3D implicit modeling applied to the evaluation of CO2 geological storage in the shales of the Irati Formation, Paraná Basin, Southeastern Brazil


 The Paris Agreement established global ambitious targets for reducing carbon dioxide (CO2) emissions, requiring the rapid and extensive development of low carbon technologies, and one of the most efficient is CO2 geological storage. Among the deep geological formations used for CO2 storage, the shale layers have been a new emerging topic showing to be efficient because they are abundant and have a high content of organic matter, being favorable for CO2 retention. However, one of the challenges in evaluating a location for possible reservoirs is the adequate geological characterization and storage volume estimates. This research evaluated the Irati Formation of the Paraná Basin, through the information from hydrocarbon exploration wells in Southeastern Brazil, where most stationary sources of carbon emissions are located. Three-dimensional (3D) implicit modeling techniques were applied not only for the volume calculation purpose, but also in the site selection stage, generating thematic 3D models of thickness, depth, structures, and distance to aquifer systems. The limestones, shales, and black shales of the Irati Formation were locally divided into six units according to geological composition and spatial continuity. The E black shale unit was considered for CO2 geological storage indicating a theoretical capacity of 1.85 Gt of CO2. The potential of the achieved capacity is promising not only for been greater than the total of CO2 locally produced but also for supporting the implantation of new projects in this region.


Introduction
In a scenario of growth in the generation and emission of CO 2 , the global emissions reached a historical record in the order of 33.1 billion tons (Gt) of CO 2 in 2018 as claimed by the International Energy Agency (IEA 2019). Brazil is among the 20 countries that have emitted the most carbon dioxide in recent years (IEA 2019). Data provided by the Sistema de Estimativas de Emissões e Remoções de Gases de Efeito Estufa (SEEG) on CO 2 emissions in the Brazilian energy sector, following the methodology of Azevedo et al. (2018), show an increase from 173 million tons (Mt) of CO 2 in 1990 to 380 Mt of CO 2 in 2018 (SEEG 2020). Brazil has also set goals to contribute to the reduction of GHG emissions, with a reduction of 37% by 2025, and 43% by 2030, based on 2005 emissions, as stated in its Nationally Determined Contribution (NDC) rati cation.
This worldwide decision requires the rapid and extensive development of low carbon technologies, and one of the most e cient is Carbon Capture and Storage (CCS). In the Brazilian context, CCS technologies are taking on a relevant space due to the company's performance in the energy sector, mainly due to the development of technologies that respond positively to global trends.
Deep geological formations are an effective alternative for CO 2 storage, but some requirements must be met to allow large amounts of CO 2 to be injected and that the gas remains trapped in the rock for a long

Hydrocarbons in the Paraná Basin
The interest in the Paraná Basin oil potential began in the late 19th century, with the rst shallow wells (less than 1,000m) being drilled close to the super cial occurrences of oil in the state of São Paulo. Between 1953 and 1979, Petrobras drilled 60 exploratory wells culminating in the discovery of four subcommercial accumulations of oil in the Santa Catarina State (Morelatto 2017). In the 1980s, between 1979and 1983 Petroleum and Paulipetro companies drilled 30 exploratory wells and discovered the sub-commercial accumulations of natural gas in Cuiabá Paulista, hosted in the Irati Formation, in the far west of São Paulo State (Morelatto 2017). In Fig. 1 and subsequent gures, the Cuiabá Paulista subcommercial gas accumulation is indicated only by the 3-CB-3-SP well, but in this position, there are also present the 2-CB-1-SP, 2-CB-1DA-SP, 3-CB-2-SP, and 3-CB-4-SP wells, that are distant from each other around 3 km. They were omitted in the gures for a scale issue, avoiding visual pollution with overlapping data.
Between 1986 and 1998 Petrobras retake exploratory activities in the Paraná Basin with the acquisition of new seismic data and with the drilling of seven exploratory wells, which reached the discovery of the Barra Bonita gas eld, a structural high trapped at the level of the Itararé Group sandstones (Campos et al. 1998) (Fig. 1). Since 1972 Petrobras operates an oil shale mine in the city of São Mateus do Sul, Paraná State (Fig. 1). The oil-rich shale of the Irati Formation occurs in two layers in the mine, the lower bituminous has 9.1% oil content and averages about 4.5 m, and the upper has 6.4% and averages about 9 m (Padula 1969;Milani et al. 2007). Currently, with the establishment of the Agência Nacional do Petróleo, Gás Natural e Biocombustíveis (ANP) new surveys and wells have been carried out and areas have been included in bidding rounds in the Paraná Basin (Morelatto 2017).

The Irati Formation
The Irati Formation is the lowermost unit of the Passa Dois Group and occurs in almost every extension of the Paraná Basin, with thickness varying around 40 m with a maximum of 70 m (Holz et al. 2010). Outcrops of the Irati Formation occur mainly along the eastern border ( Fig. 2) and at the northern and southern ends of the basin. The lower contact with the sandstones and mudstones of the subjacent Palermo Formation is given by a dark mudstone (Holz et al. 2010). In its top contact, the Irati Formation is superimposed by the Serra Alta Formation shales, and by the Teresina Formation sandstones and siltstones. Abundant sills and dykes generally attributed to the Serra Geral magmatism cut or occur intercalated with Irati Formation rocks.
The composition of the Irati Formation consists of interbedded shales, organic-rich shales, limestones, dolomites, siltstones, and mudstones, with an irregular distribution, distinct lateral continuity, and varying thicknesses along the entire basin (Padula 1968). A subdivision of Irati Formation in two members considers shallow marine siltstones and shales called Taquaral Member in the base, and bituminous black shales intercalated with dolomites, restricted only to the portion northward the Paraná Basin, called Assistência Member in the top (Barbosa and Gomes 1958). Subdivisions of the Irati Formation in seven to eight similar chemostratigraphic units based on geochemical characteristics (identi ed as A to H units from bottom to top) have been recognized at different locations in the eastern portion of the Paraná Basin (Alferes et al. 2011;Euzébio et al. 2016;Reis et al. 2018;Martins et al. 2020).

DATABASE
The database is composed of hydrocarbon exploration well data provided by ANP in December 2019. The wells were carried out by Petrobras Company during the 1950s until the 2000s. The original individual text le format (.txt) of each well was converted to Excel format (.xlsx) generating a drillhole format database with collar, survey, lithology, stratigraphy, and hydrocarbon spreadsheets. Conventional wireline logs with gamma-ray, resistivity, density, neutrons, and sonic log of each well were also provided in a reading format le (.pdf). Tables with scanning geochemical data, when available, were provided in a reading format le (.pdf). A lter was applied in a total of 123 wells in the Paraná Basin resulting in 32 wells in the study area (Fig. 1). From these wells, all have stratigraphy data, 29 have some lithology data, and 15 have some hydrocarbon indication data. Almost all ltered wells are vertical with only one having survey data. Then, comma-separated value (.csv) les with the wells data were loaded in the Leapfrog Geothermal® software for performing three-dimensional (3D) geological modeling. The Leapfrog software use radial basis functions (RBFs) to perform surfaces by the implicit modeling method (Cowan et al. 2003). A regional topographic surface was generated from the GTOPO30 data (https://earthexplorer.usgs.gov/). Surface geology data (lithology and structures) were obtained from Companhia de Pesquisa de Recursos Minerais (CPRM) (Lopes et al. 2004) and regional geological structures are from Zalán et al. (1990) and were initially treated in the free software QGIS Desktop before importing into Leapfrog platform.

THE IRATI FORMATION IN THE STUDY AREA
In the study area, outcrops of the Irati Formation are restricted to the southeast portion only (Fig. 2), so the internal unit subdivision came mainly from well data. All wells in the area reached and exceeded the horizon of the Irati Formation, except for the 1-QT-1-PR well, which is very shallow, and the 1-TI-1-SP well, which presents an extensive interval of diabase dike between the Teresina and Palermo Formations ( A 3D interpretation of the lithology data together with some TOC data enabled a local subdivision of the Irati Formation in units following the subdivision of Euzébio et al. (2016), Reis et al. (2018), and Martins et al. (2020). Due to the erratic availability of samples with TOC analysis, the three A, B, and C base units were undifferentiated and referred to as a single A/B/C shale unit (Fig. 3). So in this work, six units were considered informally named A/B/C shale, D limestone, E black shale, F shale, and H black shale from the bottom to the top (Fig. 3), in order to adapt to the existing division in the literature (Reis et al. 2018). For the interpretation of the units, well intervals less than one meter thick were incorporated (diluted) into the larger adjacent intervals. Thus, shaly units can contain thin and discontinuous levels of limestone, as well as carbonate units, can contain some thin shale levels. The A/B/C unit is a shale layer with local carbonate layers and a high thickness variation from 30 m in the middle of the area to only 3 m in the eastern portion. The sequence from D to G units is quite continuous laterally and occurs in the entire E-W section (Fig. 3) along the study area for more than 300 km. The D unit consists of limestones, with dolostones locally, which vary from calcilutite to calcarenite with thickness in the wells varying between 1 and 16 m. The thickness from the E black shale unit varies from 2 to 20 m. Both E and F units are interbedded with a diabase sill in the eastern portion. The F unit is composed of shales and has high variable thickness throughout the E-W section (1 to 14 m). The G unit is composed of limestones and dolostones with grain size varying from calcilutite to calcarenite, and with thickness in the wells from 2 to 11 m. The topmost H unit is a black shale layer that occurs only locally in the west of the area. The local Irati Formation subdivision used in this work is similar to the ones seen in outcrops (Fig. 4) and presented in the literature (Reis et al. 2018), although it is not exactly equivalent.
INITIAL CRITERIA FOR SELECTING THE CO 2 STORAGE LOCATION CO 2 stationary sources and some legal aspects The macro study area ( Fig. 1) already takes into account some criteria for the selection of the CO 2 reservoir location, as the region with the highest occurrence of CO 2 stationary sources and some legal aspects. Excluding the land use, land-use change, and forestry (LULUCF) sector, which is the main net emitter of CO 2 in Brazil, the second sector with the highest emission is energy. The São Paulo State is the state with the highest CO 2 emissions in the energy sector and the highest concentration of thermoelectric power plants in the country (Fig. 1). In 2018 from a total of 380 Mt of CO 2 emitted in the Brazilian energy sector, 80 Mt (21%) refer to the São Paulo State (data from SEEG 2020). The Paraná State also has a high concentration of thermoelectric plants. However, the Paraná State Law 19.878 of 2019 prohibits the production of unconventional natural gas through hydraulic fracking (Ramos et al. 2020), although the state has no oil and gas production activity, not even conventional. The Barra Bonita in the Paraná State is a conventional gas eld and is not in production. The legal context of shale gas in the Paraná Basin is better and deeply discussed by Lenhard et al. (2018) and Ramos et al. (2020). The São Paulo State has no legal restrictions or speci c legislation for CCS, as well as all over Brazil (Almeida et al. 2017;Costa and Musarra 2020). Therefore, areas inside the São Paulo State were prioritized in this study, mainly because there is also a high concentration of bioenergy industries related to sugar-cane ethanol plants (da Silva et al. 2018).

Regional geological aspects
The Paraná Basin is tectonically related to a more than 10,000km-long divergent margin originated by the Gondwana paleocontinent break-up and the separation of the African and South American plates (Milani and Thomaz Filho 2000), with no relation with fold belts, thus being in a tectonically favorable location for CO 2 reservoir, as recommended by IEA-GHG (2009).
The Paraná Basin in the southeast of Brazil presents a low level of seismic activity, as it is a typically intra-plate region. Only ve earthquakes with m b magnitude above 5.0 (with two being of largemagnitude: 6.3 and 6.8) have occurred in the studied region during the past 220 years according to Berrocal et al. (1996) and the bulletins of Centro de Sismologia da Universidade de São Paulo (USP), Brazil (http://moho.iag.usp.br/eq/bulletin). This scenario characterizes a favorable region for CO 2 reservoir, following IEA-GHG (2009) Depth, thickness and distance to groundwater aquifers To evaluate the subsurface depth, thickness, and distance to groundwater aquifers (IEA-GHG 2009), 3D thematic models were built in the Leapfrog platform using lithology data from wells. These models were generated using the Vein type software tool. For the depth of the Irati Formation (Fig. 5), all intervals immediately above the formation were selected, and then the model was generated. For the Irati Formation thickness model (Fig. 6) were used the selected wells intervals of the formation itself. For a model of the distance of the CO 2 reservoir until the aquifer systems in the area (Bauru, Serra Geral, and Guarani), it was considered the deepest of them: the Guarani Aquifer, so using the lithology intervals from the top of Irati Formation to the base of the Botucatu and Piramboia Formations (Fig. 7). The depth of the Irati Formation increases in the western direction towards the depocenters of the basin and can reach more than 2,800 m (Fig. 5). The Irati Formation thickness varies from 22 to 64 m in the wells inside the study area (Fig. 6). In nine wells the Irati Formation thickness has shown outlier values from 94 to 265 m. The respective Irati Formation intervals in the database include some diabase intervals. In the case of these outliers, the data were reclassi ed considering only the intervals above the diabase. The distance between the top of the Irati Formation and the base of Guarani Aquifer increases in the west direction and can reach more than 1,400 m outside the state of Sao Paulo in the east of the Mato Grosso do Sul state (Fig. 7). 3D IMPLICIT GEOLOGICAL AND STRUCTURAL MODELING Regional-scale 3D geological and structural model The 3D geological implicit modeling was built initially on a regional-scale using the previously interpreted stratigraphy data of gamma-ray log, comprising data of geological formations and groups, inside the limits of the study area (Fig. 2). The following 12 geological units were considered in the model, stratigraphically from base to top: Paraná Group, Itararé Group, Rio Bonito Formation, Palermo Formation, Irati Formation, Serra Alta Formation, Teresina Formation, Rio do Rasto Formation, Piramboia Formation, Botucatu Formation, Serra Geral Formation, and Bauru Group. Data from Corumbataí Formation were grouped with Teresina Formation data. The geological contacts from the CPRM geological map were used together with the well data in the East region of the study area, where the geological formations crop out (Lopes et al. 2004). The wells intervals were individually selected and the contact surfaces between each geological formation were generated through the Deposit type software tool. Detailed descriptions of the cited software tools are available on the Seequent website (https://help.leapfrog3d.com/Geothermal/). Avoiding any bias or subjective interpretations, the lithological model was exclusively based on well data, with no use of trends or other software arti ces. This approach was useful also to identify possible abrupt changes in the stratigraphic sequence that could indicate fault displacements.
The structural model was generated from the surface fault traces (Zalán et al. 1990;Lopes et al. 2004) that have been transformed into fault surfaces assuming a vertical dip for all faults. This assumption is based on a characteristic of extensional regimes, which is the dominance of steeply deep normal faults (Etheridge et al. 1985), following the evolution of the Paraná Basin (Milani and Ramos 1998;Milani and Thomaz Filho 2000). The initial analysis for the site selection looked through regions with less incidence of fractures or faults, avoiding possible CO 2 leaks. The surfaces of the geological contacts were rst generated from the contact points of the formations, and then after the fault system was activated in the software generating fault blocks. Due to a large number of faults and lineaments, a process with an empirical approach was applied to activate some faults and analyzing the displacement of blocks case by case. The goal was to locate possible structural traps for CO 2 storage. The nal con guration of the structural model generated seven fault blocks (Fig. 8), considering six faults: 1) The Guaxupé Fault with an approximate N60E direction; 2) The Mogi Lineament with an E-W direction; 3) The São Jerônimo Fault with an approximate N40W direction; 4-5) The Santo Anastácio and Guapiara Faults both with a N60W direction, and 6) A local fault with N10E direction limited by the previous two faults.
A structural and stratigraphic trap was identi ed in Fault Block 4, where there is a structural high with the Irati Formation rocks in contact with the above mudstone-dominated Serra Alta and/or Teresina Formations (Figs. 9 and 10). Inside the domain of Fault Block 4, there is no incidence of any other faults that could allow some possible CO 2 leakage. The Irati Formation in the Fault Block 4 has just over 1,800 km², an average thickness of 38 m, an average depth of 2,640 m, and it is at an average distance of 920 m from the base of the Botucatu Formation (Guarani Aquifer), which is a safe distance because it is greater than the distance of 588 m that was considered by Davies et al. (2012) as the maximum for the propagation of fractures in hydraulic fracking processes, based on thousands of fracture operations performed in the shales of Marcellus, Barnett, Eagle Ford among other elds. The sequence immediately overlying the Irati Formation is the shales and mudstones of the Serra Alta Formation, and then over it, the mudstones interlaminated with ne sandstones of the Teresina Formation (Holz et al. 2010). Considering only the Serra Alta Formation, the caprock thickness is around 65 m, if considered both, Serra Alta and Teresina Formation, the caprock thickness is more than 750 m. Local 3D geological model Two wells are located in Fault Block 4: the 1-TI-1-SP and the 2-TB-1-SP. A possible CO 2 reservoir geological model was considered within the Fault Block 4 re ning the Irati Formations domain (meshes), considering the previously de ned A/B/C to H units (Fig. 12), and a diabase dike that was intercepted by the 1-TI-1-SP well (from 2,854 to 3,140 m). the dike was interpreted as subvertical with the N55W direction, which is the main direction of magnetic lineaments in the area, based on the reduced total eld image of magnetic airborne survey (Lopes et al. 2004), and also because it is the preferred direction for intrusions (Zalán et al. 1990). Therefore, the diabase dike divides the Fault Block 4 in two (Fig. 11). The site considered for the possible CO 2 reservoir is on the eastern side of the diabase dike (Fig. 11). Thus, the CO 2 reservoir is bounded by the Mogi Lineament to the north, by the Guapiara Fault to the northeast, by the diabase dike to the southwest, and by a local N10E fault to the southeast, with an area of approximately 1,200 km².

LOCAL ASPECTS OF THE CO 2 RESERVOIR
Porosity and permeability Table 1 shows the porosity (Ф) and permeability (K) calculated from 2-TB-1-SP well log data for each geological unit of Irati Formation. Porosity derived from the sonic log was based on an equation from Richardson and Tassinari (2019). The values computed for porosity were used to predict the permeability of the layers based on the rede ned equations for shale and limestone units (Richardson and Tassinari 2019).
The porosity of the A/B/C and F shale units varies from 8.0 to 16.7%, and the permeability from 0.542 to 75.11 mD, while the porosity of the E and H black shale units have values from 6.1 to 9.2%, and permeability values from 0.088 to 1.383 mD. In the D and G limestone units, the porosity is between 12.6 and 69.0% and the permeability is between 0.567 and 28.41 mD.

CAPACITY ESTIMATION AND CLASSIFICATION
Among the current CO 2 storage resources, the shale formations have been shown to have the greatest potential, in particular the black shales with TOC content higher than 2% (Levine et al. 2016). The available data of TOC in Irati Formation shale intervals of the 2-TB-1-SP well varies between 0.52 to 9.62% in black shale samples ( Table 2). Although the H black shale unit presents higher TOC values (8.45 and 9.62%) than the E unit (0.52 to 7.36%), this last one is considered for capacity calculation due to its greater thickness (20.00 m against 2.00 m) ( Table 1 and Fig. 3).
The DOE NETL's equation for shales was utilized for the estimation of the CO 2 mass storage (Levine et al. 2016). Nevertheless, several challenges regarding the storage resource estimation of shale reservoirs are pointed out by Azenkeng et al. (2020). The key factors that control the CO 2 storage in organic-rich shales are matrix pore spaces, and natural and induced fractures, and it is very di cult to differentiate them (Azenkeng et al. 2020). Table 3 summarizes the variables that were used for calculating the capacity of the E unit shale of Irati Formation inside the Fault Block 4. The calculated porosity of 6.1% from the 2-TB-1-SP well for the E unit (Table 1) refers to natural fractures and pore spaces. The approach assumed for this initial assessment of the Irati Formation shale reservoir was quite conservative, although this study considers hydraulic fracturing to generate fractures and therefore increases porosity, only natural porosity data was used in the calculation. The CO 2 supercritical density of 842.3 kg/m 3 was calculated from the geothermal gradient of 20.4 ºC/km (Gomes 2009)  The volume was directly obtained from the 3D solid of the E black shale unit from the 3D implicit geological model (Table 3). No e ciency factor for the area (E A ) was been applied since some legal surface constraints have been applied previously in the site selection, and by the structural constraints of the limiting faults. The e ciency factor for thickness (E h ) has also not been applied. Although almost all geological parameters such as porosity, temperature, thickness, and TOC come from a single exploration well (2-TB-1-SP), the volume has a considerable level of reliability to be derived from a 3D geological model that was constructed considering also the data of all other 31 wells, geological mapping, topography, and local and regional structures together.
The two e ciency factors from DOE NETL's equation (E Ф and E s ) were assumed from the simulations of Myshakin et al. (2018), which was based on 60 years of CO 2 injection. The most conservative parameters of P 10 probability values of 0.15 for E Ф , and 0.11 for E s , was considered, reaching 1.85 billion tons of total capacity for the E unit (Table 3). If considered the P 90 probability values of 0.36 for , and of 0.24 for from Myshakin et al. (2018), the total capacity would be 4.44 billion tons.
For comparison with other organic-rich shale formations, in the Devonian shale in Kentucky, a capacity of 27.7 Gt was estimated at least 304 m deep and 15 m thick (Nuttal et al. 2005), while in the Marcellus Shale a total theoretical capacity of 171.2 Gt was estimated for a depth greater than 915 m (Godec et al. 2013a). Bachu et al. (2007) proposed a classi cation considering technical and economic aspects in four categories: theoretical, effective, practical, and matched capacity, according to a gradual level of uncertainty of storage potential. The present study is the initial assessment of the shales of Irati Formation as a CO 2 reservoir, taking into account mainly geological aspects and aiming to provide the basis for a future numerical model study. We assume a Theoretical classi cation since only a few parameters have already been well de ned, such as the seal, depth, geothermal gradient, temperature, and distance to aquifers, which will not vary with the progress of research, all other parameters still need to be re ned.

Discussion
Site selection Table 4 compiles the main parameters used in this study for the determination of the site location of a potential CO 2 reservoir, and evaluation of this reservoir. Concerning seismicity, location in fold belts, complex stratigraphy, and the presence of reservoir-seal pairs for the location of a CO 2 reservoir site (IEA-GHG 2009;Smith et al. 2011;Miocic et al. 2016), the Paraná Basin in Southeastern Brazil presents preferred or favorable results in all aspects. Although in this study, limitations about legal aspects, such as the state legislation of Paraná that prevents hydraulic fracturing had been applied, other topics, like social and environmental issues (Ciotta et al. 2020), still need to be addressed in that region.
The 3D implicit modeling is more robust and has better representativeness of structure and geology than other surface-based methods, improving the modeling accuracy (Cowan et al. 2003;Wellmann and Caumon 2018). The methodology of implicit modeling to build thematic models for depth and thickness of Irati Formation sequence, and for evaluating the distance to protect groundwater, the Guarani Aquifer system, in the study area scale, proved satisfactory and with fast results of easy interpretation (Figs. 5, 6,   and 7). The 3D structural model divided the study area into seven fault blocks and de ned a structural high (Fault block 4) that conditions a structural and stratigraphic trap (Figs. 8, 9, and 10). These geological aspects of the selected site are favorable for storing carbon because, in addition to having the sealed siltstones of Serra Alta and Teresina formations above, these siltstones stand side by side with the Irati Formation rocks also acting as lateral barriers for possible gas leaks. The 3D geological models were generated based only on well data and surface or regional geological and structural mapping data, and in that site may exist faults or other structures not visible in the current work scale. Seismic data should be incorporated into that model in future studies to better detail the structural framework. simple regression, the CO 2 site location could support up to ve times more than the current capacity considering a regional industrial park installation in a long-term horizon of about 68 years, taking into account also the location and the privileged infrastructure of the region. A possible disadvantage concerning the large shale area presented, when projected on the surface, would be the need for extensive terrestrial monitoring, as discussed by Brennan and Burruss (2006) in their comparative study with coal seams versus saline aquifers or depleted reservoirs.

Physical and chemical trapping mechanisms
To maximize storage capacity and the buoyant, the CO 2 is generally injected as a supercritical uid and is trapped by different physical and chemical trapping mechanisms according to the geological characteristics of each site. In a CO 2 shale reservoir, an adsorption trapping mechanism can permanently store the CO 2 and the shale itself could act as an impermeable barrier and prevent leakage (Kang et al. 2011;Mohagheghian et al. 2019). In addition, to ensure CO 2 retention for a long time, there are immediately above the Irati Formation the shale layers of the Serra Alta Formation, with more than 70 meters thick, which will function as a low permeability sealant layer.
The present study assumes a CO 2 reservoir composed of a unique black shale layer (E unit), however, future studies may consider the entire Irati Formation as a unique hybrid reservoir, contemplating limestones intercalations, with different trapping mechanisms and injectivity conditions. Nevertheless, this evaluation is much broader and requires more extensive detailed characterization studies, which are outside the scope of this research. The focus of this work was to delimit spatially the potential units into the Irati Formation in a 3D environment and provide the basis for subsequent evaluations, what we believe we have achieved. The reached results will be the basis of subsequent studies of mineral characterization of each Irati Formation subdivision de ned here and numerical simulations.
Concluding Remarks 1. The Paraná Basin in Southeastern Brazil meets most of the international regional requirements in terms of seismicity, no fold belt, uniform stratigraphy without complex lateral variations, and the presence of reservoir-seal pairs in multi-layered systems.
2. The application of 3D implicit modeling to build thematic models for depth, thickness, structural geology, and distance to protect groundwater, proved satisfactory, and with fast results of easy interpretation. The thematic 3D models were used integrated to de ne the CO 2 reservoir site.

Supplementary Information
Instructions to visualize the 3D model.
To visualize the surfaces in Leapfrog Viewer (as in Fig. 10 Total capacity estimated to be much larger than the total amount produced from the CO 2  Geological map of the study area with local structures (Lopes et al. 2004), and main regional structures (Zalán et al. 1990). Location of the schematic geological section (A'-A") of