Disclosure. Available at:
http://cogcc.state.co.us/Announcements/Hot Topics/Hydraulic Fracturing/COGCC%20Hydraulic%20Fractu
ring%20Rules.htm
COGCC (Colorado Oil and Gas Conservation Commission). 2013. COGCC CIS online. Available at:
http://cogcc.state.co.us/Home/gismain.cfm
COGCC (Colorado Oil and Gas Conservation Commission). 2014a. Annual well starts by county. Oil & Gas Staff
Report (July 28, 2014): 21. Available at: http://cogcc.state.co.us. Query: Staff reports (Staff RPT).
COGCC (Colorado Oil and Gas Conservation Commission). 2014b. COGCC reports portal. Available at:
http://cogcc.state.co.us/COGCCReports/production.aspx?id=MonthlyWaterProdBvCountv.
Colorado Geological Survey. 2003. Ground water atlas of Colorado, Chapter 6.2 Piceance Basin. Special Publications
53. http://coloradogeologicalsurvey.org/water/groundwater-atlas/.
Cueto-Felgueroso, L., and R. Juanes. 2013. Forecasting long-term gas production from shale. Proceedings of the
National Academy of Sciences 110(49): 19660-19661.
CWCB (Colorado Water Conservation Board). 2007. Colorado's water supply future. Statewide Water Supply
Initiative-Phase 2. Available at: http://cwcb.state.co.us/public-
information/publications/Documents/ReportsStudies/TechnicalRoundtableReportFinalDraft.pdf.
CWCB (Colorado Water Conservation Board). 2011. Colorado's Water Supply Future—Statewide Water Supply
Initiative 2010. Available at: http://cwcb.state.co.us/water-management/water-supply-
planning/pages/swsi2010. aspx
CWCB (Colorado Water Conservation Board), 2014. Colorado's Water Plan. Available at:
https://www.colorado.gov/pacific/cowaterplan/framework-documents-and-draft-chapters.
Davis, C. 2012. The Politics of 'Tracking": Regulating natural gas drilling practices in Colorado and Texas. Review
of Policy Research 29 (2): 177-191.
Dennison, D. 2005. Drilling 101: Drilling of a natural gas well and natural gas production in the Piceance Basin.
Garfield County Oil & Gas Liaison. Available at: http://www. garfield-countv.com/oil-
gas/documents/drilling 101 .pdf
Dietrich, J. D., and R. C. Johnson. 2013. Simplified stratigraphic cross sections of the Eocene Green River formation
in the Piceance Basin, Northwestern Colorado. U.S. Geological Survey Open-File Report 2013-1012. 23 pp.
Dott, R. H., Jr., and R. L. Batten. 1981. Evolution of the Earth. McGraw-Hill, New York. 579 pp.
Dripps, W. R., R. J. Hunt, and M. P. Anderson. 2006. Estimating recharge rates with analytic element models and
parameter estimation. Ground Water 44(1): 47-55.
Dunlap, K. 2011. Shale gas production and water resources in the eastern United States. U.S. Senate Committee on
Energy and Natural Resources Subcommittee on Water and Power.
Dunne, T., and L. B. Leopold. 1978. Water in environmental planning. W.H. Freeman and Company, New York. 818
pp.
Eckhardt, K. 2008. A comparison of baseflow indices, which were calculated with seven different baseflow separation
methods. Journal of Hydrology 352(1-2): 168-173.
Entrekin, S., M. Evans-White, B. Johnson, andE. Hagenbuch. 2011. Rapid expansion of natural gas development
poses a threat to surface waters. Frontiers in Ecology and the Environment 9(9): 503 -511.
Erenpreiss, M. S., L. H. Wickstrom, C. J. Perry, R. A. Riley, D. R. Martin, and others. 2011. Regional organic-
thickness map of the Marcellus Shale with addition organic-rich shales beds in the Hamilton Group included
102
-------
Water Acquisition for Hydraulic Fracturing May 2015
for New York, Pennsylvania, and West Virginia. Ohio Department of Natural Resources, Division of
Geological Survey.
FEMA (Federal Emergency Management Agency). 2014. Hydro logic models meeting the minimum requirement of
National Flood Insurance Program: Current nationally accepted hydrologic models. Available at:
http://www.fema.gov/national-flood-insurance-program-flood-hazard-mapping/hvdrologic-models-meeting-
minimum-requirement
FracFocus. 2014. Chemical disclosure registry. Available at: http://fracfocus.org/
FracTracker Alliance. 2014. Available at: http://www.fractracker.org/
Freeze, R. A., and J. A. Cherry. 1979. Groundwater. Prentice-Hall, Inc., Englewood Cliffs, NJ. 604 pp.
Freyman, M. 2014. Hydraulic fracturing & water stress: Water demand by the numbers—shareholder, lender, &
operator guide to water sourcing. CERES. Available at: http://www.ceres.org/resources/reports/hydraulic-
fracturing-water-stress-water-demand-bv-the-numbers/
GAO (U.S. Government Accounting Office). 2003. Freshwater supply: States' view of how federal agencies could
help them meet the challenges of expected shortages. GAO-03-514. Washington, DC.
GAO (U.S. Government Accountability Office). 2014. Freshwater: Supply concerns continue, and uncertainties
complicate planning. GAO-14-430. Washington, DC.
Gilmore, K., R. Hupp, and J. Glathar. 2014. Transport of hydraulic fracturing water and wastes in the Susquehanna
River Basin, Pennsylvania. Journal of Environmental Engineering 140(SPECIAL ISSUE: Environmental
impacts of shale gas development).
Grantham, J. 2011. Synopsis of Colorado Water Law. Report for the Colorado Division of Water Resources.
Available
at: http://water.state.co.us/DWRIPub/DWR%20General%20Documents/SynopsisOfCOWaterLaw.pdf
Gregory, K. B., R. D. Vidic, and D. A. Dzombak. 2011. Water management challenges associated with the production
of shale gas by hydraulic fracturing. Elements 7: 181-186.
Grubb, R. G. 1986. Soil survey of Bradford and Sullivan Counties, Pennsylvania. U.S. Department of Agriculture,
Soil Conservation Service. 135 pp.
GWPC (Groundwater Protection Council). 2009. State oil and natural gas regulations designed to protect water
resources. U.S. Department of Energy, Office of Fossil Energy, Washington, DC.
Haitjema, H. M. 1995. Analytic element modeling of groundwater flow. Academic Press, Inc., San Diego, CA. 394
pp.
Hansen, E., D. Mulvaney, M. Betcher. 2013. Water resource reporting and water footprint from Marcellus Shale
development in West Virginia and Pennsylvania. Earthworks Oil and Gas Accountability Project. 88 pp.
Available at: http://www.downstreamstrategies.com/documents/reports_publication/marcellus_wv_pa.pdf
Heisig, P. M. 2012. Hydrogeology of the Susquehanna River valley-fill aquifer system and adjacent areas in eastern
Broome and southeastern Chenango Counties, New York. U.S. Geological Survey Scientific Investigations
Report 2012-5282. 19pp.
Hettinger, R. D., and M. A. Kirschbaum. 2002. Stratigraphy of the Upper Cretaceous Mancos Shale (upper part) and
Mesaverde Group in the southern part of the Uinta and Piceance basins, Utah and Colorado. U.S. Geological
Survey Geological Investigations Series 1-2764. 21 pp.
Hill, M. 2013. Source water protection of the Colorado River partnership: Source water protection plan Garfield
County, Colorado. Colorado Rural Water Association. 87 pp.
Hunt, C. B. 1974. Natural regions of the United States and Canada. W. H. Freeman and Company, San Francisco, CA.
725 pp.
Hunt, R. J. 2006. Ground water modeling applications using the analytic element method. Ground Water 44(1): 5-15.
Kurd, B., N. Leary, R. Jones, and J. Smith. 1999. Relative regional vulnerability of water resources to climate change.
Journal of the American Water Resources Association 35: 1399-1409.
Ivahnenko, T., and J. L. Flynn. 2010. Estimated withdrawals and use of water in Colorado, 2005. USGS Scientific
Investigations Report 2010-5002. Reston, VA. 71 pp.
Johnson, C., and M. Mifflin. 2006. The AEM and regional carbonate aquifer modeling. Ground Water 44(1): 24-34.
103
-------
Water Acquisition for Hydraulic Fracturing May 2015
Johnson, N. 2010. Pennsylvania energy impacts assessment report 1: Marcellus Shale: Natural Gas and Wind. The
Nature Conservancy. Available at: http://www.nature.org/media/pa/tnc_energy_analysis.pdf
Johnson, R. C. 1989. Geologic history and hydrocarbon potential of Late Cretaceous-age, low-permeability reservoirs,
Piceance Basin, western Colorado. U.S. Geological Survey Bulletin 1787-E. 51 pp.
Johnson, R. C., and S. B. Roberts. 2003. The Mesaverde total petroleum system, Uinta-Piceance Province, Utah and
Colorado. U.S. Geological Survey Digital Data Series DDS-69-B. 68 pp.
Juckem, P. F., 2009. Simulation of the regional ground-water-flow system and ground-water/surface-water interaction
in the Rock River Basin, Wisconsin. U.S. Geological Survey Scientific Investigations Report 2009-5094. 38
pp.
Kargbo, D. M, R. G. Wilhelm, and D. J. Campbell. 2010. Natural gas plays in the Marcellus Shale: Challenges and
potential opportunities. Environmental Science and Technology 44: 5679-5684.
Kenny, J. F., N. L. Barber, S. S. Hutson, K. S. Linsey, J. K. Lovelance, and M. A. Maupin. 2009. Estimated use of
water in the United States in 2005. U.S. Geological Survey Circular 1344. 52 pp. Available at:
http://pubs.usgs.gov/circ/1344/.
Kottek, M., J. Grieser, C. Beck, B. Rudolf, and F. Rubel. 2006. World map of the Koppen-Geiger climate
classification updated. Meteorologische Zeitschrift 15(3): 259-263.
Leonard Rice Engineers. 2009. Historic crop consumptive use analysis, Colorado River Basin, final report. Prepared
for the Colorado Water Conservation Board. 44 pp. Available at:
http://cwcbweblink.state.co.us/WebLink/ElectronicFile.aspx?docid=146047&searchid=6a9fd4ff-d203-4864-
a8f5-4fO 14c64994a&dbid=0
Lukas, J., J. Barsugli, N. Doesken, I. Rangwala, and K. Wolter. 2014. Climate change in Colorado—a synthesis to
support water resources management and adaption. Prepared by the Cooperative Institute for Research in
Environmental Sciences, University of Colorado Boulder, for the Colorado Water Conservation Board.
University of Colorado Boulder. 114 pp.
Maloney, K. O., and D. A. Yoxtheimer. 2012. Production and disposal of waste materials from gas and oil extraction
from the Marcellus Shale play in Pennsylvania. Environmental Practice 14(4): 278-287.
Maupin, M.A., J.F. Kenny, S.S. Huston, J.K. Lovelace, N.L. Barber, and K.S. Linsey. 2014. Estimated water use in
the United States in 2010. U.S. Geological Survey Circular 1405. 52 pp. Available at:
http://dx.doi.org/10.3133/cirl405,
McGonigal, K. H. 2005. Nutrients and suspended sediment transported in the Susquehanna River Basin, 2004, and
trends, January 1985 through December 2004. Publication 241. Available at:
http://www.srbc.net/pubinfo/techdocs/Publication_241/techreport241.htm.
Milici, R. C., and C. S. Swezey. 2006. Assessment of Appalachian Basin oil and gas resources: Devonian Shale-
Middle and Upper Paleozoic Total Petroleum System. U.S. Geological Survey Open-File Report 2006-1237.
70 pp. Available at: http://pubs.usgs.gov/of/2006/1237/
Mitchell, A. L., M. Small, and E. A. Gasman. 2013. Surface water withdrawals for Marcellus Shale gas development:
Performance of alternative regulatory approaches in the Upper Ohio River Basin. Environmental Science and
Technology 47(22): 12669-12678.
Montgomery, C. T., and M. B. Smith. 2010. Hydraulic fracturing—history of an enduring technology. Journal of
Petroleum Technology 62(12): 26-32.
Moriasi, D. N., J. G. Arnold, M. W. VanLiew, R. L. Bingner, R. D. Harmel, and T. L. Veith. 2007. Model evaluation
guidelines for systematic quantification of accuracy in watershed simulations. Transactions of the AS ABE
50(3): 885-900.
Murray, K. E. 2013. State-scale perspective on water use and production associated with oil and gas operations,
Oklahoma, U.S. Environmental Science and Technology 47(9): 4918-4925.
Narasimhan, B., R. Srinivasan, J. G. Arnold, and M. Di Luzio. 2005. Estimation of long-term soil moisture using a
distributed parameter hydrologic model and verification using remotely sensed data. Transactions of the
ASAE48(3): 1101-1113.
Nash, J. E., and J. V. Sutcliffe. 1970. River flow forecasting through conceptual models part I—a discussion of
principles. Journal of Hydrology 10(3): 282-290.
104
-------
Water Acquisition for Hydraulic Fracturing May 2015
Nicot, J.-P., and B. R. Scanlon. 2012. Water use for shale-gas production in Texas, U.S. Environmental Science and
Technology 46: 3580-3586.
Nicot, J.-P., B. R. Scanlon, R. C. Reedy, andR. A. Costley. 2014. Source and fate of hydraulic fracturing water in the
Barnett shale: A historical perspective. Environmental Science and Technology 48: 2464-2471.
NOAA (National Oceanic and Atmospheric Administration). 2013. Climate data online. Available at:
http ://www. ncdc .noaa. gov/cdo-web/
PADCNR (Pennsylvania Deptartment of Conservation and Natural Resources). 2014a. Pennsylvania Groundwater
Information System (PaGWIS). Available at:
http://www.dcnr.state.pa.us/topogeo/groundwater/pagwis/index.htm
PADCNR (Pennsylvania Deptartment of Conservation and Natural Resources). 2014b. Pennsylvania Internet Record
Imaging System/Wells Information System (PA*IRIS/WIS): Gateway to detailed oil and gas well
information data. Available at: http://www.dcnr.state.pa.us/topogeo/econresource/oilandgas/pa_iris_home/
PADEP (Pennsylvania Department of Environmental Protection). 2013a. Oil and gas reports. Available at:
http://www.portal.state.pa.us/portal/server.pt/communitv/oil and gas reports/20297.
PaDEP (Pennsylvania Department of Environmental Protection). 2013b. Marcellus Shale development. Fact Sheet
0100-FS-DEP4217. Available at: http://www.elibrary.dep.state.pa.us/dsweb/Get/Document-97683/0100-FS-
DEP4217.pdf.
PADEP (Pennsylvania Department of Environmental Protection). 2014a. State Water Plan. Available at:
http://www.pawaterplan.dep.state.pa.us/StateWaterPlan/WaterDataExportTool/WaterExportTool.aspx.
Query: Primary facility report by year and county.
PADEP (Pennsylvania Department of Environmental Protection). 2014b. State Water Plan. Available at:
http://www.pawaterplan.dep.state.pa.us/StateWaterPlan/WaterDataExportTool/WaterExportTool.aspx.
Query: Chapter 110 (Act 220) registration.
Padowski, J. C., and J. W. Jawitz. 2012. Water availability and vulnerability of 225 large cities in the United States.
Water Resources Research 48(12):W12529.
Perrone, D., J. Murphy, and G. M. Hornberger. 2011. Gaining perspective on the water-energy nexus at the
community scale. Environmental Science and Technology 45: 4228-4234.
Pranter, M. J., and N. K. Sommer. 2011. Static connectivity of fluvial sandstones in a lower coastal-plain setting: An
example from the Upper Cretaceous lower Williams Fork formation, Piceance Basin, Colorado. AAPG
Bulletin 95(6): 899-923.
R Core Team. 2013. R: A language and environment for statistical computing. R Foundation for Statistical
Computing, Vienna, Austria.
Rahm, B. G., and S. J. Riha. 2012. Toward strategic management of shale gas development: Regional, collective
impacts on water resources. Environmental Science and Policy 17: 12-23.
Reeves, H.W. 2010. Water availability and use pilot—A multiscale assessment in the U.S. Great Lakes Basin. U.S.
Geological Survey Professional Paper 1778, 105p.
Reilly, T. E., K. F. Dennehy, W. M. Alley, and W. L. Cunningham. 2008. Ground-water availability in the United
States. U.S. Geological Circular 1323. 70 pp. Available at: http://pubs.usgs.gov/circ/1323/
Richardson, N., M. Gottlieb, A. Krupnick, and H. Wiseman. 2013. The state of state shale gas regulation. RFF Report
(June 2013). Available at: http://www.rff.org/Publications/Pages/default.aspx
Richter, B. 2014. Chasing water. Island Press, Washington D.C.
Risser, D.W., R. W. Conger, J. E. Ulrich, and M. P. Asmussen. 2005. Estimates of ground-water recharge based on
streamflow-hydrograph methods: Pennsylvania. U.S. Geological Survey Open-File Report 2005-1333. 30 pp.
Robson, S.G, and E.R. Banta. 1995. Groundwater atlas of the United States, Arizona, Colorado, New Mexico, Utah,
Colorado Plateau aquifers. U.S. Geological Survey HA 730-C. http://pubs.usgs.gov/ha/ha730/ch c/C-
text8.html
Roy, S. B., P. F. Ricci, K. V. Summers, C.-F. Chung, andR. A. Goldstein. 2005. Evaluation of the sustainability of
water withdrawals in the United States, 1995 to 2025. Journal of the American Water Resources Association
41(5): 1091-1108.
105
-------
Water Acquisition for Hydraulic Fracturing May 2015
Rutledge, A. T. 1998. Computer programs for describing the recession of ground-water discharge and for estimating
mean ground-water recharge and discharge from streamflow data—update. U.S. Geological Survey Water-
Resources Investigations Report 98-4148. 43 pp.
Schlumberger 2014. Directional Drilling. Oilfield glossary.
http://www.glossary.oilfield.slb.com/en/Terms/d/directional_drilling.aspx
Slonecker, E. T., L. E. Milheim, C. M. Roig-Silva, A. R. Malizia, D. A. Marr, and G. B. Fisher. 2012. Landscape
consequences of natural gas extraction in Bradford and Washington Counties, Pennsylvania 2004-2010. U.S.
Geological Survey Open-File Report 2012-1154. 36 pp.
Soeder, D. J., and W. M. Kappel. 2009. Water resources and natural gas production from the Marcellus Shale. U.S.
Geological Survey Fact Sheet 2009-3032. West Trenton, NJ.
Spahr, N. E., L. E. Apodaca, J. R. Deacon, J. B. Bails, N. J. Bauch, C. M. Smith, and N. E. Driver. 2000. Water
quality in the Upper Colorado River Basin, 1996-98. U.S. Geological Survey Circular 1214. Available at:
http://pubs.usgs.gov/circ/circl214/#pdf
SRBC (Susquehanna River Basin Commission). 2003. Guidelines for using and determining pass-by flows and
conservation releases for surface-water and ground-water withdrawal approvals. Susquehanna River Basin
Commission, Harrisburg, PA.
SRBC (Susquehanna River Basin Commission). 2005. Groundwater management plan for the Susquehanna River
Basin. Publication No. 236.
SRBC (Susquehanna River Basin Commission). 2012. Technical guidance for low flow protection related to
withdrawal approvals under Policy No. 2012-01. Susquehanna River Basin Commission, Harrisburg, PA.
Available at: http://www.srbc.net/policies/lowflowpolicy.htm
SRBC (Susquehanna River Basin Commission). 2013a. Public information Available at:
http://www. srbc.net/pubinfo/index.htm. Query water source acquisition volume.
SRBC (Susquehanna River Basin Commission). 2013b. Policies, guidances & regulations. Available at:
http://www.srbc.net/policies/policies.htm
SRBC (Susquehanna River Basin Commission). 2013c. Water Resource Portal Overview. Available at:
http ://www. srbc.net/wrp/
SRBC (Susquehanna River Basin Commission). 2013d. Public information. Available at:
http://www.srbc.net/pubinfo/index.htm. Query: Gas well consumptive use.
SRBC (Susquehanna River Basin Commission). 2013e. Natural gas well development in the Susquehanna River
Basin. Susquehanna River Basin Commission Information Sheet. Available at:
http://www.srbc.net/programs/docs/NaturalGasInfoSheetJan2013.PDF
SRBC (Susquehanna River Basin Commission). 2014. Approved water sources for natural gas development.
Available at: http://www.srbc.net/wrp/ApprovedSourceListReport.aspx
Stantec. 2013. Evaluation of water yields of the Getty Oil Company water system water rights in the Colorado River
Basin, Colorado. Prepared for Chevron U.S. A Inc. JN 87300002. Denver, CO.
Stuckey, M. H. 2006. Low-flow, base-flow, and mean-flow regression equations for Pennsylvania streams. USGS
Scientific Investigations Report 2006-5130. 84 pp.
Tidwell, V. C., P. H. Kobos, L. A. Malczynski, G. Klise, and C. R. Castillo. 2012. Exploring the water-thermoelectric
power nexus. Journal Water Resources Planning and Management 138(5): 491-501.
URS. 2008. Energy development water needs assessment (Phase 1 Report). Prepared for the Colorado,Yampa, and
White River Basin Roundtables. Glenwood Springs, CO.
U.S. Census Bureau. 2010. 2010 Census home. Available at: http://www.census.gov/201 Ocensus/
U.S. EIA (Energy Information Administration). 2009. U.S. crude oil, natural gas, and natural gas liquids proved
reserves, 2009. Available at:
http://www.eia.gov/pub/oil gas/natural gas/data_publications/crude oil natural gas reserves/historical/200
9/pdf/geodistribution.pdf
U.S. EIA (Energy Information Administration). 201 la. The geology of natural gas resources. Today in Energy
(February 14). Available at: http://www.eia.gov/todayinenergy/detail.cfm?id=110
106
-------
Water Acquisition for Hydraulic Fracturing May 2015
U.S. EIA (Energy Information Administration). 201 Ib. North American shale plays. Available at:
http://www.eia. gov/oil_gas/rpd/northamer_gas.jpg
U.S. EIA (Energy Information Administration). 2012. Annual energy outlook 2012. DOE/EIA-0383(2012). U.S.
Department of Energy, Washington, DC. Available at: http://www.eia.gov/forecasts/aeo/
U.S. EIA (Energy Information Administration). 2013a. Annual energy outlook 2013. DOE/EIA-0383(2013). U.S.
Department of Energy, Washington, DC. Available at: http://www.eia.gov/forecasts/aeo/
U.S. EIA (Energy Information Administration). 2013b. Drilling often results in both oil and gas production. Available
at: http://www.eia.gov/todayinenergy/detail.cfm?id=13571#
U.S. EIA (Energy Information Administration). 2014a. Drilling productivity report. Available at:
http://www.eia.gov/petroleum/drilling/. Query: Report data (aggregated by region); July 2014.
U.S. EIA (Energy Information Administration). 2014b. Data for the US shale plays map; Data for the tight gas plays
map. Available at: http://www.eia.gov/pub/oil gas/natural gas/analysis_publications/maps/maps.htm#field
U.S. EIA (Energy Information Administration). 2014c. Marcellus region production continues growth. Available at:
http://www.eia. gov/todayinenergy/detail.cfm?id= 17411
U.S. EIA (Energy Information Administration). 2014d. Colorado State energy profile. Available at:
http://www.eia. gov/state/print.cfm?sid=co
U.S. EPA (Environmental Protection Agency). 2012a. Study of the potential impacts of hydraulic fracturing on
drinking water resources: Progress report. EPA/601/R-12/011. Washington, DC. Available at:
http://www2.epa.gov/sites/production/files/documents/hf-report20121214.pdf
U. S. EPA (Environmental Protection Agency). 2012b. Plan to study the impacts of hydraulic fracturing on drinking
water resources, Quality Management Plan, Revision No. 1. Available at:
http://www2.epa.gov/hfstudv/qualitv-management-plan-revision-no-l-plan-studv-potential-impacts-
hydraulic-fracturing
U.S. EPA (Environmental Protection Agency). 2012c. Safe Drinking Water Information System—federal version
(SDWIS/FED). Available at: http://water.epa.gov/scitech/datait/databases/drink/sdwisfed/index.cfm
U.S. EPA (Environmental Protection Agency). 2013a. Watershed modeling to assess the sensitivity of streamflow,
nutrient, and sediment loads to potential climate change and urban development in 20 U.S. watersheds.
EPA/600/R-12/058F. Washington, DC.
U.S. EPA (Environmental Protection Agency). 2013b. Modeling the impact of hydraulic fracturing on drinking water
resources based on water acquisition scenarios: Phase 2, EPA hydraulic fracturing study quality assurance
project plan (QAPP). Athens, GA. Available at: http://www2.epa.gov/hfstudv/qapp-modeling-impact-
hydraulic-fracturing-drinking-water-resources-based-water-acquisition
U.S. EPA (Environmental Protection Agency). 2014. Study of the potential impacts of hydraulic fracturing for oil and
gas on drinking water resources, Quality Management Plan, Revision No. 2.
U.S. EPA (Environmental Protection Agency). 2015. Analysis of hydraulic fracturing fluid data from the FracFocus
Chemical Disclosure Registry 1.0. Office of Research and Development, Washington D.C. EPA/601/R-
14/003.
USFWS (U.S. Fish and Wildlife Service). 2008. Programmatic biological opinion for water depletions associated with
Bureau of Land Management's Fluid Mineral Program within the Upper Colorado River Basin in Colorado.
December 19. Grand Junction, CO.
USGS (U.S. Geological Survey). 1982. Guidelines for determining flood flow frequency. Bulletin #17B. U.S.
Geological Survey, Office of Water Data Coordination, Reston, VA. Available at:
http://water.usgs.gov/osw/bulletinl7b/dl flow.pdf
USGS (U.S. Geological Survey). 2003. Aquifers. National Atlas of the Unites States, Map Maker. Retrieved
September 16, 2014 from http://www.nationalatlas.gov/mapmaker
USGS (U.S. Geological Survey). 2014a. USGS water use data for the nation. Available at:
http://waterdata.usgs. gov/nwis/wu
USGS (U.S. Geological Survey). 2014b. U.S. Geological Survey surface-water use data for the nation. Available at:
http://waterdata.usgs.gov/nwis/sw
USGS (U.S. Geological Survey). 2014c. WaterWatch. Available at: http://waterwatch.usgs.gov
107
-------
Water Acquisition for Hydraulic Fracturing May 2015
USGS Uinta-Piceance Assessment Team. 2003. Executive summary—assessment of undiscovered oil and gas
resources of the Uinta-Piceance Province of Utah and Colorado, 2002. U.S. Geological Survey Digital Data
Series DDS-69-B.
Vengosh, A., R. B. Jackson, N. Warner, T. H. Darrah, and A. Kondash. 2014. A critical review of the risks to water
resources from unconventional shale gas development and hydraulic fracturing in the United States.
Environmental Science and Technology 48(15): 8334-8348.
Weiskel, P.K. R.M. Vogel, P.A. Steeves, PJ. Zarriello, L.A. DeSimone, andK.G. Ries III. 2007. Water use regimes:
Characterizing direct human interaction with hydrologic systems. Water Resources Research 43:
doi: 10.1029/2006WR005062
Williams, J. H., L. E. Taylor, and D. J. Law. 1998. Hydrogeology and groundwater quality of the glaciated valleys of
Bradford, Tioga, and Potter Counties, Pennsylvania. Pennsylvania Geological Survey, 4th ser, Water
Resource Report 68. 89 pp.
Wilson, J. M, and J. M. VanBriesen. 2012. Oil and gas produced water management and surface drinking water
sources in Pennsylvania. Environmental Practice 14: 288-300.
Yager, R. M., and C. J. Neville. 2002. GFLOW 2000: An analytic element ground water flow modeling system.
Ground Water 40(6): 574-576.
Zhou, Y. 2009. A critical review of groundwater budget myth, safe yield and sustainability. Journal of Hydrology 370:
207-213.
108
-------
Water Acquisition for Hydraulic Fracturing May 2015
GLOSSARY
(numbers in parentheses are references at end of glossary)
Acid mine drainage: Drainage of water from areas that have been mined for coal of other mineral ores. The water has
a low pH because of its contact with sulfur-bearing material and is harmful to aquatic organisms. (2)
Analysis of existing data: The process of gathering and summarizing existing data from various sources to provide
current information on hydraulic fracturing activities. (8)
API number: A unique identifying number for each oil/gas well drilled in the United States. The system was
developed by the American Petroleum Institute. (1)
Aquifer: An underground geological formation, or group of formations, containing water. A source of groundwater
for wells and springs. (2)
Consumptive use: The part of water withdrawn that is evaporated, transpired, incorporated into products or crops,
consumed by humans or livestock, or otherwise removed from the immediate water environment. Consumptive-use
estimates were included in some previous water use U.S. Geological Survey circulars, but were omitted beginning in
2000. Also referred to as water consumed. (6)
Contaminant: A substance that is either present in an environment where it does not belong or is present at levels that
might cause harmful (adverse) health effects. (2)
Conventional reservoir: A reservoir in which buoyant forces keep hydrocarbons in place below a sealing caprock.
Reservoir and fluid characteristics of conventional reservoirs typically permit oil or natural gas to flow readily into
wellbores. The term is used to make a distinction from shale and other unconventional reservoirs, in which gas might
be distributed throughout the reservoir at the basin scale, and in which buoyant forces or the influence of a water
column on the location of hydrocarbons within the reservoir are not significant. (5)
Conveyance loss: Water lost in transit from a pipe, canal, conduit, or ditch by leakage or evaporation. Generally, the
water is not available for further use; however, leakage (e.g., from an irrigation ditch) may percolate to a groundwater
source and be available for further use. (6)
Domestic water use: Water used for indoor household purposes such as drinking, food preparation, bathing, washing
clothes and dishes, flushing toilets, and outdoor purposes such as watering lawns and gardens. Domestic water use
includes water provided to households by a public water supply (domestic deliveries) and self-supplied water. (6)
Drinking water resource: Any body of water, ground or surface, that could (now or in the future) serve as a source
of drinking water for public or private water supplies. (8)
Flowback: After the hydraulic fracturing procedure is completed and pressure is released, the direction of fluid flow
reverses, and water and excess proppant flow up through the wellbore to the surface. (3)
Fluid formulation: The entire suite of products and carrier fluid injected into a well during hydraulic fracturing. (8)
Formation: A geological formation is a body of earth material with distinctive and characteristic properties and a
degree of homogeneity in its physical properties. (2)
Formation water: Water that occurs naturally within the pores of rock. (5)
109
-------
Water Acquisition for Hydraulic Fracturing May 2015
FracFocus: National registry for chemicals used in hydraulic fracturing, jointly developed by the Ground Water
Protection Council and the Interstate Oil and Gas Compact Commission. Serves as an online repository where oil and
gas well operators can upload information about the chemical compositions of hydraulic fracturing fluids used in
specific oil and gas production wells. Also contains spatial information for well locations and information on well
depth and water use. (8)
Freshwater: Water that contains less than 1,000 milligrams per liter (mg/L) of dissolved solids. Generally, water with
more than 500 mg/L of dissolved solids is undesirable for drinking and many industrial uses. (6)
Geographic information system (GIS): A computer system designed for storing, manipulating, analyzing, and
displaying data in a geographic context, usually as maps. (2)
Groundwater: All water found beneath the surface of the land. Groundwater is the source of water found in wells and
springs and is used frequently for drinking. (2)
Horizontal drilling: The intentional deviation of a wellbore from the path it would naturally take to a horizontal
trajectory. A subset of the more general term "directional drilling," used where the departure of the wellbore from
vertical exceeds about 80 degrees. Horizontal lateral sections can be designed to intersect natural fractures or simply
to contact more of the productive formation. (5)
Hydraulic fracturing: The process of using high pressure to pump proppant (e.g. sand) along with a base (e.g. water)
and other fluids into subsurface rock formations in order to improve flow of oil and gas into a wellbore. (8)
Hydraulic fracturing fluid: Specially engineered fluids containing chemical additives and proppant that are pumped
under high pressure into the well to create and hold open fractures in the formation. (8)
Hydraulic fracturing wastewater: Flowback and produced water, where flowback is the fluid returned to the surface
after hydraulic fracturing has occurred but before the well is placed into production and produced water is the fluid
returned to the surface after the well has been placed into production. (8)
Hydraulic fracturing water cycle: The cycle of water in the hydraulic fracturing process, encompassing the
acquisition of water, chemical mixing of the fracturing fluid, injection of the fluid into the formation, the production
and management of flowback and produced water, and the ultimate treatment and disposal of hydraulic fracturing
wastewaters. (8)
Hydraulic gradient: Slope of a water table or potentiometric surface. More specifically, change in the hydraulic head
per unit of distance in the direction of the maximum rate of decrease. (2)
Hydrocarbon: An organic compound containing only hydrogen and carbon, often occurring in petroleum, natural
gas, and coal. (2)
Industrial water use: Water used for fabrication, processing, washing, and cooling. Includes industries such as
chemical and allied products, food, paper and allied products, petroleum refining, wood products, and steel. (6)
Instream use: Water that is used, but not withdrawn, from a surface water source for such purposes as hydroelectric
power generation, navigation, water quality improvement, fish propagation, and recreation. (6)
Irrigation water use: Water that is applied by an irrigation system to assist crop and pasture growth, or to maintain
vegetation on recreational lands such as parks and golf courses. Irrigation includes water that is applied for pre-
irrigation, frost protection, chemical application, weed control, field preparation, crop cooling, harvesting, dust
suppression, leaching of salts from the root zone, and conveyance losses. (6)
Kerogen: The naturally occurring, solid, insoluble organic matter that occurs in source rocks and can yield oil upon
heating. Kerogen is the portion of naturally occurring organic matter that is nonextractable using organic solvents.
Typical organic constituents of kerogen are algae and woody plant material. (5)
110
-------
Water Acquisition for Hydraulic Fracturing May 2015
Livestock water use: Water used for livestock watering, feedlots, dairy operations, and other on-farm needs. Types of
livestock include dairy cows and heifers, beef cattle and calves, sheep and lambs, goats, hogs and pigs, horses and
poultry. (6)
Mining water use: Water used for the extraction of naturally occurring minerals including solids (such as coal, sand,
gravel, and other ores), liquids (such as crude petroleum), and gases (such as natural gas). Also includes uses
associated with quarrying, milling, and other preparations customarily done at the mine site or as part of a mining
activity according to the U.S. Geological Survey. Does not include water associated with dewatering of the aquifer
that is not put to beneficial use. Also does not include water used in processing, such as smelting, refining petroleum,
or slurry pipeline operations. The U.S. Geological Survey includes these processing uses in industrial water use. (6)
Monte Carlo simulation: A technique used to estimate the most probable outcomes from a model with uncertain
input data and to estimate the validity of the simulated model. (8)
Natural gas or gas: A naturally occurring mixture of hydrocarbon and non-hydrocarbon gases in porous formations
beneath the Earth's surface, often in association with petroleum. The principal constituent of natural gas is methane.
(5)
Permeability: Ability of rock to transmit fluid through pore spaces. (1)
Play: A set of oil or gas accumulations sharing similar geologic and geographic properties, such as source rock,
hydrocarbon type, and migration pathways. (1)
Porosity: Percentage of the rock volume that can be occupied by oil, gas, or water. (1)
Produced water: After the drilling and fracturing of the well are completed, water is produced along with the
resource. Some of this water is returned fracturing fluid and some is natural formation water. These produced waters
move back through the wellhead with the gas. (4)
Proppant/propping agent: A granular substance (sand grains, aluminum pellets, or other material) that is carried in
suspension by the fracturing fluid and that serves to keep the fractures open when fracturing fluid is withdrawn after a
fracture treatment. (8)
Public supply water use: Water withdrawn by public and private water suppliers that furnish water to at least 25
people or have a minimum of 15 connections. Public suppliers provide water for a variety of uses, such as domestic,
commercial, industrial, thermoelectric power, and public water use. (6)
Public supply deliveries: Amount of water delivered from a public supplier to users for domestic, commercial,
industrial, thermoelectric power, or public-use purposes. (6)
Public water system (PWS): A system that provides water to the public for human consumption through pipes or
other constructed conveyances. A PWS, per EPA's definition, must have at least 15 service connections or regularly
serve at least 25 people. (2)
Public water use: Water supplied from a public supplier and used for such purposes as firefighting, street washing,
flushing of water lines, and maintaining municipal parks and swimming pools. Generally, public-use water is not
billed by the public supplier. (6)
Q?,io: The average minimum streamflow that can be expected for seven consecutive days once every 10 years,
computed from measured flow records.
Safe yield: Commonly used in efforts to quantify sustainable groundwater development. The term should be used
with respect to specific effects of pumping, such as water-level declines, reduced streamflow, and degradation of
water quality. (7)
Scenario evaluation: Exploration of realistic, hypothetical scenarios related to hydraulic fracturing activities using
computer models. Used to identify conditions under which hydraulic fracturing activities may adversely impact
drinking water resources. (8)
m
-------
Water Acquisition for Hydraulic Fracturing May 2015
Science Advisory Board: A federal advisory committee that provides a balanced, expert assessment of scientific
matters relevant to EPA. An important function of the Science Advisory Board is to review EPA's technical programs
and research plans. (2)
Self-supplied water use: Water withdrawn from a groundwater or surface water source by a user rather than being
obtained from a public supply. (6)
Service company: A company that assists well operators by providing specialty services, including hydraulic
fracturing. (8)
Shale: A fine-grained sedimentary rock composed mostly of consolidated clay or mud. Shale is the most frequently
occurring sedimentary rock. (5)
Source water: Water withdrawn from surface or ground water, or purchased from suppliers, for hydraulic fracturing.
(8)
Statistical analysis: Analyzing collected data for the purposes of summarizing information to make it more usable
and/or making generalizations about a population based on a sample drawn from that population. (2)
Surface water: All water naturally open to the atmosphere (rivers, lakes, reservoirs, ponds, streams, impoundments,
seas, estuaries). (2)
Surfactant: Used during the hydraulic fracturing process to decrease liquid surface tension and improve fluid passage
through the pipes. (8)
Tight sands: A geological formation consisting of a matrix of typically impermeable, non-porous tight sands. (8)
Total dissolved solids: The quantity of dissolved material in a given volume of water. (2)
Unconventional resource: An umbrella term for oil and natural gas produced by means that do not meet the criteria
for conventional production. What has qualified as unconventional at any particular time is a complex function of
resource characteristics, the available exploration and production technologies, the economic environment, and the
scale, frequency, and duration of production from the resource. Perceptions of these factors inevitably change over
time and often differ between users of the term. At present, the term is used in reference to oil and gas resources
whose porosity, permeability, fluid trapping mechanism, or other characteristics differ from conventional sandstone
and carbonate reservoirs. Coalbed methane, gas hydrates, shale gas, fractured reservoirs, and tight gas sands are
considered unconventional resources. (5)
Water use: Pertains to the interaction of humans with and influence on the hydrologic cycle; includes elements such
as water withdrawal, delivery, consumptive use, wastewater release, reclaimed wastewater, return flow, and instream
use. (6)
Water withdrawal: Water removed from the ground or diverted from a surface water source for use. (6).
Well files: Files that generally contain information on all activities conducted at an oil and gas production well. These
files are created by oil and gas operators. (8)
Well operator: A company that ultimately controls and operates oil and gas wells. (8)
112
-------
Water Acquisition for Hydraulic Fracturing May 2015
Glossary References
1. Oil and Gas Mineral Services. 2010. Oil and gas terminology. Available at: http://www.mineralweb.com/librarv/oil-
and-gas-terms/
2. U.S. Environmental Protection Agency. 2006. Terminology services: Terms and acronyms. Available at:
http://iaspub.epa. gov/sor_internet/registry/termreg/home/overview/home.do
3. New York State Department of Environmental Conservation. 2011. Supplemental generic environmental impact
statement on the Oil, Gas and Solution Mining Regulatory Program: Well permit issuance for horizontal
drilling and high-volume hydraulic fracturing to develop the Marcellus Shale and other low-permeability gas
reservoirs. Revised draft. Available at: ftp://ftp.dec.state.ny.us/dmn/download/OGdSGEISFull.pdf
4. Ground Water Protection Council and ALL Consulting. 2009. Modern shale gas development in the United States:
A primer. Prepared for the U.S. Department of Energy. Available at:
http://energv.gov/sites/prod/files/2013/03/fO/ShaleGasPrimer Online 4-2009.pdf
5. Schlumberger. The oilfield glossary: Where the oilfield meets the dictionary. Available at:
http://www.glossary.oilfield.slb.com/
6. Kenny, J. F., N. L. Barber, S. S. Hutson, K. S. Linsey, J. K. Lovelance, and M. A. Maupin. 2009. Estimated use of
water in the United States in 2005. USGS Circular 1344. 52 pp. Available at: http://pubs.usgs.gov/circ/1344/
7. Reilly, T. E., K. F. Dennehy, W. M. Alley, and W. L. Cunningham. 2008. Ground-water availability in the United
States. USGS Circular 1323. 70 pp. Available at: http://pubs.usgs.gov/circ/1323/
8. U.S. Environmental Protection Agency. 2012. Study of the potential impacts of hydraulic fracturing on drinking
water resources: Progress Report. EPA/601/R-12/011. Available at:
http://www2.epa.gov/sites/production/files/documents/hf-report20121214.pdf
113
-------
Water Acquisition for Hydraulic Fracturing May 2015
[This page intentionally left blank/
114
-------
Water Acquisition for Hydraulic Fracturing May 2015
Appendix A. Compendium of Data Sources
Appendix A-l
-------
Water Acquisition for Hydraulic Fracturing May 2015
[This page intentionally left blank/
Appendix A-2
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Appendix A lists the data sources used in this study. The data sources have been organized into three tables. Table A-l contains the national data sources used for
one or both of the study areas. Table A-2 contains the data sources specific to the Colorado study site. Table A-3 contains the data sources specific to the
Pennsylvania study area.
Table A-1. National Data Sources
National Data Source
USGS National Water Information
System (NWIS) Surface Water Data
http://waterdata.usgs.gov/nwis/sw
USGS National Water Information
System (NWIS) Ground water Data
http://waterdata.usgs.gov/nwis/gw
USGS National Water Information
System (NWIS) Water Use Data
http://waterdata.usgs.gov/nwis/wu
NOAA National Climatic Data Center
(NCDC)
http://www.ncdc.noaa.gov/cdo-web/
U.S. Energy Information
Administration (EIA)
http://www.eia.gov/
U.S. Energy Information
Administration (EIA)
http://www.eia.gov/
as compiled by the U.S. EPA
FracFocus Database
as compiled by The Cadmus Group,
Inc.fortheU.S. EPA
FracFocus 2.0
http://fracfocus.org/
FracTracker Alliance
http://www.fractracker.org/
EPA Safe Drinking Water Information
System-Federal version (SDWIS/FED)
http://water.epa.gOV/scitech/datait/d
atabases/drink/sdwisfed/index.cfm
Last
Access
Date
9/8/2014
2/26/2014
3/14/2014
11/7/2013
5/1/2014
9/12/2012
11/22/2013
5/31/2014
2/21/2014
6/1/2012
Information Obtained
- location of stream gages in the Susquehanna River Basin
(SRB) and the Upper Colorado River Basin (UCRB)
- streamflow time series for SRB and UCRB
- location of water wells
— well depth and aquifer water level elevations in wells
— 2005 water use data by county
— location and elevation of weatherstations in SRB and UCRB
- meteorological data (precipitation, airtemperature,
evapotranspiration) for SRB and UCRB
— location of shale and tight gas basins in the United States
— projections of unconventional gas annual drilling rates
generated with the National Energy Modeling System
(NEMS)— Annual Energy Outlook 2012 (AEO2012) reference
case
— data on total fluid volume used for O&G wells between
January 2011 and February 2013
— data on fluid volume in O&G wells fractured after 2010
(location of wells, fracture date, total fluid volume used)
- data on permitted oil and gas wells and drilled wells (date,
location, and number of wells)
- location of public drinking water systems in SRB
Use in Report
- deterministic hydrologic modeling
- baseflow separation analysis
- estimation of average thickness of drinking water aquifer in
SRB and UCRB
— county-level estimation of water use in SRB
- deterministic hydrologic modeling
— input for SRB regional equations
— general reference; cartography
- hydraulic fracturing future drilling projections
- computation of average flu id volumes used per well for
hydraulic fracturing in Garfield, Mesa, and Rio Grande
Counties in Colorado
- computation of average flu id volumes used per well for
hydraulic fracturing in Garfield, Mesa, and Rio Grande
Counties in Colorado
- well counts
— mapping location of oil and gas activity in UCRB
- mapping the general location of public water suppliers in SRB
Appendix A-3
-------
Water Acquisition for Hydraulic Fracturing
May 2015
National Data Source
USDA National Agricultural Statistics
Service
http://www.nass.usda.gov/
Multi-Resolution Land
Characteristics Consortium (MRLC)
http://www.mrlc.Hov/index.php
National Elevation Dataset (NED)
http://ned.usHS.Hov/
GeoCommunity
— :
USGS Historical Topographic Map
storical/index.html
National Hydrography Dataset (NHD)
http://nhd.usHS.Hov/index.html
EPA Center for Exposure Assessment
Modeling (CEAM)
assessment-models/whaem2000-
bbm-files-us
Better Assessment Science
Integrating Point and Nonpoint
Sources (BASINS)
http://water.epa.Hov/scitech/datait/
models/basins/index.cfm
USDA Natural Resources
Conservation Service
l/nrcs/detail/soils/survev/?cid=nrcsl4
2p2 053629
Last
Access
Date
2/1/2014
(provided
bySRBC)
8/22/2013
3/31/2014
12/10/2013
5/22/2014
9/6/2013
11/4/2013
8/28/2012
3/19/2014
(accessed
through
ArcSWAT)
Information Obtained
- 2010 crop-specific land cover data in raster format at 30
meters spatial resolution (2010 Cropland Data Layer)
— 2007 Census of Agriculture statistics
- 2006 land cover classification in raster format at 30 meters
spatial resolution (National Land Cover Database [NLCD]
2006) for SRB and UCRB
— ground elevation data in rasterformat at 1/3 arc-second
(about 10 meters) and 1 arc-second (about 30 meters)
resolution for SRB and UCRB
(approximately 3 meters) resolution for SRB
- medium resolution (1:100,000 scale) vector elevation
contours (USGS Digital Line Graphs — Hypsography)forSRB
and UCRB
- drainage networks from digital version of historical
— general site information
- high-resolution (1:24,000 or larger scale) vector stream
network for SRB and UCRB
— medium-resolution (1:100,000 scale) vector surface
in binary base map (BBM) format for SRB and UCRB
— low-resolution (1:500,000 scale) vector stream network —
Reach File 1 (RF1)
- generalized (1:250,000 scale) polygon-based soils information
(STATSGO2) for SRB and UCRB
Use in Report
— surface water use intensity index (SUI) calculations in SRB
— deterministic hydrologic modeling
- input for SRB regional equations
- deterministic hydrologic modeling
- input for SRB regional equations
— extraction of elevations for stream channels
- cartography
- extraction of elevations for stream channels
- reference stream network when generating the ArcSWAT
stream network
- evidence of perennial stream channels in Colorado
— cartography
Appendix A-4
-------
Water Acquisition for Hydraulic Fracturing
May 2015
National Data Source
United States Census Bureau
http://www.census.gov/
United States Census Bureau —
American Factfinder
http://factfinder2.census.gOV/faces/n
av/isf/pages/communitv facts.xhtml
tfnone
Last
Access
Date
11/13/2013
2/1/2014
(provided by
SRBC)
04/16/2014
Information Obtained
- 2010 population data and census block shapefiles
(TIGER/Line) for Bradford, Lycoming, Sullivan, and Tioga
Counties in Pennsylvania and Garfield, Mesa, Pitkin, and Rio
Blanco Counties in Colorado
- 2010 population data and housing unit counts by census
block (TIGER/Line shapefile) forTowanda Creek basin in
Pennsylvania
- 2012 population estimates for Colorado municipalities
(Parachute and DeBeque); only 2010 data available for
Battlement Mesa, Colorado
Use in Report
— estimation of private water well use
- SUI calculations in SRB
Table A- 2. Colorado Data Sources
Colorado Data Source
Colorado's Decision Support
Systems (CDSS)
http://cdss.state.co.us/onlineTools
/Pages/OnlineToolsHome.aspx
Colorado's Decision Support
Systems (CDSS)
http://cdss.state.co.us/GIS/Pages/
GISDataHome.aspx
Colorado Municipal League
Water & Wastewater Su rvey
http://www.cml.org/water-
wastewater
Colorado Division of Water
Resources (DWR)
http://water.state.co.us/Home/Pa
ges/default.aspx
Last Access
Date
4/19/2014
3/25/2014
3/11/2014
2/21/2014
Information Obtained
— records of water rights and administrative structures in
UCRB
- water withdrawal history from administrative structures in
UCRB
— shapefiles of water structures and water management
boundaries in UCRB
- information on selected Colorado public drinking water
systems from 2012 water and wastewater survey
— Colorado water use yearly statistics by water division
(roughly equivalent to a major river basin)
— location of drilled household wells
Use in Report
- quantification of water use in various areas of UCRB;
identification of administrative structures owned by oil and gas
- reference; locate diversion structures in UCRB
- background information
- SUI calculations in UCRB
Appendix A-5
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Colorado Data Source
Colorado Oil and Gas
Conservation Commission
(COGCC)
http://cogcc.state.co.us/
StreamStats (Colorado
application)
http://water.usgs.gov/osw/stream
stats/colorado.html
Last Access
Date
6/4/2014
04/01/2014
Information Obtained
- location of permitted oil and gas wells in UCRB
- number of well completions priorto 2010
— number of producing wells
— produced water volumes
— names of oil and gas operators in UCRB
— line shapefile representing directionally drilled oil and gas
wells in Colorado
- drainage area delineations for selected water diversion
structures in Parachute and Roan Creeks, Colorado
Use in Report
— general analysis of hydraulic fracturing use metrics
- SUI calculations in UCRB
- cartography
- help identify administrative structures own by oil and gas in
Garfield, Mesa, Rio Blanco, Moffat, Routt, and Jackson Counties
in Colorado
— mapping of well directional lines in Parachute Creek upper valley
- area-weighted extrapolation to ungaged diversion structures in
UCRB
Table A- 3. Pennsylvania Data Sources
Pennsylvania Data
Source
Susquehanna River Basin
Commission (SRBC)
Personal communication
SRBC River Basin Commission
http://www.srbc.net/
Pennsylvania Department of
Environmental Protection
(PADEP) State Water Plan
http://www.pawaterplan.dep.stat
e.pa.us/StateWaterPlan/WaterDat
aExportTool/WaterExportTool.asp
X
Last Access
Date
4/2/2014
2/4/2014
4/15/2014
Information Obtained
- well pad consumption volumes in SRB
— water withdrawal volumes and location of permitted SRB
surface water withdrawal sites
- data on oil and gas withdrawal permits in SRB (permit
requirements including passbys and capacity)
— location, total depth and pump ratings of permitted
freshwater wells in SRB
— annual water use reports from public water systems in
Pennsylvania, including amounts sold to the oil and gas
industry
Use in Report
- generate hydraulicfracturing use scenarios in SRB
— generate hydraulicfracturing use scenarios in SRB
— SUI calculations
- area-weighted extrapolation to ungaged SRB withdrawal sites
- general analysis of hydraulicfracturing use metrics in SRB
- mapping of self-supplied well fields
— groundwater impact analysis
Appendix A-6
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Pennsylvania Data
Source
Last Access
Date
Information Obtained
Use in Report
Pennsylvania Department of
Environmental Protection
(PADEP) North-Central Regional
Office, Williamsport
1/20/2014
depth of fresh groundwater (DFGW) in oil and gas wells
reported by industry (paper records)
- estimation of average thickness of fresh groundwater in SRB
Pennsylvania Department of
Environmental Protection
(PADEP) Oil and Gas Reports
http://www.portal.state.pa.us/por
tal/server.pt/communitv/oil and
gas reports/20297
7/29/2013
location of unconventional drilled hydraulicfracturing wells
in Pennsylvania
mapping of unconventional drilled hydraulic fracturing wells in
Pennsylvania
PADCNR—Pennsylvania
Groundwater Information System
(PaGWIS)
http://www.dcnr.state.pa.us/topo
geo/groundwater/pagw is/index.ht
5/22/2014
drilled depth, static water elevations in water wells, pump
ratings in SRB
estimation of thickness of shallow aquifers
groundwater modeling
PADCNR—The Pennsylvania
Internet Record Imaging
System/Wells Information System
(PA*IRIS/WIS)
http://www.dcnr.state.pa.us/topo
geo/econresource/oilandgas/pa ir
is home/
2/6/2014
- DFGW in oil and gas wells in SRB
— estimation of average thickness of fresh groundwater in SRB
StreamStats (Pennsylvania
application)
http://water.usgs.gov/osw/stream
stats/pen nsvlvania.html
02/07/2014
drainage area delineations for selected oil and gas
withdrawal sites in SRB
— area-weighted extrapolation to ungaged SRB withdrawal sites
Appendix A-7
-------
Water Acquisition for Hydraulic Fracturing May 2015
[This page intentionally left blank.
Appendix A-8
-------
Water Acquisition for Hydraulic Fracturing May 2015
Appendix B. Surface Water Hydrology
Appendix B-l
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Contents
Overview
Deterministic Watershed Modeling.
Description of Modeling Steps
Pennsylvania: Susquehanna River and Towanda Creek 8
Study Area Hydrometeorology 8
Deterministic Watershed Modeling: Towanda Creek 8
Empirical Streamflow Estimation: Susquehanna River Basin 17
Colorado: Parachute and Roan Creeks 19
Study Area Hydrometeorology 19
Deterministic Watershed Modeling: Upper White River, Parachute Creek, and Roan Creek 20
Empirical Streamflow Estimation: Parachute and Roan Creeks 26
References.....,,. 27
Overview
This research required measurements or estimates of Streamflow at
numerous points throughout the Pennsylvania and Colorado study areas.
While all available U.S. Geological Survey (USGS) Streamflow data in both
locations were incorporated, active USGS gages tend to be located on
streams and rivers draining larger watersheds, and there was limited data
availability at oil and gas (O&G) water withdrawal sites, particularly on
tributary streams. For quantification of water availability at und sites, a
combination of empirical and deterministic modeling methods were
implemented to generate estimates of daily Streamflow (Booker and Woods
2014).
Physically-based, deterministic watershed modeling relies on mathematical
formulations of watershed processes within a given system. Deterministic
models are typically complex software tools that reproduce key components
of the hydrologic cycle, incorporating a combination of mathematical
representations of watershed processes, along with varying degrees of
locally-observed spatial and time series data (Fig. B-l). These models rely
on observed Streamflow data only to compare simulated flows for
calibration and validation. Empirical hydrologic methods use recorded
Streamflow observations from local and regional hydrology to simulate
Streamflow. Typical empirical methods involve either direct extrapolation
of data from gaged streams to ungaged streams, based on similarities in
watershed characteristics, or statistical methods of varying sophistication.
This project acquired Streamflow data using all approaches depending on
specific applications as discussed in this appendix and in the main body of the
report. Generally, deterministic modeling generated longer-term records for
Resalts
Figure B-l. Mechanistic watershed
modeling: General conceptual
diagram (U.S. EPA 2011).
Appendix B-2
-------
Water Acquisition for Hydraulic Fracturing
May 2015
surface water use intensity (SUI) calculations and scenario analysis.
Deterministic Watershed Modeling
Analyses of SUI in Pennsylvania and Colorado required streamflow estimation at all permitted O&G water
withdrawal sites within the Susquehanna River Basin (SRB). Additionally, streamflow was estimated at a large
number of random locations within the focus study watersheds, including many very small subwatersheds. The
watershed modeling approach used herein was designed for maximum repeatability— preference was given to
commonly used and freely available technologies (Soil & Water Assessment Tool, or SWAT; and Hydrological
Simulation Program—Fortran, orHSPF), while detailed, region-specific scenarios were avoided. While many of the
same methodologies were implemented in both locations, pronounced regional differences in data availability dictated
what methods could be used for extrapolating streamflow from gaged to ungaged locations in the individual study
areas.
*| Transmission Losses |
dTlPond/Resevoir Water Balance
Revap | Seepage | | Return Flow |— -
Figure B-2. SWAT conceptual diagram (Reprinted with permission: Neitsch eta/. 2011).
Appendix B-3
-------
Water Acquisition for Hydraulic Fracturing May 2015
Table B-l. HSPF and SWAT input data requirements
Input Format HSPF SWAT Source*
Topography spatial DEM (30m) DEM (30m) National Elevation Dataset
Soils spatial - STATSGO2 Natural Resources Conservation Service
Land Use spatial NLCD(30m) NLCD(30m) Multi-Resolution Land Characteristics Consortium
NOAA National Climatic Data Center:
Temperature time series max and min hourly mean daily USC00361212 (Canton PA), USC00368905 (Towanda PA),
USC00050214 (Altenbern CO) ,USC00057031 (Rifle CO),
USROOOOCRIF (Rifle CO), USC00053359, (Glen wood Springs
Precipitation time series hourly total daily total CO),Center:USS0007K02S (Burro Mountain CO),
USC00055484 (Meeker CO), USS0007J05S (Ripple Creek CO)
*see Appendix Afor details on data sources
SWAT is a widely used, freely available, process-based, deterministic, semi-distributed watershed model, developed
by Texas A&M University and the U. S. Department of Agriculture (Neitsch et al. 2011). Minimal model inputs
include land use, topography, and soil coverages, along with daily mean temperature and total precipitation (Table B-
1). Pre-processing of spatial data is most commonly performed using the ArcSWAT extension for ArcGIS (SWAT
2014). Details of specific pre-processing steps are presented in the following section. SWAT uses a semi-distributed
approach to simulate water quantity and quality. Within each topographically defined subbasin, the following
variables are lumped into discontiguous Hydrologic Response Units (HRUs): spatially explicit land use, topography,
and soils data. Each HRU is composed of members sharing similar rainfall-runoff responses. These aggregation
processes are user-guided, allowing some degree of control over the model's spatial resolution (Fig. B-2). SWAT is
primarily an infiltration-excess flow model that separates precipitation into two components: that which infiltrates to
shallow and deep subsurface storage, and that which occurs as overland flow. To partition runoff vs. non-runoff flow
components, SWAT uses the SCS curve number II (Hawkins et al. 2009). SWAT has successfully been used to assess
water quantity response to human watershed manipulation in many settings, across a range of watershed scales
(Ahmad et al. 2012; Faramarzi et al. 2013; Gabriel et al. 2014; Price et al. 2014). SWAT's default operation uses a
daily timestep.
HSPF is a widely used, freely available, process-based, deterministic, semi-distributed watershed model for
streamflow and water quality simulation that is endorsed by the Federal Emergency Management Agency (Atkins et
al. 2005; FEMA 2013; U.S. EPA 2013; Bicknell 1997). HSPF water balance processes distinguish between surface
runoff, throughflow, and groundwater storage, which are determined by processes of infiltration, loss to deeper
groundwater, and soil storage (Golden et al. 2014). The model inputs are digital representations of land use and
topography, time series of meteorological data for simulation forcing, and observed streamflow time series for model
calibration (Table B-l). HSPF is a "semi-distributed parameter" model, meaning there is limited spatial discretization
of watershed processes (Johnson et al. 2003). It has been established as a reliable water quantity modeling tool for a
wide range of settings and applications (Buchanan et al. 2013; He et al. 2013; Johnson et al. 2003; Kim and Chung
2014). Like SWAT, HSPF is primarily an infiltration-excess model that separates precipitation into infiltrated/non-
infiltrated components, but HSPF uses the Philip method (Philip 1957) as opposed to curve number (Fig. B-3).The
model works via three major modules: (1) PERLND, which simulates terrestrial processes on pervious land areas; (2)
IMPLND, which simulates terrestrial processes on impervious surfaces; and (3) RCHRES, which simulates linkages
between the stream network and terrestrial segments, and routes streamflow through water bodies (U.S. EPA 2013).
This multi-module configuration is partitioned via area-weighting of watershed characteristics, but does not directly
model watershed processes in a spatially explicit manner.
Appendix B-4
-------
Water Acquisition for Hydraulic Fracturing
May 2015
HSPF requires hourly maximum and minimum temperature and hourly precipitation totals, but this study used
companion programs to disaggregate daily data to an hourly timestep. Preprocessing for HSPF is most commonly
performed using BASINS, an open-source program that facilitates data acquisition and model setup for multiple
models (U.S. EPA 2013). However, for the sake of consistency, this study used the processed spatial data from the
SWAT preprocessing steps described herein to initialize HSPF.
Actual ET
3rd-«
5th
4th
1st
Potential ET
Precipitation
Temperature
Radiation
Wind, Dewpoint
HSPF PERLND Module
7CTD
LZSN
Delayed Infiltration
Zone
Storage
AGWETP
, ;TO
Figure B-3. HSPF pervious surface conceptual diagram (Atkins et al. 2005)
Description of Modeling Steps
Of the seven steps detailed below, steps 1-4 were followed for setup and initialization of both SWAT and HSPF.
After initializing both models, preliminary simulations and calibrations were performed using HSPF and SWAT, but
detailed SUI analyses were ultimately performed using only HSPF due to significantly faster run times and slightly
better simulation accuracy. All of the following steps (1-7) were followed for HSPF setup, calibration, and
simulation, with details provided for each study area in subsequent sections of this appendix:
1. Setup of Model Inputs: Standard model inputs of topography, land cover, soils, meteorology, and observed
streamflow were obtained for all study watersheds (Table B-l).
2. Stream Network Definition and Subbasin Delineation: All spatial preprocessing was performed using ArcSWAT,
which provides a user-friendly environment for customized configuration of stream network and subbasin delineation.
Equivalent tools are available for HSPF in BASINS (U.S. EPA 2013). Stream network definition is achieved by
estimating a draining area threshold required for perennial streamflow, and subbasins are delineated upstream of every
node in the stream network topology. Given the importance of low flows to these research objectives, establishment of
a simulation stream network that closely approximated the actual perennial stream network was deemed essential. The
National Hydrography Dataset (NHD) high-resolution network was used as a baseline, and any additional regionally
available perennial streamflow data were incorporated. Stream networks were iteratively generated in ArcSWAT
Appendix B-5
-------
Water Acquisition for Hydraulic Fracturing May 2015
using a range of accumulation area thresholds, until a visual match between observed and generated networks was
achieved. In ArcSWAT, subbasins are delineated upstream of every node in the stream network; changes to subbasin
delineation alters the stream network and vice-versa, which is key for model operation of both HSPF and SWAT.
Of the seven steps detailed below, steps 1-4 were followed for setup and initialization of both SWAT and HSPF.
After initializing both models, preliminary simulations and calibrations were performed using HSPF and SWAT, but
detailed SUI analyses were ultimately performed using only HSPF due to drastically faster run times and slightly
better simulation accuracy. All of the following steps (1-7) were followed for HSPF setup, calibration, and
simulation, with details provided for each study area in subsequent sections of this appendix:
3. Pour Point Designations. SUI calculations were required across a wide range of watershed areas. Thus, subbasins
were aggregated into a series of larger-order watersheds based on a Strahler ordering scheme performed on the
ArcSWAT-delineated stream network. Most outlets of these aggregated streams were retained as simulation pour
points ("pour point" is an ArcGIS term for the point of accumulated flow when delineating drainage basins using the
Watershed tool, i.e., flow simulation locations at subbasin outlets). Subbasins were omitted as pour points if they (1)
contained defined stream channel for less than 20% of basin length, (2) contained major ponds or impoundments
proportional to watershed area, and (3) were tributary watersheds to other subbasins of the same order (i.e., did not
flow directly into a subbasin of higher order).
4. Sensitivity Analysis, Preliminary Calibration, and Watershed Model Selection. Deterministic watershed models
generally contain "free parameters," whose values can be mathematically optimized to best match observed
streamflows. Automated parameter-fitting procedures are widely used to streamline the process of identifying the best
combination of parameters. For example, streamflow simulation using SWAT allows for modification of 27 free
parameters, whose values can be optimized to best match observed streamflows during a given period of record.
Sensitivity analysis and preliminary calibration were performed in SWAT-CUP, a companion program to SWAT
(Abbaspour 2009).
The drawback of automated procedures is that they may introduce physically unlikely parameter combinations among
the multiple mathematical solutions that exist to optimize the parameter set. Such concerns are also known as
equifinality and overfitting (Beven 2006; Matott et al. 2009). To minimize overfitting, we retained a small number of
parameters for calibration: the most sensitive parameters, ranked by the amount of variability they explained, stopping
once (1) 90% of total sensitivity was explained and (2) a minimum of six model parameters were included. These
stopping rules represent a compromise between model parsimony and exploring the full calibration space among the
models' available free parameters.
For HSPF, sensitivity was calculated with PEST, a model-independent parameter sensitivity and optimization program
(Watermark Numerical Computing 2005). PEST uses a Jacobian matrix approach to sensitivity estimation, which is a
vector calculus, partial-derivative determination approach. Sensitivity is quantified using the following equation:
EquatlonB-1
m m
where, st = composite sensitivity of ith parameter, / = Jacobian matrix, Q = diagonal matrix whose diagonal element
is the square of the weight (all weight is unity in this case), m = number of modeled value, Mj: y'th modeled value,
and xt = value of ith parameter.
For preliminary calibration and model selection, agreement was evaluated between simulated and observed
streamflow using the standard Nash-Sutcliffe (Nash and Sutcliffe 1970) efficiency criterion (NS):
NS = 1-
5; - O)2
Equation B-2
Appendix B-6
-------
Water Acquisition for Hydraulic Fracturing May 2015
where /' represents each timestep in the series, O = observed streamflow, and gages = simulated streamflow. While
there are limits to implementation of NS, it is a widely used fit statistic and is available in both SWAT-CUP and
PEST. For these preliminary analyses, a minimum of 500 Latin hypercube-derived parameter combinations were
tested.
Simulations for the calibration period were more than 50 times longer with SWAT than HSPF (~45 minutes vs. < 1
minute per run), an impediment to using advanced calibration procedures, which require many thousands of runs for
each watershed. The prohibitively long run times for calibrating SWAT using the Monte Carlo approach and
somewhat greater simulation accuracy using HSPF (NS values of 0.75and 0.60 or HSPF and SWAT, respectively),
lead the project to conclude that HSPF was the better alternative for deterministic modeling in this particular
application (Price et al. in preparation).
5. Calibration. Next, the reduced parameter set was used for a rigorous calibration of HSPF to identify the optimized
parameter set that best matched simulated and observed streamflow at the USGS-d outlets. The models were
calibrated and validated using a standard split-sample approach (Andreassian et al. 2009), meaning that part of the
observed record was reserved for independent model evaluation. From the total observed record of 1987-2012, the
calibration period used was 1997-2012, while!987-1996 was withheld as a validation period.
While NS was well-suited toward preliminary model comparison, there were concerns about using NS for the model
application itself, given its bias toward fitting flood peaks at the expense of low flows (Krause et al. 2005; Price et al.
2012). Taking the NS of log-transformed flow values (NSta) is an established method of removing flood bias in model
calibration. As it was desired to retain the influence of low flows, while not completely losing the water balance
considerations associated with proper estimation of flood magnitudes, NS and NSta were combined to form a
Weighted Nash Sutcliffe (WNS) fit criterion:
„,. _ , _ Lt=i\.™MJ ~ m^tjj Equation B-3
VVJ^ J. ,_,„ r, f m N , x—-.in
Zr=iDn(Oi) - ln(0)]2
M/W5 = WNS • NS + (1 - Mfo) • W5(n Equation B-4
where /' represents each timestep in the series, O is the observed streamflow, gages is the simulated streamflow, and W
is the weighting factor (determined a posteriori by minimizing the root-mean-square error, or RMSE, among
weighting scenarios).
A Latin hypercube-derived Monte Carlo analysis (10,000 15-year simulations on a daily timestep) was used to
identify the parameter sets associated with the highest WNS scores in each watershed. Additionally, calibrations were
cross-validated by applying the parameterizations to simulate streamflow at a nearby USGS gaging station and
evaluating model fit.
6. Streamflow Simulation. Once models were calibrated and parameter sets were established, streamflow was
simulated across the full range of pour point drainage areas in each of the three study watersheds, Towanda Creek in
the SRB and Parachute and Roan Creeks in the Upper Colorado River Basin (UCRB). The input parameters values of
land use, vegetation and soils were tailored to each subbasin. Daily flows were simulated for the period spanning
1986-2012 because this water the period of available weather data. Simulations for 1986 were discarded as model
spin-up, and 1987-2012 were retained for further analysis.
7. Simulation Uncertainty. Simulation uncertainty was characterized by identifying the range of flows and flow
statistics in the subset of simulations that produced daily WNS of 0.3 or greater. Additionally, any simulations for
which annual water yield (as mean streamflow) differed from observed by more than 15% were excluded.
Appendix B-7
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Pennsylvania: Susquehanna River and Towanda Creek
Study Area Hydrometeorology
Bradford County and the Towanda Creek watershed lie entirely within Pleistocene glacial margins. This area is
underlain by shales and sandstones, without significant occurrence of karst. Soils are primarily well-drained loamy
inceptisols and entisols forming in glacial and alluvial parent material (Grubb 1986). Topographic relief in the
Towanda Creek watershed is moderate, with elevations ranging from 770 feet above sea level (ASL) at the outlet, to
2436 feet ASL along the northwestern divide.
The climate of Bradford County is classified as "cold humid" (Df) in the revised Koppen-Geiger classification system
(Kottek et al. 2006). Long-term (1981-2010) annual average precipitation is 35.7 inches in Towanda, PA (station ID
= USC00368905; station elevation = 751 feet ASL), with the lowest monthly average (1.94 inches) occurring in
February and highest monthly average (3.60 inches) occurring in July (Fig. B-4). Minimum and maximum monthly
mean temperatures occur in January (24.8°F) and July (70.4°F), respectively (Arguez et al. 2012).
Towanda Creek above Monroeton, PA (USGS 01532000, 215 mi2) consists of two watersheds and drains into the
Susquehanna River in Towanda, PA. USGS-observed streamflow during the modeling and hydraulic fracturing
activity periods demonstrate the absence of major snowpack/snowmelt cycles and general hydrological stationarity
during the study period (Fig. B-5A). Lowest flows typically occur in late fall. Remnants of major Atlantic and Gulf
storms occasionally affect the region during tropical storm season (NOAA Coastal Services Center 2014), but the
influence of these is not frequent enough to substantially alter the low-flow season as a whole. Comparison of study-
period streamflows with long-term records shows that the hydraulic fracturing analysis period includes one very wet
year and one very dry year (Fig. B-5B).
Figure B-4. Towanda Creek
watershed mean annual
precipitation, 1981-2010 National
Climatic Data Center gages are
shown for reference as black
triangles (CN = Canton; TW =
Towanda 1S). Precipitation
interpolations shown were
derived from PRISM data (PRISM
Climate Group 2013)
Deterministic Watershed Modeling: Towanda Creek
Background and general explanations for each modeling step are provided in the Overview section above.
1. Setup of Model Inputs. Standard model inputs of land cover and topography were obtained for Towanda Creek
(Table B-l). Two meteorology stations were available for model forcing data (Fig. B-6). One USGS streamflow
gaging station was available for the entire calibration period (USGS 1532000, Towanda Creek at Monroeton, PA, 215
mi2). A second site with a shorter period of record was available internal to the Monroeton (Fig. B-6).
Appendix B-8
-------
Water Acquisition for Hydraulic Fracturing
May 2015
A)
Towanda Creek Observed Flow (1997-2012)
10,000-
fi
I
100 -
1 -
' "^ "HW
- daily prerip
daily flow
365-day mean flow
f '
i i r
. 4-0
- 6?
i i i i i i i i i r
f / / / / / / / / / / /
B)
Annual deviation from long-term mean flow: 1934-2012
Towanda Creek near Monroeton, PA: USGS 01 532000
wet years
2011
o
arf
ro
.. i ..II,. 11 nl . II ml, II. I ill.
I IN i I Wl1' II'MITM"1 ni
I 201
dry years
i r r r
Figure B-5. A) Towanda Creek observed streamflow: 1997-2012. Towanda Creek near Monroeton, USGS
01532000. The hydrograph and hyetograph show hydrologic conditions during the HSPF calibration period. The
running 365-day mean streamflow is shown in dark blue on the hydrograph, and the detailed hydraulic
fracturing activity period (2009-2012) is highlighted in green. These plots show that conditions were stationary
over the calibration period, while including a range of wet and dry flows. B) Annual deviations from long-term
conditions. Positive deviations indicate wet years, while negative deviations indicate dry years. Surface water
use intensity (SUI) values are highest when flows are lowest.
2. Stream Network Definition and Subbasin Delineation. The NHD high-resolution network was used as a baseline,
with field data collection of stream network locations additionally incorporated. The headwaters extent of the
perennial stream network was verified through field survey during the low-flow season (late fall) of 2013 after a
period of no rain for more than one week. The timing of the field surveys provided an excellent snapshot of perennial
flow conditions in the region (Fig. B-7A). Two lines of information were collected at various points, in association
with GPS coordinates: (1) presence/absence of flowing water where the perennial stream network had been mapped
Appendix B-9
-------
Water Acquisition for Hydraulic Fracturing
May 2015
(39 points), and (2) the exact location of flow initiation in first-order streams (24 points). This information was then
used to fine-tune the accumulation area threshold and stream network in ArcSWAT by iteratively generating stream
networks using a range of accumulation area thresholds, until a visual match between observed and generated
networks was achieved (Fig. B-7B). The accumulation area that best matched the NHD and observed perennial stream
networks was 0.35 mi2 in the Towanda Creek watershed, but resultant subbasin size ranged considerably. While
accumulation area provides a baseline for stream network definition, topographic slope is also important in
determining stream network locations. As a result, not all subbasins are equal in size. To achieve a consistency of
subbasin areas, which helps ensure realistic hydrologic processes are dominating the streamflow modeling,
anomalously small and large subbasins were grouped or split accordingly (Gabriel et al. 2014). Within the Towanda
Creek watershed, all subbasins less than 0.19 mi2 were merged into adjacent subbasins, and subbasins greater than 1.9
mi2 were split to create multiple subbasins, ensuring that all modeled subbasin areas were within an order of
magnitude.
1
Figure B-6. Towanda Creek study
area. Black triangles indicate
National Climatic Data Center
gages (CN = Canton; TW =
Towanda 1 S). The stream
network shown is the perennial
stream network used in SWAT
and HSPF modeling.
3. Pour Point Designations. The aforementioned automated delineation processes resulted in 288 subbasins of
Towanda Creek. These subbasins (each of which corresponds to an individual stream segment) were aggregated into
watersheds based on a Strahler stream ordering scheme, delineating the entire contiguous area upstream of each
subbasin outlet. Of the original 288 subbasins, 168 locations were retained as simulation pour points. Of these, 134 are
first-order stream segments, 26 are second-order, and five are third-order pour points; both the North and South Forks
of Towanda Creek are fourth-order pour points, and the main Towanda Creek pour point at Monroeton is a fifth-order
stream. Simulated pour point drainage areas ranged from 0.35 to 215 mi2 in Towanda Creek.
4. Sensitivity Analysis. The most sensitive parameters were retained based on rank of explained variability stopping
once 90% of total sensitivity was explained. This resulted in eight calibration parameters for Towanda Creek (Table
B-2).
Appendix B-l 0
-------
Water Acquisition for Hydraulic Fracturing
May 2015
A)
1USGS
USGS 01532000 Towanda Creek near Monroeton, PA
2089 2009 2010 2010 2011 2011 2812 2012 2013 2013
Daily nean discharge •— Period of approved data
— Estinated daily nean discharge — Period of provisional data
_ Flowing water
• No flowing water
NHD-HR stream network
Figure B-7. Ground-truthingof the perennial stream network by field survey. A) The sampling period
was during baseflow-only, seasonal low-flow conditions. B) Sampled locations where flowing water
(perennial flow) was observed (blue triangles) or not observed (red circles). The yellow-highlighted
inset corresponds to water table elevations used in Appendix C.
Appendix B-ll
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Table B-2. Sensitive HSPF parameters for Towanda Creek. "Qualified values" refers to simulations with weighted
Nash-Sutcliffe (WNS) fit criteria greater than 0.3 and annual water yields within 15% of observed. "Best set"
indicates the parameter values associated with the highest WNS score.
Qualified values
Parameter
KMELT
INFILT
AGWRC
DEEPFR
BASETP
INTFW
IRC
LZETP
Definition*
Degree dayfactor
Soil infiltration rate
Groundwater recession rate
Ground water fraction lost to deep storage
Fraction of evapotranspiration from baseflow
Interflow inflow
Interflow recession
Lowerzone evapotranspiration
Initial
range
0-10
0.0001 - 2
0.85-
0.999
0-1
0-1
0-
0.0001
0-
10
-0.999
1.5
Min
0.003
0.0059
0.85
0.0002
0.0001
0.001
0.0002
0.0005
Max
9.997
1.9986
0.9989
0.908
0.9989
9.991
0.9989
1.5
Mean
4.9069
0.7874
0.9186
0.1668
0.4999
4.6989
0.4565
0.6718
Std
2.8845
0.5606
0.0416
0.1239
0.2913
2.8766
0.2799
0.4329
Best set
0.314
0.0341
0.9282
0.0428
0.5041
4.5555
0.7732
1.0917
* Source: EPA Off ice of Water 2000
5. Calibration. HSPF was calibrated by identifying an optimized parameter set that best matched simulated and
observed streamflow at the USGS-d Towanda Creek outlet. Overall, calibrations were reasonably successful, with
optimized WNS values greater than 0.7. WNS scores will, by definition, be lower than either an optimized raw NS or
NSln, because it is simultaneously optimizing two separate functions. The scores achieved in these calibrations exceed
common performance thresholds set for raw NS (Narasimhan 2005; Moriasi et al. 2007). All calibration period,
validation period, and cross-site validation WNS and R2 scores for Towanda Creek are presented in Table B-3, and
optimized parameter values are shown in Table B-2.
Towanda Creek at Monroeton
3000 -
2500 -
2000 -
O
rci
1500 -
1000 -
500 -
o H
|—I all simulated
flows (n = 2116)
^_ best simulation
(WNS = 0.74)
O observed flow
O
-o
I
5th% 50th% 95th%
low flow median flow high flow
Figure B-8. HSPF simulation uncertainty across flow
magnitudes. Boxplots represent all successful HSPF
simulations from the 10,000 run Monte Carlo
analyses, with "successful" defined as weighted
Nash-Sutcliffe (WNS) fit criterion greater than 0.3
and annual water yield within 15% of observed. The
boxes themselves indicate the inner-quartile range,
with whiskers extending to the 5th and 95th
percentiles of the distributions. Boxes for Towanda
Creek are based on 2,116 successful simulations.
The three boxes represent three flow magnitudes
calculated separately for each simulation. For
example, the 5% box shows the distribution of the
5th percentile flow value, across all 2,116
simulations. The orange line and yellow circle
indicate the value associated with the optimized
parameter set and the observed value, respectively.
These figures show that HSPF tended to
overestimate low flows and underestimate high
flows. However, the optimized parameter sets (the
highest single WNS score within 15% of annual
water yield) approximated observed flow
magnitudes reasonably well.
Appendix B-l 2
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Table B-3. Fit statistics for HSPF calibration, validation, and cross-site 6. Simulation Uncertainty. Uncertainty
comparisons. was characterized by identifying the range
of flows and flow statistics in the subset of
simulations that produced daily WNS 0.3
or greater and annual water yields within
15% of observed flows. The boxplots in
Fig. B-8 show streamflow magnitudes (at
low, median, and high flow) calculated for
each main stem Towanda Creek simulation
meeting the criteria. The wide range of
values for each magnitude indicates large
uncertainty across the simulation set. In
general, HSPF tended to overestimate low
flows and underestimate high flows, but
this calibration approach identified an
optimized parameter set for each watershed that agrees well with observed flows across a range of magnitudes. The
range of simulated flows across a 5th-95th% confidence interval (CI) are shown for a representative range of
hydrometeorological conditions (2009-2010) in Fig. B-9.
7. Streamflow Simulation. Once calibrated parameter sets were established, streamflow was simulated across all 168
pour points in Towanda Creek. These flows were used as the water availability portion of the SUI calculations, results
of which are presented in the main body of this report.
Calibration
Period
Towanda Cr.
White R.
WNS
R2
WNS/BE
R2
0.74
0.75
0.75
0.75
Validation
Period
0.57
0.54
0.69*
0.60*
Cross-site
Comparison
0.86
0.82
0.48*t
0.99*
*Without irrigation adjustment to observed flows
tBenchmark Efficiency (Eq B.7)
Towanda Creek Simulation Uncertainty (2009-2010)
8000 -
2" _
5,
finnn —
> wvVJv
O -
t^
£ 4000 -
S
t
1/1 2000 -
0 -
|,p|,,,. ,.,,.
J precipitation
simulated flow (5-95*%)
,^-" best simulation
- — observed flow
i
^jM^jL^^^
1 1 1 1 1 1 1 1 1 1 1
-pi _pi JDi Oi
\JO >{O vQ \Q
O
1 1 1 1 1 1 1
i
1
^.Vv. ^
1
ll
u -,"
- , R
2 -a
-4i
NN ^N ^N XN\N N^N
Figure B-9. Example of HSPF simulation uncertainty: 2009-2010. These hydrographs are subset from our
simulation period to illustrate the ranges of "successful" model runs, where successful is defined as
weighted Nash-Sutcliffe fit criterion of 0.3 or greater, and annual water yield within ± 15% of observed. The
blue band indicates the 5th-95th percentile range across all successful simulations, with the orange and
black lines indicating the optimized and observed simulations. Flow was from USGS gage 01532000.
Appendix B-l 3
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Table B-4. Part 1. U.S. Geological Survey streamflow gaging stations used in area-weighted extrapolation
USGS Station #
USGS04233300
USGS01521500
USGS01529950
USGS01531000
USGS01518700
USGS01516350
USGS01503000
USGS01515000
USGS01531500
USGS01534500
USGS01534300
USGS01538000
USGS01533400
USGS01544500
USGS01550000
USGS01547950
USGS01544000
USGS01447720
USGS01549500
USGS01553700
USGS01546500
USGS01541303
USGS01542500
USGS01547200
USGS01541500
USGS01440485
USGS01443900
USGS01510000
USGS01540500
USGS01545500
USGS01520000
USGS01500000
USGS01531908
USGS01555000
Location
Lat42°24'll", Long 76°26'07"
Lat42°23'45", Long 77°42'42"
Lat 42°08'47", Long 77°03'28"
Lat 42°00'08", Long 76°38'06"
Lat 41'57'09", Long 77'06'56"
Lat 41'47'49", Long 77'04'50"
Lat 42°02'07", Long 75°48'12
Lat 41°59'05", Long 76°30'05"
Lat 41'45'55", Long 76'26'28"
Lat 4r30'16", Long 75'32'33"
Lat 41'40'47", Long 75'28'20"
Lat 4r03'33", Long 76'05'38"
Lat 4r36'26", Long 76'03'02"
Lat 41'28'33", Long 77'49'34"
Lat 41'25'06", Long 77'01'59"
Lat 4r06'42", Long 77'42'09"
Lat 41'24'06", Long 78'01'28"
Lat 4r05'05", Long 75'36'21"
Lat 41'28'25", Long 77'13'52"
Lat 4r03'42", Long 76'40'50"
Lat 40'53'23", Long 77'47'40"
Lat 4ro0'16", Long 78'27'25"
Lat 4r07'03", Long 78'06'33"
Lat 40'56'35", Long 77'47'12"
Lat 40'58'18", Long 78'24'22"
Lat 4r05'38", Long 75'19'21"
Lat40°58'50", Long 75°02'21"
Lat42°32'28", Long 75°54'00"
Lat 40'57'29", Long 76'37'10"
Lat 4ri9'28", Long 77'45'03"
Lat 41'59'48", Long 77'08'25"
Lat 42°20'00", Long 75°14'07"
Lat 41'41'52", Long 76'34'43"
Lat 40'52'00", Long 77'02'55"
Drainage
area (mi )
39.0
30.6
2006.0
2506.0
446.0
153.0
2232.0
4773.0
7797.0
108.0
38.8
43.8
8720.0
136.0
173.0
152.0
245.0
118.0
37.7
51.3
87.2
474.0
1462.0
265.0
371.0
6.6
5.3
147.0
11220.0
2975.0
298.0
103.0
112.0
301.0
Name
Sixmile Creek at Bethel Grove NY
Canisteo River at Arkport NY
Chemung River at Corning NY
Chemung River at Chemung NY
Tioga River at Tioga Junction PA
Tioga River near Mansfield PA
Susquehanna River at Conklin NY
Susquehanna River Near Waverly NY
Susquehanna River at Towanda PA
Lackawanna River at Archbald PA
Lackawanna River near Forest City PA
Wapwallopen Creek near Wapwallopen
PA
Susquehanna River at Meshoppen, PA
Kettle Creek at Cross Fork PA
Lycoming Creek near Trout Run PA
Beech Creek at Monument PA
First Fork Sinnemahoning Cr near
Sinnemahoning PA
Tobyhanna Creek near Blakeslee PA
Blockhouse Creek near English Center PA
Chillisquaque Creek at Washingtonville PA
Spring Creek near Axemann PA
West Branch Susquehanna River at Hyde
PA
WB Susquehanna River at Karthaus PA
Bald Eagle Creek bl Spring Creek at
Milesburg PA
Clearfield Creek at Dimeling PA
Swiftwater Creek at Swiftwater PA
Yards Creek near Blairstown NJ
Otselic River at Cincinnatus NY
Susquehanna River at Danville PA
West Branch Susquehanna River at
Renovo PA
Cowanesque River near Lawrenceville PA
Ouleout Creek at East Sidney NY
Towanda Creek near Franklindale, PA
Penns Creek at Penns Creek PA
Period of Record
March 1995 to current year
January 1937 to current year
1941, 1968-69. October 1974 to current year
September 1903 to current year
July 1976 to current year
July 1976 to current year
November 1912 to current year
February 1937 to March 1995, April 1995 to September 2000
October 1913 to current year
October 1939 to current year
October 1958 to current year
October 1919 to current year
October 1976 to current year
October 1940 to current year
December 1913 to current year
October 1968 to current year
October 1953 to current year
October 1961 to current year
October 1940 to current year
May 1979 to current year
October 1940 to current year
October 1978 to current year
October 1995 to present. February 1940 to September 1995
October 1955 to current year
October 1913 to current year
September 21, 1994 to April 18, 2001 (measurements only);
April 2001 to current year
October 1966 to current year
June 1938 to September 1964, October 1969 to current year
March 1899 to current year
October 1907 to current year
June 1951 to current year
August 1940 to current year
July 2010 to current year
October 1929 to current year
Appendix B-l 4
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Table B-4. (continued). U.S. Geological Survey streamflow gaging stations used in area-weighted extrapolation
USGS Station #
USGS01541200
USGS01549700
USGS01552500
USGS01516500
USGS01525981
USGS01429500
USGS01432900
USGS01551500
USGS01518862
USGS01531325
USGS01532000
USGS01428750
USGS01542810
USGS01541000
USGS01531250
USGS01534000
USGS01572950
USGS01548500
USGS01543500
USGS01415000
USGS01557500
USGS0142400103
Location
Lat 40°57'41",
Lat41°16'25",
Lat41°21'25",
Lat 41°47'27",
Lat 42°04'20",
Lat 41°36'26",
Lat 41°40'05",
Lat 41°14'10",
Lat 41°55'23",
Lat 41°45'38",
Lat 41°42'25",
Lat 41°40'28",
Lat 41°34'44",
Lat 40°53'49",
Lat 41°50'25",
Lat 41°33'30",
Lat 40°26'20",
Lat41°31'18",
Lat 41°19'02",
Lat 42°07'12",
Lat 40°41'01",
Lat 42°10'25",
Long78°31'10"
Long 77°19'28"
Long 76°32'06"
Long 77°00'54"
Long77°17'56"
Long 75°16'03"
Long 74°46'50"
Long 76°59'49"
Long77°31'56"
Long 76°40'30"
Long 76°29'06"
Long 75°22'35"
Long 78°17'34"
Long 78°40'38"
Long 76°49'38"
Long 75°53'42"
Long 76°35'55"
Long 77°26'52"
Long78°06'12"
Long 74°49'07"
Long 78°14'02"
Long 75°16'46"
Drainage
area (mi2)
367.0
944.0
23.8
12.2
102.0
64.6
76.6
5682.0
90.6
93.6
215.0
40.6
5.2
315.0
8.8
383.0
5.5
604.0
685.0
33.2
44.1
20.2
Name
Wb Susquehanna River Near Curwensville
PA
Pine Creek Bl L Pine Creek Near Waterville
PA
Muncy Creek Near Sonestown PA
Corey Creek Near Mainesburg PA
Tuscarora Creek Above South Addison NY
Dyberry Creek Near Honesdale PA
Mongaup River At Mongaup Valley NY
Wb Susquehanna River At Williamsport PA
Cowanesque River At Westfield PA
Sugar Creek At West Burlington PA
Towanda Creek Near Monroeton PA
West Branch Lackawaxen River Near
Aldenville PA
Waldy Run Near Emporium PA
West Branch Susquehanna River At Bower
PA
Nb Sugar Creek Trib Near Columbia Cross
Roads PA
Tunkhannock Creek NearTunkhannock PA
Indiantown Run Near Harper Tavern PA
Pine Creek At Cedar Run PA
Sinnemahoning Creek At Sinnemahoning
PA
Tremper Kill Near Andes NY
Bald Eagle Creek At Tyrone PA
Trout Creek Near Trout Creek NY
Period of Record
October 1955 to current year
October 1957 to current year
October 1940 to current year
May 1954 to current year
Annual maximum only-October 1989 to September 2000;
October 2000 to current year
October 1943 to current year
Occasional low-flow and/or miscellaneous discharge
measurements, water years 1949, 1957-61, 1965, 1970, 1973-
74. October 2002 to current year
March 1895 to current year
August 1983 to current year
July 2010 to current year
February 1914 to current year
Occasional discharge measurements and annual maximums,
water years 1975-86. October 1986 to current year. Published as
station number 01427950, 1975-88
Occasional discharge measurements and annual maximum,
water years 1963-64. September 1964 to current year
October 1913 to current year
September 1962 to September 1968; October 1968 to March
1981(partial record); May 13, 2010 to current year
February 1914 to current year. Prior to October 1965, published
as "at Dixon"
August 2002 to current year
July 1918 to current year. Prior to October 1918 monthly
discharge only, published in WSP 1302
July 1938 to current year. Prior to October 1938 monthly
discharge only, published in WSP 1302
February 1937 to current year. Published as "near Shavertown"
1937-67
October 1944 to current year. Prior to October 1967, published
as South Bald Eagle Creek at Tyrone
June 1952 to June 1967, annual maximum only-1996, maximum
only—November 1996, December 1996 to current year. Prior to
November 1996, published as Trout Creek near Rockroyal
(01424000).
Appendix B-l 5
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Empirical Streamflow Estimation: Susquehanna River Basin
Area-Weighted Extrapolation of Regional
Data. SUI was calculated at the locations of
permitted O&G water withdrawal sites, none of
which is co-located with a USGS gaging
station. However, there were an ample number
of regional USGS gages elsewhere in the SRB
that could be used for empirically-based data
extrapolation at known withdrawal sites.
Following standard USGS methods (Ries
2007), availability was estimated at 95
permitted sites using an area-weighting factor
based on 56 nearby gages of similar size. Und
withdrawal sites were subjectively paired with
gaged sites based on nearness and similarity of
basin area. Regional gages included are
presented in Table B-4.
Watershed areas above withdrawal points
ranged from 1.8 - 10,547 mi2, corresponding to
USGS gaged watershed areas of 5.2 - 11,220
mi2. This area-weighting method was cross-
validated using five pairs of USGS gages. In
each pair, flow was estimated for one based on
observed flow from the other, replicating the
method used for each permit site/ pair. These
extrapolations were then compared with the
actual flow measured at the cross-validation
using five pairs of USGS gages. In each pair,
flow was estimated for one gage based on
observed flow from the other gage, replicating
the method used for each permit site/gage
pair. These extrapolations were then
compared with the actual flow measured at
the cross-validation gage. In most cases,
results were excellent, with NS and NSta
values as high as 0.96 and 0.98, respectively
(Table B-5). From these results, it can be
concluded that the estimated streamflows
applied at ungaged sites using this area-
weighting method are indeed representative of
actual streamflows.
Figure B-10. Relationship between extreme low
flow (Q7,io, and mean flow with drainage area,
produced by empirical modeling (regional
equations) and mechanistic modeling (HSPF)
approaches. The 90% confidence interval is
shown for both methods. The black line
represents the single best HSPF simulation,
based on optimized weighted Nash-Sutcliffe fit
criterion and annual water balance.
Pair
1
2
3
4
5
Gage ID Area (mi2) Distance (mi) NS
USGS 01541500
USGS 01541303
USGS 01532000
USGS 01531908
USGS 01541000
USGS 01541200
USGS 01553700
USGS 01552500
USGS 01503000
USGS 01515000
371
474
215
112
315
367
51
24
2232
4773
3
4
9
22
36
0.83
0.96
0.76
0.76
0.96
NSln
0.91
0.95
0.97
0.7
0.98
Table B-5. Cross-validation of area-weighting method.
Each pair represents two gaged watersheds on which the
area-weighting method was tested. Flow for the second
member of each pair was estimated by extrapolating
values from the first member of the pair, with the NS and
NS|n scores indicating the fit between this estimation and
the actual observed flows. "Distance" is the straight-line
distance between the two gaging stations.
1000
5 o.ooi
_o
u. 0.0001
0.00001
1000
-ii 100
HI
tuo
ro
—
T3
Best Fit (HSPF)
90% Cl (HSPF)
90% PI (Regional)
100 1000 10000
Drainage Area (ha)
100000
90% Cl (HSPF)
90% PI (Regional)
100 1000 10000
Drainage Area (ha)
100000
Appendix B-l 6
-------
Water Acquisition for Hydraulic Fracturing May 2015
Regional Equations. Multiple regression equations are frequently developed from gaged stream locations to
estimate streamflow in und locations, based on watershed geologic, topographic, and land use characteristics.
Many managers can more easily access regional equations than estimates derived from deterministic models, given
the time, expertise, and computational resources necessary for rigorous calibration and simulation. These
equations are not used to generate continuous time series, as is the case with deterministic modeling and area-
weighted extrapolation methods, but are instead used to estimate key flow magnitudes, such as extreme low flows
and median flows.
Regional equations were previously developed for low and median flows in five regions of Pennsylvania by USGS
and the Pennsylvania Department of Environmental Protection (Stuckey 2006). Multivariate predictive equations
were developed for various metrics of flow magnitude, linking watershed characteristics to each flow magnitude
based on observed data from 293 regional streamflow gages.
Agreement was evaluated between flow statistics derived from these regional equations and the deterministic
modeling for ungaged subwatersheds of the Susquehanna River. This served two objectives: 1) to confirm that the
deterministic models were reasonably representative of low flow magnitudes, given their importance in this impact
assessment, and 2) to determine whether regional regression techniques could be used for such impact
assessments, independent of deterministic modeling. Both methods suffer from lack of measured streamflow in
very small streams.
Stuckey's (2006) regression analysis showed that the most important watershed characteristics for predicting low
and mean flow statistics in this part of the Susquehanna River System are:
• Drainage area (DA, mi2)
• Mean annual precipitation (Ppt, inches)
• Percent watershed area that was glaciated (Gla, %)
• Percent watershed area in forest cover (F, %)
• Mean elevation (El, ft.)
• Percent watershed area in urban land use (U, %)
The importance of drainage area in these equations for determining streamflow is a key theme of the SUI analyses
presented throughout this report. From the six low-flow metrics evaluated in Stuckey's (2006) regional regression
analysis, two commonly used indices were calculated: mean annual flow (Qm) and the Q710 for exploration in this
analysis. Mean annual flow is simply the mean instantaneous streamflow (in cubic feet per second, cfs) occurring
at each analysis point. Q7)i0 is the lowest seven-day average flow that occurs on average once every 10 years, and
is a common benchmark of a very low flow metric in watershed management (Mitchell et al. 2013).
The variables in these equations were calculated from readily available data sources. Drainage area (DA mi2) and
mean elevation (El, ft) were determined via digital elevation model (DEM) analysis, forest (F, %) and urban (U,
%) land cover percentages were derived from the National Land Cover Database (NLCD), and mean annual
precipitation (Ppt, inches) was obtained from regional gages (location of digital data sources is further listed in
Appendix A). Percent of watershed glaciation (Gla, %) is an important variable for statewide Pennsylvania
predictive equations, because the Pleistocene ice margin transects the state.
Qm = i0-32363£)A10081£-/01283Ppt17949(l + 0.01F)04136(1 + O.Olf/)04130 Equation B-5
Q7.10 = iQ-12.22164D^i.27803ppt5.43i65 (1 + Q.OlGto)183875 (1 + 0.01F)415769 Equation B-6
However, the entirety of the Middle and Upper Susquehanna River Basins were entirely glaciated during the
Pleistocene (Engel et al. 1996), and a value of 100% was used for the glaciation variable in all cases.
Comparison of Empirical and Deterministic Methods. Statistical relationships between drainage area and flow
magnitude were established for Qm and Q7j0, using 1) the 2116 HSPF Monte Carlo simulations meeting the pre-
established fit criteria, at all 168 pour points (described above in "Deterministic Watershed Modeling"), and 2) the
regional regression equations shown previously at each of the permitted sites (Stuckey 2006). Results showed that
Appendix B-l 7
-------
Water Acquisition for Hydraulic Fracturing
May 2015
the statistics estimated within the 90% CI by the regional equations agreed reasonably well with the 90% CI of the
HSPF simulations (Fig. B-10). The best-fit HSPF Q710 estimates agree well with the regional equation-derived
estimates, particularly in smaller watersheds, which were the focus of much of this analysis. HSPF tends to
estimate lower Q710 flow than the regional equations, particularly in larger watersheds. Mean flows across a wide
range of drainage areas were all very close using these two methods.
Colorado: Parachute and Roan Creeks
Study Area Hydrometeorology
Within the Parachute and Roan watersheds (198 and 509 mi2), surface bedrock is sandstone and shale, with no
Quaternary glacial influence. These watersheds show high relief and pronounced dissection, with elevations
ranging from 4890 feet ASL at the confluence of Roan Creek and the Colorado River, to 9299 feet ASL along
Parachute Creek's eastern divide. Soils are primarily loamy, well-drained aridisols and mollisols forming in
colluvium, loess, and rock outcrop, with some poorly-drained entisols occurring in alluvial valleys (Harman and
Murray 1985).
10-15
15-20
^20-25
25-30
HI 30-35
1] 35-40
I 40-45
45-50
Figure B-ll. Upper Colorado River
Basin annual mean precipitation,
1981-2010. National Climatic Data
Center gages are shown for
reference as black triangles (AL =
Altenbern Ranch; BM = Burro
Mountain; GS = Glenwood Spgs #2;
MK = Meeker; RF = Rifle), but
precipitation interpolations shown
were derived from PRISM data
(PRISM Climate Group 2013).
Thirty-year annual average precipitation in Altenbern, CO (station ID = USC00050214 located in the Roan Creek
watershed; station elevation = 6800 feet ASL) is 17.9 inches (Fig. B-ll). The lowest mean monthly precipitation
occurs in June (0.98 inches), while the highest occurs in October (2.00 inches). Snowpack typically develops at
higher elevations (BLM 2006). Minimum and maximum monthly mean temperatures occur in January (24.4°F)
and July (69.7°F), respectively (Arguez et al. 2012). Garfield, Mesa, and Rio Blanco Counties are all classified as
'arid steppe' (Bs) (Kottek et al. 2006).
The stationarity of hydrological conditions during the study period is demonstrated by the nearby USGS on the
Colorado River near Cameo (Fig. B-12A). The second highest annual average flow occurred in 2011 and the third
lowest occurred in 2012. Due to significant snowpack development along the Rocky Mountain crest in the eastern
part of the UCRB, there is dominant snowpack/snowmelt cycle evident in the mainstem Colorado River. A
persistent but thinner snowpack develops on top of the Roan Plateau that forms the headwaters the Parachute and
Roan watersheds.
Appendix B-l 8
-------
Water Acquisition for Hydraulic Fracturing
May 2015
A)
I
CC
E
-------
Water Acquisition for Hydraulic Fracturing
May 2015
sites were available but the measurement period was too short to be useful. The USGS gage on the White River at
Meeker, CO (09304500) (760 mi2) was deemed the better choice for calibration among the few available gages in
this area that had a sufficiently long record of flow (Figure B-13). The calibration gage is 30 miles from the study
area and White River tributaries border the study watersheds. The contributing watershed area is similar to Roan
and Parachute Creeks. Geomorphology and climate in the lower White River is reasonably similar to the study
area, although the White River heads in the mountainous alpine climate. Calibrated streamflow was cross-
validated using gages on the mainstem UCR and to the observed flow in Parachute Creek.
There were additional challenges for creating the
streamflow record. The hydrologic system of the
UCRB is intensively managed, and there are
hundreds of diversion structures used for water
withdrawal for municipal, industrial, and
agricultural purposes. Importantly, streamflow at
gaged sites is affected by diversion of significant
volumes of flow for irrigation during the period
April to November. The project required a
"natural" streamflow record with the effects of
irrigation diversions removed for calculation of
SUI. Flows were adjusted for irrigation after
calibration at the White River gage. Finally, the
streamflow was checked against the observed
discontinuous flow record from Parachute Creek
as an additional validation step. The details of
calibration and flow adjustment steps are
presented in the following sections
Preliminary calibration and simulations were
again performed with both HSPF and SWAT,
with HSPF chosen for detailed analyses based on
shorter runtimes and greater simulation accuracy.
1. Setup of Model Inputs. Standard model inputs
of land cover and topography were obtained for
the area covered by the upper White River
watershed and the portion of the UCR between
Glenwood Springs and Cameo (Fig. B-13).
Calibration was performed on the White River,
which had three available meteorology stations,
and one USGS streamflow gaging station located
near Meeker, CO (USGS 09304500, elevation
6,300 feet ASL, 760 mi2).
Figure B-13. White and Colorado Rivers, Colorado. NCDC
gages are shown as black triangles (AL = Altenbern Ranch;
BM = Burro Mountain; GS = Glenwood Spgs #2; MK =
Meeker; RF = Rifle). HSPF calibration was performed on the
White River above Meeker and cross-site valid validation
was performed between the two USGS gages shown on the
UCR. Our focus study areas are the Parachute and Roan
Creek watersheds within the UCRB.
2. Stream Network Definition and Subbasin Delineation. The NHD high-resolution network was used as a
baseline, incorporating field observations and perennial flow data from CODWR. Winter road conditions
prevented detailed field mapping of the stream network, but six observations of presence versus absence of
flowing water were collected at locations where mapped perennial stream networks disagreed. All available
information was then used to fine-tune the accumulation area threshold and stream network in ArcSWAT.
Networks were iteratively generated using a range of accumulation area thresholds, until a visual match between
observed and generated networks was achieved. The accumulation area that best matched the NHD and observed
perennial stream networks was 3.9 mi2 in the White River watershed, but again, resultant subbasin size ranged
considerably. To achieve a consistency of subbasin areas, anomalously small and large subbasins were grouped or
split accordingly (Gabriel et al. 2014). These delineation processes were originally performed on the upper White
River for model calibration, and were replicated for the UCR between Glenwood Springs and Cameo, including
Parachute and Roan Creeks.
Appendix B-20
-------
Water Acquisition for Hydraulic Fracturing
May 2015
3. Pour Point Designations. Model calibration in the upper White River and cross-validation in the UCRB were
based solely on simulated and observed flows at the gaged watershed outlets, meaning no additional pour point
designation steps were required. However, it was desirable to establish a wide range of watershed areas for SUI
calculation in the focus study areas of Parachute and Roan Creeks, similar to what was achieved in Towanda
Creek. The original delineation resulted in 80 subbasins of the Parachute and Roan watersheds, which were
aggregated to sequentially larger-order watersheds. Subbasins were then eliminated following the criteria detailed
previously (see "Description of Modeling Steps"), with 48 simulation pour points retained. Of these 48, there were
35 first-order pour points, nine second-order, three third-order (including the outlet of Parachute Creek), and one
fourth-order pour point at the outlet of Roan Creek. Simulated pour point drainage areas ranged from 4.4 to 509
mi2 in this study area.
Table B-6. Sensitive HSPF parameters: White River. "Qualified values" are all parameter values associated with
simulations meeting the threshold criteria of weighted Nash-Sutcliffe (WNS) fit criterion of 0.3 or greater and
annual water yield within 15% of observed. "Best set" indicates the parameter values associated with the
highest WNS score.
Qualified values
Parameter
KMELT
KVARY
AGWRC
DEEPFR
BASETP
CEPSC
Definition*
Degree day factor
Nonlinear groundwater recession rate
Linear groundwater recession rate
Groundwater fraction lost to deep storage
Fraction of evapotranspiration from baseflow
Canopy interception
Initial range
0
0
0
0
0
0
-10
-5
.85-0.999
-1
-1
-1
Min
0.010
0.015
0.951
0.294
0.0003
0.0001
Max
0.
0.
0.
0.
212
,981
,999
,501
0.197
0.199
Mean
0.050
0.377
0.989
0.405
0.077
0.080
Std
0.034
0.
0.
0.
0.
0.
.274
.009
.055
.054
.060
Best
set
0.035
0.194
0.987
0.412
0.075
0.009
* Source: EPA Office of Water 2000
4. Sensitivity Analysis. The most sensitive parameters by rank of variability explained were retained for HSPF
calibration. Only three parameters explained 90% of sensitivity, which is less than the minimum of six established
for this methodology. Thus, the six most sensitive parameters were retained, together accounting for 95% of total
streamflow sensitivity (Table B-6).
5. Calibration. There are no long-term gaging records available for Parachute or Roan Creek. The UCR has two
gages in the river in this area. However, these gages were not ideal for calibrating HSPF for Parachute or Roan for
the following reasons: the large size of the UCRB at this location, its flow dependency on remote alpine regions
with far greater precipitation, and flow regulation structures. Thus, calibration was performed on the White River
watershed above Meeker and applied to neighboring Parachute and Roan Creeks (Fig. B-13), following a cross-
site validation step. This White River gage is located about 20 miles from Parachute Creek, and is in a much
smaller watershed, but it too originates in the alpine headwaters in the Rocky Mountains. The best available cross-
validation option near the Parachute and Roan watersheds was the Colorado River at Cameo, which drains a very
large watershed and requires a great deal of input data and computation time. Thus, flow was simulated between
the Cameo gage and the next upstream gage, which is the Colorado River near Glenwood Springs (Fig. B-13).
Because the added flow volume between these two gages was a small proportion of the total river flow, use of the
WNS score was inappropriate. Instead, an alternative metric for this cross-site validation was used, the benchmark
efficiency criterion (BE). BE is analogous to NS, but the user specifies the benchmark model, which is the overall
mean in NS (Schaefli and Gupta 2007). For this application, the mean difference between the two gaging stations
was used as the benchmark model.
N
Equation B-7
Appendix B-21
-------
Water Acquisition for Hydraulic Fracturing
May 2015
where, qin (t) is observed inflow from upstream gaging station at time t.
6. Simulation Uncertainty Estimation. Following calibration, simulation uncertainty was characterized by
identifying the range of flows and flow statistics in the subset of simulations that produced daily WNS > 0.3 and
annual water yields within 15% of observed flows. The boxplots in Fig. B-14 show streamflow magnitudes (at
low, median, and high flow) calculated for each White River simulation meeting the criteria. The wide range of
values for each magnitude indicates large uncertainty across the simulation set. In general, HSPF tended to
overestimate low flows and underestimate high flows, but calibration identified an optimized parameter set for
each watershed that agrees well with observed flows across a range of magnitudes. The range of simulated flows
across a 5^-95*% confidence interval are shown for a representative range of conditions (2009-2010) in Fig. B-15.
White River above Meeker
1200 -
1000 -
r soo -
O 600 H
a;
400 -
200 -
0 -
all simulated
flows (n= 161)
best simulation
(WNS = 0.71)
observed flow
5th% 50th% 95th%
low flow median flow high flow
Figure B-14. HSPF simulation uncertainty
across flow magnitudes. These boxplots
represent all "successful" HSPF simulations
from our 10,000 run Monte Carlo analyses,
with successful defined as weighted Nash-
Sutcliffe (WNS) fit criterion of 0.3 or greater,
and annual water yield within 15% of
observed. The boxes themselves indicate the
inner-quartile range, with whiskers extending
to 5th-95th percentiles of the distributions,
based on 134 successful simulations.
White River Simulation Uncertainty (2009-2010)
4000 -
3000 -
O
2000 -
1000 -
i i r
•^
''T
|| T precipitation
simulated flow (5-95th%)
-. best simulation
-— raw observed flow
irrigation-adjusted flow
lh -
- 0.5 X
E' ?
- 1.5
\ I I I I I I
\ I I I I I I I T
c4N
OJ OJ
Figure B-15. Example of HSPF simulation uncertainty: 2009-2010. These hydrographs are subset from the
simulation period to illustrate the ranges of 'successful' model runs, where successful is defined as weighted
Nash-Sutcliffe fit criterion of 0.3 or greater and annual water yield within ± 15% of observed. The blue band
indicates the 5th-95th percentile range across all successful simulations, with the orange and black lines
indicating the optimized and observed simulations. As seen in Figure B-14, despite model uncertainty,
calibration identified optimized parameter ranges that were able to replicate observed streamflow
reasonably well.
Appendix B-22
-------
Water Acquisition for Hydraulic Fracturing May 2015
These calibration results tend to support the use of the Monte Carlo parameter selection method, at the parameter
set that achieved the overall best match to observed flow (the red line matches the yellow circle in Fig. B-15) was
found outside the central tendency of the parameters (e.g. the 5th percentile low flow and 95th percentile high flow
bars). It is evident from this figure that uncertainty was far greater in the Colorado simulations than in Towanda
Creek. This greater uncertainty is likely due to the combined challenges of modeling in this mountainous setting,
with its snowpack/snowmelt cycles, sharp transitions between steep hillslopes and alluvial valleys, and highly
variable precipitation associated with steep elevation gradients.
Streamflow in Parachute and Roan Creeks was simulated with the weather data from Altenburn Ranch The
streamflow record generated by the model calibrated to the White River gage at Meeker was considered the best
representation of the daily fluctuations of flow to local weather. However, streamflow at the Meeker gage, like
most in the region, is heavily affected by irrigation. The SUI analysis required an "irrigation-free" record, thus the
simulated flow was manually adjusted to remove the effects of diversions on streamflow, as described in the next
step.
7. Finalized Simulated Natural Flows. Water is diverted from rivers through ditches that inefficiently convey
water for irrigation. Water leaks from the ditches and drains back into the floodplain returning water to the
streams. Some irrigation water is consumed by crops and lost to the system. The combined crop and structure loss
(termed "structure inefficiency") has been quantified by Leonard Rice Engineers (2009) for water supply planning
in the StateMod modeling system (CODWR 20 14e). Just 3 1% of diverted water is delivered while 82% of applied
irrigation water is consumed by crops (Lin and Garcia 2012), varying monthly. The streamflow record was
adjusted by adding the appropriate volume of water that was diverted but returned to the stream system back into
the streamflow record.
Two HSPF models with unique parameter sets were produced (each calibrated on flow data from the White River
at Meeker, as described above) to serve as lower and upper flow bounds. The lower bound model was the original
flow record at the gage, while the upper bound model added the diversion volume back into the observed
streamflow series. The two records were compared on a daily basis and the difference was found to vary from day
to day due to the nonlinearity of some model processes.
The difference between the records was multiplied by the structural inefficiency factors that varied monthly to
adjust the daily streamflow:
Equation B-8
Flow A = O.Sl^Flowu-FlowJ + FlowL
where Flowuis the record with diversion added and FlowL is the actual hydrologic record.
This adjusted record was considered to be the "natural" flow with appropriate daily fluctuations based on the
White River record.
Comparison to Historic Observations. Once a single adjusted flow series had been created, the average monthly
flows of this series (spanning the period 1987-2012) were compared to observed monthly flows in Parachute
Creek during three historic periods (1921-1927, 1948-1954, and 1974-1982). Some differences in monthly
averages would be expected due to different time frames represented in the data. Nevertheless, monthly averages
were reasonably close during the summer months but were in error during the winter and spring months (Table B-
7).
Flows were significantly underestimated during the spring snowmelt months and overestimated during the winter
months from January to March. This pattern suggests winter precipitation was applied as rainfall rather than
stored in a snowpack delivered later during the spring. The weather data was collected 3,000 feet lower and annual
rainfall is 3 inches less than in the headwaters (BLM 2006). There was no representative data from the plateau
available that would improve snow representation. Rather than trying to adjust weather data, the flow record itself
was adjusted. The Parachute Creek streamflow was constrained to match the observed flow. A monthly
adjustment factor was computed by comparing observed to predicted (Table B-7) and applied to each daily value.
This increased flow during the spring months, lowered flow from January to March, and left the remaining months
unchanged. The resulting streamflow record was closer to observed, as indicated in the boxplot of modeled and
observed datasets in Figure B-16.
Appendix B-23
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Table B-7. Correspondence between average monthly simulated (1987-2012) and observed (1921-1927,1948-
1954,1974-1982) flows in Parachute Creek. Units are cubic feet per second (cfs).
Month
200
150-
.100-
50-
o-
Modeled
Observed
Ratio
January
February
March
April
May
June
July
August
September
October
November
December
17.4
20.6
22.1
24.4
20.4
15.1
10.3
S.S
9.2
11.2
12.2
13.6
11.8
13.3
17.3
62.7
171.1
33.9
10.1
8.7
7.4
10.3
14.8
12.3
0.68
0.65
0.78
2.57
8.37
2.58
0.98
0.99
0.80
0.92
1.22
D.9D
E-3 Modeled
A Observed
In conclusion, the simulated Parachute
Creek streamflow record was a much
manipulated data set. The daily
streamflow matched the daily
fluctuations observed at weather
stations, represented "natural" flows
without the influence of irrigation
diversions, and was constrained to
match the range of flow observed in
Parachute Creek, but outside the
calibration period of the streamflow
simulation. As such, there is lower
confidence in the streamflow record
used for SUI and scenario assessment
intheUCRB. Nevertheless, the
streamflow record represents the
general volume of flow in the stream
and may be somewhat high given the
adjustment factors.
op -' •$>
Month
Figure B-16. Boxplots of average monthly flows for the adjusted model results (blue) and observed flows in
Parachute Creek over three historic periods (orange).
Appendix B-24
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Empirical Streamflow Estimation: Parachute and Roan Creeks
In addition to modeled pour points, SUI was also calculated at diversion structures on Parachute and Roan Creeks
known to be used as withdrawal points by the O&G industry (Fig. B-17). Again, none of these locations was
directly gaged, and area-weighting was used to estimate flow at each structure for SUI calculation. While USGS
regional gages were used for this extrapolation process in Pennsylvania, where there is a large number of gaging
stations across a wide range of watershed sizes, few gages exist in UCRB. Instead, the area-adjusted simulated
flows from the closest HSPF sub-basin outlet was applied to each diversion structure in the study area. Given the
small differences in drainage area, it can be assumed that these interpolated and extrapolated estimates are of
similar to accuracy to the directly simulated HSPF pour points.
O&G withdrawal slrucluie
Stream
Modeled sub-basin
Main &tern Colorado River
Figure B-17. O&G owned diversion structures in Parachute and Roan Creeks.
Most are used for irrigation and not HF.
Appendix B-25
-------
Water Acquisition for Hydraulic Fracturing May 2015
References
Abbaspour, K. C. 2009. SWAT-CUP2: SWAT calibration and uncertainty programs—a user manual. Eawag,
Swiss Federal Institute of Aquatic Science and Technology, Department of Systems Analysis, Integrated
Assessment and Modelling, Dtibendorf, Switzerland. 95pp.
Ahmad, Z., M. Hafeez, and I. Ahmad. 2012. Hydrology of mountainous areas in the upper Indus Basin, northern
Pakistan with the perspective of climate change. Environmental Monitoring and Assessment 184(9):
5255-5274.
Andreassian, V., C. Perrin, L. Berthet, N. Le Moine, J. Lerat, C. Loumagne, L. Oudin, T. Mathevet, M. -H.
Ramos, and A. Valery. 2009. Crash tests for a standardized evaluation of hydrological models. Hydrology
and Earth System Sciences 13(10): 1757-1764.
Arguez, A., I. Durre, S. Applequist, R. S. Vose, M. F. Squires, X. Yin, R. R. Heim Jr., and T. W. Owen. 2012.
NOAA's 1981-2010 U.S. climate normals: An overview. Bulletin of the American Meteorological
Society 93(11): 1687-1697.
Atkins, J. T., Jr., J. B. Wiley, and K. S. Paybins. 2005. Calibration parameters used to simulate streamflow from
application of the Hydrologic Simulation Program—FORTRAN Model (HSPF) to mountainous basins
containing coal mines in West Virginia. USGS Scientific Investigations Report 2005-5099. 79 pp.
Beven, K. 2006. A manifesto for the equifinality thesis. Journal of Hydrology 320(1-2): 18-36.
Bicknell, B.R, J.C. Imhoff, J.L Kittle, A.S. Donigian, and R.C. Johanson (1997) Hydrological Simulation
Program-Fortran: User's manual for version 11. (U.S. Environmental Protection Agency, Athens, GA), p
755.
Booker, D. J. and R. A. Woods. 2014. Comparing and combining physically-based and empirically-based
approaches for estimating the hydrology of ungauged catchments. Journal of Hydrology 508: 227-
Buchanan, C., H. L. N. Moltz, H. C. Hay wood, J. B. Palmer, and A. N. Griggs. 2013. A test of the Ecological
Limits of Hydrologic Alteration (ELOHA) method for determining environmental flows in the Potomac
River basin, U.S.A. Freshwater Biology 58(12): 2632-2647.
Engel, S. A., T. W. Gardner, andE. J. Ciolkosz. 1996. Quaternary soil chronosequences on terraces of the
Susquehanna River, Pennsylvania. Geomorphology 17(4): 273-294.
Faramarzi, M., K. C. Abbaspour, S. A. Vaghefi, M. R. Farzaneh, A. J. B. Zehnder, R. Srinivasan, and H. Yang.
2013. Modeling impacts of climate change on freshwater availability in Africa. Journal of Hydrology
480: 85-101.
FEMA (Federal Emergency Management Agency). 2013. Hydrologic models meeting the minimum requirement
of National Flood Insurance Program: Current nationally accepted hydrologic models. Available at:
http://www.fema.gov/national-flood-insurance-program-flood-hazard-mapping/hvdrologic-models-
meeting-minimum-requirement
Gabriel, M., C. Knightes, E. Cooler, and R. Dennis. 2014. The impacts of different meteorology data sets on
nitrogen fate and transport in the SWAT watershed model. Environmental Modeling and Assessment
19(4): 301-314.
Golden, H. E., C. R. Lane, D. M. Amatya, K. W. Bandilla, H. R. Kiperwas, C. D. Knightes, and H. Ssegane. 2014.
Hydrologic connectivity between geographically isolated wetlands and surface water systems: A review
of select modeling methods. Environmental Modelling and Software 53: 190-206.
Grubb, R. G. 1986. Soil survey of Bradford and Sullivan Counties, Pennsylvania. U.S. Department of Agriculture,
Soil Conservation Service. 135 pp.
Harman, J. B., and D. J. Murray. 1985. Soil survey of Rifle area, Colorado: Parts of Garfield and Mesa Counties.
U.S. Department of Agriculture, Soil Conservation Service. 161 pp.
Hawkins, R. H., T. J. Ward, D. E. Woodward, and J. A. Van Mullem. 2009. Curve number hydrology: State of the
practice. American Society of Civil Engineers, Reston, VA. 116 pp.
He, Z., Z. Wang, C. J. Suen, and X. Ma. 2013. Hydrologic sensitivity of the Upper San Joaquin River Watershed
in California to climate change scenarios. Hydrology Research 44(4): 723-736.
Appendix B-26
-------
Water Acquisition for Hydraulic Fracturing May 2015
Johnson, M. S., W. F. Coon, V. K. Mehta, T. S. Steenhuis, E. S. Brooks, and J. Boll. 2003. Application of two
hydrologic models with different runoff mechanisms to a hillslope dominated watershed in the
northeastern US: a comparison of HSPF and SMR. Journal of Hydrology 284(1): 57-76.
Kim, Y. and E.-S. Chung. 2014. An index-based robust decision making framework for watershed management in
a changing climate. Science of the Total Environment 473: 88-102.
Kottek, M., J. Grieser, C. Beck, B. Rudolf, and F. Rubel. 2006. World map of the Koppen-Geiger climate
classification updated. Meteorologische Zeitschrift 15(3): 259-263.
Krause, P., D. P. Boyle, and F. Base. 2005. Comparison of different efficiency criteria for hydrological model
assessment. Advances in Geosciences 5: 89-97.
Lin, Y., andL. A. Garcia. 2012. Assessing the impact of irrigation return flow on river salinity for Colorado's
Arkansas River Valley. Journal of Irrigation and Drainage Engineering 138: 406-415.
Matott, L. S., J. E. Babendreier, and S. T. Purucker. 2009. Evaluating uncertainty in integrated environmental
models: A review of concepts and tools. Water Resources Research 45(6).
Mitchell, A. L., M. Small, and E. A. Gasman. 2013. Surface water withdrawals for Marcellus Shale gas
development: Performance of alternative regulatory approaches in the Upper Ohio River Basin.
Environmental Science and Technology 47(22): 12669-12678.
Moriasi, D. N., J. G. Arnold, M. W. VanLiew, R. L. Bingner, R. D. Harmel, and T. L. Veith. 2007. Model
evaluation guidelines for systematic quantification of accuracy in watershed simulations. Transactions of
the ASABE 50(3): 885-900.
Morway, E. D., T. K. Gates, and R. G. Niswonger. 2013. Appraising options to reduce shallow groundwater tables
and enhance flow conditions over regional scales in an irrigated alluvial aquifer system. Journal of
Hydrology 495: 216-237.
Narasimhan, B., R. Srinivasan, J. G. Arnold, and M. Di Luzio. 2005. Estimation of long-term soil moisture using a
distributed parameter hydrologic model and verification using remotely sensed data. Transactions of the
ASAE48(3): 1101-1113.
Neitsch, S. L., J. G. Arnold, J. R. Kiniry, J. R. Williams. 2011. Soil and Water Assessment Tool: Theoretical
Documentation, Version 2009. Texas A&M University System, College Station, TX.
NOAA Coastal Services Center. 2014. Historical Hurricane Tracks. Available at:
http://www.coast.noaa. gov/hurricanes/#
NRCS (Natural Resources Conservation Service). 2006. U.S. general soil map (STATSGO2). U.S. Department of
Agriculture, NRCS National Cartography and Geospatial Center.
Philip, J. R. 1957. The theory of infiltration: 1. The infiltration equation and its solution. Soil Science 83(5): 345-
358.
Price, K., K. Kim, T. Hong, M. C. Gabriel, S. T. Purucker, C. D. Knightes, andK. Sullivan, (in preparation). A
comparison of SWAT and HSPF parameter sensitivity and streamflow simulation accuracy: Two case
studies from active hydraulic fracturing regions.
Price, K., S. T. Purucker, S. R. Kraemer, and J. E. Babendreier. 2012. Tradeoffs among watershed model
calibration targets for parameter estimation. Water Resources Research 48(10).
Price, K., S. T. Purucker, S. R. Kraemer, J. E. Babendreier, and C. D. Knightes. 2014. Comparison of radar and
gauge precipitation data in watershed models across varying spatial and temporal scales. Hydrological
Processes 28(9): 3505-3520.
PRISM Climate Group. 2013. 30-year normal precipitation normal: annual; period: 1981-2010. Oregon State
University.
Ries, K. G., III. 2007. The National Streamflow Statistics Program: A computer program for estimating
streamflow statistics for ungaged sites. USGS Techniques and Methods 4-A6. Reston, VA. 37 pp.
Schaefli, B. and H. V Gupta. 2007. Do Nash values have value? Hydrological Processes 21(15): 2075-2080.
Stuckey, M. H. 2006. Low-flow, base-flow, and mean-flow regression equations for Pennsylvania streams. USGS
Scientific Investigations Report 2006-5130. 84 pp.
SWAT. 2014. ArcSWAT. Available at: http://swat.tamu.edu/software/arcswat/
Appendix B-27
-------
Water Acquisition for Hydraulic Fracturing May 2015
U.S. EPA (Environmental Protection Agency). 2000. BASINS technical note 6: Estimating hydrology and
hydraulic parameters for HSPF. EPA-823-ROO-012. U.S. EPA Office of Water. 32 pp.
U.S. EPA (Environmental Protection Agency). 2011. Automated Geospatial Watershed Assessment: AGWA 2.0
product release. Available at: http://www.epa.gov/esd/land-sci/agwa/arcgis/prod-release.htm
U.S. EPA (Environmental Protection Agency). 2013. BASINS (Better Assessment Science Integrating point and
Non-point Sources). Available at: http://water.epa.gov/scitech/datait/models/basins/index.cfm
USGS (U.S. Geological Survey). 2012. The StreamStats program. Available at:
http://water.usgs.gov/osw/streamstats/
Watermark Numerical Computing. 2005. PEST model-independent parameter estimation: User manual. 5th
edition
Appendix B-28
-------
Water Acquisition for Hydraulic Fracturing May 2015
Appendix C. Groundwater Methods
Appendix C-l
-------
Water Acquisition for Hydraulic Fracturing May 2015
Contents
Overview 2
Step-Wise and Progressive Groundwater Modeling Approach 3
Appropriateness of GFLOW™ and GMS™-MODFLOW for Project Use 4
GFLOW 4
GMS-MODFLOW 7
Testing Model Parameters with Pumping Test Data 7
Field Application: Towanda Creek Watershed, Bradford County, Pennsylvania 14
PART Baseflow Separation at USGS at Monroeton 16
GFLOW Calibration at the USGS Bradford County Observation Well 20
GFLOW Model of the Towanda Creek Watershed 22
GFLOW Model of the Towanda Creek Watershed Municipal Wellfield 23
Field Application: Colorado River Watershed Between Glenwood Springs and Cameo, Colorado 25
Baseflow Separation USGS Gage at Cameo 25
GFLOW Model of the Colorado River Between Glenwood Springs and Cameo 25
GFLOW Model Groundwater Depot at Parachute Confluence 26
References 29
Overview
Groundwater methods used in analyses of the potential impact of large-volume water acquisition by the oil and gas
(O&G) industry on drinking water aquifers are described below. O&G uses these large volumes of water in hydraulic
fracturing of tight/shale gas. EPA recognized that the investigations should assess the potential for impact at the
individual site of extraction, or the pumping wellfield. Also, the investigations needed to account for cumulative
impacts at the appropriate "groundwatershed" scale. Evaluations of impacts included investigations into the effects of
groundwater pumping on streamflow depletion, lowering of piezometric heads, and lowering of the water table.
Groundwater computer models were used to test and demonstrate hypotheses regarding potential for impact associated
with pumping wells screened in the hidden subsurface. The use of computer models allowed statements beyond basic
water balance and empirical observations.
The study areas for this project included the aquifers associated with: (1) the Susquehanna River watershed that overlies
the Marcellus shale of Pennsylvania, focusing on the Towanda Creek watershed in Bradford County, Pennsylvania; and
(2) the Upper Colorado River watershed that overlies the tight gas units of the Piceance structural basin, focusing on the
Parachute Creek watershed in Garfield County, Colorado. Both Bradford County and Garfield County have been among
the top producers of natural gas from unconventional reservoirs in the United States.
The supporting methods used in this project included: (1) baseflow separation, to estimate average annual groundwater
recharge at the catchment scale; (2) regional groundwater flow modeling, to estimate spatially averaged hydraulic
conductivity and generate water table contour maps of aquifers; and (3) local-scale groundwater flow modeling for
mapping drawdown of the water table and the source water zone, as well as streamflow capture associated with pumping
wellfields. Estimates of baseflow and groundwater recharge informed the groundwater modeling; the calibration and
solutions from the regional groundwater modeling informed the local-scale modeling. A step-wise and progressive
modeling approach, as described in the next section, was applied at each of the study areas.
Appendix C-2
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Step-Wise and Progressive Groundwater Modeling Approach
The groundwater impact investigation used a step-wise and progressive modeling approach incorporating hand
calculations, empirical and spreadsheet analyses, and mechanistic groundwater simulation modeling. This investigation
dealt with shallow and unconfined aquifers, and involved water balance and flow issues only. The analytic element
method (Strack and Haitjema 1981a, 1981b; Stack 1989) and the GFLOW model (Haitjema 1995) were selected for
characterization of averaged steady-state conditions. The finite difference model MODFLOW (Harbaugh 2005) was used
to represent transient flow solutions.
A practical advantage of the analytic element method is operational efficiency. While groundwater models implementing
numerical solutions (e.g., finite differences and finite elements) deal with grids or meshes, the geohydrologist building an
analytic element model works with hydrologic features. For example, representation of streams by strings of straight line
elements and lakes by polygons is an intuitive task. The standard analytic elements, including elements representing
wells, rivers, lakes, inhomogeneities, and recharge, are shown in Fig. C-l. For initial modeling runs, a limited set of
surface water features may be introduced.
Later, when insight into the groundwater
flow regime increases, more data may be
added to refine the model.
The stepwise and progressive groundwater
modeling approach is not new. The
ensuing discussion draws from Henk
Haitjema's description
(http://www.haitjema.com). Ward and
others applied what they called a
telescopic mesh refinement modeling
approach (TMR) to the Chem-Dyne
hazardous waste site in southwestern Ohio
(Ward et al. 1987). However, Ward et al.
had to use three different computer models
for the three different scales at which they
were modeling. Conditions on the grid
boundary of the local scale were obtained
from the regional-scale modeling results,
while, similarly, the conditions on the grid
boundary of the site scale were obtained
from the local-scale modeling results. In
contrast, the analytic element method
allows these different scales to be treated
within the same model by locally refining
the input data, thus avoiding transfer of
conditions along artificial boundaries from
one model into the other.
Standard Analytic Elements
Well
River
Lake
Recharge
Elementary solutions
superimposed to obtain complete
description of flow system...
Inhomogeneity
Figure C-l. The suite of standard analytic elements available for
superposition in the model domain to create a site specific model.
The influence of the element on the hydraulic head contours and
gridded surface and the velocity vectors is shown (Source: Craig 2014).
The analytic element modeling approach allows progression from simple to more complex representations in order to test
understanding. A suite of simple models with few measurable parameters is preferred over a multi-parameter model that
may better fit the data (Kelson et al. 2002). Simple models are used within a deterministic approach in this investigation;
a stochastic approach would require more field data than were available.
While especially suitable for groundwater flow modeling at different scales, analytic element modeling does have some
limitations. For instance, both transient flow and three-dimensional flow are only partially available. While an analytic
element model can represent macro-scale heterogeneities (such as the difference in hydraulic conductivity associated
with alluvium and hard-rock aquifers) in a piece-wise manner, the models do not currently represent gradually varying
aquifer properties. The representation of multi-layer aquifer flow is an advanced technique.
Appendix C-3
-------
Water Acquisition for Hydraulic Fracturing May 2015
Simple hand calculations can help guide the complementary use of steady-state and transient groundwater models
(Haitjema 2006). When needed in our investigation, the finite difference numerical model MODFLOW was extracted
from the regional analytic model GFLOW to facilitate a local-scale transient representation of the groundwater system.
This hybrid approach is well documented (Dunning et al. 2004; Feinstein et al. 2003). As explained in the next section,
the complexity of the modeling used in this project was tested to demonstrate whether it was appropriate for the uses
intended.
Appropriateness of GFLOW™ and GMS™-MODFLOW for Project Use
Both field cases for this investigation involved shallow, unconsolidated valley-fill aquifers containing perennial
groundwater-supplied creeks surrounded by tighter rock units with topographic relief. The GFLOW and MODFLOW
models were evaluated and deemed appropriate for representing steady and transient flow to pumping wells in these
single-layer hydrogeologic systems. Model performance was demonstrated using pump test data at a private groundwater
supply in Wyoming County, Pennsylvania, which is also a permitted private supplier of groundwater to the O&G
industry.
GFLOW
The GFLOW computer program (v.2.1.2; July 8, 2007) was used in this project to solve for regional and steady
groundwater flow in single-layer aquifers (Haitjema 1995). GFLOW is well documented and accepted within the
groundwater modeling community (Hunt 2006; Yager and Neville 2002), particularly when applied to shallow
groundwater flow systems involving groundwater/surface water interactions (Johnson and Mifflin 2006; Juckem 2009)
and for recharge estimation (Dripps et al. 2006). The mathematical foundations of the model include equations that
express the physics of steady advective groundwater flow within a continuum; continuity of flow and Darcy's law (water
flows down the hydraulic potential gradient) are satisfied at the mathematical elementary volume. GFLOW solves the
regional steady-state groundwater flow equations using the analytic element method (Haitjema 1995; Strack 1989) based
on the principle of superposition of elements—line-sink elements represent streams, point-sink elements represent wells,
line-doublet polygon elements represent discontinuities of aquifer properties (such as hydraulic conductivity, base
elevation, and no-flow boundaries), and area elements represent aquifer recharge. The influences of these elements on
the regional flow field are shown in Fig. C-l. GFLOW includes standard example run files to test proper model
installation.
In practice, the basic steps for building a GFLOW groundwater flow model are to:
1. Collect data for model building and testing, including U.S. Geological Survey (USGS) stream gage data for
baseflow characterization and static water levels in wells; USGS digital elevation maps (DEM) and digital
raster graphic (DRG) topographic maps; and USGS digital line graph (DLG) maps of hydrography.
2. Build the model base map for hydrography and geology. Assign labels of topographic elevation (with respect to
the mean sea level datum) along stream reaches.
3. Create the elements using line-sink strings to represent streams, point elements for wells, and area element
polygons for various aquifer properties (recharge, hydraulic conductivity, aquifer base).
4. Run the GFLOW model and conduct manual or automated calibration, minimizing residuals (model simulated
water table elevations compared to observed elevations; model line-sink network cumulative baseflow to field
observed baseflow at the watershed outlet).
5. Refine the local scale, adding wellfields and conducting drawdown analyses and source water zone mapping.
The areas of interest for GFLOW models in this project ranged in scale from full groundwatershed aligned with the
surface watershed down to an individual groundwater depot (e.g., pumping wellfield). Theoretically, analytic element
solutions are spatially infinite, and good modeling practice typically represents both a far field, with coarse
representation of elements and geohydrologic features, and a near field at higher resolution.
To create a bounded flow solution in GFLOW assigned to a topographically defined surface watershed, a closed string of
no-flow line elements was placed on the perimeter of the surface watershed. Even though the static no-flow boundary is
an artificial one (not actually occurring in the natural system), the setup is justified in geohydrologic systems where the
shape of the shallow water table tends to follow the shape of the surface topography, permitting the assumption that
groundwater fluxes in and out of this boundary are insignificant. Also, the base of the single-layer aquifers are assumed
to be horizontal and to constitute a no-flow boundary—indeed, it was evident that deep leakage in both case studies was
Appendix C-4
-------
Water Acquisition for Hydraulic Fracturing May 2015
minimal. GFLOW can represent flow in the aquifer as either unconfmed or confined, or both. The present models
represent shallow, unconfined aquifers. The bounded solution setup simplifies the calibration of a water balance
associated with a surface watershed.
Shallow groundwater flow systems are intimately linked with surface drainage. The perennial stream network is
understood to be flowing year round. In contrast, the ephemeral stream network is dry most of the year, only flows
during intense rainfall events, and contributes to rapid surface runoff. The intermittent stream network is understood to
be supported by shallow drainage of the unsaturated soil horizon. For a stream to be flowing when it has not rained for
many days, the source of the river water is subsurface groundwater drainage, also called baseflow. The distinction on the
landscape of perennial, intermittent, and ephemeral flow is dynamic and dependent on antecedent soil moisture
conditions.
Field evidence of a snapshot of the topographically defined drainage network, including ephemeral, intermittent, and
perennial channels, appears on USGS topographic maps (the dashed lines are assigned to intermittent channels, the solid
blue lines to perennial channels). For the maps in our study area, and based on field reconnaissance, the "blue lines"
were confirmed to give a reasonable first estimate of the perennial stream network. The perennial stream network was
used as a calibration target in the GFLOW model. Granted, the transition point on the landscape will move up and down
the stream segment depending on groundwater recharge and the movement up and down of the shallow aquifer water
table, the USGS blue line is hypothesized to be an effective representation of average drainage conditions.
The perennial stream network defines an internal boundary condition for GFLOW, and the network of line-sinks
integrates and routes drainage from recharge to baseflow discharge at the groundwatershed outlet (Mitchell-Bruker and
Haitjema 1996). The nominated stream locations from USGS topographic maps or digital elevation models (DEM) were
translated into GFLOW line-sink representations of streams. Head at a location on the landscape is understood to be the
elevation at which water saturates an open pipe piezometer driven into the aquifer. The strength (or inflow/outflow per
unit length) of the line-sink is determined in the analytic element solution by maintaining a specified head in the center
of the line-sink element. A combination of methods was used to estimate the land surface elevation at select locations on
the GFLOW base map (Fig. C-2A-C-2D): (1) labeling elevations where elevation contour lines from the USGS map
crossed the stream channel; and/or (2) labeling elevations at selected points on the landscape using a 30m resolution
DEM. The GIS software Arc View 3.3 with the Spatial Analyst plug-in and an Avenue Script (sppntzVal.ave) were used
to select the specific points along the drainage network where elevations were labeled. The GFLOW line-sinks were then
manually superimposed on the base map, ensuring that vertices at the end of line-sink strings corresponded with points
of known head/elevation from the USGS sources (see Fig. C-2D). The head at the center of each of the line-sink strings
is calculated through linear interpolation.
The GFLOW conjunctive groundwater-surface water solution integrates the baseflow in the network of tributary streams
represented by line-sinks to the watershed outlet, and through numerical iteration results in a flow solution that defines
an active line-sink network. Headwater line-sinks that appeared above the water table in the model were allowed to dry
up (Fig. C-2D). The GFLOW recharge parameter was adjusted and associated with the areal element (inhomogeneity)
and the baseflow was summed in the activated line-sink network to match the inferred baseflow observed at the USGS
stream gage at the watershed outlet. The USGS computer program PART and chemical methods were used for baseflow
separation. PART uses streamflow partitioning to estimate daily groundwater discharges under the streamflow record
(Rutledge 1998). The method designates groundwater discharge to be equal to streamflow on days that fit a requirement
of antecedent recession, linearly interpolates groundwater discharge for other days, and is applied to a long period of
record to obtain an estimate of the mean rate of groundwater discharge. If one assumes there is no deep groundwater
leakage and no subsurface flux of groundwater across the watershed boundary, the average groundwater recharge rate for
the time period can be translated as the volume of baseflow distributed over the watershed area.
Appendix C-5
-------
Water Acquisition for Hydraulic Fracturing
May 2015
A)USGS1:100KTopoMap
B) USGS 1:24K Topo Map
S
USGS ephemeral/perennial
transition
C) Extraction of elevations along stream channels
D) GFLOW line-elements
Figure C-2. Example of steps to parameterize the GFLOW line-elements. A) The USGS 1:100K topographic maps
suggest the active stream channels. B) The USGS 1:24K topographic maps show stream channels as ephemeral
(dashed blue lines) and perennial (solid blue lines), and were confirmed by EPA field reconnaissance as wet or dry.
(See Appendix B, "Surface Water Hydrology," Fig. B-7B). C) The labeling of points on the base map for extraction
of the elevation either manually from the contour line or digitally using a background DEM. D) Assignment of
topographic elevations to the vertices of the line-elements resulting in the conjunctive groundwater-surface
water solution resulting in line elements contributing to baseflow (wet, colored as aqua) and not contributing to
baseflow (dry, colored as gray).
Appendix C-6
-------
Water Acquisition for Hydraulic Fracturing May 2015
Another output of the GFLOW regional groundwater model is a continuous surface representing piezometric head, or
groundwater flow potential, for the unconfmed aquifer. This surface of heads is the same as the water table surface for
unconfined aquifers. The water table solution depends on the aquifer recharge rate and the aquifer transmissivity (or
hydraulic conductivity times aquifer thickness). Assuming a constant transmissivity, the higher the recharge rate, the
higher the model-predicted elevation of the water table. Conversely, assuming a higher recharge rate, the higher the
aquifer transmissivity, the lower the water table will be. Once the recharge rate is known after conducting baseflow
separation as described above, the model can be calibrated to "fit" the observed water table elevations at points by
varying the aquifer transmissivity, and monitoring the model-predicted water table at monitoring wells where the water
table elevation is measured. Additionally, comparing model predictions of the geographic transitions between
perennial/intermittent streams may provide additional opportunities to compare model predictions to field observations.
In summary, the two calibration targets, baseflow at the watershed outlet and observed elevations of the water table in
unconfined aquifers, allow for the parameterization of the average recharge and transmissivity of the regional steady-
state aquifer flow system equations in the GFLOW model.
For the shallow aquifers of Pennsylvania and Colorado, it is recognized that the groundwater systems are dynamic and
responsive to changes in recharge and evapotranspiration, surface water boundary conditions, and water pumping. A
visual inspection of the observed water table at the USGS Bradford County Observation Well shows that there is
seasonal periodicity to the water table response, and in the long term, the range of water table change is approximately
10 feet above and below a mean (Fig. C-7). A long-term record of a well screened in the alluvium of the Colorado River
was not available, but it is expected that the aquifer responds quickly and periodically about a mean. In this project,
GFLOW is used to represent annual (or longer-term) averaging. The model can represent the long-term average mean,
or, equally interesting, the long-term average low flow condition of the system. The veracity of the averaging
assumptions is tested in the next sections.
GMS-MODFLOW
The USGS MODFLOW model was used to represent transient groundwater responses to pumping wells. The
MODFLOW-2005 (Harbaugh 2005) open source groundwater solver is included in the Groundwater Modeling System
(GMS) version 10.0 (Aquaveo 2014). MODFLOW is the most widely applied groundwater modeling flow model in the
United States. It has undergone 30 years of development and quality testing by USGS. GMS includes standard
MODFLOW example run files to confirm proper model installation. In addition to facilitating a standard cell-based
interface to the MODFLOW finite difference grid, GMS includes a geohydrological conceptual design environment
much like GFLOW. Therefore, the data preparation and regional solution that were previously described for GFLOW
were used.
Testing Model Parameters with Pumping Test Data
To test the appropriateness of conceptual assumptions of single-layer aquifers and averaged steady flow, a data-rich field
site was selected for model testing. The publicly available pump test data were acquired for the private groundwater
supply wellfield (SID 3711/3712) in Wyoming County, Pennsylvania from the Susquehanna River Basin Commission
(SRBC). The data had been submitted in support of the permitting of three pumping wells (Casselberry and Associates
2009, 2010, 2014). The request was approved to sell freshwater and effluent to the O&G industry. The private water
supply wells are on the Bowman Creek floodplain, which delineates the valley-fill glacial outwash deposits bounded by
bedrock uplands having high topographic relief (Fig. C-3). The bedrock is composed of very fine-grained sandstone,
siltstone, mudstone, and shale belonging to the Catskill formation. The essentially horizontal outwash aquifer is
approximately 40-60 feet thick (Fig. C-4).
Appendix C-7
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Figure C-3. The setup for the 600-gallon-per-minute pump test of
December 21-24, 2009, at the Wyoming County, Pennsylvania, private
water supply wellfield — three pumped wells (Wl, W2, W3) and five
monitoring wells. The outwash deposits of Bowman Creek are bounded by
bedrock uplands of high topographic relief (approximate limits shown as a
tight dashed line) (after Casselberry and Associates 2009, with
permission).
Section A - A' Drawn Along the Valley Axis
Section B - B' Drawn Perprndicular to Valley Axis
Figure C-4. Geologic cross-sections through Bowman Creek Valley at the private
water supply location (after Casselberry and Associates 2009, with permission).
Appendix C-&
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Days Pumping
Figure C-5. Typical monitoring well response to the pumping test, including
static water level in the well prior to initiation, initial rapid response to
maximum drawdown (DD in ft), with equally quick recovery. Data from
Casselberry and Associates (2010), with permission.
A 72-hour pumping test was
initiated by Casselberry and
Associates on December 21,
2009, at 11 a.m.; the pumping
concluded on December 24 at 11
a.m. The 20 horsepower pump is
at Well #1 and a suction line
connects to Well #2 and Well #3.
Interestingly, Well #3
experiences the greatest water
level drawdowns (Casselberry
and Associates 2010). There
were five monitoring wells, as
shown in Fig. C-3. The typical
water level response to the
pumping test is shown for
monitoring well #2 in Fig. C-5,
with the pre-pumping static
water level rapidly responding to
the 600 gallons per minute (gpm)
pumping to a maximum
drawdown, and followed by an
equally rapid recovery once the
pumping stops. Based on water level response at monitoring well #1, a transmissivity was estimated to be 264,000
gpd/ft, and storativity (specific yield) was estimated at 0.025 (Casselberry and Associates 2010).
The data were used to do a quick back-of-the-envelope calculation to characterize the transient response of the aquifer.
Townley (1995) introduced a dimensionless response time of an aquifer to transient recharge:
T =
TP
where S (-) is the aquifer storativity, L (ft) is the average distance between the stream and a water divide or effective no-
flow boundary associated with the rock outcrop, T (ft2/d) is the aquifer transmissivity, and P (d) is the period of the
recharge forcing. If T < 1, then transient groundwater flow can be approximated with successive steady-state solutions
(Haitjema 2006). The estimate for dimensionless response time for the wellfield is much less than one (T is «1) given
S= 0.025, T= 35,291.66 ft2/d, the width of the outwash valley is approximately 1,200-3,300 feet, and recharge annual
forcing is 365 days. Therefore, the steady-state model is expected to approximate the average pre-pumping and the
average maximum pumping conditions.
This is further demonstrated with the GFLOW and MODFLOW models.
The regional-scale GFLOW model was parameterized for the wellfield in the steps described above. The perennial
stream network that surrounded the wellfield was represented by constant head line-sinks. The head assigned to the
center of each line-sink was estimated from elevations provided by USGS topographic maps. The boundary of the
outwash aquifer was inferred from USGS topographic maps, and the GFLOW inhomogeneity element was associated
with enhanced transmissivity and recharge. The value of the enhanced recharge was provided by baseflow separation
using the USGS PART computer program and the USGS annual discharge recorded at the Tunkhannock Creek USGS
stream gage (01534000) near Tunkhannock, Pennsylvania. The observed static and pumping water levels at the wells
were used to parameterize the hydraulic conductivity of the outwash aquifer.
The USGS PART-modeled average baseflow and watershed recharge for the period 2009 and 2013 was 11.19 in/yr
(0.0025519051 ft/d). The recharge was assigned to the inhomogeneity element representing the valley (Fig. C-6).
Appendix C-9
-------
Water Acquisition for Hydraulic Fracturing
May 2015
A) GFLOW
B) MODFLOW
inhomoge
line-sinR
Figure C-6. Layout of numerical features of the two groundwater flow models for the Wyoming County,
Pennsylvania, private water supply case study. A) GFLOW analytic elements. The inhomogeneity element controls
the recharge and the hydraulic conductivity of the valley-fill outwash.The line-sink elements represent the
perennial creeks. The background map has USGS 1:100,000 scale hydrography. The nearfield includes local detail
and the wellfield. B) GMS-MODFLOW has 15,198 cells, showing grid refinement associated with the pumping
wells. The purple cells are head specified. The heads of cells associated with Bowman Creek were informed by the
U.S. Geological Survey 7.5-minute topographic maps. The heads on the model (grid) boundary were supplied by
the GFLOW model. The cells in the yellow region are associated with the properties of the outwash aquifer.
Table C-l. GFLOW calibration for steady-state model of Wyoming County private water supply pumping test of
December 2009. Drawdown is relative to the no-pumping solution. Observed heads were measured at the three
pumping wells and the five monitoring wells.
Well
W1
W2
W3
Observed
Drawdown
(ft)
3.40
5.08
9.96
GFLOW Well
Discharge
(ft3/d)
14,774
23,777
77,010
Maximum Difference Fj~4
Minimum Difference l.rj 2
Average Difference |Q 4 -§
Median Difference fg~2
Mean Absolute |rj 5
Root-Mean-Square |g 7
V
/
/
1
/
-1-
/
1 1
/
/
z
71 0 71 2 71 4 71 6 71 8 720 722 724 726
Sum of Squared [37 Observed Head
Differences
For the steady-state GFLOW model calibration, the observed drawdown associated with the December 2009 pumping
test was enforced, and a uniform assigned hydraulic conductivity of 248 ft/d to the valley inhomogeneity element
resulted in the 600 gpm (115,500 ft3/d) of pumping to be distributed to the three wells, with acceptable model error, as
Appendix C-10
-------
Water Acquisition for Hydraulic Fracturing May 2015
shown in Table C-l. The hydraulic conductivity of the bedrock was assigned a value of 0.4ft/d, consistent with the
Towanda Creek GFLOW model (described below) and effectively minimizing the impact of the surrounding tight rock
formations on the valley aquifer flow system.
The GFLOW steady-state model was transferred to GMS MODFLOW for pre-pumping (Fig. C-7) and pump test (Fig.
C-8) scenarios. The GFLOW heads on the boundary were extracted to the GMS MODFLOW boundary. These artificial
boundary conditions are far enough away from the pumping center to have no impact on local solution.
GMS MODFLOW was run in full transient mode, and model predictions of drawdown at the five monitoring wells were
compared to the observed drawdown during the 72-hour, 600 gpm pumping test. This is different from the previous
comparison, where the MODFLOW-predicted heads were compared to the GFLOW-predicted heads at the monitoring
points. In this comparison the MODFLOW-predicted drawdowns are compared to the drawdowns observed during the
2009 pumping test. See Fig. C-9. The initial condition was set to the steady-state pre-pumping condition. The shape of
the drawdown curve, which is controlled by the specific yield, is effectively represented in the MODFLOW model. The
magnitude of the drawdown is controlled by the hydraulic conductivity represented in MODFLOW, and the model error
ranges fromO.l feetatMW#4to 1.2 feet at MW#2 and 1.4feetatMW#5. The typical valley-fill outwash unconsolidated
deposits are expected to be stratified and heterogeneous. While the monitoring wells have 30 feet of total depth, they are
only screened in the last 10 feet. Both GFLOW and MODFLOW assume a homogeneous single-layer aquifer without
resistance to vertical flow (the so-called Dupuit-Forchheimer assumption) and are not expected to represent actual
measured heads in a given interval of outwash. Since the observed drawdowns at depths of 30-40 feet appear somewhat
larger than the average drawdowns represented by the model, it is reasonable to expect that the drawdowns higher up in
the aquifer are then lower than predicted by the model.
Thus, the model slightly overestimates the water table decline due to pumping. The transient MODFLOW model
accurately predicted the rapid drawdown during pumping, and rapid recovery once the pumps were turned off. The
MODFLOW demonstration supported the use of the steady-state GFLOW model going forward to meet the project's
conceptual demonstration and objectives.
Appendix C-ll
-------
Water Acquisition for Hydraulic Fracturing
May 2015
C)
Observations 'g
Maximum Difference |g 2
Minimum Difference jg g
Average Difference
Median Difference
Mean Absolute
Difference
Root-Mean-Square
Difference
Sum of Squared
Differences
D)
"D
ro
0)
g
LJ_
Q
O
oo
726
725
£ 722
717 718 719 720 721 722 723 724 725 726
Observed Head
GFLOW™ head (ft amsl)
Figure C-7. Pre-pumping steady-state solutions for A) GFLOW and B) MODFLOW. The constant heads on the
boundary of the GFLOW model were extracted to the boundary of the MODFLOW model. The difference in the
simulated heads at the well points is insignificant.
Appendix C-12
-------
Water Acquisition for Hydraulic Fracturing
May 2015
C)
Observations
M aximum D if f erence j g 2
Minimum Difference |grj
Average Difference
Median Difference
Mean Absolute
Difference
Root-Mean-Square
Difference
Sum of Squared
Differences
[01
[01
-
D)
"i/i
E724 -
724
ro
T3 o 72a
(D *
o
— ' 712 -
Q
V
X
X
^
X
/
X
X
OO
Observed Head
GFLOW™head(ftamsl)
Figure. C-8. End of pumping steady-state solutions for A) GFLOW and B) MODFLOW. The difference in the
simulated heads at the well points is insignificant.
Appendix C-13
-------
Water Acquisition for Hydraulic Fracturing
May 2015
726-
723-
J722-
5
|721-
m
720-
719-
718-
717-
-
;
; ;
M
^r-
-^
H
M
•— —
f
r^
S~~.
1
' 1
• •• moi
lei
"ve(
1
MW#
j
MW#
MW#
MW#
1 ivi VVff
4
5
1
2
3
Figure C-9. Comparison between MODFLOW
drawdown and observed drawdown at the
Wyoming County private water supply monitoring
wells during the December 21, 2009, 72-hour pump
test. The differences between the modeled
drawdown and the mean drawdown are
highlighted in green. The vertical whisker lines span
the 95% confidence interval. (Data source:
Casselberry and Associates 2010.)
Field Application: Towanda Creek Watershed, Bradford County, Pennsylvania
The Towanda Creek watershed (215.6 mi2) of southeastern Bradford County, Pennsylvania, is the focus of our
groundwater availability investigation. Bradford County is in the heart of the Marcellus Shale O&G activity, and the
Towanda Creek watershed has been a hot spot of O&G drilling since 2009. The watershed was used for the previously
described assessment of surface water impact. The watershed contains a public groundwater system that has registered
sales to O&G. The watershed outlet is associated with the USGS gage at Monroeton (01532000) (Fig. C-10).
The aquifers in the study area are associated with the sedimentary rocks of the Pennsylvanian, Mississipian, and
Devonian periods, and the unconsolidated sediments associated with the retreat of the glaciers (Fig. C-10). The rocks
include sandstones, siltstones, and claystones, and the fracturing and bedding provide secondary porosity and
permeability supporting freshwater aquifers (Fig. C-l 1). The unconsolidated deposits include alluvium, valley-fill, and
till, associated with the topographic lows and streams.
The Towanda Creek groundwatershed model
was built following the step-wise and
progressive approach. An inspection of the
shallow geology and topography of the area
informed a two-zone conceptual model (Fig. C-
12). At the full Towanda Creek
groundwatershed scale, GFLOW represents the
rock aquifers associated with the Pottsville
formation and the Chemung formation using
inhomogeneity polygon
Figure C-10. The Towanda Creek watershed.
The surficial geology is based on Lohman
(1939). Also shown are the U.S. Geological
Survey (USGS) stream gage at Monroeton, the
USGS Bradford County observation well, and
the Pennsylvania Geological Survey/USGS
Gleason test hole.
» N o
16 Mile,
j Alleghenyformation X jPoconoand others
Che mungand other
Pottsvi I le f ormatio
n ix Catskill and other
I 1
Appendix C-14
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Depth Age Form. Thickness and Geologic Description Gamma
Depth (bgs)
50 CPS 300
Litho gamma
0
100
200
,„„
AOO
4-V7V7
500
600
700
800
900
1000
1 100
1 1V7V7
1200
1300
1400
1500
1600
-,
1
|
fc
C3
'&,
'1
"i
§
g
TO
0
Q
.§
o
Q
1
a
a
1
s
•S
3
P
o
iS
j?
1
ffi
rt
a
o
96 feet thick
2 to 98 feet bgs
5 84 feet thick
98 to 682 feet
bgs
982 feet thick
682 to 1664
feet bgs
Pale-orange to yellowish gray, fine to medium grained
sandstone, cross-bedded with iron staining along bedding
planes and iron pitting. Few interbedded claystone layers,
claystone rip-up clasts within sandstones and minor mica.
Several thin clay layers, 0.25 to 0.50 feet thick, pale-olive
to pale-yellowish orange were noted.
Buff to greenish-gray, very fine to medium-grained,
poorly sorted sandstone: thin, planar to cross-bedded
micaceous, iron staining along bedding planes, blackish
brown speckles, rip-up clasts. Grayish olive-green and
greenish-gray siltstones and claystones are finely
laminated, plant material and carbonaceous layers
common with average thicknesses of 0.5 inches and can
have pyrite claystones, siltstones, and minor very fine-
grained sandstones as thick as 50 feet contain burrows,
root casts, and clay slickensides.
Grayish-red, gray, or mottled red and gray, interbedded,
micaceous siltstones, claystones, and sandstones. Upper
portion is overall sandier and characterized by fining
upward sequences grading from intra-formational
conglomerate with clay rip-up clasts to claystones.
Sandstones are commonly calcareous with low-angle
cross-beds, and ripple marks. Silstones and micaceous,
planar bedded, finely laminated, commonly bioturbated
with burrows and root casts. Claystones have calcareous
white nodules, slickensides, root casts. Carbonaceous
layers of large-bladed plant fossils are found in the
siltstones and claystones. Fossil fish scales, bones, and
plates are common in the red units.
J?jr —
ifi=
i _a~ i
^
•*•?*
-'•^j?
* y_
.'<£
^-T*
r^r^f
fir"*
JHfe
~_~H
^•\
^-
g
K
§
s^-
5^^
^j_*
*^r
n
8
^s
g-S=J3g'
r_?
_ -•>
- — r
Figure C-ll. Description of the stratigraphy/lithology of the Gleason Test Hole. The natural gamma log is shown.
Depth is feet below ground surface. After Risser et al. (2013).
Appendix C-15
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Bradford County
Observation Well
Figure C-12. The GFLOW modeling strategy for the Towanda Creek watershed. Separate
submodels were constructed about the Gleason Test Hole representative of the
Pottsville formation (green areas) and the Bradford County Observation Well
representative of the Chemung formation (red areas). The effective hydraulic properties
of the submodels were transferred to the full groundwatershed-scale model.
elements. The equivalent hydraulic conductivity of the two zones was estimated from two separate calibrations: a
calibration associated with the Gleason Test Hole for the Pottsville zone and a calibration associated with the Bradford
County Observation well for the Chemung zone. The full Towanda Creek groundwatershed model was assembled using
no flow elements associated with the topographically defined catchment, and line-sink elements representing the
perennial stream network. The average annual recharge was supplied by an inhomogeneity element and parameterized
based on baseflow separation. The Towanda Creek GFLOW model facilitated the characterization of available
groundwater storage at the catchment scale. The Towanda Creek groundwatershed model provided the basis for the local
scale GFLOW model that includes the public water supply well. The following sections detail the approach and
methods, starting with generation of the baseflow calibration target.
PART Baseflow Separation at USGS at Monroeton
Baseflow separation was used to extract the groundwater component of the daily streamflow hydrograph recorded at the
USGS stream gage at Monroeton, which defines the outlet of the watershed. The USGS computer program PART uses
streamflow partitioning to estimate a daily record of groundwater discharge under the streamflow record (Rutledge
1998). The method designates groundwater discharge to be equal to streamflow on days that fit a requirement of
antecedent recession, linearly interpolates groundwater discharge for other days, and is applied to a long period of record
to obtain an estimate of the mean rate of groundwater discharge. If no deep groundwater leakage is assumed and no
subsurface flux of groundwater along the watershed boundary is assumed, the average groundwater recharge rate for the
time period can be translated as the volume of baseflow distributed over the watershed area. PART has the advantage of
being automated and uses standard USGS daily flow records.
Appendix C-16
-------
Water Acquisition for Hydraulic Fracturing
May 2015
The entire flow record of the USGS gage at Towanda Creek was evaluated using PART baseflow separation (Fig. C-13).
The average baseflow leaving the catchment for 1915-2013 is 155.6 cfs (380,687 m3/d), equivalent to an average annual
recharge of 9.8 in/yr (0.000682 m/d).
o
in
fO
USGS Towanda Creek @ Monroeton
PART Annual Bas
mow
1
Year
Figure C-13. Annual PART baseflow at the U.S. Geological Survey gage at Monroeton. Baseflow is
expressed as volume of flow divided by contributing watershed area.
The observed fluctuations of the water table at the nearby USGS Bradford County Observation Well suggest an annual
and cyclic frequency (Fig. C-14). The annual time period was assumed for characterization of central tendency of
recharge (e.g., the annual average). In fact, 2011 appears to be the wettest year on record, with PART-computed annual
average recharge of 17.91 in/yr. The calculated recharge rates for a variety of time periods of interest to the project are
summarized in Table C-2.
Table C-2. PART baseflow separation at the U.S. Geological Survey gage at
Monroeton and equivalent recharge at the catchment scale.
Time Period
1915-2013
2000-2011
2009-2013
2011
2013
Baseflow (cfs)
155.63
174.66
166.26
284.26
119.10
Recharge (in/yr)
9.806
11.00
10.475
17.91
7.504
Appendix C-l 7
-------
Water Acquisition for Hydraulic Fracturing
May 2015
ZUSGS
USGS 414330076280501 BR 92 Bradford County Observation Well
744
2088
2882
2984
2006
Daily naninun depth to Hater level
Daily nininun depth to uater level
Daily nean depth to uater level
Period of approved data
Period of record naxinun depth belou land surface.
Figure C-14. Water level hydrograph for the U.S. Geological Survey Bradford County observation
well (433007680501) for 2000 to 2012. (Source: http://waterdata.usgs.gov.)
The GFLOW layout of elements and solution for the hydraulic head contours are shown in Fig. C-15 A. This is the near
field used in the calibration. The surrounding line-elements in the far field associated with perennial streams are not
shown. Areal recharge was set to 11 in/yr (0.002511 ft/d) based on the PART baseflow separation for the 2000-2011
period. Drainage to perennial flow in creeks was included with line-sink strings to the north and south of the Gleason test
hole. Three points of observed or inferred water table elevation were used to calibrate the GFLOW model and to
characterize the hydraulic conductivity of the shallow bedrock aquifer. Note that this assumes perfect communication
between the shallow aquifer and the active creeks—there is no representation in the model of a clay layer in the creek
beds providing resistance to flow. Field survey revealed mostly rocky stream beds. The elevation of the stream is equal
to the elevation of the water table at the creek channel location. The elevation of the water table at Points 1 and 2 was
inferred based on evidence of a transition from intermittent to perennial flow, as designated in the USGS Gleason and
Canton 7.5 minute quad maps (dashed blue line to solid blue line representation of the creeks). Point 3 is associated with
the water table observed at the Gleason Test Hole.
During calibration, the hydraulic conductivity of the rock aquifer was varied to minimize residuals (the differences
between model-predicted heads and observed heads at the three points). "Head" is another name for piezometric head, or
groundwater flow potential, and includes both elevation and pressure components. If a hollow stand pipe is driven into a
shallow aquifer, or piezometer, the elevation to which the water rises in the pipe is a measure of head. And groundwater
always flows from higher head to lower head, or down the hydraulic head gradient. For the unconfmed aquifer, the
elevation of the water table is a measure of head, and direction of groundwater flow can be inferred from a map of the
water table surface. Given a constant recharge rate, which for this model was assumed to equal 11 in/yr (0.002511 ft/d), a
lowering of the hydraulic conductivity of the rock aquifer would result in a rise in the regional water table. Likewise,
raising the hydraulic conductivity of the aquifer would cause a lowering of the water table. The goal of the calibration
was to match the water table of the model at the three observation points. The result of the calibration process minimized
the measures of difference (Fig. C-15B), and resulted in a rock hydraulic conductivity of k = 0.086 ft/d (0.0262 m/d).
Knowing a saturated thickness of 2,095 feet amsl at the Gleason Test Hole, aquifer transmissivity (hydraulic
conductivity times aquifer thickness) of 180.2 ft2/d (16.74 m2/d) was associated with the mapped areas of the outcrop of
the Pottsville geologic formation (Figs. C-10, C-12). This rock transmissivity was used in the GFLOW model of the
Towanda Creek watershed, as described below.
Appendix C-18
-------
Water Acquisition for Hydraulic Fracturing
May 2015
A)
B)
• GFeason Test Hole
_~^
Towanda Creek watershed ) ,
Rashbone Creek
'
Head Calibration Statistics
File Edit
Piezometers
Observations [3
Maximum Difference [o~4
Minimum Difference |_33
Average Difference pf^
Median Difference [04
Mean Absolute Pfs
Difference
Root-Mean-Square [20
Difference
Sum of Squared jl~2~2
Differences
I Gages J Lake Stages
ff" ScaHefPlof ' ' Cumulat
i
E
3
Y
2090 21
/
/
/
ve Probability Plot
/
-2
30 2110 2120 2130 21
Observed Head
z
/
10 2150 2160
001
Figure C-15. A) GFLOW model for the Gleason Test Hole, showing the layout of
elements and contours of solution for hydraulic heads. B) GFLOW calibration
statistics at points 1, 2, and 3. Heads are in feet above mean sea level.
Appendix C-19
-------
Water Acquisition for Hydraulic Fracturing
May 2015
GFLOW Calibration at the USGS Bradford County Observation Well
A GFLOW groundwater flow model was constructed in association with the USGS Bradford County Observation Well
(Fig. C-12). The purpose of this model was to assist in the characterization of the permeability of the rock aquifers of this
area. The observations of the water table at the Bradford County Observation Well provided a valuable calibration target
for the groundwater flow model. Once developed, this local-scale model informed the full Towanda Creek
groundwatershed flow model.
The USGS Bradford County Observation Well was drilled to 117 feet below land surface and completed in the Lock
Haven rock formation, which underlies the Catskill formation and is associated with the Chemung and other formations
(Fig. C-10). The maximum fluctuation of the water table is approximately 10 feet, and there is an annual cycle of low
water tables in the late summer and high water tables in early spring (Fig. C-14).
The GFLOW groundwater model represents the long-term average geohydrologic state of the aquifer system. Four points
of observed or inferred water table elevation were used to calibrate the model in the geographic area surrounding the
USGS Bradford County Observation Well (Fig. C-16A). The USGS Bradford County observation well, shown as Point
4, provided an average static water table elevation of 8.261 feet below ground surface from 2000 to 2011 (Table C-3).
The land surface elevation at the USGS well is 741.6 feet amsl.
Therefore, the average water table elevation for 2000-2011 at the
USGS observation well is 733.3 feet amsl (223.5 m). The heads (or
water table elevation) at Points 1, 2, and 3 were assigned to the
points of transition from intermittent to perennial flow in creeks in
the area, as inferred from the Monroeton and Ulster USGS 7.5-
minute quad maps (dashed blue lines transitioning to solid blue
lines) (Fig. C-16B).
Table C-3. Static water table elevations in the U.S.
Geological Survey Bradford County Observation
Well (2000-2009).
Static water Head (ft above mean
elevation (ft sea level)
below ground
surface)
The GFLOW model represented two major geology zones: the
glacial valley fill (alluvium) and the surrounding rock aquifers. The
model has line-sinks for streams, area elements for recharge (2000-
2011, 11 in/yr or 0.002511 ft/d), and inhomogeneity elements to
represent the jump in hydraulic conductivity between the alluvium
from the surrounding bedrock. The base of the aquifer was set at sea
level. During calibration, the hydraulic conductivity was varied to
minimize residuals between model-predicted heads and observed
heads. The solution was insensitive to the hydraulic conductivity of
the alluvium which was set to 100 ft/d, typical of sand and gravel.
The best fit parameterization that resulted in the lowest measures of
difference (Fig. C- 16B) has hydraulic conductivity of the rock at k
= 0.38 ft/d (0.116 m/d). This rock and hydraulic conductivity was
associated with the mapped outcrop of the Chemung and other
formations (Figs. C-10, C-12), and approximated a transmissivity
(hydraulic conductivity times thickness) of 278.7 ft2/d (25.9 m2/d) at
the USGS Bradford County Observation well.
2000
2001
2004
2006
2007
2009
2010
2011
Average
8.294
9.088
7.093
8.080
8.769
8.755
8.785
7.220
8.261
733.3
732.5
734.5
733.5
732.8
732.8
732.8
734.4
733.3
The next section presents a full-scale Towanda Creek groundwatershed model based on the GFLOW models for the
Gleason Test Hole and the Bradford County Observation Well, and the estimates of the hydraulic conductivity of the two
major rock geology zones.
Appendix C-20
-------
Water Acquisition for Hydraulic Fracturing
May 2015
A)
USGS Bradford County
Observation Well
B)
File Edit
Piezometers
Observations [4
Maximum Difference [^4"
Minimum Difference [^5
Average Difference pfg
Median Diffetence pu
M ean Absolute [2*4
Difference
Root-Mean-Square [27
Difference
Sum of Squared |29.6
Differences
I Gages j Lake Stages
^
ff Scatter Plot T Cumulative Probability Plot
3
1
/
-K4
/
/
700 800 9C
Gt
lj :^E- »l
2
2
/
/
/
^
0 1000 1100 1200
served Head
Figure C-16. A) Layout of GFLOW elements and solution showing
the contouring of heads. B) GFLOW calibration statistics for the
model of the USGS Bradford County well. Heads in ft above mean
sea level.
Appendix C-21
-------
Water Acquisition for Hydraulic Fracturing
May 2015
GFLOW Model of the Towanda Creek Watershed
The Towanda Creek GFLOW groundwater model includes the complete stream channel network that drains to the
watershed outlet at the USGS gage at Monroeton (Fig. C-10). The creeks were represented with line-sinks and an area
element was used to represent recharge over the catchment; no-flow elements were used to represent an artificial
boundary condition, effectively aligning the groundwatershed with the surface watershed.
A two-zone geologic model (two rock types) was tested for its ability to capture the essence of the groundwater system.
At this scale, it the alluvium was assumed to be unimportant to the water balance. The hydraulic conductivity from the
Gleason Test Hole GFLOW model was mapped to the Pottsville affiliated formations, and the hydraulic conductivity of
the USGS Bradford County observation well GFLOW model was mapped to the Chemung affiliated formations, both
from previously described calibrations (Fig. C-12).
The conjunctive groundwater-surface solution using GFLOW is achieved through iteration, and the solution integrates
the cumulative baseflow in the line-sink network for comparison to the discharge observed at the USGS gage at
Monroeton (the outlet of the Towanda Creek watershed). The GFLOW model removes from the solution the headwater
stream channels that are above the water table, and thus do not receive any groundwater, that is, GFLOW allows the non-
contributing creeks to go dry. The Towanda Creek GFLOW solution is shown in Fig. C-17, based on the area! recharge
of 11 in/yr. The two zones of hydraulic conductivity are the Pottsville zone inhomogeneity element (green zone) and the
Chemung zone areas outside of the Pottsville zone (red zone). Also shown are the continuous water table surface,
represented by the contours of hydraulic head, and the cumulative baseflow, as represented by the thickness of the line-
sinks representing the creeks. The dried-up creeks are grayed out.
Figure C-17. GFLOW model of Towanda Creek aquifer system, showing hydraulic head
contours and cumulative baseflow in the line-sink network. GFLOW represents two zones
of hydraulic conductivity.
Appendix C-22
-------
Water Acquisition for Hydraulic Fracturing
May 2015
To this point, a Towanda Creek groundwatershed model has been presented based on annual average area! recharge and
two zones of rock transmissivity, resulting in an output of a smooth and continuous water table surface that honors the
elevations of the perennial stream network. The surface can be used for estimating the aquifer thickness and estimating
available groundwater in storage.
GFLOW Model of a Towanda Creek Watershed Municipal Wellfield
A local-scale groundwater model was built centered on a municipal wellfield (PA2080003) within the Towanda Creek
watershed to demonstrate the potential impact of water withdrawals. The municipal wellfield sells surplus groundwater
to the O&G industry. The stratified-drift unconfined aquifers in the valleys and the post-glacial alluvium associated with
the streams in the area are the most important sources of shallow and available groundwater (Fig. C-18).
The hydrogeologic system in the area
was represented in GFLOW as a two-
geologic-zone model with
transmissivity associated with the
bedrock and the valley-fill (Fig. C-
19A). The wellfield is within the
model near-field with detailed
representation of point-sinks (wells),
line-sinks (streams), and
inhomogeneities (aquifer types with
relevant hydraulic conductivity). The
model far-field extends the elements
in much coarser expression and far
enough away from the wellfield so
that its influence on the near-field
solution is insignificant. The
rectangular areal element providing
recharge extends into the far-field.
The represented line-sinks in the far-field are informed by the previously described Towanda Creek watershed GFLOW
model, with its perennial stream network associated with observed baseflows (Fig. C-19B).
The municipal wells were drilled to a depth of 120 feet from the land surface elevation of 1,128 feet amsl, and the pumps
are rated at 350 and 305 gpm, or a combined 655 gpm (3,570 m3/d). The two wells are represented as a single pumping
center in the GFLOW model. The base of the aquifer beneath the wellfield was set at 307.2 meters amsl.
Three observation points of static water levels close were used to inform the near-field calibration. Calibrating the two-
zone GFLOW model involved varying the hydraulic conductivity of the rock and alluvium to minimize the difference
between the observed head and the model-predicted heads at the locations of the three observations of static water levels.
An additional requirement was that the hydrogeologic system had to support a wellfield drawdown to the base of the
aquifer (307.2 m) at the maximum pumping rate of the wells (3,570 m3/d). The hydraulic conductivities of krock= 0.113
m/d and kaiiuvmm = 6.554 m/d met the requirements. The solution is shown in Fig. C-20.
The GFLOW groundwater model allowed the mapping of the cone of depression of the water table, the zone contributing
recharge at the maximum supported pumping rate, and the local streamflow capture. The discussion of the Marcellus
Shale/Susquehanna River Basin in Chapter 4 of this report includes analysis of groundwater use intensity.
Figure C-18. The principal aquifer of the municipal wellfield is
associated with the post-glacial valley-fill deposits of Towanda Creek.
(Image source: Williams et al. 1998.)
Appendix C-23
-------
Water Acquisition for Hydraulic Fracturing
May 2015
A)
B)
A alluvium
F alluvial fan
T glacial till
VSG valley-bottom sand and gravel
ISG ice-contact sand and gravel
TL glacial till and lake sediments
area-element => recharge - line-element => ere
inhomogeneity-element => alluvium
Figure C-19. A) The distribution of alluvium based on the surficial geology map of Sevon and Braun (1997). B) Layout of
analytic elements in the GFLOW model of the municipal wellfield.
0.3
0.3
0.6 Miles
Figure C-20. GFLOW solution showing hydraulic head or water table contours for the Towanda
Creek watershed municipal wellfield for the maximum supported pumping rate.
Appendix C-24
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Rio Blanco County
,.
Field Application: Colorado River Watershed Between Glenwood Springs and Cameo,
Colorado
The project study area for groundwater impact included the watershed of the Colorado River in Garfield County between
the USGS stream gage below
Glenwood Springs and the USGS
stream gage near Cameo (drainage area
1,975 mi2) (Fig. C-21). Of particular
interest was the confluence of the
Parachute Creeks, which is the location
of an oil & gas structure, a significant r
groundwater wellfield that is sourcing
the O&G industry. The investigation
included the use of chemical baseflow
separation to estimate the areal
recharge rate, the use of the recharge
rate to build a GFLOW Colorado
River/Garfield County
groundwatershed model to estimate the
regional rock transmissivity, and the
use of both the recharge and the rock
transmissivity to build a near-field
GFLOW model of the Parachute Creek
confluence. The Parachute Creek
confluence GFLOW model was used to
model and map the localized cone of
depression and streamflow capture
associated with the O&G wellfield.
20
20
40 Miles
Baseflow Separation USGS Gage
at Cameo
Figure C-21. The Colorado River watershed between Glenwood Springs
and Cameo, and the Parachute Creek confluence, location of a major
water supply wellfield supplying oil and gas.
The graphic method for baseflow
separation using USGS PART did not
prove effective for this snowmelt-
dominated watershed. There were
sufficient data for Miller et al. (2014)
to perform a chemical baseflow separation method using stream water conductance measurements to distinguish runoff
from baseflow in the Upper Colorado River Basin. The analysis was associated with the Colorado River watershed
draining to the USGS gage at Cameo (7,986 mi2 drainage area), and mean annual stream discharge for the 2007-2012
period of 4,061 ft3 per second (dfs). The chemical data suggested annual average baseflow of 44% of discharge (1,892 ft3
per second); snowmelt period baseflow 29% of discharge; and low flow period baseflow 72% of discharge. Adjusting for
the drainage area between the USGS gage at Glenwood Springs and Cameo, the effective annual average recharge for
2007-2012 was 1.03 in/yr.
GFLOW Model of the Colorado River Between Glenwood Springs and Cameo
The purpose of the GFLOW Colorado River groundwater flow model was to distribute baseflow spatially to the
perennial stream network of the catchment of the Colorado River between Glenwood Springs and Cameo. The model
represented the creeks with line-sinks, used an area element to represent recharge over the catchment, and no-flow
elements to represent an artificial boundary condition, effectively aligning the groundwatershed with the surface
watershed. The heads associated with the line-sinks were informed by USGS topographic maps and DEM. The two-
zone (alluvium, rock) model performed better than a single-zone model (rock) based on difference statistics comparing
Appendix C-25
-------
Water Acquisition for Hydraulic Fracturing
May 2015
model-predicted water table to average observed water table. The GFLOW solution is shown in Fig. C-22. The average
baseflow at the Parachute Creek confluence for 2007-2012 was 485,063 ft3/d (5.614 cfs).
Parachute Creek
Confluence
;/;:-; USGS
I/-/- Glenwood
!,; i Springs
USGS
Cameo
Figure C-22. GFLOW solution showing contouring of hydraulic heads, and cumulative baseflow in the streams.
The five USGS observations wells are represented as triangles, with the orientation and color based on the
sign of the residual (difference between model and observed: red triangle with point down means model head
too low, green triangle pointing up means model head too high), and the size representative of the total
difference between modeled and observed. The average difference was 9.5 feet.
GFLOW Model Groundwater Depot at Parachute Confluence
An example of a private groundwater depot or structure to supply fresh water to the O&G industry in the study area
located at the confluence of the west fork, the middle fork, and the east fork of Parachute Creek in Garfield County.
flat alluvial valley is surrounded by the steep slopes of the mountains (Figs. C-23 A, C-23B).
is
The
B) Ground-level view looking north
Figure C-23. Aerial and ground-level views of the confluence of the Parachute Creeks (© 2014 Google Earth, Street
View Image Landsat © USFWS).
Appendix C-26
-------
Water Acquisition for Hydraulic Fracturing
May 2015
The O&G Water Well No. 1 (structure 395298) in the Parachute Creek confluence area provided source groundwater to
industry from 2007 to 2010, and the O&G Well No. 1A (structure 395301) provided source groundwater to industry
from 2011 to 2012. The depth of both wells is 57 feet and the elevation of the land surface at the wellfield is 5,790 feet
amsl.
The hydrogeological conceptual model for the Parachute Creek area shows that the fractured rock of the mountain areas
provides focused discharge to the alluvium of the creek valleys (Fig. C-24).
-ii£ • I
WEST
/ / h' f * ' i, f'{/hi EAST
Unita Formation a
upper par- of
Green Rivet Formation
X
Lowef pa't of Green River Formation
X
W'asatch Forrnatioo
Not to scale
Figure C-24. Generalized geologic cross-section of Parachute Creek. (Source: Adams et a/.
1986.)
The GFLOW model of the Parachute Creek confluence includes line-sink representation of the perennial creeks, an area
element to provide the recharge, and line-elements to differentiate the mountain rocks from the creek valley alluvium.
The wellfield is represented by a single point-sink pumping center (Fig C-25). The transmissivities from the regional
model, with horizontal base of the aquifer set at sea level (0 ft amsl) and regional model hydraulic conductivities of the
mountain and alluvium, were translated to the local model hydraulic conductivities, assuming a horizontal base 5,730
feet amsl and static (no pumping) water levels at the wellfield and the saturated thickness of the aquifer (59.8 ft).
Appendix C-27
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Oil and Gas Water Wells
No. 1 and 1A
Figure C-25. GFLOW model of the Parachute confluence. The shallow bedrock geology
comes from Hail et al. (1989), showing valley-fill alluvium in yellow, and rock
formations marlstone, claystone, and mudstone in shades of orange. The GFLOW
analytic elements include line-sinks for creeks and inhomogeneities for areal recharge
and alluvium.
The O&G wellfield rated pumping is 45,430 ft3/d. A model with a rock hydraulic conductivity of 1.1115 ft/d
alluvium 22.2 ft/d supported the maximum pumping rate, with drawdown reaching the base of the aquifer (5,
the wellfield location. The contouring of
hydraulic head for the GFLOW solution
associated with the maximum rated
pumping is shown in Fig. C-26.
The Parachute confluence GFLOW
groundwater model allows mapping of the
cone of depression of the water table at the
maximum supported pumping rate at the
O&G structure, knowing the pre-pumping
water table and the pumping influenced
water table. The Parachute confluence
GFLOW also allows modeling and mapping
of the zone contributing water to the
pumping wellfield, including capture of
baseflow in the nearby creeks. The analysis
of groundwater use intensity is included in
the discussion of the Piceance structural °-g^^.^^^^^-^^-^^^^^_^^^^^j|^^^^^^^_
basin/Upper Colorado River Basin in
Chapter 5 of this report.
Figure C-26. GFLOW model solution of the Parachute Creek
confluence supporting the maximum rated pumping.
and
730 ft) at
Appendix C-2&
-------
Water Acquisition for Hydraulic Fracturing May 2015
References
Adams, D. B., K. E. Goddard, R. O. Patt, and K. C. Galyean. 1986. Hydrologic data from Roan Creek and Parachute
Creek basins, Northwestern Colorado. USGS Open-File Report 83-859.
Aquaveo. 2014. MODFLOW modeling with GMS. Available at: http://www.aquaveo.com/software/gms-modflow
Casselberry & Associates. 2009. Sugar Hollow Trout Park & Hatchery long-term aquifer testing plan, Wyoming County,
Monroe Township, Pennsylvania. Report submitted to the Susquehanna River Basin Commission.
Casselberry & Associates. 2010. Sugar Hollow Trout Park &0 Hatchery water-supply well evaluation, Eaton Township,
Wyoming County, Pennsylvania: December 2009 aquifer testing results, sustained yield assessment and
evaluation of the potential impacts of pumping on surface-water features and groundwater supplies. Report
submitted to the Susquehanna River Basin Commission.
Casselberry & Associates. 2014. Sugar Hollow Trout Park and Hatchery, October 2013 aquifer testing results, Eaton
Township, Wyoming County, Pennsylvania. Report submitted to the Susquehanna River Basin Commission.
Craig, J. 2014. University of Waterloo lecture notes for course E661: Analytical methods in mathematical geology.
Available at: http://www.civil.uwaterloo.ca/jrcraig/pdf/EARTH66 l_AEMLecture.pdf
Dripps, W. R., R. J. Hunt, and M. P. Anderson. 2006. Estimating recharge rates with analytic element models and
parameter estimation. Ground Water 44(1): 47-55.
Dunning, C. P., D. T. Feinstein, R. J. Hunt, and J. T. Krohelski. 2004. Simulation of ground-water flow, surface-water
flow, and a deep sewer tunnel system in the Menomonee Valley, Milwaukee, Wisconsin. USGS Scientific
Investigations Report 2004-5031. 48 pp.
Feinstein, D., C. Dunning, R. J. Hunt, and J. Krohelski. 2003. Stepwise use of GFLOW and MODFLOW to determine
relative importance of shallow and deep receptors. Ground Water 41(2): 190-199.
Hail, W. J., Jr., R. B. O'Sullivan, and M. C. Smith. 1989. Geologic map of the Roan Plateau Area, Northwestern CO.
USGS Miscellaneous Investigations Series Map 1-1797-C. Available at:
http://ngmdb.usgs.gov/Prodesc/proddesc 9919.htm
Haitjema, H. M. 1995. Analytic element modeling of groundwater flow. Academic Press, Inc., San Diego, CA. 394 pp.
Haitjema, H. 2006. The role of hand calculations in ground water flow modeling. Groundwater 44(6): 786-791.
Harbaugh, A. W. 2005. MODFLOW-2005. The U.S. Geological Survey modular ground-water model —the Ground-
Water Flow Process. USGS Techniques and Methods 6-A16.
Hunt, R. J. 2006. Ground water modeling applications using the analytic element method. Ground Water 44(1): 5-15.
Johnson, C., and M. Mifflin. 2006. The AEM and regional carbonate aquifer modeling. Ground Water 44(1): 24-34.
Juckem, P. F. 2009. Simulation of the regional ground-water-flow system and ground-water/surface-water interaction in
the Rock River Basin, Wisconsin. USGS Scientific Investigations Report 2009-5094. 38 pp.
Kelson, V. A., R. J. Hunt, and H. M. Haitjema. 2002. Improving a regional model using reduced complexity and
parameter estimation. Ground Water 40(2): 132-143.
Lohman, S. W. 1939. Ground water in north-central Pennsylvania. Pennsylvania Geological Survey in cooperation with
the U.S. Geological Survey. Water Resources Report 6. Reprinted 1973. Plate 01. Geologic map of north-
central Pennsylvania compiled by Geo. W. Stose and O. A. Ljungstedt.
Miller, M. P., D. D. Susong, C. L. Shope, V. M. Heilweil, andB. J. Stolp. 2014. Continuous estimation of baseflow in
snowmelt-dominated streams and rivers in the Upper Colorado River Basin: A chemical hydrograph separation
approach. Water Resources Research 50(8): 6986-6999.
Mitchell-Bruker, S., and H. M. Haitjema. 1996. Modeling steady-state conjunctive groundwater and surface water flow
with analytic elements. Water Resources Research 32(9): 2725-2732.
Risser, D. W., J. H. Williams, K. L. Hand, R.-A. Behr, and A. K. Markowski. 2013. Geohydrologic and water-quality
characterization of a fractured-bedrock test hole in an area of Marcellus Shale gas development, Bradford
County, Pennsylvania. Pennsylvania Geological Survey. 4th ser. Open-File Report OFMI 13—01.1. 49 pp. 4
appendices.
Appendix C-29
-------
Water Acquisition for Hydraulic Fracturing May 2015
Rutledge, A. T. 1998. Computer programs for describing the recession of ground-water discharge and for estimating
mean ground-water recharge and discharge from streamflow data—update. USGS Water-Resources
Investigations Report 98-4148. 43 pp.
Sevon, W. D., and D. D. Braun. 1997. Surficial geology of the Towanda 30 x 60 minute quadrangle, Pennsylvania.
Pennsylvania Geological Survey Open-File Report 97-03.
Strack, O. D. L. 1989. Groundwater Mechanics. Prentice Hall, Englewood Cliffs, NJ. 732 pp.
Stack, O. D. L., and H. M. Haitjema. 1981a. Modeling double aquifer flow using a comprehensive potential and
distributed singularities: 1. Solution for homogenous permeability. Water Resources Research 17(5): 1535-
1549.
Stack, O. D. L., and H. M. Haitjema. 1981b. Modeling double aquifer flow using a comprehensive potential and
distributed singularities: 2. Solution for inhomogeneous permeabilities. Water Resources Research 17(5): 1551-
1560.
Townley, L. R. 1995. The response of aquifers to periodic forcing. Advances in Water Resources 18(3): 125-146.
Ward, D. S., D. R. Buss, J. W. Mercer, S. S. Hughes. 1987. Evaluation of a groundwater corrective action at the Chem-
Dyne hazardous waste site using a telescopic mesh refinement modeling approach. Water Resources Research
23(4): 603-617.
Williams, J. H., L. E. Taylor, and D. J. Low. 1998. Hydrogeology and ground water quality of the glaciated valleys of
Bradford, Tioga, and Potter Counties, Pennsylvania. Pennsylvania Geological Survey. 4th ser. Water Resource
Report 68. 89 pp.
Yager, R. M., and C. J. Neville. 2002. GFLOW 2000: An analytic element ground water flow modeling system. Ground
Water 40(6): 574-576.
Appendix C-30
-------
Water Acquisition for Hydraulic Fracturing May 2015
APPENDIX D. Water Use Estimation Methods
Appendix D-l
-------
Water Acquisition for Hydraulic Fracturing May 2015
Contents
Overview 2
Data Sources 2
Estimation of Hydraulic Fracturing Water Use 3
Susquehanna River/Towanda Creek 4
Individual Hydraulic Fracturing Well Consumption 4
Water Withdrawals from Permit Sites 6
Hydraulic Fracturing Scenarios in the Susquehanna River Basin 6
Background Consumption 6
Upper Colorado River Basin 7
Individual Hydraulic Fracturing Well Consumption 8
Water Withdrawals from Structures 8
Hydraulic Fracturing Scenarios 8
Background Consumption 9
Colorado Division of Water Resources: StateMod 10
Scenario Analyses 13
References 15
Overview
This research required quantifying water use at individual withdrawal locations in the Susquehanna River Basin
(SRB) and Upper Colorado River Basin (UCRB) study areas to determine water use intensity from hydraulic
fracturing water acquisition. Water use data were compiled and summarized to develop the factual foundation for
understanding patterns and volumes of use. They were then used to apply the water use intensity analytical approach
to quantify the water balance effects of observed withdrawals on water bodies. Lastly, scenario analyses were used to
fill in gaps related to water bodies, climate conditions, or levels of activity that were not well-represented in observed
withdrawals. In these analyses, withdrawal assumptions, developed from observed use, were applied to lengthy
hydrologic records. This appendix discusses the data sources used in these analyses, and describes the basis for, and
derivation of, water use assumptions for scenario analyses.
Data Sources
Data relevant to Appendix D were obtained from sources listed in Tables D-l and D-2. Information primarily
involved hydraulic fracturing activities (fluid use for well drilling and fracturing, produced water, numbers of wells
drilled, etc.) and water withdrawal volumes. A detailed description of all data sources used throughout the report is
given in Appendix A.
Appendix D-2
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Table D-l. Data sources used for use intensity assessment in the Susquehanna River Basin study area (also listed
in Table 4-2 in the main report).
Source
PADEP
(2014)
PADEP
(2013)
SRBC
(2013)
Data Type
a. Annual PWS water use
report
b. PWS facility information
Well drilling reports
a. Water site water use
b. Miscellaneous reports,
policies, maps
a. Well consumptive Use
c. Docketed permits
Location
http://www.pawaterplan.dep.state.pa.us/State
WaterPlan/WaterDataExportTool/WaterExport
Tool.aspx
http://www.portal.state.pa.us/portal/server.pt/
community/oil_and_gas_reports/20297
http://www.srbc.net/pubinfo/index.htm
http://www.srbc.net/publicinfo/index.htm
http://www.srbc.net/pubinfo/index.htm
http://www.srbc.net/wrp/
Query
Primary Facility Report, by year and
county
Chapter 110 (Act 220) Registration
Wells drilled by county
Provided by SRBC by written request
Website search
Provided by SRBC by written request
Water Resource Portal/Search for
Projects
Table D-2. Data sources used for use intensity assessment in the Upper Colorado River Basin study area (also
listed in Table 5-2 in the main report).
Agency/Organization
Colorado Division of
Water Resources
(CODWR 2014)
Colorado Oil and Gas
Conservation
Commission
(COGCC 2014)
FracFocus(2014)
Description
a. Water rights
information
b. Structure
information
c. Structure water use
d. StateMod water
planning program
a. Well starts and
completions
b. Produced Water
Well fluid volumes
Source/query
http://cdss.state.co.us/onlineTools/
Pages/WaterRights.aspx
http://cdss.state.co.us/onlineTools/
Pages/StructuresDiversions.aspx
http://cdss.state.co.us/onlineTools/
Pages/StructuresDiversions.aspx
(query structures for diversion reports)
http://cdss.state.co.us/Modeling/
Pages/SurfaceWaterStateMod.aspx
http://cogcc.state.co.us
Query: staff report
http://cogcc.state.co.us/COGCCReports/
production.aspx?id=MonthlyWaterProdByCounty
http://www.fracfocusdata.org/DisclosureSearch/
(query by county, look at individual well reports)
Data Use
Priority, decreed use
Locations, history,
ownership
Daily, monthly,
annual volumes used
Scenario analysis of
use and structure
priority
Well counts
Estimates of HF
wastewater reuse
Total well
consumption, counts,
timing
Estimation of Hydraulic Fracturing Water Use
In both study areas, data were obtained for:
• Individual hydraulic fracturing well consumption
• Water withdrawals from sources (streams, ponds, reservoirs, etc.)
• Hydraulic fracturing scenarios: estimating annual hydraulic fracturing activity
• Background water consumption (domestic, agricultural, etc.)
Appendix D-3
-------
Water Acquisition for Hydraulic Fracturing May 2015
Susquehanna River/Towanda Creek
The approach for estimating daily hydraulic fracturing demand in modeled subbasins in Towanda Creek was to divide
individual hydraulic fracturing well demand by an estimated average number of days used to fracture the well. That
daily demand was added to daily background demand, a sum that could then be compared to daily availability
(streamflow) to calculate surface water use intensity (SUI) values.
Individual Hydraulic Fracturing Well Consumption
SRBC provided a database of water used from September 2008 to December 2011 in hydraulic fracturing operations
at well pads. This dataset of 944 records was considered a large statistical sample of the entire population of hydraulic
fracturing wells in the study area. It contained the following pertinent data fields:
• ABR (a unique ID given to each individual well pad)
• Project sponsor (generally, the name of the pad owner)
• Stimulation start date
• Stimulation end date
• Total fluids injected
• Wastewater injected
• Flowback injected
• Freshwater injected
• Flowback recovered
Data were reviewed and 196 records removed according to the following criteria:
• All data fields identical to another record (duplicate records)
• Zero volume of injected freshwater
• Injected freshwater volume greater than total injected fluid volume
A total of 748 records remained. Their distribution of injected freshwater is shown in Fig. D-l. The mean use was
approximately 4 million gallons of freshwater per well. Other statistical information for the sample is shown in Table
D-3.
Table D-3. Information on water usage at hydraulic fracturing wells in the Susquehanna River Basin from 2008 to
2011. (Data source: SRBC 2013a.)
Hydraulic Fracturing Well Facts:
Susquehanna River Basin Value
Number of wells in sample 748
Average total injected volume per well (MG) 4.25
Average total freshwater volume per well (MG) 3.85
% Freshwater volume per well 87%
% Wastewater 0.1%
% Flowback water injected 13%
% Flowback water returned 7%
Appendix D-4
-------
Water Acquisition for Hydraulic Fracturing
May 2015
C.9
0.8
0.2 -
0.1 -
0
160 -,
140 -
1ZD -
100 -
1 80-
50 -
40 -
20 -
n -
I — ,
4 6
Freshwater Used (MG)
D-l 1-2 2-3 3-4 4-5 5-6 6-7 7-B B-9
Freshwater Used (MG)
Figure D-l. The cumulative distribution function (A) and histogram (B) of injected freshwater volumes at 748
hydraulic fracturing wells across the Susquehanna River Basin. (Data source: SRBC 2013a.)
Since this dataset also provided starting and ending dates of stimulation of each well, the length of the fracturing event
was derived by calculating the number of days between them. This period may not accurately represent the duration of
stimulation, however, due to work stoppages for reasons such as weekends and holidays, equipment availability, and
equipment failure. Additionally, companies sometimes begin stimulation to meet permit requirements, then complete
the process later when personnel and equipment are readily available (SRBC staff, personal communication, April
2014). Therefore, events lasting more than 30 days were discarded. The cumulative distribution function (CDF) and
histogram of stimulation lengths for the remaining records are shown in Fig. D-2. The mean and median of this
distribution were nine and eight days, respectively. Seven days was assumed to be a reasonable estimate for the length
of hydraulic fracturing activity at an individual well.
1 -,
0.9 -
0.8 -
£• 0.7 -
15
ra 0.6 -
_Q
O
ol
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Water Withdrawals from Permit Sites
SRBC provided a database of water removal from permitted sites across the Susquehanna for 2009-2013. The original
database had 35,011 non-zero-volume withdrawal records and the following pertinent data fields:
• Docket # (the number given to an individual withdrawal location)
• Source ID (the number given to a specific entity withdrawing from a certain location)
• Date of withdrawal
• Gallons taken
The data were sorted by source type (stream, pond, reservoir, groundwater well, etc.) and records corresponding to
non-stream sites were discarded, leaving 29,907 records of withdrawals for 97 permitted streams. The CDF and
histogram of these removal records are shown in Fig. D-3. The median withdrawal was 0.185 MG, and the 95th
percentile was 1.0 MG.
_
1.0 1.5 2.0
Withdrawal (MG)
2.5
3.0
Withdrawal (MG)
Figure D-3. The cumulative distribution function (A) and histogram (B) of water volumes removed at permitted
sites in the Susquehanna River Basin, 2008-2013. (Data source: SRBC 2013a).
Hydraulic Fracturing Scenarios in the Susquehanna River Basin
Hydraulic fracturing scenarios were designed to test all subbasin streamflows during the modeling period (1987-
2012) in proportion to their occurrence. Each day, a volume of water representing the sum of hydraulic fracturing and
background use was withdrawn. That estimation is discussed in the next section. Three daily hydraulic fracturing
demand volumes were investigated. Two reflect current conditions in the SRB: the median (0.185 million gallons per
day, or MOD) and mean (0.31 MOD) of the data, presented in Fig. D-3. The third represented a future peak drilling
rate in the Towanda Creek basin of 0.57 MOD. That rate was determined using an estimate of 52 hydraulic fracturing
wells per year (Fig. D-2) at an average of 4 MG per well. The daily withdrawal volumes in all three scenarios were
routinely observed (Fig. D-3B). Scenario analysis assumed that all hydraulic fracturing wells consume water from one
source across all 26 years, rather than distributing withdrawals among modeled subbasins.
Background Consumption
The SRBC is conducting a detailed study to estimate cumulative water use in basins across the Susquehanna
watershed. Although it has not yet published a final report on methodology used to derive these estimates, draft
documents detailing the study are online at http://www.srbc.net/planning/cwuas.htm. Data used by SRBC are shown
in Table D-4. SRBC provided daily water use estimates for the scenario test basin. Residential usage was estimated at
0.11 MOD, livestock at 0.16 MOD, and crop irrigation at 0.215 MOD. These figures were based on county-level data
Appendix D-6
-------
Water Acquisition for Hydraulic Fracturing
May 2015
on water use coefficients in residential areas (USGS), population information (U.S. Census Bureau), livestock
populations (U.S. Department of Agriculture, or USDA) and per-acre irrigation volumes for various crops (USDA).
Estimates took into account the amount of water used in each category that is actually consumed (lost to the system)
versus water eventually returned to the stream network.
Table D-4. Data sources used by the Susquehanna River Basin Commission to estimate non-hydraulic-fracturing
water use in Towanda Creek, Pennsylvania.
Organization
U.S. Geological
Survey
U.S. Census
Bureau
U.S. Department
of Agriculture
Description
Per capita water use
coefficients
Population density by
county
Livestock population
by county
Per-animal water use
coefficients
Crop acreage by
county
Water use per acre
by crop type
References
Shaffer, K.H., and Runkle, D.L. 2007. Consumptive Water-Use
Coefficients for the Great Lakes Basin and Climatically Similar
Areas: U.S. Geological Survey Scientific Investigations Report
2007-5197.
United States Department of Commerce, Census Bureau, Geography
Division. 2010. 2010 Census Population & Housing Unit Counts— Blocks.
TIGER/Line Shapefile. Available at: http://www.census.gov/geo/maps-
data/data/tiger-data.html
U.S. Department of Agriculture. 2007. Tables 11-17. In: Census of
Agriculture. Volume 1, Chapter 2: County data.
Jarrett, A.R. 2002. Estimation of agricultural animal and irrigated-crop
consumptive water use in the Susquehanna River Basin forthe years
1970, 2000, and 2025. PSU Department of Agricultural Engineering.
U.S. Department of Agriculture, National Agriculture Statistics Service.
2010. Cropland data layer. Available at:
http://www.nass.usda.gov/research/Cropland/Release/.
U.S. Department of Agriculture, National Agriculture Statistics Service.
2007. Tables 10, 26-29, 31, and 33; Appendix B-35. In: Census of
Agriculture. Volume 1, Chapter 2: County Data (Maryland, New York,
Pennsylvania).
U.S. Department of Agriculture, National Agriculture Statistics Service.
2008. Table 28: Estimated quantity of water applied and primary method
of distribution by selected crops harvested: 2008 and 2003. In: Farm and
Ranch Irrigation Surveys. Available at:
http://www.agcensus.usda.gov/Publications/2007/Online Highlights/Fa
rm and Ranch Irrigation Survey/index. php.
Use
Residential
water use
Residential
water use
Livestock
water use
Livestock
water use
Irrigation
water use
Irrigation
water use
Daily background use estimates in the three categories were distributed among the modeled subbasins in Towanda
Creek using categories of the 2006 National Land Cover Database: agricultural row crops, grass/pasture, and
residential/urban (see Appendix A for additional information on data references). The area of a specific cover type
within a certain subbasin was compared to the area of that cover type within the entirety of Towanda Creek. This
determined the percentage of total daily use volume assigned to that unit. This method produced a unique background
use estimate for each subbasin. For example, if a subbasin had 440 acres of grass/pasture and the total grass/pasture
area of Towanda Creek was 29,400 acres, that subbasin received 1.5% (440/29,400) of the livestock background water
use estimate.
Upper Colorado River Basin
The approach for estimating daily hydraulic fracturing demand for modeled subbasins in the UCRB was to multiply
per-well freshwater usage by number of wells developed annually to arrive at annual hydraulic fracturing demand.
Annual demand was then converted to daily demand by assuming that hydraulic fracturing activities occur 365 days a
year. Daily demand was added to daily background demand, a sum that could then be compared to daily availability
(streamflow) to calculate SUI values.
Appendix D-7
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Individual Hydraulic Fracturing Well Consumption
Since there is no direct report of how much freshwater is used for hydraulic fracturing wells in databases managed by
the state of Colorado, estimates were developed using information provided to federal agencies by hydraulic
fracturing operators and data sources listed in Table 5-2. Oil and gas (O&G) companies have reported that freshwater
is used only for drilling and its associated activities. According to local operators and agencies, high reuse rates are
possible because nearly all hydraulic fracturing fluid injected into directionally drilled tight gas wells in the Williams
Fork Formation returns to the surface within a few months after fracturing.
Operators report that drilling of directional wells and related development consume 0.25 MG (0.77 ac-ft) per well.
WPX Corporation has begun to drill horizontal wells into the Mancos Shale in recent years, reporting that 1.0 MG
(3.2 ac-ft) is needed for these deeper, longer wells (site interview, January 8, 2014).
Water Withdrawals from Structures
Daily (2008-2013) and total monthly (1950-2013) withdrawal volumes from 21 active diversion structures (Fig. D-4)
in Parachute Creek, and 12 in Roan Creek, were obtained from the Colorado Division of Water Resources:
http://cdss.state.co.us/onlineTools/Pages/StructuresDiversions.aspx. For SUI calculations, daily structure withdrawals
in both basins were compared to estimated streamflow at these structures, while historical monthly structure
withdrawals were used in a modeling exercise, StateMod, discussed in Section 3.
Figure D-4. Head gates used to control flows into a diversion (© 2014 Farmers Conservation Alliance, used with
permission, http://farmerscreen.org/screen-proiects/featured-proiects/).
Hydraulic Fracturing Scenarios
Two sets of two hydraulic fracturing withdrawal scenarios (four in all) were examined for Roan and Parachute
watersheds (Table D-5). We define current use as annual mean hydraulic fracturing drilling rates in Garfield County
during 2011-2013, and peak use as the highest historic drilling rate, which occurred in 2008 (see main text, Table 5-
3). Within both current and peak scenarios, SUI values were computed by assuming all completed wells were
directional (0.25 MG fresh water per well) and all new wells were horizontal (1.05 MG fresh water per well). Note:
horizontal drilling has just begun in this area; from 2008 to 2013, completed wells were overwhelmingly directional.
Appendix D-8
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Table D-5. Daily withdrawal rates (in million gallons per day) under the four hydraulic fracturing drilling scenarios.
Scenario
Current Drilling
Directional
Horizontal
Peak Drilling
Directional
Horizontal
Division 5
Drilling Rate
(Wells/Year)
600
1,700
Parachute
Contribution
(Wells/Year)
300
850
Parachute
Total Water
Need (MGY)
75
315
213
893
Parachute
Groundwater
Contribution
(MGY)
41
41
41
41
Parachute
Surface
Water
Contribution
(MGY)
34
274
172
852
Parachute Surface
Water Withdrawal
Rate (MGD)
0.093
0.75
0.47
2.33
Current Drilling. From 2011 to 2013, approximately 600 wells were initiated annually in Garfield County (Table 5-
3); CODWR water use data suggested that Parachute Creek supplied the water needed for half of them (Fig. 5-18).
Structure data indicate that groundwater sources in Parachute supplied 41 MGY of freshwater for hydraulic fracturing
use, and it was assumed that this would remain constant across all scenarios. The remaining volume would be
supplied by surface waters in Parachute, a magnitude that would depend on the types of wells being developed (Table
D-5). The 100% directional well scenario equated to a 0.093 MGD withdrawal rate, assuming withdrawal occurred
uniformly for 365 days per year. The 100% horizontal well scenario withdrawal rate equated to 0.75 MGD.
Peak Drilling. During peak hydraulic fracturing activity in Garfield County in 2008, almost 1,700 wells were initiated
(Table 5-3). Using the same calculations presented in the Current Drilling scenario, the daily withdrawal rate was
estimated to be 0.47 MGD under the 100% directional well scenario and 2.33 MGD for the 100% horizontal well
scenario.
Background Consumption
For Parachute and Roan Creeks, background water consumption estimates were derived from 2005 data (Table D-6)
on water use in Garfield County (Ivahnenko and Flynn 2010). Only water for domestic (1.2 MGD) and livestock (0.3
MGD) purposes was considered relevant. Though water volumes used for crop irrigation are quite large, the irrigation
season coincides with the "call period," which historically runs from mid-April through October. Water use at
structures with junior rights may cease if a senior right places a "call" for water. SUI values at withdrawal structures
were explicitly calculated for this period, so SUIs for modeled subbasins were only calculated during the "free-river
period" (November through mid-April). Municipal water use was excluded because public water supplies in Parachute
and Roan are primarily derived from groundwater wells or the Colorado River (see Chapter 5). Industrial use, which
includes the hydraulic fracturing industry, was excluded from background use calculations. There are no mining
operations in the Parachute and Roan watersheds.
Table D-6. Daily volumes of water use in Garfield County by category (Data source: Ivahnenko and Flynn 2010).
Water Use
Amount (MGD)
Used
Assumption
Domestic
Livestock
Irrigation
Municipal
Industrial
Mining
1.2
0.3
334
15
0.5
0.1
Yes
Yes
No
No
No
No
Relevant
Relevant
Impact directly evaluated using structure data
Water supply derived primarily from Colorado River
Hydraulic fracturing water use was explicitly
estimated
Does not occur in the study area
Appendix D-9
-------
Water Acquisition for Hydraulic Fracturing May 2015
County-wide background use was applied to Parachute and Roan Creeks by area. Since these basins constitute 25% of
Garfield County, 25% of domestic and livestock water use was included in the SUI analyses. Roan is approximately
510 mi2 and Parachute is 200 mi2, so these areas were used to apportion total background use between the watersheds.
Livestock water use was distributed to subbasins using their relative acreage of hay/pasture (from the 2006 National
Land Cover Database; see Appendix A for data reference). The same approach was taken in distributing domestic
water use to subbasins, but using residential land use relative to acreage instead.
Colorado Division of Water Resources: State Mod
The Colorado Division of Water Rights has a modeling system called StateMod
(http://cdss.state.co.us/software/Pages/StateMod.aspx') to help water planners assess impacts of planned actions on
water allocation in the rights system. The user can apply scenarios to determine how much water may actually be
acquired from modeled structures, given streamflow. Simulation results are based on
• Location of each withdrawal structure on a stream network
• Priorities of the various water rights held at each structure
• A demand scenario identifying the structure-specific volume of water that may be withdrawn
• A time series of streamflow that represents water availability
StateMod uses this information to route streamflow through the network of structures, removing a portion of flow at
each, while maintaining all water rights, limitations, and priorities associated with them. A representation of the
StateMod network built for Parachute Creek is shown in Fig. D-5.
This project applied two demand scenarios:
• Per-month median of total monthly withdrawals at each structure since 1950 (MD)
• Per-month total decreed amount of water, summed across all rights at each structure (DD)
These scenarios were calculated from withdrawal records and decreed volumes for primary structures in Parachute
Creek obtained from the Colorado Water Conservation Board
(http://cdss.state.co.us/onlineTools/Pages/StructuresDiversions.aspxX Median and decreed monthly withdrawals for
four example structures are shown in Fig. D-6. Note: the DD scenario is the maximum volume that could theoretically
be taken under unlimited water availability (i.e., very high streamflow).
StateMod was run under two extreme flow regimes:
• 2011 (second-highest annual streamflow in the Colorado River near Cameo since 1934)
• 2012 (third-lowest annual streamflow in Colorado River near Cameo since 1934)
Four sets of StateMod results were thus produced:
• Median withdrawals in a dry year
• Median withdrawals in a wet year
• Decreed withdrawals in a dry year
• Decreed withdrawals in a wet year
For each set, StateMod's simulated volume of water taken at each structure was compared to volume specified by the
demand scenario. Deficiency was defined as:
Deficiency = 1 - (Water Taken/Water Demand) Equation D-l
If a structure received water equal to the scenario demand, deficiency was 0. If the structure received no water,
deficiency was 1. A deficiency map for each scenario is shown in the main text (Fig. 5-25).
Appendix D-l 0
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Colorado River
Figure D-5. Schematic representation of the structures on Parachute Creek simulated in StateMod. Numbers
inside the shapes indicate relative priority (1 for highest priority, etc.) of the senior water right at each structure.
Inset map depicts location of these structures in the watershed.
Appendix D-ll
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Demand Scenarios - Structure 798
Demand Scenarios - Structure 562
70 -,
60 -
< 50 -
15 40 -
ra 30 -
13
£ 20 H
§ 10 H
0
sec -
700 -
6OO -
IT 500 -
400 -
300 -
2OQ -
100 -
c
re
(TJ
lilu
Decreed ~J Median
Decreed e Median
Demand Scenarios - Structure 635
Demand Scenarios - Structure 610
1500 -
15 1000 H
I
2 500 -
lllll.
—, 1500 -
I loco
to
i_
T3
•£ 500 -
inn.
Decreed • Median
Decreed ; Median
Figure D-6. Demand scenarios (median and decreed) for example structures in Parachute Creek. Note the different
magnitude of y-axes from plot to plot. Information from Colorado Division of Water Resources database on
structure withdrawals, 1950-2013.
Fig. D-7 illustrates scenario results and provides some insight into how the water allocation system locally allocates
water during shortages. In the very wet year, the median 60-year irrigation demands (top left in Fig. 5-25) were met at
all structures. Indeed, there was no "call" placed on the Colorado River in this year. In the dry year, historical water
demand (top right in Fig. 5-25) could not be satisfied at many of the structures. In the decreed demand scenario, no
structure receives all the water it expects. Some of the O&G structures supplying hydraulic fracturing freshwater
would receive at least some of their request while some downstream structures would not. This example demonstrates
that the system is dependent on water supply and that water demand locally exceeds supply.
Appendix D-l 2
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Median Demand, Wet
Median Demand, Dry
Decreed Demand, Wet
Decreed Demand, Dry
Water demand (acre-feet)
O <25
O 25-200
(_) 200-1000
C J 1000-2000
2000-10000
O
Figure D-7. Structure deficiency in Parachute Creek for two demand and meteorological scenarios applied at each
structure: (1) the median of annual withdrawal since 1950 and (2) decreed demand is the total of all water rights.
The size of each circle represents the categorical water demand volume; color indicates the proportion of demand
not met—"deficiency"—given streamflow. Deficiency is estimated with CODWR's water planning model StateMod
(CODWR 2014e). The wet year and dry year were historical high and low flow years at the stream gage in the
Colorado River at Cameo: 2011 and 2012, respectively.
Scenario Analyses
The EPA progress report detailing hydraulic fracturing impacts on drinking water supplies (U.S. EPA 2012) outlined a
strategy for capturing future conditions and the impact of changes on water availability in the SRB and the UCRB.
Scenario modeling would ensure that analyses reflected the most important factors affecting water use and availability
in the study areas vis-a-vis current practices and possible future practice:
• Hydraulic fracturing activity
• Hydraulic fracturing water management (primarily recycling of hydraulic fracturing fluids)
• Land use change (population growth and public water supplier (PWS) demand)
• Meteorology variability
• Source locations
A summary of these factors and how each was addressed in this study's current scenario ("Business as Usual," U.S.
EPA 2012) and high-end scenario ("Energy Plus," U.S. EPA 2012) appears in Table D-7. Hydraulic fracturing activity
was addressed using current conditions and peak levels that could occur in the next 30 years, based on recent drilling
trends and projections of natural gas production. Hydraulic fracturing water management looked at potential
reductions in the percentage of recycled hydraulic fracturing fluid due to changes in geology of tapped plays.
Population growth was simulated by increasing PWS demand, where relevant. Meteorological variability was
addressed using a 26-year precipitation series (1987-2012) that captured a range of conditions for each study area.
Watershed modeling allowed SUI estimation at smaller basin scales than could be thoroughly investigated with
empirical data.
Appendix D-l 3
-------
Water Acquisition for Hydraulic Fracturing
May 2015
Table D-7. Important factors affecting water availability and hydraulic fracturing water use in the Susquehanna River Basin (SRB) and Upper Colorado River
Basin (UCRB) study areas.
Scenario
Current
High-End
Hydraulic Fracturing Activity
SRB
Median/mean
withdrawals from
actual sources
Peak annual
projected drilling
rate
UCRB
Current annual
drilling rate,
2011-2013
Peak annual
drilling rate, 2008
Hydraulic Fracturing Water
Management
SRB
Current recycling %
No change:
% recycling already
low
UCRB
Current
recycling %
Reduced
recycling rate
Population Growth
SRB
Current
population
No change:
population
growth not
expected
UCRB
Current
population
Increased
public water
supplier
demand
Meteorology
SRB
1987-2012
No change
UCRB
1987-2012
No change
Source Locations
SRB
Modeling:
0.35-215 mi2
basins
No change
UCRB
Modeling:
4.5-510 mi2
basins
No change
Appendix D-l 4
-------
Water Acquisition for Hydraulic Fracturing May 2015
References
Ivahnenko, T., and J. Flynn. 2010. Estimated withdrawals and use of water in Colorado, 2005. USGS Scientific
Investigations Report 2010-5002. Reston, VA.
Jarrett, A.R. 2002. Estimation of agricultural animal and irrigated-crop consumptive water use in the Susquehanna
River Basin for the years 1970, 2000, and 2025. PSU Department of Agricultural Engineering.
PADEP 2013. http://www.portal.state.pa.us/portal/server.pt/community/oil_and_gas_reports/20297
PADEP 2014. http://www.pawaterplan.dep.state.pa.us/StateWaterPlan/WaterDataExportTool/WaterExportTool.aspx
Shaffer, K.H., and Runkle, D.L. 2007. Consumptive Water-Use Coefficients for the Great Lakes Basin and
Climatically Similar Areas: U.S. Geological Survey Scientific Investigations Report 2007-5197.
Susquehanna River Basin Commission. 2013. http://www.srbc.nei/pubinfo/index.htm
U.S. Department of Agriculture. 2007. Tables 11-17. In: Census of Agriculture. Volume 1, Chapter 2: County data.
U.S. Department of Agriculture, National Agriculture Statistics Service. 2010. Cropland data layer. Available at:
http://www.nass.usda.gov/research/Cropland/Release/.
U.S. Department of Agriculture, National Agriculture Statistics Service. 2007. Tables 10, 26-29, 31, and 33; Appendix
B-35. In: Census of Agriculture. Volume 1, Chapter 2: County Data (Maryland, New York, Pennsylvania).
U.S. Department of Agriculture, National Agriculture Statistics Service. 2008. Table 28: Estimated quantity of water
applied and primary method of distribution by selected crops harvested: 2008 and 2003. In: Farm and Ranch
Irrigation Surveys. Available at:
http://www.agcensus.usda.gov/Publications/2007/Online Highlights/Farm and Ranch Irrigation Survey/in
dex.php.
U. S. Department of Commerce, Census Bureau, Geography Division. 2010. 2010 Census Population & Housing Unit
Counts—Blocks. TIGER/Line Shapefile. Available at: http://www.census.gov/geo/maps-data/data/tiger-
data.html
U.S. EPA (Environmental Protection Agency). 2012. Study of the potential impacts of hydraulic fracturing on
drinking water resources: Progress report. EPA/601/R-12/011. Washington, DC. Available at:
http://www2.epa.gov/sites/production/files/documents/hf-report20121214.pdf
Appendix D-l 5
-------
Water Acquisition for Hydraulic Fracturing May 2015
[This page intentionally left blank.]
Appendix D-16
-------
Water Acquisition for Hydraulic Fracturing May 2015
APPENDIX E.
QUALITY ASSURANCE AND QUALITY CONTROL
Appendix E-l
-------
Water Acquisition for Hydraulic Fracturing May 2015
Contents
Overview 2
Data Acquisition 2
Analytics 3
Product Review 3
References 3
Overview
This project has followed quality assurance procedures described in the Quality Assurance Project Plan titled
Modeling the Impact of Hydraulic Fracturing on Drinking Water Resources Bases on Water Acquisition Scenarios:
Phase 2 Version 1.0 (US EPA, 2013b). The quality assurance project plan (QAPP) addresses:
• Data source quality and documentation
• General analytical approach applied to acquired data,
• Modeling quality assurance, and
• Data management and project archival record keeping, and
• Product review.
Project implementation involved gathering information and data from state and federal agency websites and
hydrologic modeling of hydraulic fracturing withdrawal scenarios. The project generated no new data though
laboratory or field projects. Acquired and modeled data were summarized according to project scientific design.
Data Acquisition
The project team gathered information on where and how much water was acquired in each study basin by querying
publicly available databases from state, regional governmental agencies, and federal data sources, augmented by
databases maintained by nonprofit, or industry organizations. A comprehensive list of data sources is provided in
Appendix A. All data acquired may not have been used in final data products presented in this report but have been
archived with project materials.
The EPA does not make any claims as to the quality or accuracy of the data gathered from the state, federal, and
industry data sources used in the project. The project team applied quality assurance and quality control measures to
acquired data to ensure that the analyses performed were properly conducted and that the data used in this report
faithfully represented the original data obtained from agency and non-governmental data sources. Acquired data
was reviewed, but was used as received. Inspection occasionally identified significant outliers that suggested
uncorrected data entry errors. The project team corrected obvious errors or consulted with source data owners to
verify or correct.
A portion of the data was spatially registered in geographic information system databases, termed secondary data.
Secondary geospatial was evaluated for completeness with the validation tool of the EPA Metadata Editor (EME)
(https://edg.epa.gov/EME/) to determine if it met the minimum requirements of the Federal Geographic Data
Committee's (FGDC) Content Standard for Digital Geospatial Metadata (FGDC, 1998) and the EPA Geospatial
Metadata Technical Specification (USEPA, 2007). The project team determined whether data could be used and the
results of the validations were documented. The EME was subsequently used to update the metadata records
addressing any validation errors previously encountered.
Appendix E-2
-------
Water Acquisition for Hydraulic Fracturing May 2015
Analytics
The project QAPP described analytical approaches for assessing and summarizing water acquisition to water
available. Three types of analyses were applied in each study basin: 1) the facts of water acquisition were
summarized, 2) data were used to systematically quantify the water balance effects of observed hydraulic fracturing
withdrawals on water bodies at the local scale, and 3) scenario analyses were applied to reduce gaps in available
data related to water bodies, climate conditions, or levels of potential hydraulic fracturing activity that were not well
represented in observed hydraulic fracturing withdrawals.
Estimates of water volumes at withdrawal sites was needed to perform water use intensity calculations. Most water
was acquired from rivers and streams and USGS streamflow data were the primary source of information on
available water volume. Various techniques were used to extrapolate streamflow volume at ungaged withdrawal
locations from observed flow at gaged sites including empirical techniques and hydrological modeling using the
Hydrological Simulation Program—Fortran (HSPF). Streamflow estimation procedures were outlined in U.S. EPA
2013b. Methods, calibration results and analysis of the precision of streamflow estimates are described in Appendix
B of this report. Groundwater pumping was assessed at well locations in each study basin using the groundwater
model GFLOW™. Modeling methods and calibration results are described in detail in Appendix C.
All statistics and graphing were performed with R Statistical Software, version 3.0.1 (R Core Team 2013) or
Microsoft Excel (2007).
Data files are managed in project electronic archives as defined in the project QAPP (U.S. EPA 2013b).
Product Review
The project quality assurance project plan (U.S. EPA 2013b) was approved by the EPA project quality assurance
manager on August 30, 2013.
The project was included in a laboratory competency audit (LCA) for the ERD/NERL Division in Athens, Georgia
on August 19-20, 2014, and no corrective actions were identified.
This report was independently peer reviewed using a contractor-led Letter Review, following procedures specified
in U.S. EPA Drinking Water Resources Quality Management Plan (Environmental Protection Agency 2012b,
Revision No. 1. (Available at: http://www2.epa.gov/hfstudv/qualitv-management-plan-revision-no-l-plan-studv-
potential-impacts-hydraulic-fracturing).
References
FGDC (Federal Geographic Data Committee). 1998. Content Standard for Digital Geospatial Metadata, FGDC-
STD-001-1998. http://www.fgdc.gov/standards/projects/FGDC-standards-projects/metadata/base-
metadata/index html.
U.S. EPA (Environmental Protection Agency) 2007. Geospatial Metadata Technical Specification Version 1.0.
Office of Information Collection Office of Environmental Information, Environmental Protection Agency,
November 2007. http://www.epa.gov/geospatial/policies.html.
U.S. EPA (Environmental Protection Agency) 2012b. U.S. EPA Drinking Water Resources Quality Management
Plan Revision No. 1. (Available at: http://www2.epa.gov/hfstudv/qualitv-management-plan-revision-no-1 -plan-
study-potential-impacts-hydraulic-fracturing).
Appendix E-3
-------
Water Acquisition for Hydraulic Fracturing May 2015
[This page intentionally left blank.]
Appendix E-4
-------
vvEPA
United States
Environmental Protection
Agency
PRESORTED STANDARD
POSTAGES FEES PAID
EPA
PERMIT NO. G-35
Office of Research and Development (8101R)
Washington, DC 20460
Official Business
Penalty for Private Use
$300
------- |