Impacts of hydraulic fracturing on surface and groundwater water resources : A case study from Louisiana, USA.

Background: Unconventional oil and gas reservoirs, frequently referred to as shale plays , have been gaining more attention in recent years. Hydraulic fracturing is performed to extract fossil fuels from unconventional reservoirs. Besides possible environmental implications, a better understanding of the potential stress that fracking may cause to water resources at the local or regional scale is still needed. The goal of this study is to assess the impact of current and projected future water demands for fracking on water resources in two main shale plays in Louisiana, USA. Methods: The analysis is performed in Louisiana’s two main shale plays, the Haynesville Shale and the Tuscaloosa Marine Shale, using the Water Supply Stress Index (WaSSI) framework. WaSSI is used to evaluate the stress at a fine watershed scale (HUC12) for annual-average conditions. The study analyzes different scenarios of historical and two future projected fracking conditions that simulate different extraction rates. In each fracking scenario, stresses on both surface and groundwater are evaluated separately. The study is based on a multitude of water supply and withdrawals datasets assembled and disaggregated to the watershed scale. Results: Under existing conditions, the impact of fracking water demands on surface water resources is within the low stress category in most watersheds in both shale plays. This impact remains low under more aggressive future fracking scenarios. In contrast, groundwater resources appear to be highly vulnerable under both the historical and projected fracking scenarios, especially in the Haynesville Shale where 20 out of the 94 watersheds become medium or highly stressed. If groundwater resources remained as a main source for fracking water, the number of stressed watersheds increased to 39 and 86 under the two projected future fracking scenarios. The least exploited Tuscaloosa Marine Shale remains mostly under low stress, except in the most aggressive future fracking scenario. Conclusions: Surface water resources in Louisiana’s shale plays seem abundant enough for fracking activities to rely on this source instead of groundwater whenever possible. Groundwater resources in Louisiana are clearly vulnerable to fracking activities, especially for the Haynesville shale play, under current and future projected fracking conditions. Water Supply Stress Index; $"# : Water Supply Stress Index for groundwater at a certain watershed (i); WaSSI !"# : Water Supply Stress Index for surface water at a certain watershed (i); WS: Total water supply; WS $" : Groundwater supply; WS !" : Surface water supply; WW: Total water withdrawal.


Background
Unconventional oil and gas reservoirs, frequently referred to as shale plays, have been gaining more attention in recent years [1], [2]. In order to extract fluids from the relatively impermeable source rock strata, hydraulic fracturing (also known as fracking) of long horizontal well sections is conducted. This fracturing involves the injection of large volumes of water amended with fracking chemicals and proppant (typically sand) into the wells at pressures large enough to fracture the rock and allow gas or oil to flow to the surface [3]. This process has created significant economic value [4], but there are also potential environmental implications. These include the possibility of groundwater contamination and the over-use of local water resources. While many of the water quality implications have been explored (e.g., [5], [6], [7], [8]), a better understanding of the potential stress that fracking may cause to water resources at the local or regional scale is still needed [9]. For example, most previous studies that analyze the costs associated with unconventional fuel production do not address the costs (real or hidden) of accounting for the water availability required to perform fracking operations (e.g., [10], [11], [12]). Although some studies have related water-stressed regions with fracking activities [6], [13], [14], [15], only a few studies [16], [17], [18], attempted to quantify the potential impact of water used for fracking on the local and regional water resources. A recent national-scale assessment on water use for hydraulic fracturing in the US [20] calls for further studies that examine regional-scale energy and water sustainability challenges.
The importance of quantifying the impact of hydraulic fracturing on water quantity was recently illustrated in the state of Louisiana where concerns over groundwater depletion lead the Louisiana Office of Conservation to issue an order of groundwater emergency in August of 2011 [21], [22] for Southern Caddo County in the Haynesville shale play. In response, the Louisiana Department of Natural Resources established hydraulic regulatory guidelines for Louisiana that encourage the use of surface water for fracking and oil/gas well drilling purposes with the intention of reducing groundwater depletion. In Louisiana, the development of unconventional gas/oil extraction has grown substantially and has the potential to further expand. Most of the unconventional wells in Louisiana are located in the Haynesville shale play ( Figure 1). Around 2,832 active wells are spread over seven counties in Louisiana [23], while many other well locations extend into Texas ( Figure 1; [24]). Additionally, the Tuscaloosa Marine shale (TMS) has 33 active wells spread throughout eight counties in east-central Louisiana [25], and an additional ~50 wells are located in southern Mississippi ( Figure 1). Recent work on the TMS has shown that there remains considerable oil and gas exploration potential [26], and new wells have recently been completed [27]. This study focuses on understanding the impact of current and projected future water demands for fracking on Louisiana's water resources. We use a water stress approach at a watershed spatial scale to quantify the added stress due to fracking activities. We chose the state of Louisiana for this study because of growing concerns on water availability for fracking and the absence of any quantitative information on the potential impact of fracking on the freshwater water system. A literature analysis revealed that the actual fracking water use in Louisiana is not documented as part of the regular state or national water use reports [28]. Including the fracking water volumes in water budget analysis is crucial to avoid inaccurate assessment of the current and future water stress and to provide reliable information for water allocation and management policies.

Methods
In this study, a water-budget analysis is used to assess the impact on water resources due to hydraulic fracturing performed in Louisiana's two main shale plays (the Haynesville Shale and the Tuscaloosa Marine Shale). The analysis is conducted using the Water Supply Stress Index (WaSSI) [29] to evaluate the stress status of a watershed over a selected timeframe. Our analysis is conducted over a fine watershed scale defined by Hydrological Unit Code 12 (HUC12) and for annual-average conditions. The HUC-12 resolution is appropriate for this analysis as it provides fine details for stressed areas that would be masked if greater scales are adopted [30], [31]. The study analyzes different scenarios of current (referred to as maximum historical fracking scenario), and two future projected hydraulic fracturing conditions (possible maximum extraction, and maximum extraction). In each fracking scenario, stresses on both surface water and groundwater are evaluated separately.

Water stress metric:
The Water Supply Stress Index (WaSSI) [29] is expressed as the ratio of water demand (withdrawals) to water supply. A modified version of WaSSI is used to examine the stress for surface water and groundwater separately. Water demands are separated into surface and groundwater based on the information available from United States Geological Survey (USGS) [28] so that the surface and groundwater stress metrics can be assessed separately. Whenever the fracking demand is incorporated into the stress metric, it is assumed that the fracking water demand is coming from one source only, either from surface or groundwater. This assumption is driven by the absence of any specific information on the source of water withdrawals for fracking operations.
The WaSSI for surface (WaSSI !"# ) and groundwater (WaSSI $"# ) were estimated for a certain watershed (i) using the following two equations: where WS is the total water supply, WW is the total water withdrawal, and ENV is an environmental flow requirement. The water stresses are calculated for the watersheds in Louisiana located within the two shale plays, the Haynesville and the Tuscaloosa. The surface water supply (WS !" ) for each watershed is calculated as the summation of two components: the equivalent volume of the annual averaged gauged (streamflow) at the watershed outlet, and the volume of water withdrawn inside each watershed. This second component is added in order to represent the total amount of flow in its natural condition (i.e., before water is removed). The groundwater supply (WS $" ) corresponds to the mean-annual natural groundwater recharge. The environmental factor (ENV) is used to ensure provision for environmental requirements in the streams [32], [33].
In this current study, a conservative 50% factor (i.e., ENV=0.5) was chosen to ensure fulfilling environmental demands.
Once the surface and groundwater WaSSI are estimated in each watershed, it can be classified into different categories similar to the concept of critical ratio [34], [35]. Following Eldardiry et al.
(2016) [19], the following thresholds were used: low stress: WaSSI < 0.50, medium stress: WaSSI between 0.50 -1.00, and high stress: WaSSI > 1.00. The WaSSI assessment will be presented only over the spatial domains of the two shale plays; a Louisiana statewide non-fracking WaSSI assessment is available in an earlier study [19].

Water Supply and Withdrawal Datasets:
The datasets used in this study are summarized in Table 1. Surface and groundwater supplies were acquired from national-scale models and datasets. Both are downscaled and projected into a HUC-12 watershed spatial scale. Surface water supply estimates are obtained from the National Hydrography Dataset (NHDPlus) [36], which reports annual average streamflow. Using streamflow routed through the flow lines network, the mean annual flow estimated for each flow line represents the average over the period from 1971-2000 [37]. Estimates of the annual average recharge rate for groundwater were taken from the USGS in a gridded format with a resolution of 1-kilometer. The USGS dataset represents the mean-annual natural groundwater recharge, which is estimated by multiplying grids of mean annual runoff values from a 1951-1980 [38] by the base-flow index (BFI) values [39]. In the case of groundwater availability, the recharge rates were averaged across the 1-km 2 grids within each watershed.
Most historical water withdrawals in Louisiana come from surface water sources and are distributed across four main use categories: agriculture, industrial, power generation, and public supply. A lesser but still important share of the withdrawals in Louisiana originate from groundwater sources, and include two additional categories, livestock and rural domestic. Water withdrawals for fracking purposes are typically quantified under the category of Mining; however, in the state of Louisiana, such information is not currently available [28]. All water withdrawal datasets used in this study, except fracking, were obtained from a public report updated every five years since 1960 by the USGS with the support of the Louisiana Department of Transportation (DOTD) [40]. According to the 2010 USGS report [28], the most recent one available at the time of this analysis, the total surface water withdrawals in Louisiana were 7,000 Million gallons per day (MGD) or 26.50 Million of cubic meters per day (MCMD), and total groundwater withdrawals were about 1,600 MGD (6.06 MCMD). All withdrawals are documented as annual values and presented for each county in the state. In the current study, the 2010 year annual-average water withdrawals dataset is used as a base year for the current and future fracking scenarios. Countyscale water withdrawals were disaggregated into a HUC-12 watershed scale using a spatial disaggregation framework developed earlier [19]. The framework is based on a geometric weighting scheme to incorporate spatially-distributed factors (e.g., crop area, surface water locations, etc.) of the water use from within each county level to a watershed scale.
Unlike other water use sectors, fracking water volumes are not reported in the DOTD and USGS annual reports. Instead, fracking water withdrawal data are available through the fracfocus website [23] for every month from year 2011 to 2016 ( Figure 1). For example, the numbers of wells in Louisiana with water demand reported in 2011 reported by Fracfocus.org were 567 wells. The fracking water is reported for each drilled well, and each well is coded by an API number which is a unique numeric identifier assigned by the American Petroleum Institute [41]. A shapefile with the specific locations of the wells, available from the Louisiana Strategic Online Natural Resources Information System (SONRIS) [42], was overlayed with the watershed map to collocate the wells with their respective HUC-12 watersheds. To calculate the annual water use for fracking, the monthly fracking water volumes were added up for all wells inside each watershed. While a well can potentially be fracked 1 to 3 times out of it is production life time [23] and the duration of fracking process may last only for few days or weeks [43], we treated the total fracking water use as if it were spread over the entire year since the time scale of the analysis is annual.   The spatial distribution of wells in the maximum extraction scenario was based on locating future wells in watersheds that do not contain any wells but are inside the boundaries of the shale play ( Figure 1). While new wells could be drilled in HUCs with existing wells, this assumption is reasonable given the lack of information on placing of future wells that would inform a more specific approach for well distribution. In the Haynesville, there are 73 watersheds that do not contain any wells. Therefore, the additional 7,168 (10,000 total minus 2,832 existing wells) were equally distributed over the 73 well-free watersheds units resulting in 98 new wells per watershed.
The same average water demand for fracking estimated previously is assigned to each new well.
Similarly, the additional (6,312-33=6,279) wells in Tuscaloosa were equally distributed along the 311 well-free watersheds, resulting in 20 wells per watershed in order to simulate a maximum extraction scenario. Table 2 summarizes the number of wells and watersheds units involved per hydraulic fracturing scenario. These new fracking water demand estimates were incorporated into the surface and groundwater WaSSI based on equations (1) and (2), assuming that fracking water is fully taken from one source only, either from surface or groundwater. The assumption of using all groundwater or all surface water is probably not realistic, but it is necessary given the fact that we cannot predict the relative distributions from these two sources. Moreover, by examining the end member scenarios of assigning 100% of fracking water volumes into surface or groundwater sources, we can examine the upper ends of the WaSSI stress for each source.

Base Case vs. Maximum historical fracking scenario:
For the Base Case scenario, where surface and groundwater stresses are assessed without including the fracking water use, the results ( Figure 3, Table 3 and Table 4) show fairly low WaSSI stress values for surface water in all the watersheds of both shale plays. As noted earlier, and due to the absence of more specific information on the source of fracking water, the stress calculations assigned the entire fracking water use to either surface or groundwater. Except for one watershed, surface water stress fell into the low stress category (WaSSI < 0.5). For groundwater, the stress results are noticeably different (Figure 4Figure 3). Although most of the watersheds are still under low stress, 12 out of the 94 watersheds (13%) in the Haynesville shale play have stresses greater than 0.50, and five (5%) are highly stressed. This indicates that groundwater resources are expected to be more vulnerable to the addition of groundwater demand for fracking purposes in the Haynesville shale play area. For the Tuscaloosa play groundwater stresses are mostly lower than 0.50, with only one unit out of the 19 watersheds having a stress ratio over 1.
By examining the maximum historical fracking scenario, and when the fracking water use was assumed to be fulfilled entirely from surface water resources, the results show that there is no significant change in the surface water stresses for most watersheds (Figure 3). However, in the case of assigning the fracking water demand to groundwater resources, the maximum historical fracking scenario shows that more watersheds are affected due to fracking operations (Figure 4).
In the Haynesville, seven watersheds are showing groundwater stress values over one, and 13 watersheds are now under medium stress ( Table 3). The assumption of fulfilling the fracking water demand from groundwater resources almost doubled the number of watersheds in the medium stress classification. On the other hand, the Tuscaloosa's watersheds were not affected measurably by fracking water demand, which is expected since only 19 wells are accounted for in this part of the study (Table 4).

Possible maximum extraction scenario
The results of the possible maximum extraction scenario are illustrated in Figure 5, Figure 6, Table   3, and Table 4. Surface water stress results showed no significant increase in either of the shale plays, except for one watershed in Haynesville ( Figure 5). While more watersheds, especially in the Haynesville, display higher stress levels, their surface WaSSI values remain below 0.1.
Nonetheless, if all the fracking water demand was attributed to groundwater use, then groundwater resources will be affected significantly (Figure 6), as 23 of the watersheds become highly stressed, and another 16 become medially stressed over the Haynesville shale play. In the Tuscaloosa, only one watershed is classified with high stress and the remaining 18 are still in the lower stress category.   (Figures 3 and 4). When water use for fracking was assumed to come from groundwater, the watersheds showed much higher levels of stress ( Figure 8). In this case 53 of the Haynesville watersheds become highly stressed, and 35 were at a medium stress level (Table 3 and Table 4).
In the Tuscaloosa over 43 watersheds showed high groundwater stress values and 27 showed medium stress values.    A summary comparison between each of the three different scenarios and the Base Case is presented in Table 5. This table reports the number of watersheds that shifted their stress levels by at least on level (i.e., from low to medium, medium to high, or low to high) compared to their without-fracking condition. A significant increase in the groundwater stress level is observed in the Haynesville shale play where more watersheds are involved in hydraulic fracturing activities.
For the maximum historical fracking scenario, 10 watersheds shifted from a lower to a higher stress level. The "possible maximum extraction" and "maximum extraction" scenarios further increased the number of watersheds migrating to higher levels of stress to 45 and 107 watersheds, respectively (Table 5). For the Tuscaloosa play, the groundwater stress levels are not significant for maximum historical fracking and possible maximum extraction scenarios considering that only 19 watersheds units contain one well or more. However, for the maximum extraction scenario, the shift from low to medium or high groundwater stress is more evident (Table 5). Maximum extraction scenario expands the area of influence of fracking activity inside the shale play area, simply due to the assumption that this scenario incorporates new wells into watershed that did not contain any wells in the other scenarios.

Conclusions and Future Work
Hydraulic fracturing activity requires large volumes of water as a primary operational component.
This can have negative implications for the sustainability of freshwater resources. In this study, the WaSSI stress analysis framework was implemented to assess the impact on current and future water resources due to hydraulic fracturing in Louisiana's two main shale plays, Haynesville and Tuscaloosa. Three fracking scenarios were simulated within the WaSSI framework to evaluate recent conditions (maximum historical fracking scenario) and two future projections that represent moderate and aggressive fracking activities. The stress analysis was performed for surface water and groundwater components separately to determine which resource is more vulnerable. Based on the results of the three scenarios, the following two main conclusions can be made (1) The surface water in the watersheds located with Louisiana's shale plays seems to be sufficient to meet water demands for hydraulic fracturing activities performed in the oil and gas extraction processes.
Therefore, fracking activities should rely on this source of water instead of groundwater whenever possible, and (2) Groundwater resources in Louisiana are clearly vulnerable to hydraulic fracturing activities. Current water uses for hydraulic fracking resulted in significant stresses on the groundwater system, especially for the Haynesville shale play. The analysis showed that such stresses are expected to increase significantly and expand spatially under future projected hydraulic fracturing activities.

Future work
State oil and gas companies will need to weigh economics of production and water sustainability in an oil boom. While the results of this study suggest that surface water resources can be adequate to sustain the current and future demands of water for hydraulic fracturing activities in Louisiana, economic factors should be taken into considerations. These include costs for developing surface water storage facilities, especially in areas with little topographic relief such as Louisiana and other coastal states, as well as costs associated with transportation of surface water to extraction sites.
Furthermore, the feasibility of other alternative resources that are not suitable for public or irrigation uses (e.g., brackish groundwater), and the potential use of recycled flowback and produced water should be assessed as viable strategies that can alleviate the impact of hydraulic fracturing on the sustainability of freshwater systems.