Renewable and Sustainable Energy State of the art review of the environmental assessment and risks of underground geo-energy resources exploitation

Geo-resources play an increasing signi ﬁ cant role in achieving a sustainable energy future. However, their exploitation is not free of environmental impacts. This paper aims to identify the lessons and knowledge gaps on understanding of the sources, mechanisms and scope of environmental consequences of underground geo- energy resources exploitation. The paper examines four underground exploitation activities: CO 2 geological storage, exploitation of shale gas, geothermal power and compressed air energy storage (CAES). Selected studies carrying out life cycle assessment (LCA) and environmental risk assessment (ERA) are structurally reviewed by applying a six steps method. Our ﬁ nding indicates that global warming potential is the major focus of examined LCA studies with relatively less attention on other impacts. Environmental impacts at the local level are less evaluated except water use for shale gas and geothermal power. Environmental impacts of exploitation with storage purposes are relatively low. For energy supply associated exploitation, the impacts largely depend on the types of underground activities and the exploited energy carriers. In the ERA studies, likelihood of a hazard occurrence is the focus of the probability assessment. There is limited information on the pathways and transport of hazard agents in the subsurface and on the relation between hazard exposure and the impacts. The leakage of the storing agents is the well-identi ﬁ ed hazard for storage associated exploitation, while the migration of ﬂ uids and exploited energy carriers are the ones for exploitation with energy supply purposes. In general, understanding of environmental risks of soil contamination are limited. Very few number of ERA studies are available for assessing a CAES. Our research points out the need for developing a framework which allows the integration between LCA and ERA in subsurface environmental management.


Introduction
Increasing energy demand, ensuring energy security, mitigating climate change and enhancing flexibility of energy systems are four key challenges for a sustainable energy future. Exploitation of geo-resources for energy purposes, which goes well beyond fossil fuels exploitation, play an important role in meeting these challenges. Current geo-resources exploitation for energy purposes can be divided into three categories: • Primary energy supply, such as oil and gas exploitation, coal mining, geothermal development, etc.
• Retrieval storage, such as compressed air energy storage, hydrogen and natural gas storage and thermal energy storage, etc.
• Permanent storage, such as radioactive waste storage and CO 2 geological storage, etc.
Geo-resources associated primary energy carriers, mostly oil, coal and natural gas, have accounted for more than 80% of the total world primary energy supply in the last four decades [1]. Future energy demand is forecasted to keep growing in the coming decades and energy security will remain an issue at both global and national levels [2,3]. Fossil geo-resources are expected to still play a significant role in the future energy mix. The US national shale gas production has increased from 1.97 tcf in 2005 to 13.34 tcf in 2015 [4]. It is expected to provide 50% of the US natural gas production in 2040 [5]. Fossil fuels are, however, not the only geo-resources with a growing trend. The global capacity of geothermal power has doubled since 1990 reaching 13.2 GW in 2015 [6]. A recent report on the potential of geothermal resources indicates an economic feasible geothermal power production in Europe at 174 TW h in 2030 and 4000 TW h in 2050 [7]. Note that the latter figure is higher than the current European electricity supply.
In addition to the activities of fuels exploitation, the use of under-ground for permanent storage is also important. CO 2 geological storage and geological disposal of radioactive wastes are two examples of using the subsurface with a permanent storage purpose. In order to meet the target of limiting the CO 2 concentration to 450 ppmv, the global cumulative CO 2 storage has been estimated at 2168 t in 2100 [8]. The global primary energy supply from nuclear power generation today is about five times as much as four decades ago [1]. Depending on the scenario, it is expected that nuclear power generation can increase from 2400 TW h in 2015 to 6500 TW h in 2050 [6,9]. Such growth will strengthen the burden of safe and long term radioactive waste disposal, for which storage in deep formation is regarded as the most promising option.
Another potential role of geo-resources is as a part of strategies to increase flexibility of future energy systems. Geo-storage, such as compressed air energy storage (CAES), underground hydrogen storage and thermal energy storage (TES), could potentially serve as a buffer to facilitate intermittent renewable energy integration. In fact, CAES systems have been proved as one of the most cost effective technologies to facilitate wind power integration [10,11]. As renewables gain a larger share in the energy mix, the need for these types of systems is very likely to increase.
Exploitation of geo-resources for energy purposes is however not free of environmental impact. The environmental effects of its life cycle chain generally include land use, atmospheric emissions, emissions to soil and water, water use and consumption, solid waste and waste heat, geological hazards as well as noise and impacts on biodiversity, etc. There are different approaches for identifying and assessing such effects. Two common ones are life cycle assessment (LCA) and environmental risk assessment (ERA). LCA is widely recognized as an effective tool to evaluate the aggregated environmental impacts over the entire life cycle of a product or service [12][13][14]. It facilitates decision making processes by allowing a quantitative comparison of environmental impacts of alternatives. ERA is a formal process for evaluating the negative environmental consequences of a hazard and their likelihoods [15].
There are several differences between LCA and ERA in terms of objectives, scope and focuses. LCA focuses on all the demands of raw materials, energies and water as well as wastes and emissions caused by the value chain of an investigated product or service. Most studies aim to compare the environmental impacts between two technologies or products under normal operation conditions. As an example, in LCA studies of CO 2 geological storage, CO 2 leakage from the reservoir is normally not considered in most studies due to it is caused by an unexpected failure. Similarly in LCA studies of shale gas, environmental consequences of discharging inappropriate treated wastewaters due to insufficient treatment capacity or leakage of on-site treatment are not included. ERA aims to assess the environmental impacts and likelihoods of a particular hazard along with the production, use and disposal of a specific substance [16]. It only focusses on the risks of potential operational failures or failure condition but does not cover the environmental impacts of all processes involved in a specific product or service. On this basis, LCA and ERA may be seen as complementary tools in providing a comprehensive picture of potential environmental consequences and thereby supporting environmental management.
Today, a large number of LCA and ERA studies of different georesource exploitation have been conducted. These studies provide valuable insights into either environmental impacts or risks of individual exploitation activities. It is however not clear what the general lessons learned are so far and how this knowledge can be applied to future exploitation activities. It is specifically true in the part of underground exploitation. An overview including the environmental consequences of both operational activities and failures would help in identifying the focuses, overlaps and potential knowledge gaps of current research.
To the best of our knowledge, such overview is missing. This paper aims to fill this gap by identifying the general lessons learned and key knowledge gaps on understanding the source, mechanism and scope of environmental consequences through evaluating the state of the art knowledge, methods and data sources applied to assess the environmental impacts and risks of underground geo-energy exploitation.

Methodology
In this paper, a six steps methodology has been applied. Fig. 1 shows a schematic diagram of the methodology.

Research scope
Shale gas, geothermal power, CO 2 geological storage and CAES were selected as the representative exploitation of the subsurface. They represent the three purposes of geo-resources exploitation and they are modern technologies with (or having the potential of) a large-scale deployment in the coming decades.
The focus of this research is the environmental impacts and risks caused by the key processes of subsurface exploitation activities. Impacts or risks caused by other activities in the life cycle chain are therefore not discussed.

Critical literature selection
About fifty LCA studies and sixty ERA studies were initial collected according to the following criteria: they were written in English; they were published between 2007 and 2015, and they are peer reviewed journal articles or peer reviewed reports.
As there are a large number of LCA studies on CCS and shale gas exploitation, a second selection round was carried out by applying two criteria. First, the environmental impacts of CO 2 storage and the underground activities of shale gas exploitation should be presented. It is because many studies only show the environmental impacts of the life cycle chain of CCS and shale gas without presenting the environmental impacts of individual phases. Second, priority was given to the studies investigating multiple impact categories. As a result, eight LCA studies on CCS and eight on shale gas exploitation were selected. Six LCA studies on geothermal power and four on CAES have been also included as they are the most recent published LCA studies on these two topics.
The first round collection of ERA studies was narrowed down according to two criteria. First, they should be quantitative studies and
Only two qualitative ERA studies on CAES were founded in open literature and were included in this analysis.

Key processes of underground geo-resources exploitation
The focuses of this step were to identify the key processes of subsurface geo-resources exploitation and to analyze the subsurface activities associated energy demands, material and water uses. Therefore, an inventory of the key processes of subsurface exploitation and associated energy demand, material and water uses was carried out. The investigation was based on the selected LCA and ERA studies.

Review application of LCA and ERA to geo-resources exploitation
As discussed in the introduction, LCA and ERA have distinctive objectives, approaches and system boundaries. LCA studies present the environmental impacts of normal operations in the entire life cycle processes of geo-resources exploitation. Potential failures, their likelihoods and corresponding environmental consequences of a given operation or process are the focus of ERA studies. On the basis of these differences, the application of LCA and ERA studies as well as the environmental impacts and risks of geo-resources exploitation were reviewed in different ways. In the case of LCA studies we: • Identified the impact assessment methods and environmental impact categories.
• Identified and reviewed key parameters and critically discussed the data sources of each parameter.
For reviewing the ERA studies we: • Reviewed the assessment stages and identified the focus of the examined studies.
• Reviewed and critically discussed the assessment methods and assessment period.

Review environmental impacts and risks of underground georesources exploitation
The underground environmental impacts and risks of geo-resources exploitation were reviewed and analyzed according to the potential spatial scale of the influence. Macro, meso and micro levels are three levels representing environmental impacts and risks at the global/ continental, regional and local levels, respectively [12].
The following steps were taken for reviewing environmental impacts of underground geo-resources exploitation from the selected LCA studies: • Categorizing the environmental impacts of the subsurface activities according to the spatial scale of the influence.
• Extracting and analyzing the environmental impacts of the subsurface activities from the results of the examined studies.
• Comparing the environmental impacts of the subsurface activities with those of the entire life cycle chain.
For the ERA studies, the risks of underground exploitation were reviewed by: • Identifying the investigated hazards and their sources and pathways. • Discussing the hazard receptors and the potential environmental consequences.
• Categorizing the environmental consequences of the hazard according to the spatial scale of the influence.

Identification of the lessons and key knowledge gaps
In general, lessons and potential knowledge gaps of environmental impacts and risks of the key processes of underground geo-resources exploitation were identified by processing, integrating and comparing reviewing results from the previous steps.
The key knowledge gaps of LCA and REA applications were identified by examining: • Whether the key processes of subsurface exploitation activities were included and evaluated.
• Assessment focuses and potential assessment weakness on different spatial scales of environmental consequences.
• Potential weak fields of applied methods, data sources and assessment periods.
The lessons of environmental impacts and risks of the key processes of underground exploitation have been learned by: • Comparing the environmental impacts and risks of underground activities in each geo-energy resource exploitation with those of the entire exploitation chain.
• Comparing the environmental impacts and risks of underground activities in one geo-energy resource exploitation with those in other exploitation.
The comparison facilitates understanding of the relative significance of environmental impacts and risks at different spatial scales. It also helps to identify potentially overlooked environmental consequences at a given spatial scale.

Geo-energy resources exploitation and potential sources of environmental impacts and risks
The first step in the analysis is to identify the key processes of subsurface exploitation activities. Here, a short description is provided and the key subsurface activities associated energy demands, materials and water uses are identified.
Geothermal power plants extract hot geothermal fluid or steam from a geothermal reservoir in the deep subsurface and convert heat to electricity. Water, heat and permeability are three key elements for forming a natural geothermal reservoir. An artificial geothermal reservoir may be created in formations where the temperature of dry and impermeable rocks is high. Permeability can be enhanced by a hydraulic stimulation (HS) process in an enhanced geothermal system (EGS). Three types of technologies exist for geothermal power generation, namely dry steam, flash and binary power plants [7].
The production of shale gas releases and collects natural gas from the shale formation with low permeability. Horizontal drilling and Hydraulic fracturing (HF) are two key processes that make shale gas exploitation different from conventional gas. Hydraulic fracturing involves the injection of fluids under great pressure to fracture the shale formation. The fluid carries and places the proppant into the fractures to keep the fractures open after the injection pumping pressure is terminated. Natural gas can then flows through the fractures into the well [17].
Suitable geological formations for CO 2 geological storage require a depth at least 800 m, good caprock integrity and good injectivity [12]. The geological reservoirs considered are (nearly) depleted or abandoned oil and gas reservoirs, deep saline aquifers and deep coal beds.
A CAES facility consists of an electricity generation system and an energy storage system. It uses off-peak electricity to compress air and store it into a subsurface reservoir. During an electricity peak demand period, compressed air will be released to a combustor in a gas turbine to generate electricity. The compressed air can be stored in under-ground salt caverns, depleted gas reservoirs and aquifers [18]. Table 1 highlights the key activities of subsurface exploitation and associated energy, material and water uses.

Impact assessment methods and impact categories
A number of impact assessment methodologies has been used to assess the environmental impacts in LCA studies. The most common ones are ReCiPe, CML 2011 and IPCC. They generally provide similar characterization results for impact categories such as global warming potential (GWP), acidification potential (AP) and eutrophication potential (EP). The impact assessment methods and impact categories of the reviewed studies are shown in Table 2.
Global warming potential is the most common mid-point impact category investigated in the reviewed studies (24 out of the 27 studies), followed by the categories related to water use, e.g. eutrophication potential (12 studies) and acidification potential (9 studies).
The LCA studies for CCS and geothermal power trend to show a more diverse mix of categories while the focus of shale gas and CAES is on GWP. Toxicity related impact categories such as HTP, TETP, MAETP are almost not examined though most concerns for stakeholders and general public are related to these categories. This is particularly the case for shale gas exploitation where a thousand of chemical additives have been recorded in the use of HF to provide different chemical and physical functions [17]. Moreover, it is noticeable that relatively low attention is given to water use (WU). WU should receive more attention in future studies of EGS exploitation due to the considerable volume of water withdrawal in the HS process.

Data sources of key parameters
The environmental impacts of geo-resources exploitation depend, to a large extent, on the geological situation, technical specification and data inventory of mass and energy flows.
In the LCA studies of CCS, parameters such as well depth and injection rate have a relatively high influence on the impacts [12]. Table 3 presents these key parameters and their data sources. Note that no site specific data was applied in the reviewed studies. Generic storage fields were analyzed and most key parameters were based on a few existing CO 2 storage projects such as the Sleipner project in Norway (the project in offshore deep saline formation) or the Weyburn project in Canada (the enhanced oil recovery (EOR) project). Moreover, key parameter data in some cases is not available [30] or the sources of key parameters are not transparent [25,44]. These make comparison of data and results difficult.
In the shale gas exploitation studies, parameters such as estimated ultimate recovery (EUR), flaring rate and episodic emission factor (of both well completion and workover) are influential to GHG emissions [22,32,35]. Water withdrawal and the recovery rate of the HF fluids are the key parameters for determining WU of subsurface exploitation. Key parameters and data sources for assessing GHG emissions and WU are listed in Tables 4 and 5. As can be seen in Table 4, a large number of data in the studies has come from the data sources [48][49][50][51], which provide technical specifications and inventories of GHG emissions of Table 1 Overview of the key processes of subsurface geo-energy sources exploitation and associated energy, material and water uses [13,[19][20][21][22][23][24]. both conventional and unconventional gas exploitation in the US. Key parameters and data sources for assessing environmental impacts of deep geothermal energy are shown in Table 6. It is observed that most of applied data in the examined studies is based on a few existing geothermal power projects. The data used in [19,20,38] is similar because it was based on the same geothermal power projects, namely the Cooper Basin EGS project in Australia and the Soultz-sous-Forêts project in France. It is also found that water withdrawal for hydraulic stimulation of an EGS is less reported in literature.

Geo-resources exploitation
Four LCA studies on CAES exploitation were reviewed. However they provide very limited data regarding the underground activities. Three reviewed studies [18,43,62] do not provide any relevant emission factors or data of energy and mass flows. The study [24] provides data of material inventory but energy and water associated data is not reported.

Stages and focuses of the risk assessment
Environmental risk assessment is a formal process of evaluating the environmental consequences of a hazard and their likelihoods [15]. A comprehensive quantitative ERA typically consists of five stages including site characterization, hazard identification, consequences assessment, probabilities assessment and risk and uncertainty characterization. The stages of hazard identification, consequences assess- Table 2 Overview of LCA impact assessment methods, impact categories and studied technologies and storage options.

Geothermal power Study
Exploitation technology Impact assessment method Impact category [19,38] Flash, binary and EGS IPCC GWP [20] Flash, binary and EGS -WU [39] EGS Ecoinvent database GWP, CED, AP, EP [40] EGS Ecoinvent database GWP, CED, AP, EP [41] Binary Simapro software GWP, CED, ADP, AP, POCP, EP Compressed air energy storage Study Storage option and exploitation technology Impact assessment method Impact category [42] Salt cavern, diabatic plant IPCC GWP [24] Salt cavern, diabatic plant ReCiPe 2008 GWP [43] Mined hard rock cavern, diabatic and adisbatic plant ReCiPe GWP, EP, FETP [18] Salt cavern, diabatic ReCiPe GWP (CED: cumulated energy demand; GWP: global warming potential; EP: eutrophication potential; AP: acidification potential; HTP: human toxicity potential; LU: land use; POCP: photochemical zone creation potential; MAETP: marine aquatic ecotoxicity potential; TETP: terrestrial ecotoxicity potential; ADP: Abiotic depletion potential; FAETP: fresh water aquatic ecotoxicity potential; WU: water use).  [30] 1000 -No relevant data and data sources are available [31] 3000 231 30 [46] Analog project-natural gas subsurface storage [23] 800 (saline aquifer); 2500 (depleted gas reservoir) 125 15 [44,47] The Sleipner project and other EOR projects Regarding the first stage site characterization, all reviewed studies (31 studies) defined systems boundaries and 24 of them conducted an in-depth system characterization and discussed potential sources of hazards. However, only 6 studies identified measures and corresponding risk thresholds for the final comparison between the evaluated risks and the risk threshold. A lack of risk thresholds in reviewed studies can be explained by the difference in research focuses. For instance, 12 studies only focused on the assessment of environmental consequences of a specific hazard while 6 studies only focused on the probability evaluation of a hazard occurrence.
Hazard analysis is the most common assessment stage. All reviewed studies conducted investigations on hazard identification and analyses on mechanisms of a hazard occurrence, hazard pathways and environmental receptors. A few studies formulated scenarios as a pre-step according to the hazard pathways or environmental receptors for either a probability assessment or consequence assessment.
After the stage of hazard analysis, 19 studies (out of the 31 studies) assessed the environmental consequences of a hazard. Twelve of these studies examined a worst case scenario assuming occurrence of a hazard without conducting a probability assessment. For example, a hypothetical methane leakage [71] and a slow CO 2 leakage from a deep saline aquifer [72] were assumed and simulated to assess their environmental impacts on the groundwater quality. Occurrence of induced seismicity was stimulated in [73] to examine the maximum magnitude of seismicity.
Eleven studies (out of the 31 studies) conducted a probability assessment. As can be seen in Table 7, there are three steps in the probability assessment. All 11 studies evaluated the probability of a hazard occurrence, however only 3 studies evaluated the possibility of a receptor exposure to the hazard and only 2 studies investigated the possibility of the receptor being affected by the exposure. A lack of sufficient assessment on the probability of a receptor exposure to a hazard indicates insufficient information on the potential pathways and transport of hazard agents between the point of release to the so-called exposure point. For example, a probability assessment of water pollution caused by five failures in shale gas exploitation was carried out in [74]. In this study, the probability of a hazard occurrence, such as well casing failure and on-site wastewater leakage, has well supporting reference. However, the probability of hazard migration (migration of fracturing fluids to an aquifer through induced fractures) was recognized as "considerable debate" and estimated in a large range between extremely rare (1 in 1 million) and relatively common (1 in 10) without any supporting reference. Regarding the third assessment step, likelihood of negative effects depends on vulnerability of the receptor, the potency of the hazard itself and the amount or extent of exposure [15]. Lacking of this step assessment indicates that potential relations between exposure and harm are limited understood.
Only 5 studies (out of the 31 studies) combined the environmental consequence of a hazard and its likelihood for comparing the risks with relevant threshold. Sensitivity assessment was carried out by 9 studies to address the issue of uncertainty. It is noted that very limited amount of ERA studies was found in terms of CAES exploitation. Two qualitative studies were found to analyze the potential consequences of the hazards without further probability assessment. Table 8 shows the approach and period in the assessment stages of consequence assessment and probability assessment. Numerical simulation and experiment approach, which is normally combined with field measurements and laboratory experiments, are two common applied methods in the consequence assessment. Probabilistic risk assessment (PRA) is the most widely applied approach for the probability assessment.

Assessment methods and periods
ERA studies have diverse assessment periods based on their research scope. In general, the studies on CO 2 storage have longer - Primary data collection, [48,50] EPA, API  assessment periods than the studies of shale gas and geothermal power. This can be explained by the different purposes of geo-resources exploitation. CO 2 is expected to be sequestrated in a geological storage for a very long period (thousands of years). While the time interval of repetitive fluids injection and withdrawal in a geothermal power plant is very short and the time period of shale gas exploitation in average is a few months. An assessment period covering entire operation of georesources exploitation is recommended for future ERA studies.

Environmental impacts and risks of underground georesources exploitation
Environmental impacts and risks of geo-resources exploitation can be divided into three levels according to the spatial scale of the influence. Macro, meso and micro levels represent environmental impacts at global/continental, regional and local levels respectively [12].
The LCA mid-impact categories at the macro level include global warming potential (GWP), abiotic depletion potential (ADP) and cumulative energy demand (CED), etc. The meso level typically consists of the mid-point impact categories such as acidification potential (AP), eutrophication potential (EP), photochemical zone creation potential (POCP) and ecotoxicity potential, etc. Human toxicity potential (HTP) and water use (WU) are considered at the micro level [12]. Environmental risks at the macro level refer to the consequences of a hazard causing negative effects on the atmospheric environment. The consequences of a hazard affecting the environmental quality of a resource such as groundwater, surface water and soil are considered at the meso level. At the micro level, the environmental risks normally refer to the consequences of a hazard threatening the local environmental quality and health of human being above the ground.

. Environmental impacts
The environmental impacts of CO 2 storage are regarded as minor at all macro, meso and micro levels comparing with the entire CCS chain. They have a share between 1% and 5% depending on the impact category (see Table 9). Note that drilling is the only underground activity that has been taken into account for determining the environmental impacts of CO 2 storage. Drilling depth is proportional to the consumed energy, water and materials. For example, two types of CO 2 storage formations were investigated in [23]. Global warming potential of storing CO 2 in a depleted gas reservoir is three times as much as that of in a saline aquifer. It is because the drilling depth of the gas reservoir was taken about three times higher than in the aquifer.

Environmental risks/hazards
CO 2 leakage is one of the most recognized hazards of CO 2 storage (see Table 10). The sources of CO 2 leakage are mostly related to operational failure such as inadequately constructed wells or geological failure such as caprock failure. These two types of failure were identified and discussed in the ERA studies.
At the macro level, the consequences of CO 2 leakage are reported as an increase of CO 2 concentration in atmosphere and accelerating climate change at the global level. The probability of CO 2 leakage through various pathways to the surface was investigated in [75,76,82]. The study [82] assessed four scenarios of CO 2 leakage and the results indicate that the probability of CO 2 migration to the surface is in generally very small, especially in the scenarios of geological failure such as failure of caprock sealing and occurrence of existing and induced faults. It concluded that the probability of CO 2 leakage due to operational failures, such as leakage through abandoned oil & gas wells is higher than that due to a geological failure. The annual occurrence frequency of CO 2 leakage through abandoned oil & gas wells was estimated between 1.0E-5 to 1.0E-3 which is higher than the one through induced or existing fault 2.0E-8. Another study indicated that the probability of CO 2 leakage to the surface due to a pre-existing fault was 0.01% [76]. The study [75] shows that the CO 2 leakage rate increases after injection and reach a peak value 0.07% at around 40 days after injection and then keeps stable afterwards. This study provides valuable information for the leakage monitoring after the injection.
At the meso level, CO 2 leakage may cause negative impacts on the quality of groundwater, soil and surface water. The study [82] concluded that likelihood of leaked CO 2 reaching a shallow aquifer was low with the annual occurrence probability between 1E-08 and 1E-05. A similar study [76] recorded the result showing the probability of shallow groundwater contaminated by CO 2 migration through a preexisting fault was 0.04%. Both numerical simulation and experiment studies were carried out to identify the change of groundwater geochemical condition caused by CO 2 leakage. Five potential consequences are: 1) pH value decreases by 1-2 magnitude [72,[77][78][79][80][81]; 2) hazard trace elements such as Lead and Arsenic can be mobilized by the intrusion of CO 2 [72,77,78]; 3) the concentration of Lead and Arsenic increases and is close to (in [77,81]) or exceeds the maximum permitted value (in [78]); 4) the concentration of Fe and Ca increases following the CO 2 injection [72,[79][80][81]; 5) CO 2 injection is also responsible for the detection of BTEX (e.g. benzene) [81]. When CO 2 is stored in a deep saline aquifer, likelihood and potential consequences of brine migration to groundwater were estimated in [83]. It shows that the lateral distance of brine migration is rather small and the chance of vertical brine migration through a sequence of layers into shallow groundwater is also small. It indicates that large-scale pressure changes may be of more concern to groundwater resources than the quality change caused by brine migration. Unlike the well discussion on the quality change of groundwater, potential impacts on soil and surface water due to CO 2 leakage received less attention. At the micro level, it was estimated that no health effects are likely to occur in the scenarios of rapid CO 2 leakage through caprock failure, injection wells and abandoned oil and gas wells [82]. Induced seismicity is another identified hazard at the micro level probably caused by the pressure change of CO 2 injection. Likelihood of induced seismicity with a magnitude larger than one was estimated as 0.3% on the condition of the CO 2 plume reaches a specific fault [76]. Table 11 presents the potential environmental impacts of shale gas exploitation. The life cycle of shale gas exploitation includes both One year [92] One year Table 9 Environmental impacts of CO 2 geological storage at the macro, meso and micro levels. • < 2% of CCS chain [30] POCP Unit Impact

Macro level
• < 1% a of CCS chain [28]; • 1% a of CCS chain [29]; • < 2% of CCS chain [30] (Note: a functional unit 1 kW h net electricity delivered to grids is used; a: the impact value refers to both CO 2 transport and CO 2 geological storage).

W. Liu, A. Ramirez
Renewable and Sustainable Energy Reviews 76 (2017) [628][629][630][631][632][633][634][635][636][637][638][639][640][641][642][643][644] upstream and downstream processes. Upstream shale gas exploitation includes the processes such as site preparation, well drilling, well casing and cementation, well completion, gas production, gas on-site process and transportation. Downstream exploitation consists of the processes such as gas combustion in a power plant. At the macro level, GWP of upstream exploitation is largely caused by methane fugitive emissions and flaming. The ratio of upstream exploitation associated GWP to the life cycle GWP was reported between 5% and 12%. Water use is the major environmental impact at the micro level. Even though the process of HF requires a large volume of water, the water use of shale gas upstream exploitation accounts only 5-10% of the life cycle demands. The cooling process of the energy conversion facility, natural gas steam turbine in [37] and combined cycle in [22], requires the largest amount of water use from a life cycle point of view. Shale gas upstream exploitation contributes to the vast majority of the life cycle environmental impacts except GWP and WU. For example, shale gas upstream exploitation contributes to 95% of the life cycle impacts in ADP, 85% in HTP and 90% in AP.   Table 12. The environmental risk at the macro level refers to methane migration to the atmosphere and contribution to climate change. Fugitive emissions caused by normal operations and their impacts at the macro level have been well discussed in the LCA studies. Potential environmental consequences of stray gas migration caused by technical and geological failure are discussed in the ERA studies only at the meso and micro levels.
At the meso level, a number of studies has been carried out to assess the risks of groundwater and surface water contamination. Hazard identification was carried out in [84] by testing and analyzing a wide range of inorganic, organic and radioactive constituents of water and waste streams. The results show that flowback waters, drilling muds and HF fluids all exceeded the limits of Safe Drinking Water Act (SDWA) to varying degrees. Regarding likelihood of the hazard occurrence, [74] indicates that the probability of surface water contamination due to wastewater disposal was several orders of magnitude larger than the one caused by stray gas migration, well casing leaks, drilling site discharge and leaks through existing faults or fractures. The study [85] developed two methane migration scenarios: communication between a shale gas reservoir and an aquifer via a connecting fracture/fault and communication via a pre-existing nearby well. It concluded that shale gas production is likely to reduce likelihood of methane migration through reduction of available free gas and lowering the reservoir pressure. In terms of the stage of consequences assessment, hypothetical leakage of methane from 1000 m below the surface was simulated in [71]. The simulation results show that the pH value of groundwater increased from about 7-9. It was suggested that 300 m below the potable-water aquifer is a critical depth for the sealing of the fracking wells. Such suggestion was given due to the fact that rising methane gas can induce fracking fluids, caused by faulty well seals, to travel up to 300 m vertically. The drinking water quality in active gas extraction areas was examined in [89]. Elevated concentrations of cations and anions including manganese, iron, bromide and chloride were identified in most drinking water wells. The average and maximum methane concentrations in drinkingwater wells near the gas well were examined as 19.2 and 64 mg CH 4 /L indicating a potential explosion hazard. These two numbers were significantly higher than the number (averaged at 1.1 mg/L) examined in neighboring non-extraction sites with the similar geological formations [89]. No evidence of contamination by deep saline brines and HF fluids was found in this study. The study [86] investigated the potential risk of shale gas wastewater disposal on the quality of surface water. Barium and radium were substantially reduced in the treatment effluents but their discharge increased downstream concentration of chloride and bromide above the background levels. Moreover, the level of radium-226 in stream sediments at the point of discharge was about 200 times greater than that of in upstream and background sediments, posing potential environmental risks of radium accumulation at the regional level.
At the micro level, potential population exposure to air toxics in shale gas exploitation regions was estimated in [87,90]. Both studies concluded that with a large number of measurements on volatile organic compounds (VOC), shale gas exploitation activities have not resulted in community-wide exposure to those VOCs at the levels that would pose a health concern. However, a significant contribution to regional VOCs from shale gas production was identified in [90].

Environmental impacts
Environmental impacts of geothermal power development at the macro level vary considerably according to the technology (see Table 13). A flash power plant has shown much higher life cycle Table 12 Environmental hazards and risks of shale gas underground exploitation at the meso and micro levels. • Accumulation of toxic and radioactive elements [71,74,84,88,89] Spills and leakage of wastewater storage • Due to shale gas drilling Soil; Surface water; Groundwater

Scale
• Contamination with organics, salts, metals and other constituents; • Accumulation of toxic and radioactive elements [74,84] Disposal of inadequately treated wastewater • Insufficient treatment capacity and ability Surface water; • Elevated the level or the concentration of salinity, toxic metals, radioactive and organic constituents [74,86] Micro level Stray gas migration GWP than other types of technologies due to the processes of geo-fluids transport and flash. Most life cycle environmental impacts of a flash power plant are contributed by the surface exploitation activities. For instance, GWP and WU relating to underground exploitation activities only account for 4% and 10% of the life cycle values. On the contrary, a binary power plant or an EGS with the binary power generation technology has a zero direct greenhouse gas (GHG) emission in the phase of power generation because of the closed circulation of geofluids. Underground exploitation activities of these two systems contribute to the most life cycle environmental impacts except WU. When only looking at the environmental impacts of subsurface activities, an EGS has higher environmental impacts than a binary and flash plant.

Environmental risks/hazards
Only two hazards were identified in the reviewed ERA studies on deep geothermal energy exploitation. They are induced seismicity and geo-fluids migration. Environmental risks of geothermal power at the macro level were not discussed in the ERA studies. The hazard of geofluids migration was identified at the meso level [99] (see Table 14). The study focused on identifying sources and pathways of geo-fluids migration in a geothermal field. It concluded that the natural upward movement of geo-fluids, the faulty reinjection application and the uncontrolled discharge of waste geo-fluids were the potential sources of groundwater contamination [99].
At the micro level, the source and mechanisms of induced seismicity, especially the one triggering a large magnitude induced seismicity (defined as induced seismicity which can be felt by people above the ground) have been investigated [73,96,98]. The presence of preexisting faults near the geothermal field was concluded as the most influential factor affecting the occurrence and magnitude of induced seismicity [73,93,94,96,97]. The volume of cold fluid injection was also identified as another important factor triggering induced seismicity [98]. The probability of induced seismicity causing by deep geothermal energy exploitation has been evaluated in [91][92][93]. The general conclusion is that even though induced seismicity has been observed in many geothermal fields, the estimated probability of large induced seismicity was low.

Environmental impacts
As shown in Table 15, the life cycle environmental impacts of a CAES system largely depend on the fuel consumed, normally natural gas, to heat up air but not due to the underground activities. An adiabatic plant has much lower life cycle GWP than a diabatic plant because the heat released by the process of air compression is captured and stored in an adiabatic plant. No fuel is then needed to heat up air afterwards.

Environmental risks/hazards
Only two qualitative ERA studies were found regarding CAES exploitation. Explosion and compressed air leakage were identified as (Note: a function unit of 1 kW h net electricity generation is used. a: the impact value refers to surface and subsurface construction and fuel production).

Table 14
Environmental hazards and risks of geothermal underground exploitation activities at the meso and micro levels. the hazards at the meso level. The mechanism of an explosion occurrence in terms of storing compressed air in depleted natural gas reservoirs was analyzed in [100]. Oxygen provided by compressed air, fuels provided by residual hydrocarbon and heat provided by either compression energy or friction were identified as the potential sources of explosion. Operational failure such as imperfect welds and construction joints, as well as the induced failures due to repetitive air compression and decompression were identified as the key sources and pathways of air leakage [101] (Table 16).

Application of LCA and ERA to geo-resources exploitation
Based on the examined studies, the potential key knowledge gaps of applying LCA and ERA to geo-resources exploitation were identified and are presented in Table 17. Fig. 2 indicates the numbers of the LCA and ERA studies focusing on different spatial scales. The environmental impacts at the macro level are the focus of examined LCA studies on geo-resources exploitation for all energy related purposes. The impacts at the meso and micro  [43] (Note: a function unit 1 kW h storage of surplus electricity is used. a: the impact value refers to both subsurface storage construction and plant construction).  The ERA studies have different spatial focuses. In general, environmental risks of geo-resources exploitation at the macro level are often less evaluated. In particular, the ERA studies on renewable georesources exploitation only assess the environmental risks at the micro level, e.g. induced seismicity. While the environmental risks at the meso level are the emphases of the studies on fossil geo-resources exploitation and permanent storage exploitation. Very limited ERA studies are currently available for retrieval storage exploitation, e.g. CAES.

Environmental impacts and risks of underground geo-resources exploitation
Within the LCA studies, the absolute impacts refer to the results of the environmental impact comparison among different geo-resources exploitation (see Table 18). The relative impacts indicate the comparison results between the environmental impacts of underground activities and the impacts of its life cycle chain.
The normal operation of underground geo-resources exploitation for the storage purpose generally has low environmental impacts. They contribute to a minor part of the life cycle environmental impacts. Drilling is the only underground activity included in the LCA studies. A widely discussed hazard of exploitation for the storage purpose is the leakage of the storing agents, such as CO 2 and compressed air. The probability of the hazard caused by an operational failure has been evaluated potentially lower than the one caused by a geological failure. The effect and magnitude of the environmental consequences caused by the leakage largely depend on the property of the storing agent and existence of environmental receptors, e.g. groundwater. Very limited studies were carried out on the environmental risks of a CAES system may partially due to the storage period of compressed air is short and the storage agent, e.g. air, is harmless.
Environmental impacts of subsurface exploitation for the energy supply purpose depend on the underground activities and exploited energy carriers. The environmental impacts of geothermal power caused by normal subsurface operation have been evaluated as low, but having a relatively large share in its life cycle impacts. Power generation processes above the surface have lower environmental impacts than the underground exploitation activities have. The situa-tion is different in the case of fossil geo-resources exploitation. The normal operation of underground shale gas exploitation generally has the highest environmental impacts among examined exploitation, largely due to the process of HF and the property of shale gas. The migration of fluids used for stimulating the exploitation processes and the migration of exploited energy carriers are identified as two major hazards for the exploitation activities with the energy supply purpose. An operational failure leads to greater likelihood of the hazard occurrence than a geological failure does. The environmental consequences of the hazard caused by fossil geo-resources exploitation is generally more serious than the one caused by renewable geo-resources exploitation.

Cooperation between LCA and ERA
LCA and ERA are two system analytical methods that provide approaches for structuring, assessing and presenting environmental information on potential impacts for (environmental) decision making. It is important to note that these two methods fulfill different purposes. LCA deals with environmental impacts caused by the full value chain of the examined service. In the context of geo-resources exploitation, the service is supplying energy or creating storage capacity. LCA uses a functional unit thereby facilitating the comparison of environmental impacts among different technologies or productions providing the same service under normal operation.
ERA deals with the environmental consequences and likelihood of a hazardous substance or event at specific time and conditions along with the exploitation activities. Unlike LCA studies investigating the demands of raw materials, energy, water as well as waste and emissions caused by normal operations, ERA focuses on addressing a hazardous substance or event due to potential operational failures or (geological) failure conditions. The application of LCA and ERA has the potential to complement each other. They have different focuses on time and site specificity, on exploitation activities and conditions, as well as on spatial scales (see Table 19). LCA could point out possible environmental issues for particular operations or sites based on emission factors and impact indicators that could be then further examined in ERA with more site specific and detailed approaches [16]. Moreover, ERA focuses on environmental impacts mostly at the regional and local levels, while LCA studies provide the impacts at the global and regional levels. The Table 18 The environmental impacts and risks of underground geo-resources exploitation. results of LCA and ERA studies may complement each other in providing an overview of environmental consequences of the subsurface activities of geo-resources exploitation.

Conclusion
This study aimed to identify general lessons and key knowledge gaps on understanding the sources, mechanisms and scope of environmental impacts of underground geo-energy resources exploitation. This was done by conducting an state of the art assessment of literature that have studied on assessing the environmental impacts and risks of georesources exploitation. CO 2 geological storage, exploitation of shale gas, geothermal power and compressed air energy storage were selected here as representative examples of different exploitation purposes. Twenty-six LCA studies and 31 ERA studies on selected geo-resources exploitation were reviewed.
The general observed trend regarding LCA application is that the environmental impacts at the macro level have received the most attention with a focus on the mid-point impact category GWP and relatively limited focus on other environmental impacts. The studies of CO 2 storage and geothermal power have relatively more investigations on the environmental impacts at the meso level, with AP and EP as the often investigated impact categories. Water use is the focus of the studies on shale gas and geothermal power at the micro level. In the cases of CO 2 storage and geothermal power exploitation, data of key parameters, such as well depth and water withdrawal are often based on a few existing projects. Limited and inconsistent data on water withdrawal is found in the studies of EGS. Moreover, the key parameters of subsurface activities in the CAES studies are often not available.
The ERA studies have similar priorities in the assessment stages with the sequence of hazard identification, consequences assessment and probability assessment. The probability of a hazard occurrence is the focus of probability assessment. In this assessment stage, there is limited information on the pathways and transport of hazard agents in the subsurface and on the relation between hazard exposure and the impacts. Numerical simulation and experiments are two widely applied approaches in the stage of consequence assessment. A relatively short assessment period is observed in both probability assessment and consequence assessment for exploitation with the energy supply purpose. An assessment period covering entire operation is recommended. Very limited ERA studies are available for the CAES systems.
According to the examined LCA studies, normal operation of georesources exploitation for the storage purpose has relatively low environmental impacts. Environmental impacts of geo-resources exploitation for the energy supply purpose depend on the type of underground activities and the exploited energy carriers. The environmental impacts of underground geothermal exploitation activities have a relatively large share in its life cycle impacts, even though absolute impacts during normal operation are evaluated as low. Normal operation of shale gas exploitation in general leads to high environmental impacts largely due to the process of HF and the property of shale gas (fossil fuel). The environmental impacts of underground share gas exploitation has accounted for a large share in its life cycle impacts, except for GWP and WU.
The environmental risks of soil contamination due to geo-resources exploitation are less understood. The leakage of the storing agents is a well-identified hazard in the ERA studies with the storage purpose. Basically, the probability of CO 2 leakage caused by an operational failure has been evaluated lower than the one caused by a geological failure. While the migration of fluids used for the exploitation processes and the migration of exploited energy carriers are identified as two major hazards for the energy supply purpose related exploitation. The effect and magnitude of environmental consequences caused by the leakage largely depends on the property of storing agents, the property of exploited energy carriers and the occurrence of environmental receptors.
There is a need for developing a framework that allows integrating the insights provided by LCA and ERA into decision making processes for subsurface environmental management. Such framework can support obtaining a more complete picture of potential environmental consequences of geo-resources exploitation caused by both normal operations and potential failures.