Permeability prediction in the South Georgia rift basin–applications to CO2 storage and regional tectonics

Absence of a permeability log necessary to assess reservoir quality and injectivity for potential CO2 storage in the heterogeneous and complex South Georgia Rift (SGR) basin provides the motivation for this study. The focus of this study was on the Triassic-Jurassic red beds buried, entrenched beneath the Cretaceous-Cenozoic Coastal Plain sediments. Moreover, the significant cost typically between $10 M and $100 M associated with drilling and logging for in situ permeability coupled with the limited resolution of existing core data further makes this work necessary. The purpose is to relate, use the interpretation of the predicted permeability distribution to assess feasibility for safe and long-term CO2 sequestration. This study also intends to establish the impacts of active and passive tectonism that has shaped and/or re-shaped the evolution of the basin on the present-day permeability. A methodology was applied that utilizes the pore space and geohydraulic properties of the reservoir from existing laboratory and well data to produce a newly derived permeability log. It shows a non-uniform distribution with depths possibly due to geologic changes in the confined and heterogeneous red beds. The derived log displays characteristics consistent with observations from the porosity and resistivity logs. The interpretation of these logs provides evidence for the presence of low permeable, tightly cemented, and compacted red beds. We conclude that the low permeability aided by the low resistivity depicted in the red beds suggests increased confining stress and reduced injectivity, and that the uncharacteristically low permeability reflects a deformed basin shaped with episodes of uplift and erosion.


Introduction
The necessity for this study derives from the need to understand and characterize permeability distribution at subsurface reservoir depths to help determine suitability for CO 2 storage in the buried Triassic-Jurassic SGR basin. The existing core-scale permeability data cannot fulfil this critical need. This offers the opportunity for this work to make an innovative contribution to the SGR basin CO 2 site characterization efforts.
According to the Intergovernmental Panel on Climate Change (IPCC 2018) and the Global Carbon Capture and Storage (CCS) Institute (Global CCS Institute 2019), CO 2 storage provides a promising, prolific pathway to fulfilling the growing global efforts and urgency for immediate, long-term, and robust plans to addressing climate change or global warming and achieving net zero emissions.
The process involves capturing, separating, transporting, and injecting CO 2 into an underground geologic formation such as a deep saline aquifer or a mature oil and gas field. The geologic site or formation is selected following a thorough and comprehensive pre-injection site characterization and must be well monitored post-injection to ensure environmental safety and permanence of the stored CO 2 . Among the geologic sequestration options for safe and permanent CO 2 storage, deep saline aquifers are considered the most viable in terms of storage capacity, closeness to emission sources, and state-of-the-art technology (Bachu et al. 1994;Hovorka et al. 2006;Lucier et al. 2006).
Deep saline aquifers lack any tangible socio-economic value for domestic and agricultural purposes because of depth of occurrence within the subsurface and the presence of high concentrations of dissolved solids. The occurrence of 1 3 391 Page 2 of 13 these deep saline aquifers that are overlain by basalt and diabase sills constitutes the main consideration and drivers for the exploratory work for long-term CO 2 sequestration within the SGR basin. The SGR saline formations also occur well below underground sources of drinking water and satisfy the depth requirement of 800 m below the surface to enable supercritical CO 2 injection (Burruss et al. 2009). In this condition, the injected CO 2 exhibits gas-like compressibility, viscosity, and surface tension with liquid-like densities.
The advantages of supercritical storage are that CO 2 can move more easily within the confined reservoir and more storage per unit volume can be achieved. This is particularly vital to the goal of achieving large-scale sequestration for carbon capture and storage to make a meaningful impact in the growing global push for net zero emissions and a decarbonized future.
Of tremendous value to the goal of a robust pre-injection characterization within the SGR basin is the necessity to quantify, evaluate the depth-dependent permeability changes within the target geologic reservoir. Additionally, a proper understanding of the tectonic setting as this relates to the impacts on permeability is particularly important as part of site characterization. This is important to understanding and assessing the role of depositional and post depositional processes on the present-day reservoir properties of interest to safe and successful CO 2 injection and storage in the buried Triassic-Jurassic SGR basin.
Furthermore, a knowledge of permeability distribution with depths is essential to understanding tectonically induced geologic processes that may be responsible for deposition, redeposition, diagenesis, lithification, and/or deformation of sediments over time.
Permeability is an important reservoir property that measures the ability of a rock to allow fluid to pass through. It is a function of the pore space and pore connectivity within a rock (Mavko et al. 2003). Its pore space property enables it to have a direct, linear relationship with porosity. A rock with high porosity would typically exhibit high permeability so long as it is characterized by large and uniformly rounded grains.
However, poor sorting and presence of fine grain materials can reduce permeability even if the porosity is high. Porosity is heavily influenced by the rock's pore space and grain size distribution. On the other hand, permeability is controlled by a combination of these factors as well as by other subsurface fluid flow properties. These properties are tortuosity, pore shape and pore throats. The importance of permeability in the evaluation of the suitability of the SGR basin for safe and permanent geologic CO 2 sequestration makes it of great interest to this study.
The SGR basin was formed about 215-175 Ma through the breakup of Pangaea and opening of the Atlantic. It is believed to be the largest and probably the most geologically complex Mesozoic graben of the Eastern North American Passive Margin (McBride et al. 1989;Chowns and Williams 1983). As shown in Fig. 1, it covers an area of about 100,000 km 2 encompassing South Carolina, Georgia, Alabama, and parts of Florida (Chowns and Williams 1983).
The basin fills consist of basalts, diabase sills and red beds. Extrusion of the basalt and intrusion of the diabase sills followed the post rifting events that occurred during the Jurassic (Chowns and Williams 1983;Ghon 1983;Ghon et al. 1983;Olsen et al. 1991). The red beds were formed through sediment deposition that accompanied the formation of the basin in late Triassic. Studies by Heffner et al. (2012), Akintunde et al. (2013a), andMcBride et al. (1989) show the SGR basin fills to be overlain by the Cretaceous-Cenozoic sediments.
This study focuses on the red beds found in the Norris Lightsey #1 well, in Northwest Colleton County, South Carolina (Fig. 1). The Norris Lightsey #1 was a wild cat well drilled in the early 1980s to explore for hydrocarbons. It is also one of the very few wells in Southern South Carolina with significant penetration of the Triassic red beds, covering a depth of about 4000 m and penetrating over 3100 m of Triassic red beds.
The lithology of the Norris Lightsey red beds consists of fine-grained sandstones that are mixed with siltstone, conglomerate, and mudstone (Fig. 2). Geological characterization of these red beds for optimum reservoir quality assessment for safe and permanent CO 2 storage will require the interpretation and correlation of a combination of relevant well logs. Core scale laboratory data are of limited resolution and are characterized by coarse sampling at depth.
On the other hand, well logs offer fine spatial sampling and continuity that are essential for a true assessment of the state and suitability of a target reservoir for CO 2 storage. Unlike core scaled data, knowledge of reservoir permeability at in situ conditions is important to dynamic reservoir modeling. Unfortunately for the study location, there is no log of permeability changes at depth to either assess the suitability of the porous red beds for injectivity or correlate with available porosity and resistivity logs to aid site characterization. This lack of a permeability log, especially at reservoir depths not sampled by the available core laboratory data (Table 1), reinforces the motivation for this study.

Objectives
The purpose this study is to develop a new permeability log for the SGR basin. A permeability log because of its continuity within the subsurface meets the optimum resolution required to better quantify, understand, and interpret depth dependent permeability changes. The existing laboratory derived core scale porosity and permeability data (Table 1) are limited in resolution. Moreover, core-based laboratory data provides subsurface measurements on the order of inch/cm (core scale) whereas well logs measure up to m/km (reservoir scale) with better resolution for subsurface characterization.
The second goal of this work is to determine the local and regional implications of the depth-dependent permeability changes for safe CO 2 storage. This is crucial to addressing any concerns about pore pressure build up, fault reactivation, and induced seismicity. Given the relative proximity of the Norris Lightsey well to the Summerville seismogenic zone of South Carolina, the decision to drill and store CO 2 will also need to address questions about environmental safety as follows. Could CO 2 storage in the confined red beds lead to reservoir overpressure capable of causing leakage or threatening the integrity of the storage reservoirs and overlying seals? How can sitespecific permeability conditions curtail or expose the risk of fault reactivation and induced seismicity with injection and storage?

Methodology
The approach utilized involves applications of the modified Kozeny-Carman relation (Gomez et al. 2010;Mavko et al. 2003) and the Flow Zone Indicator (FZI) technique developed by Amaefule et al. (1993). The modified Kozeny-Carman relation described below in Eq. 1 computes permeability k from porosity for a rock with predetermined pore space and geometrical properties.
where d Mean is the mean grain size; τ is tortuosity, ϕ is the total porosity and ϕ p is the percolation porosity. These properties can be obtained from laboratory measurements on rock samples. Percolation porosity is the porosity when the pore is disconnected and does not contribute to flow. It   is generally between 1 and 3% (Gomez, et al., 2010, andMavko et al., 2003). Further discussion on the Kozeny-Carman relation including its derivation can be found in Mavko et al. (2003) and Schon (2011). The FZI as used in this study and previous research by Alam et al. (2011) and Prasad (2003) adapts and extends the Kozeny-Carman relation to enable better characterization of the spatial distribution of permeability in a reservoir characterized by presence of heterogeneities. It allows for an assessment of the petrophysical response and sensitivity to dynamic and depth dependent reservoir changes in a way similar to the applications of geophysical well logs for reservoir characterization. Its relationship to porosity and permeability k , which represents the application to this study, can be seen in the below Eq. 2.
In the above equation, 0.0314 is a constant that accounts for the pore size, tortuosity, pore shape, and the pore throat to pore-body ratio (Prasad, 2003). ϵ (Eq. 3) is the ratio of the pore volume to grain volume.
Further discussion and derivation of this equation can be found in Schon (2011), Prasad (2003, and Amaefule et al. (1993). The application to a porous reservoir is based on the premise that reservoir units with FZI values within a narrow range belong to one hydraulic unit. The implication of this is that these have similar pore throats and therefore constitute a flow unit. The step-by-step procedure for the implementation of the Kozeny Carman and the FZI technique to predict and provide depth-varying permeability changes in the heterogeneous red beds reservoir are discussed as follows.
1. Development of a porosity-permeability transform for the study area based on the Kozeny-Carman approach. 2. Development of FZI from the core derived laboratory measurements in Table 1 to allow for red beds with similar pore throats to be grouped as a single flow unit. 3. The use of the porosity-permeability relationship in step 1 to convert the porosity from the Norris Lightsey #1 well to reservoir scale permeability at target depths for potential CO 2 injection. 4. Incorporation of the results from step 3 into Eq. 2 using the computed FZI values from step 2 to produce the permeability log.
In both the Kozeny-Carman and FZI applications for this study, we utilized the existing well data, and the core laboratory measurements for the Norris Lightsey #1 well and other locations with penetrations of the South Georgia Rift red beds (Table 1). The Norris Lightsey #1 well has the deepest penetration of the red beds, covering a depth greater than 800 m below the surface to maintain supercritical CO 2 injection (Akintunde et al. 2013a). The variations in the porosity and permeability data from these locations are due to the influence of depositional environments (Akintunde et al. 2013b). In addition, these study locations share similar lithologic composition, age, geologic history, and tectonic setting with red beds recovered from several wells within the basin (Heffner et al. 2012;Chowns and Williams 1983;Gohn, 1983;Marine and Siple 1974).

Results
The permeability-porosity relationships based on linear correlation and the Kozeny-Carman relation are shown in Fig. 3. A grain size of 250 μm was used based on relevant information from literature review (Mavko et al. 2003;Gomez et al., 2010), and the subsequent testing and comparison with the direct correlation approach (Fig. 3). The predicted permeability using the linear regression method is less accurate (with R 2 of 0.52) when compared to the measured values and the prediction from the modified Kozeny-Carman relation. The direct correlation is based solely on pore size. Permeability depends not just on the pore space property, it also depends on the geometrical and fluid flow properties of the rock. This is the reason why the Kozeny-Carman porosity-permeability relationship yields a more accurate prediction of permeability from porosity than the linear correlation approach (Fig. 3). The Kozeny-Carman relationship considers the grain size and tortuosity of the rock, whereas the direct correlation does not account for these. The advantage of the FZI over the direct correlation approach is that it adapts and extends the predicted permeability from Kozeny Carman to provide estimates of permeability within definable flow units within the reservoir. The error and uncertainty in the direct correlation approach due to the flow properties of the rock that are not accounted are compensated for in the FZI technique.
The FZI allows for a division of the core-derived porosity and permeability data into flow zone units (Fig. 4). It performs this by treating the assigned porosity and permeability contributing to the same flow unit as one FZI value. The consequence of this is that reservoir units with FZI values within a narrow range have similar pore throats and therefore constitute a single hydraulic or flow unit. The distribution of the FZI shows that a large concentration of the data falls within FZI of 0.35 with a significantly high correlation (R 2 of 0.98). Plugging this value into Eq. 2 and substituting the well log derived porosity values into the derived porosity-permeability transform (in Fig. 3) allow to produce the new permeability log in Fig. 5.
The permeability log signatures are consistent with the trends exhibited by the in-situ porosity log as should be expected given the contribution from porosity. This provides a measure of the reliability of the Kozeny-Carman prediction and the computed FZI values. It is remarkable to note that while the porosities are high, the permeability values are low. This is significant as it shows that factors responsible for porosity such as pore size and grain size distribution are not the most dominant controls on permeability.
Permeability also depends on the rocks geometrical and fluid flow properties such as tortuosity, pore shape and pore throats. The strong correlation (R 2 of 0.98) observed in the FZI of 0.35 lends credibility to the accuracy of the predicted low permeability.
Furthermore, analytical comparison with studies carried out by Marine and Siple (1974) and Marine (1974) Table 1 is plotted on the left. The computed flow zone indicators are shown on the right. The flow zone intervals are shown by different colors in the color bar analog red beds from Dunbarton South Carolina provides validation for the results from the FZI approach. We also checked for its geo-reservoir consistency with known reservoir data from the same location by comparison with the field data from the down hole, reservoir porosity measurements (Fig. 5) in which the trends from the derived permeability log correlate significantly, statistically well based on the FZI of 0.35 (with R 2 of 0.98). This gives the needed confidence regarding the reliability of the FZI method.
In addition to this statistical check, the comparison with the thin sections in Figs. 6, 7, 8, and 9 offers another georeservoir validation and verification check. These thin sections were prepared from in-situ, down hole red beds cores recovered at subsurface depths of interest to this study.
Petrological and petrographic evidence from the thin sections (Figs. 6,7,8,and 9) show degradation in grain size distribution, sorting, pore sizes and shapes that are responsible for permeability. The alterations in the mineralogical composition that includes quartz, chlorite, claystone, and presence of mudstone casts offer factual geologic evidence to support the kind of permeability changes illustrated by the newly derived permeability log.

Discussion
The derived permeability log (Fig. 5) manifests the following characteristics: (1) uniform and non-uniform distribution with depths (2) noticeable spikes or increases at depth intervals 1395-1440 m, 1438-1445 m, and 1458-1465 m, (3) generally low permeability values that are less than 2 mD, and (4) vertical distribution which is consistent with the trends of the porosity log. The correlation of the permeability log with the porosity and resistivity logs allows for easy recognition of the highly resistive and non-porous diabase sills at 1410-1424 m.
Within the confined and heterogeneous red beds that are both above and below the impermeable diabase sills (Figs. 2 and 5), non-uniform permeability distribution is observed. We interpret the permeability variations with depth to be due to geologic changes and the presence of fluids in the red beds. These geologic changes involve key controls on permeability such as sorting, pore shape, pore throats and tortuosity. Analysis of photomicrographs (Figs. 6, 7, 8, and 9) of thin sections on red bed cores recovered from the Rizer #1 Test Borehole in Collenton County, South Carolina, provides evidence for these geologic changes. The Rizer #1 borehole, drilled in spring 2012, is within 5 km to the Norris Lightsey well (Fig. 1). The similarities in depositional environment and lithologic composition of the Rizer #1 well red beds with the Norris Lightsey lacustrine red beds provide the basis for the use of these thin sections. Analysis reveals cemented and lithified red beds with abundant quartz overgrowths and calcite cement (Figs. 6, 7, 8, and 9). The exposure to increased compaction and possibly periods of sustained subsidence during sediments deposition has significantly altered reservoir properties responsible for permeability. This is supported by the presence of clasts and small  Fig. 9 Photomicrograph of a thin section of a much fined grained red bed at the Rizer #1 Test Borehole at 1862.33 m, and net confining stress 13.10 MPa. It is well sorted but its lithified and compacted nature is not helpful to permeability. Changes in color are due to variations in mineralogy (after Weatherford Laboratories Report 2014) pore sizes and pore throats seen in the thin sections. These photomicrographs also show irregular pore shapes and sizes in the tectonically deformed red beds which may be responsible for the non-uniform distribution of permeability with depths (Fig. 5). The low resistivity in the red bed units is indicative of fluid saturated red beds. This is because the observed log resistivity values ranging from as low as 0.05 to less than 100 Ω m fall within the range of known resistivity values for water and saltwater reported in Telford et al. 2001. Chemical analysis conducted by Marine and Siple (1974) on pore water from a Dunbarton well with penetration of the red beds found dissolved solid content of approximately 11,000 mg/L that supports the interpretation of brine-saturated red beds. Their study also revealed much higher chloride in the red beds (6720 mg/L) in comparison with water from the Cretaceous-Cenozoic coastal plain sediments (1.5 mg/L) and the crystalline metamorphic rock (1260-1400 mg/L).
Also, there is no gas in the red beds as this would have caused an increase in resistivity. On the other hand, the overlying and non-porous diabase sills are completely dry. This explains the virtually non-existent permeability and the unusually high resistivity of the sills. The interpretation of a brine saturated reservoir is consistent with the plan for a deep saline CO 2 storage system for the South Georgia Rift basin.
The consistency in the observed trends of the porosity and permeability logs provides validation for the reliability of the new permeability log. Their disproportionate values relative to each other however show that the key controls responsible for both are not mutually identical. The correlation with the resistivity log also provides an additional validation of the permeability log. Both the red beds and the diabase sills are easily delineated from the resistivity log.
The ensuing question from the signatures of the resistivity and permeability logs is what does this mean about the state of the reservoir? The resistivity of a formation based on Archie (1942) varies with porosity depending on the nature and degree of fluid saturation as well as on the rock's cementation and tortuosity. For a fully brine saturated reservoir exhibiting the kind of depth varying porosities shown in Fig. 5, the consistently low resistivity suggests a tightly cemented, compacted rock. This is because the presence of brine exposes the red beds to chemical dissolution and geochemical reactions that contribute to their cementation and compaction. And with increasing confining stress from burial depths, compaction is further aided. The thin sections (Figs. 6, 7, 8, and 9) support the inference for a tightly cemented, compacted rock.
The process of compaction or lithification in a reservoir closes pores and/or restricts the interconnectivity between pores that are responsible for permeability (Figs. 6,7,8,and 9). The effect of increasing confining stress is to reduce the pore pressure by closing openings in a rock responsible for fluid movement (permeability).
Apart from geologic changes in the reservoir, the reservoir response to the permeability log signatures may be stress induced. Burial depth, age, geologic history, and composition are additional factors that influence low permeability. The regional implication of the low permeability is that the South Georgia Rift red beds and possibly the ones encountered in other buried Triassic-Jurassic basins in the Southeastern United States are most likely to be low permeable rocks in view of the similarities in age, geologic history, and composition (Akintunde et al. 2013b;Marine 1974, Marine andSiple 1974).

Implications for CO 2 storage and regional tectonics
In assessing the implications for CO 2 storage, we ask these questions. What does the low permeability mean for reservoir quality determination? How would this affect subsurface suitability for CO 2 storage? How would this impact CO 2 migration and containment in the red beds? Hydrogeologically, the injection, movement and storage of fluids are most effective in underground formations with high porosity and permeability. The direct consequence for low permeability is a reduction in fluid flow and movement even if the porosity or pore space distribution favors substantial fluid storage. Low permeability would impact the degree and effectiveness of injectivity for CO 2 sequestration in the porous formation. Whether or not a reservoir would be viable for long term CO 2 storage depends not only on the storage capacity but also on the quality of reservoir injectivity. A preliminary petrophysical investigation by Akintunde et al. (2013b) demonstrates that the confined South Georgia Rift red beds in the Norris Lightsey do exhibit porous intervals with the potential for substantial CO 2 storage capacity that far exceeds the 30 million tons set by the Department of Energy. Also, a preliminary reservoir modeling of CO 2 injection in the Norris Lightsey red beds by Brantley et al. (2015) demonstrates feasibility for injection of at least 30 million tons of CO 2 at a rate of 1 million tons per year for 30 yr. The overarching issue for the kind of subsurface distribution depicted by the permeability log in Fig. 5 is the impact on the degree of reservoir injectivity. The desirability for an effective CO 2 storage is to have sufficient injectivity to allow seamless fluid flow, movement, and containment without any fear of reservoir failure should the pore pressure exceeds the reservoir capacity. Conceptually, low permeability suggests reduced injectivity which in turn would impact the effectiveness of fluid flow and movement in the heterogeneous, porous red beds. Ideally, an increase if fluid flow and concentration will increase the pore pressure. The implication of increasing pore pressure with fluid injection is to counteract the effect of increased confining stress thereby opening pores and interconnectivity that could enhance permeability or fluid movement. Unless the pre-injection permeability can be physically or geo-mechanically enhanced, the predictably low injectivity will not be promising for effective fluid flow and storage.
Low permeability may help with safety and security of storage since the chances of sudden and unsafe pore pressure build up capable of either triggering induced seismicity or threatening the caprock integrity are unlikely with low injectivity. We understand from Zoback and Gorelick (2012) that increasing pore pressure with CO 2 especially in the vicinity of preexisting potentially active faults and considering the critically stressed nature of the crust were likely to increase the potentials for earthquake triggering. It is also inferred from Brantley et al. (2016) that the presence of an active fault with a permeability as low as 1 mD can cause significant CO 2 leakage. With a properly planned injection that incorporates and implements the applicable geological framework as well as robust monitoring and management techniques, the risks of faulting and induced seismicity can be mitigated. A study by Talwani et al. (2007) showing related seismogenic permeability values that are unlikely to cause induced seismicity with fluid injection also lends credence to the potential for safe storage in the low permeable SGR red beds. Ideally, CO 2 injection will increase the pore pressure leading to opening of closed pores. So long as this effectively balances the effect of increasing confining stress with depths and does not alter the differential stress equilibrium, the chances of unsafe seismicity within and around the injection reservoir are very unlikely. Moreover, the confining nature of the red beds together with the presence of the impermeable diabase caprocks would ensure containment of the injected CO 2 .
With adequate pore pressure monitoring before, during, and after injection, the risks to safe CO 2 storage may be quickly detected and averted. 4D seismic monitoring can help with understanding and quantifying dynamic reservoir changes to assess storage efficiency as well as the integrity of the overlying cap rocks. The current permeability log would provide the baseline information necessary for the next steps involving reservoir modeling and simulations, seismic modeling, and imaging, as well as field testing to assess the impacts of enhanced permeability on long term storage, the integrity of the overlying diabase sills, and monitor the efficiency and safety of injection and storage.
In terms of the application to regional tectonics, the predicted low permeability at depth reflects a compacted, deformed basin with a history of uplift and erosion. Corroborating evidence from the analysis of the thin sections (Figs. 6,7,8,and 9) shows that the low permeability is a direct consequence of poor sorting and small pore throats resulting from the effect of tectonically induced and post depositional processes such as compaction and diagenesis.
These thin sections on recovered red beds from the Rizer #1 Test Borehole in Collenton County, South Carolina, manifest fine to medium to coarse grained red beds that are interbedded with mudstone clasts. The mineral composition depicted by these thin sections as shown in Figs. 6, 7, 8, and 9 predominantly consists of quartz, calcite, biotite, feldspar, and chlorite. The dark color reflects claystone and silty claystone casts, while the light red color is indicative of calcite. Apart from the observed changes in colors because of the variations in mineralogy, the interpretation of the thin sections indicates lithified, compacted sediments. This is supported by the presence of claystone, and mudstone casts as well as the irregular, small pore shapes and pore throats. The textural and mineralogical characteristics exhibited in the thin sections are similar to that of the Clubhouse Crossroads red beds encountered near Charleston, South Carolina, in which the physical properties such as sorting, pore shape, pore throats, and tortuosity have been altered, impacted by tectonically induced post depositional processes such as diagenesis, compaction, and uplift that have re-shaped the tectonic evolution of the SGR basin following the major phase of rifting (Ghon 1983;Ghon et al. 1983). The occurrence within the Triassic red beds of pore-filling clasts as demonstrated in the thin section analysis indicates possible textural and mineralogical alterations that retard the effectiveness of pores and other interconnectivity within the rock that are responsible for permeability. The effect of erosion, during late Triassic to early Jurassic, as well as tectonically induced events, such as uplift and subsidence that preceded sedimentation near the Triassic-Jurassic boundary may have also contributed to the alterations and diagenesis within these red beds. In light of the above, the textural and mineralogical characteristics illustrated by the interpretation of the thin sections substantiate evidence for the low permeability of the tectonically impacted, altered red beds. Comparison with the petrological and textural analysis of similar red beds reported in Gohn et al. (1983) supports this interpretation. The buried Triassic-Jurassic rocks in the study area are among the Atlantic-type passive continental rift basins. In view of this, there is no indication that the sedimentary materials in this region at present are what they were during the early cycle of sedimentation associated with rifting of sediments and continental break up. It is possible based on evidence from Heffner et al. (2012) involving the analysis of multiple seismic lines and several well data with significant penetration of the SGR basin that erosion during Jurassic removed large amounts of rocks that include the Triassic sediments encountered in this study. Subsequent uplift of sediments and emplacement of the underlying mafic igneous deposits forming pore-filling casts in the deformed, depleted sediments over time. The thin sections confirm the presence of these pore-filling casts that over time result in closing of cracks, fissures, and pores that are responsible for permeability.

Conclusions
The decision to drill and store CO 2 will depend on the quality of the reservoir and the safety of injection and containment. Of importance to reservoir quality are the in-situ porosity and permeability that determine the storage capacity and injectivity. Knowledge of the permeability regime is an extremely valuable rock property that dictates and determines the progress and efficiency of injection and storage. Its correlation and interpretation with porosity and resistivity logs to better understand and characterize the state of the red beds reservoir for CO 2 storage provides the motivation for this study.
Permeability is most relevant for correlation with these logs, because of its strong connection to fluid saturation, grain size, pore shapes, cementation and tortuosity that are key controls on porosity and resistivity. Core based laboratory data do not have the resolution, scale, and continuity required for correlation and interpretation with well logs. Consequently, a significant, new contribution from this work is the development of a permeability log for the study area based on a robust methodology involving applications of the Kozeny-Carman relation and the Flow Zone Indicator technique. The rationale for the use of these two approaches was to ensure reliable permeability prediction and distribution that considers the pore space and geometrical properties of the target red beds.
The development of this new permeability log offers an alternative way to save time and significant cost associated with expensive well drilling and logging for in situ permeability measurements for reservoir characterization. It would also aid dynamic reservoir modeling of the distribution of fluid flow to better characterize the CO 2 injection distribution and efficiency for the purpose of storage optimization and management.
The interpretation of the permeability log supported by the correlation with the porosity and resistivity logs shows non-uniform distribution with depths possibly caused by geological and stress induced changes in the heterogeneous red beds. Moreover, the petrophysical responses in both the resistivity and permeability logs are generally low. We interpret this in conjunction with the porosity distribution to suggest: (1) the South Georgia Rift is a tightly cemented and compacted reservoir, and (2) a reservoir exposed to increased confining stress. Increasing confining stress closes and/or restricts reservoir openings responsible for porosity and permeability.
On the other hand, increasing pore pressure with CO 2 injection has the potential to counteract the effects of increased confining stress by opening closed pores or enhancing weak pores for efficient fluid movement and storage over time. However, low permeability will reduce injectivity that is a key requirement for the efficiency of CO 2 injection and storage. We also conclude that the predicted low permeability distribution with depth is a function of the active and passive post-tectonic depositional processes that have impacted the physical properties of the Triassic red beds. Future directions would focus on dynamic reservoir simulations to evaluate the integrity and capacity of the diabase sills as CO 2 caprocks. The dynamic reservoir simulations would also assess the immediate, short-term, and long-term injectivity capacities of the porous red beds using the newly derived permeability log.