Will Water Scarcity in Semiarid Regions Limit Hydraulic Fracturing of Shale Plays? Supporting Information

Transcription

1 Will Water Scarcity in Semiarid Regions Limit Hydraulic Fracturing of Shale Plays? Bridget R. Scanlon, Robert C. Reedy, and Jean-Philippe Nicot Bureau of Economic Geology, Jackson School of Geosciences, The University of Texas at Austin; Burnet Road, Austin, TX, USA 78758; Supporting Information Contents: 52 pages 27 figures 13 tables 1

2 Table of Contents 1. Acronyms Units Terminology Reporting of Water Sources for Hydraulic Fracturing Background Information on Eagle Ford Shale Play Data Sources and Analyses Comparison of Water Demand for Hydraulic Fracturing with Other Sectors Projected Water Demand for Hydraulic Fracturing and Other Sectors Water Supply Sources for Hydraulic Fracturing from Aquifer Storage Legal issues related to water use for hydraulic fracturing Impacts of HF and Other Water Uses on Water Resources Water for gas extraction versus thermoelectric generation Water produced by natural gas combustion Alternative Water Sources Characterization of flowback/produced water volumes Characterization of fresh and saline groundwater availability Municipal water purchased to support hydraulic fracturing Water Sources for Hydraulic Fracturing in the Permian Basin Summary of symposium on use of brackish and recycled produced water in the Permian Basin Water Sources for Hydraulic Fracturing in the Bakken Play References

3 List of Figures Figure S1. Mean annual precipitation in the Eagle Ford play area... 8 Figure S2. Generalized land use / land cover in the Eagle Ford play area... 9 Figure S3. Surface geology and hydrostratigraphy in the Eagle Ford play area Figure S4. Groundwater conservation district boundaries and major rivers in the Eagle Ford play area.. 11 Figure S5. Oil and gas, public, and irrigation water supply wells in the Eagle Ford play region Figure S6. HF well density in the Eagle Ford play Figure S7. HF water demand for HEWD at Eagle Ford play maturity Figure S8. Water demand for the next 20 years for irrigation and municipal uses Figure S9. Schematic diagram representing wells completed in unconfined and confined aquifers Figure S10. Groundwater recharge rates in the Eagle Ford play region based on GAMs Figure S11. Total groundwater storage in the Eagle Ford play region based on GAMs Figure S12. Monthly mean discharge for the Rio Grande River at Laredo, Texas Figure S13. Water level hydrographs for groundwater wells Figure S14. Modeled water level changes for the Carrizo-Wilcox aquifer Figure S15. Flow duration curve analysis for stream flows monitored by USGS stream gages Figure S16. Electrical power generation in Texas by fuel type for Figure S17. Flowback-produced (FP) water volumes Figure S18. Monthly and cumulative median ratio of FP to HF Figure S19. Distribution of groundwater TDS concentrations in the Carrizo aquifer Figure S20. Probability of groundwater TDS concentration 1,000 mg/l in the Queen City aquifer Figure S21. Probability of groundwater TDS concentration 1,000 mg/l in the Sparta aquifer Figure S22. Probability of groundwater TDS concentration 1,000 mg/l in the Yegua-Jackson aquifer Figure S23. Distribution of groundwater TDS concentrations in the Jasper aquifer Figure S24. Groundwater TDS concentrations and cross section of the the Carrizo-Wilcox aquifer Figure S25. Major and minor aquifers underlying the Permian Basin in western Texas Figure S26. Locations of HF oil wells and mean annual precipitation for the Bakken Shale play Figure S27. Total daily storage for Lake Sakakawea

4 List of Tables Table S1. Mean water withdrawals by sector for the 16 counties in the Eagle Ford play Table S2. Mean water consumption by sector for the 16 counties in the Eagle Ford play Table S3. Numbers of wells and total HF water Table S4. Projected 20-year HF, irrigation, and municipal water demand vs available water storage Table S5. Groundwater recharge rates in the Eagle Ford play area Table S6. Available fresh groundwater storage in the Eagle Ford play region Table S7. Mean aquifer storativity values in the Eagle Ford region Table S8. Annual total water diversions from the Rio Grande River for mining purposes Table S9. Summary of USGS stream flow monitoring records for gages in the Eagle Ford play Table S10. Estimated brackish water storage for aquifers in the Permian Basin region Table S11. Estimated brackish water storage for aquifers in the Permian Basin region Table S12. Hydrocarbon production and disposed water volumes in the Permian Basin region Table S populations, areas, and population densities for Bakken Shale play area counties

5 1. Acronyms AF acre-foot FP flowback-produced water GCD Groundwater Conservation District GMA Groundwater Management Area HEWD highest established well density HF hydraulic fracturing IHS name of database provider SI Supporting Information TDS Total dissolved solids 2. Units 1 gallon = liters 1 mgal = 1 million gallons = 3.1 acre feet = 3785 m 3 1 barrel = 42 gallons = 159 liters Btu = 1 million British thermal units = Joules = giga Joules, GJ 1 Mcf of gas = 1000 ft 3 of gas = Btu of energy 1 TWh of electricity = Wh = Btu energy or TBtu of energy 1 TBtu energy / power plant efficiency [e.g for natural gas combined cycle plants in Texas] = TBtu of natural gas required 3. Terminology Conventional versus Unconventional Oil and Gas Production Conventional production generally refers to production from high permeability reservoirs where oil and gas migrates from the source to the reservoir and are trapped in the reservoir. Because of the generally high permeability the oil or gas readily flows to the well bore. Unconventional production in this context refers to production from very low permeability reservoirs, such as shales, tight sands or tight limestones. Because of the low permeability, special stimulation techniques, such as hydraulic fracturing are required. In order to maximize oil and gas production, horizontal drilling is often combined with hydraulic fracturing. Flowback - Produced Water Because of difficulties of distinguishing flowback water from produced water, we used the collective term FP water. Flowback is generally defined as fluids that are geochemically similar to that of the HF fluid, whereas produced water is generally derived from the brine or saline fluid residing in the formation. Flowback is also sometimes operationally defined as production water during the first two or three weeks or until gas or oil is actually produced. In most cases, the transition period between the two end members is long and complex. Ratio of FP to HF The FP/HF ratio refers to the ratio of the cumulative combined amount of flowback and produced water (FP water) to the amount of water used in the HF process. The ratio generally increases initially but stabilizes after a few years at a near constant value as the amount of produced water decreases. Injection wells Although, from an operational standpoint, water is injected into the formation during the hydraulic fracturing step, from a Underground Injection Control (UIC) regulatory standpoint, these wells are 5

6 production wells being developed and stimulated, not injection wells. Throughout this article, we reserved the term injection well for UIC Class II injection wells, typically used for disposal of flowback / produced water in addition to spent drilling fluids. 6

7 4. Reporting of Water Sources for Hydraulic Fracturing Texas does not require reporting of water sources for unconventional production. However, W. Virginia and Pennsylvania, under new laws unacted in 2011 (WV) and 2012 (PA) require, as part of the permit process, a Water Management Plan that identifies water sources for unconventional development. The volume of water sourced is required to be monitored and reported. In West Virginia, water withdrawals, water used in development of horizontal Marcellus wells and disposal of flowback fluid are regulated by two acts and their associated rules: the Natural Gas Horizontal Well Control Act and the Water Resources Protection and Management Act. The Natural Gas Horizontal Well Control Act [W.Va. Code 22-6A] passed by the West Virginia Legislature in Applicable language is: For all surface water withdrawals, a water management plan is required. Withdrawing water from surface waters of the state by methods deemed appropriate by the secretary, so as to maintain sufficient in-stream flow immediately downstream of the withdrawal location. In no case shall an operator withdraw water from ground or surface waters at volumes beyond which the waters can sustain. As in West Virginia, data collection and reporting requirements for Marcellus operations have also evolved in recent years in Pennsylvania, particularly with the passage of the Oil and Gas ACT, also known as PA ACT 13, in guidelines/20306 Applicable language: PA ACT 13 (2012); proposed water withdrawal should be approved by and operated in accordance with conditions established by the Susquehanna River Basin Commission, the Delaware River Basin Commission or the Great Lakes Commission, as applicable. 7

8 5. Background Information on Eagle Ford Shale Play Figure S1. Mean annual precipitation in the Eagle Ford play area ( , PRISM, Mean precipitation ranges from 19.5 in/yr (495 mm/yr) in the west to 38.6 in/yr (980 mm/yr) in the east. Locations of wells for hydraulic fracturing and county names in the Eagle Ford play area are shown for reference. The areal extent of the Eagle Ford play area is indicated by the black line. 8

9 Figure S2. Generalized land use / land cover, consisting of 45% shrubland, 25% pasture, 8% grassland, 7% forest, and 5% cropland in the Eagle Ford play area (Fry et al., 2011). The areal extent of the Eagle Ford play is indicated by the black line. The Winter Garden region in the west (Dimmit, Frio, LaSalle, and Zavala counties) is a major irrigated region. Major rivers include, from west to east, Rio Grande, Nueces, Frio, Atascosa, San Antonio and Guadalupe rivers. County names are shown in italic font and town names are shown in regular font. The named towns shown within or adjacent to the Eagle Ford play have populations generally ranging from ~1,000 to ~10,000. Based on 2010 US Census data, the total population in the play boundary is estimated to be 202,600, which is divided between 84,300 (42%) residing in the named communities shown and 118,300 (58%) living either in smaller communities or in rural areas. This analysis does not include the San Antonio metropolitan area (pop. 2,200,000), Laredo (pop. 236,000), or Eagle Pass (pop. 26,000). 9

10 Figure S3. Surface geology and hydrostratigraphy in the Eagle Ford play area. Locations of cross sections A-A and B-B shown in Figures 2b and 2c, respectively, are shown for reference. The Queen City and Sparta sands, which are considered aquifers east of the Frio River, are respectively equivalent to the Upper Bigford/Lower El Pico Clay and Lower Laredo Formations west of the Frio River while the remaining formations are continuous across the region (Geologic Atlas of Texas, The Jasper and Evangeline aquifers, along with the younger Chicot aquifer (not shown) together constitute the Gulf Coast aquifer system. Older and younger units than those outcropping in the Eagle Ford play region are shown in gray. 10

11 Figure S4. Groundwater conservation district (GCD) boundaries and major rivers in the Eagle Ford play area. The GCDs include some multicounty areas (Evergreen and Wintergarden GCDs) and several single county areas (Bee, Gonzales, Lavaca, Live Oak, McMullen, and Pecan Valley GCDs). The areal extent of the Eagle Ford play area is shown by the black line. The predominant Groundwater Management Area (GMA) is GMA13, which covers much of the play area, along with smaller areas in GMAs 15 and 16. Surface water is owned and managed by the State whereas groundwater is owned by individual landowners. 11

12 6. Data Sources and Analyses Water use data for hydraulic fracturing were obtained from two sources, including (by license) the commercially available IHS database at and the publically available FracFocus database at Hydraulic fracturing water use data were processed and summarized as described in Scanlon, et al. (2014). Gridded hydraulic property data used to estimate groundwater storage volumes were obtained from the TWDB for the Groundwater Availability Models (GAMs) relevant to the study area. Gridded GAM data files are provided upon request at Data on groundwater quality samples used to estimate fresh and brackish groundwater resources were obtained from the Texas Water Development Board (TWDB) groundwater database available at Data on groundwater wells used to supply water for hydraulic fracturing were obtained from the Texas Department of Licensing and Regulation (TDLR) Submitted Driller s Report Database. The database is maintained by the TWDB and is available at Data on annual total groundwater and surface water use by sector and county area were obtained from the TWDB at Data on stream flows except for the Rio Grande River were downloaded from the U.S. Geological Survey (USGS) National Water Information System (NWIS) database at Information on Rio Grande River flow rates were obtained from the International Boundary & Water Commission, United States and Mexico, at Information on diversions from the Rio Grande River for hydraulic fracturing and other sectors was obtained from the Rio Grande Water Master through phone and contact. Information on electrical power generation and power plant fuel types were obtained from the U.S. Energy Information Agency (EIA) at Information on land use (NLCD 2006) was obtained from the Multi-Resolution Land Characteristics Consortium at Information on mean precipitation was obtained from the PRISM Climate Group at To assess HF water sources from municipalities, we identified 20 utilities out of the 128 water providers located within the 16 county area of the play as listed by the Texas Commission on Environmental Quality (TCEQ). Utilities were selected based on their physical location above the play and service area size. Contact was made by phone with follow up s. Of the 20 municipal utilities selected, 14 did not sell bulk water to O&G producers, two sold water for residential use only where man camps supporting O&G producers are found, and the remaining two did not record data. The City of Cuero sells water to oil field related entities but does not record volumes. The City of Jourdanton sells water and meters the fire hydrant sales. The cities of Laredo and Seguin have contracts for municipal waste water as discussed in the paper. Maverick County Utility Board voted down an offer to sell irrigation water and convert it to mining. 12

13 7. Comparison of Water Demand for Hydraulic Fracturing with Other Sectors Table S1. Mean total annual water withdrawals (both surface water and groundwater) by sector for the 16 counties in the Eagle Ford play for the period and for 2010 (wet year), 2011 (dry year), and 2012 (normal year) (estimated from Texas Water Development Board). Values represent countywide totals that are generally greater than those for the play footprint Sector % of % of % of % of bgal bl bgal bl bgal bl bgal bl total total total total Irrigation Municipal Steam Electric Mining Livestock Manufacturing Total TWDB mean mining water use for and annual HF water use (bold) from this study for 2010, 2011, and Table S2. Estimated mean annual consumptive water use (both surface water and groundwater) by sector for the 16 counties in the Eagle Ford play for the period and for 2010 (wet year), 2011 (dry year), and 2012 (normal year). Consumptive use is estimated as 85% of irrigation, 40% of municipal, and 100% of remaining categories shown in Table S1. Values represent county-wide totals that are generally greater than those for the play footprint Sector % of % of % of % of bgal bl bgal bl bgal bl bgal bl total total total total Irrigation Municipal Steam Electric Mining Livestock Manufacturing Total TWDB mean mining water use for and annual HF water use (bold) from this study for 2010, 2011, and

14 Figure S5. Locations and depths of oil and gas (O&G) water supply wells (round symbols) in the Eagle Ford play region in relation to depths of a) public water supply wells (square symbols) and b) Class II injection wells (triangle symbols). The legend color vs depth interval scale is consistent between all well types shown. Shown are O&G water supply wells completed since 2009 based on the Texas Department of Licensing and Regulation (TDLR) well report database maintained by the TWDB ( Public water supply well data were obtained from the Texas Commission on Environmental Quality (TCEQ) by special request. Class II injection well data were obtained by license from the commercial IHS database at 14

15 8. Projected Water Demand for Hydraulic Fracturing and Other Sectors Future water demand for hydraulic fracturing was estimated by assuming that drilling would increase to achieve the (current) highest established well density (HEWD) based on the approach described in Gong et al. (2013). Two different methods were used: (1) HEWD at the intersections of the seven production zones (oil, volatile oil-condensate, condensate, and dry gas, including W and E zones for O, VOC, and C zones) and counties for a total of 40 regions and (2) the overall HEWD in each of the seven production zones was used to develop county estimates of future drilling. Results are summarized by county area in Table S3 with method (1) resulting in the minimum and method (2) in the maximum estimates shown. The results indicate that an estimated range of ~42,000-62,000 wells remain to be completed by play maturity (total wells at HEWD minus 8300 wells drilled from 2009 to 2013). An initial base map depicting horizontal well section ground surface traces was converted to an HEWD map with a 0.25 mile (400 m) grid with values representing the line density function of all well traces within a 1 mile (1600 m) radius. Figure S6. Well density in the Eagle Ford play based on all wells laterals completed through Dec Insets show example areas with red lines representing the surface traces of well lateral sections. 15

16 Table S3a. Numbers of wells completed and total HF water use during the study period and corresponding values for estimated total numbers of wells assuming ultimate well densities in production zones are equal to the current highest established well density (HEWD) in each production zone (oil, volatile oil-condensate, condensate, and dry gas). Total HF Remaining Remaining Area Existing HF to date Total No. of Wells at County/Region mi 2 at HEWD HF (min) HF (max) Wells bgal HEWD bgal bgal bgal Atascosa ,287-5, Bee De Witt ,237-2, Dimmit 1,324 1, ,648-12, Fayette Frio ,264-4, Gonzales ,035-3, Karnes 617 1, ,040-4, La Salle 1,444 1, ,510-10, Lavaca Live Oak ,109-1, Maverick , Mc Mullen ,599-5, Webb 1, ,919-8, Wilson , Zavala ,956-4, West 6,895 4, ,311-53, East 2,546 3, ,396-17, Total 9,441 8, ,707-70,

17 Table S3b. Numbers of wells completed and total HF water use during the study period and corresponding values for estimated total numbers of wells assuming ultimate well densities in production zone are equal to the current highest established well density (HEWD) in each production zone (oil, volatile oil-condensate, condensate, and dry gas). Metric equivalent to Table S3a. Total Wells at Total HF Remaining Remaining Area Existing HF to date County/Region kmi 2 HEWD at HEWD HF (min) HF (max) Wells bl bl bl bl Atascosa 1, ,287-5, Bee De Witt ,237-2, Dimmit 3,429 1, ,648-12, Fayette Frio 1, ,264-4, Gonzales 1, ,035-3, Karnes 1,598 1, ,040-4, La Salle 3,740 1, ,510-10, Lavaca Live Oak ,109-1, Maverick , Mc Mullen 2, ,599-5, Webb 3, ,919-8, Wilson , Zavala 1, ,956-4, West 17,858 4, ,311-53, , ,027.6 East 6,595 3, ,396-17, Total 24,453 8, ,707-70, , ,

18 Figure S7. Spatial distribution of the ultimate HF water demand expressed as an equivalent depth, which assumes a total of ~70,000 HF wells at play maturity after ~20 years based on the current HEWD for the seven production zones in this study. Figure S8. Spatial distribution of water demand for 20 years of combined irrigation and municipal use expressed as an equivalent depth, assuming an average annual demand equal to that of the period

19 Table S4a. Estimated total water demand (withdrawal) by sector for the next 20 years, total availabilities of fresh and brackish groundwater, and ratios of water demand to availability. Values represent estimates for the play plus a 5-mile wide buffer zone area and in most cases do not include all of the county areas (Figure S7 and Figure S8). HF demand is estimated from HEWD for the seven production zones in this study (Table S3) and demand by other sectors is based on annual mean withdrawals for the period (Table S1, TWDB). Highlighted values indicate that the estimated total 20-year demand for fresh water in Frio and Zavala counties exceeds 100% due to irrigation demands and that HF demand exceeds modeled available groundwater storage in Maverick County. County Water demand for next ~20 years Available water storage Ratios (demand / storage) HF Irrigation Municipal Other Total Fresh Brackish Total HF/ Fresh HF/ Brackish Total/ Fresh Total/ Brackish bgal bgal bgal bgal bgal bgal bgal bgal % % % % % Atascosa ,139 10,044 11, Bee De Witt Dimmit ,394 1, Fayette , Frio ,164 1, Gonzales ,184 4,876 8, Karnes ,164 5, La Salle ,178 18, Lavaca Live Oak ,588 4, Maverick McMullen ,198 19, Webb ,916 11, Wilson ,863 3, Zavala Total ,413 9,784 80,308 90, Total/ Total 19

20 Table S4b. Estimated total water demand (withdrawal) by sector for the next 20 years, total availabilities of fresh and brackish groundwater, and ratios of water demand to availability. Values represent estimates for the play plus a 5-mile wide buffer zone area and in most cases do not include all of the county areas (Figure S7 and Figure S8). HF demand is estimated from HEWD for the seven production zones in this study (Table S3) and demand by other sectors is based on annual mean withdrawals for the period (Table S1, TWDB). Highlighted values indicate that the estimated total 20-year demand for fresh water in Frio and Zavala counties exceeds 100% due to irrigation demands and that HF demand exceeds modeled available groundwater storage in Maverick County. Metric equivalent to Table S4a. County Water demand for next ~20 years Available water storage Ratios (demand / storage) HF Irrigation Municipal Other Total Fresh Brackish Total HF/ Fresh HF/ Brackish Total/ Fresh Total/ Brackish bl bl bl bl bl bl bl bl % % % % % Atascosa ,309 38,018 42, Bee ,075 1, De Witt , , Dimmit ,660 5,277 6, Fayette ,296 3,217 5, Frio 106 1, ,542 1,392 4,404 5, Gonzales ,050 18,455 30, Karnes ,129 19,546 21, La Salle ,909 68,804 70, Lavaca ,601 1,495 3, Live Oak ,324 17, Maverick McMullen ,256 72,696 74, Webb ,317 41, Wilson ,795 10,838 13, Zavala , ,321 1, Total 1,260 3, ,348 37, , , Total/ Total 20

21 9. Water Supply Sources for Hydraulic Fracturing from Aquifer Storage For hydraulic fracturing we are mostly interested in recoverable or accessible water held in storage. Water storage varies markedly between two basic aquifer types, unconfined and confined (Alley et al., 1999). A geologically unconfined aquifer is also categorized as a water table aquifer, where the elevation of the water table is at atmospheric pressure. The water yielding capacity of this type of aquifer can be characterized by the calculated value of specific yield (Sy). In unconfined aquifers in aquifer outcrop zones, recoverable storage or drainable storage is estimated from the specific yield times saturated aquifer thickness. In an unconfined aquifer the specific yield is also known as drainable porosity and the value approaches the effective porosity of the aquifer. Effective porosity is that portion of the porous media that consists of interconnected pore space that allows for flow, not to be confused with total porosity. Typical values for effective porosity are <5% for a clay and 35% for a coarse sand. The effective porosity excludes non-connected pore space and is used to describe water storage capacity in unconfined aquifers (MacDonald et al., 2013). Potentiometric surface Water table CS = S h h D DS = Sy D Unconfined Confined Figure S9. Schematic diagram representing wells completed in unconfined and confined aquifers. In unconfined aquifers, water is released by gravity drainage and drainable storage (DS) = specific yield (Sy) water level decline (D). In confined aquifers, the aquifer is bounded by low permeability units. Water in the well rises above the top of the aquifer and this level is termed the potentiometric surface. Compressible storage (CS) is calculated as storativity (S) potentiometric surface height (h). Water level declines are up to three orders of magnitude greater in confined versus unconfined aquifers because of the much lower storativity of confined aquifers. For example, a 1 ft decline in an unconfined aquifer with a specific yield of 0.15 would result in an equivalent depth of water of 0.15 ft whereas a 1 ft decline in the potentiometric surface of a confined aquifer with a storativity of would result in an equivalent depth of ft of water. In confined aquifers, the aquifer is overlain by a low permeability confining layer and water in the aquifer is under pressure. The elevation to which water rises in the well is the local potentiometric elevation, which may extend above land surface resulting in a flowing (artesian) well. The recoverable 21

22 water is the compressible storage, which is calculated from storativity times the vertical distance between the potentiometric surface and the top of the aquifer. Storativity is defined as the volume of water released from storage per unit aquifer area per unit decline in head, and is dimensionless. Pumping in artesian aquifers reduces the pressure, allowing both the stored water and the aquifer matrix to expand. Typical values of storativity, estimated from pumping tests, are up to three orders of magnitude less than specific yield of unconfined aquifers, and are typically The cone of depression for a single pumping well in an aquifer of infinite extent can be estimated using an analytical solution by Theis (1940) and is much larger both in depth and areal extent than cones of depression in unconfined aquifers. The ratio of the volumetric extent of cones of depression in confined aquifers to those in unconfined aquifers is equivalent to the ratio of specific yield to storativity, about a factor of 1000 in many cases (Alley et al., 1999). If the potentiometric elevation drops below the top of the confining layer, the aquifer is no longer confined and the drainable storage is calculated from the specific yield times the remaining aquifer saturated thickness. Conversion from confined to water table conditions generally only occurs near aquifer outcrop zones where air is free to enter the aquifer matrix to replace drained water. While this section focuses on water derived from aquifer storage, pumped water can also be derived from recharge from the outcrop or confined sections and cross formational flow from adjacent aquifers. In this study we estimated groundwater storage based on 1-mile square gridded data from Groundwater Availability Models (GAMs) sponsored by the Texas Water Development Board. Drainable storage in unconfined aquifer outcrop (water table) areas was estimated as the difference between water table and base of aquifer elevations multiplied by the specific yield (=0.15 for most aquifers in the play area). Compressible storage in confined aquifer areas was estimated as the difference between the potentiometric surface elevation and top of (confined) aquifer elevation multiplied by the storage coefficient (storativity). We did not estimate drainable storage for confined aquifers areas because of uncertainties in well yields and technical and economic recoverability of water. Estimates of water storage derived from the GAMs were combined with secondary gridded data sets representing groundwater quality based on TWDB groundwater sample data to produce estimates of both fresh and brackish water storage in the outcrop and confined areas of each aquifer. Fresh water is herein defined has having a total dissolved solids (TDS) concentration 1000 mg/l and brackish water as having a TDS concentration >1000 mg/l. One of two kriging methods was used to estimate water quality for each of the aquifer systems, depending on water quality data availability. Ordinary kriging was used for the Carrizo and Gulf Coast aquifers to predict actual TDS concentration distributions due to abundant data. The Carrizo aquifer results herein also include data for the Wilcox aquifer in their combined outcrop areas. The Gulf Coast aquifer system is represented almost entirely by the Jasper aquifer in the Eagle Ford play area and this analysis does not include any volumes associated with other units of the Gulf Coast aquifer system. All ordinary kriging results are based on log-transformed TDS concentrations. Due to sparse data over much of the region, indicator kriging was used for the remaining aquifers (Queen City, Sparta, and Yegua-Jackson) to predict the spatial probability of TDS exceeding 1,000 mg/l. Where indicator kriging was used, areas were defined as brackish water where this probability exceeds 75%. The Wilcox aquifer has of Upper, Middle, and Lower units. Water quality data are extremely limited for the confined zones of the Wilcox aquifer and water in these zones is assumed for this study to be entirely brackish based on geophysical log interpretation (Figure S24). The Carrizo grades into the Upper Wilcox and becomes brackish downdip (Figure S24). 22

23 Table S5. Groundwater recharge rates in the Eagle Ford play area based on groundwater availability models (GAMs), the PCR Global Water Balance (PCR GlobWB) model (Wada et al., 2010)], and the chloride mass balance approach applied to groundwater chloride data (Reedy et al., 2009). GAMs PCR GlobWB Reedy et al. Area or Aquifer bgal/yr bl/yr Regional 1 Play 2 Regional 1 Play 2 bgal/yr bl/yr bgal/yr bl/yr Play Footprint All Outcrops Carrizo-Wilcox Queen City Sparta Yegua-Jackson Jasper (Gulf Coast) recharge in outcrop areas within the play and up-dip of the play 2 recharge in outcrop areas within the play only Figure S10. Distribution of ground water recharge rates (inches/yr) in the Eagle Ford play region based on Groundwater Availability Models (GAMs). 23

24 Figure S11. Distribution of total groundwater storage in the Eagle Ford play region expressed as an equivalent depth based on Groundwater Availability Models (GAMs). Storage values include the sum total of both fresh and brackish water in both unconfined (outcrop) and confined regions of the Carrizo- Wilcox, Queen City, Sparta, Yegua-Jackson, and Gulf Coast (Jasper) aquifers in the area shown, which includes the Eagle Ford play core area and a surrounding 5-mile wide buffer region. 24

25 Table S6a. Available fresh (TDS 1,000 mg/l) groundwater storage in the Eagle Ford play and surrounding 5-mile wide buffer. Storage volumes are based on GAM hydraulic parameters with superimposed kriged maps based on water quality information from the TWDB database. GAM hydraulic parameters include [specific yield saturated thickness] in unconfined (UC) areas and [storativity potentiometric surface height] in confined (C) areas. Two kriging methods were used depending on water quality data availability. Ordinary kriging was used for the Carrizo and Gulf Coast aquifers to predict actual TDS concentration distributions due to abundant data. Indicator kriging was used for the remaining aquifers to predict the probability of TDS exceeding 1,000 mg/l due to sparse data over much of the region. The Carrizo data here include Wilcox data for the Wilcox outcrop area. There are very limited Wilcox water quality data elsewhere and Wilcox water in the confined zone is assumed to be brackish. The Carrizo grades into the Upper Wilcox downdip. Volumes associated with indicator kriging represent regions where there the probability is 75% that TDS exceeds 1,000 mg/l. Aquifer > Carrizo Queen City Sparta Yegua-Jackson Jasper Kriging method > Ordinary Indicator Indicator Indicator Ordinary Aquifer section> UC C C UC C UC C UC C UC C Total County/Region bgal bgal bgal bgal bgal bgal bgal bgal bgal bgal bgal bgal Atascosa ,124 1,139 Bee De Witt Dimmit Fayette Frio Gonzales , , ,184 Karnes La Salle Lavaca Live Oak Maverick McMullen Webb Wilson Zavala West 404 1, ,353 2,997 East , , ,376 1,411 6,787 Total 404 2, , , ,020 3,764 9,784 All 25

26 Table S6b. Available fresh (TDS 1,000 mg/l) groundwater storage in the Eagle Ford play region. Metric equivalent to Table S6a values. See Table S6a caption for more details. Aquifer > Carrizo Queen City Sparta Yegua-Jackson Jasper Kriging method > Ordinary Indicator Indicator Indicator Ordinary Aquifer section> UC C C UC C UC C UC C UC C Total County/Region bl bl bl bl bl bl bl bl bl bl bl bl Atascosa - 2,951 1, ,254 4,309 Bee De Witt , , ,045 Dimmit 1, , ,660 Fayette , , ,296 Frio - 1, ,312 1,392 Gonzales - 1, , ,418 1,632 12,050 Karnes , ,129 La Salle - 1, ,799 1,909 Lavaca , , ,601 Live Oak Maverick McMullen - 1, ,893 2,256 Webb Wilson , ,571 1,225 2,795 Zavala West 1,527 6,728 2, ,437 8,905 11,343 East - 3, , , ,347 5,342 25,689 Total 1,527 10,431 2, , , ,785 14,247 37,032 All 26

27 Table S7. Mean Groundwater Availability Model (GAM) storativity values for aquifers in the Eagle Ford region. Lower Middle Upper Queen County/Region Carrizo Sparta Yegua Jackson Jasper Wilcox Wilcox Wilcox City Atascosa Bee De Witt Dimmit Fayette Frio Gonzales Karnes La Salle Lavaca Live Oak Maverick McMullen Webb Wilson Zavala West East All

28 Table S8. Annual total water diversions from the Rio Grande River for mining purposes, including oil and gas extraction, for counties in the Eagle Ford play (Texas Commission on Environmental Quality, Rio Grande Watermaster). County bgal bl bgal bl bgal bl bgal bl Webb Maverick Figure S12. Monthly mean discharge for the Rio Grande River at Laredo, Texas (International Boundary & Water Commission, United States and Mexico). 28

29 Legal issues related to water use for hydraulic fracturing Texas Water Code (TWC) Chapter includes the descriptions of all groundwater production activities that are exempt from GCD permitting requirements. For example, it is clear that a well that is solely for domestic use, or for providing water for livestock or poultry, if the well is located on a tract of land larger than 10 acres, and is incapable of producing more than 25,000 gallons per day, then the well is exempt from GCD rules related to permitting (TWC (b) (1) (A B). Water wells drilled in support of surface mining activities are also exempt from GCD permitting rules (TWC (b) (3)). However, the exemption for water wells used for HF activities is not as clear. Generally, entities involved in HF, including the Railroad Commission of Texas, are of the opinion that the exemption in TWC (b) (2) which states, the drilling of a water well used solely to supply water for a rig that is actively engaged in drilling or exploration operations for an oil or gas well permitted by the Railroad Commission of Texas provided that the person holding the permit is responsible for drilling and operating the water well and the well is located on the same lease or field associated with the drilling rig means that groundwater produced for HF activities is exempt for GCD permitting. Some GCDs, however, come to a different conclusion with regards to this exemption, because the terms drilling and exploration are not defined in statute, and based on a conclusion that once drilling operations for the oil or gas well have been completed, then exploration activities have been completed and that subsequent water use for HF is a component of hydrocarbon production activities, and thus not exempt from GCD permitting requirements. Later in TWC (l), the final qualifier regarding GCD exemptions states that This chapter applies to water wells, including water wells used to supply water for activities related to the exploration or production of hydrocarbons or minerals. This chapter does not apply to production or injection wells drilled for oil, gas, sulphur, uranium, or brine, or for core tests, or for injection of gas, saltwater, or other fluids, under permits issued by the Railroad Commission of Texas. This provision is very confusing and a source of great contention between legal professionals in the field of natural resources. As a result of this confusing/conflicting language, the one absolute conclusion one may draw is that there is no clear consensus as to whether or not groundwater produced for HF activities is exempt from GCD regulation. As a result, certain GCDs have adopted a regulatory framework based on a conclusion that groundwater for HF activities is not exempt and subject to all permitting and reporting rules of the GCD, whereas others are silent on the issue. Several GCDs, including all those in the Eagle Ford play have adopted rules requiring reporting of all groundwater produced for oil and gas activities, as allowed in this section of TWC Chapter 36. Pecan Valley Groundwater Conservation District Rules Related to Brackish Water (Excerpted from the Pecan Valley GCD website The amount of the annual maximum groundwater production specified in the operating permit for a non-exempt well or well field, issued under these rules, shall not exceed one-half (1/2) acre-foot per contiguous surface acre owned (0.5 ft = 0.15 m) or controlled by the applicant. Unless the following conditions are met: 29

30 Wells drilled deeper than 700 ft (213 m) with no well screens above 500 ft (152 m) shall have one acre foot per contiguous acre (1 ft = 0.3 m) owned or controlled by the applicant. Wells drilled deeper than 700 ft (213 m) and no well screens above 500 ft (152 m) and have a TDS (total dissolved solids) reading of 1,000-4,999 ppm per liter shall have 2 AF per contiguous acre (2 ft = 0.6 m) owned or controlled by the applicant. Wells drilled deeper than 700 and no well screens above 500 and have a TDS reading of 5,000-10,000 ppm per liter shall have 5 AF per contiguous acre owned or controlled by the applicant. Wells drilled deeper than 700 ft (213 m) and no well screens above 500 ft (152 m) and have a TDS reading of 10,000 ppm per liter or higher shall have 10 AF per contiguous acre (10 ft = 3 m) owned or controlled by the applicant. 30

31 10. Impacts of HF and Other Water Uses on Water Resources Figure S13. Water level hydrographs for groundwater wells shown in Figure 7. Information for each well includes the well ID number, primary well use, producing aquifer, and well depth. Data were obtained from the Texas Water Development Board. 31

32 Figure S14. Modeled water level changes from pre-development to 1999 for the Carrizo-Wilcox aquifer resulting primarily from irrigation pumping in the Winter Garden area (Dimmit, Frio, La Salle, and Zavala counties) based on Huang et al. (2012). Points represent locations of Eagle Ford HF wells completed and show that much of the hydraulic fracturing is downgradient from the most intense aquifer depletion from irrigation. Cross hatched area represents the Carrizo-Wilcox outcrop area. Figure S15. Flow duration curve analysis results (i.e., percentage of the flow record period or record having non-zero flow rates) for stream flows monitored at 20 USGS stream gages in the Eagle Ford play region. Numbers shown indicate the USGS stream gage ID number. Values are summarized in Table S9. 32

33 Table S9. Summary of USGS stream flow monitoring records for gages in the Eagle Ford play, including mean non-zero flow rates, flow duration (percent of monitored period with non-zero flow), and mean Base Flow Index (BFI, estimated percentage of total flow contributed by groundwater). BFI was not estimated for streams with flow duration < ~85% as these are presumed to be losing streams. Water Course Nueces River Frio River Atascosa River San Antonio River Gage ID Drainage area Monitoring record Mean flow Flow duration mi 2 km 2 From To Span Years ft 3 /sec m 3 /sec % of time Mean BFI % ,830 4, ,100 5, ,080 10, ,170 13, ,090 20, ,560 22, ,400 39, , ,430 8, ,490 11, ,490 14, , ,170 3, ,740 4, ,110 5, Cibolo Creek , Guadalupe River ,490 9, ,930 12, , Sandies Creek , Peach Creek ,

34 Water for gas extraction versus thermoelectric generation The amount of natural gas required to generate a certain amount of electricity can be determined using the following calculations. In 2012, natural gas combined cycle (NGCC) power plants in Texas generated Watt hours (terawatt hour or TWh) of electricity which corresponds to Btu of energy (3.412 Btu/Wh for natural gas). Analysis of NGCC plants in Texas indicated that they are 44% fuel efficient, i.e. 44% of the energy results in electricity and the other 66% goes to waste heat based on analysis of NGCC plants in Texas (Scanlon et al., 2013) = Btu of natural gas energy = ft 3 of natural gas (1 Btu = 1 ft 3 of natural gas) This study shows that 1.5 gal is required to extract 10 6 Btu of natural gas considering the estimated ultimate recovery of dry gas wells in the Eagle Ford (Scanlon et al., 2014); Btu 1.5 gal/btu = or gal H 2 0 Previous studies in Texas showed that consumptive cooling water requirements for NGCC plants was 0.19 gal/kwh (Scanlon et al., 2013). Therefore, 170 TWh of electricity consumes Wh 0.19 gal/kwh for cooling = gal Therefore, water consumed during natural gas extraction (2 bgal) = ~6% of water consumed during power generation from NGCC plants (32 bgal). Figure S16. Electrical power generation in Texas by fuel type for 2012 (U.S. EIA). Other category includes landfill and other gasses (0.77%), biomass (0.39%), petroleum (0.35%), hydroelectric (0.14%), and solar (0.03%). 34