Fault‐controlled dolostone geometries in a transgressive–regressive sequence stratigraphic framework

This study investigates the geometries of fault‐controlled dolostone geobodies and their structural and sequence stratigraphic controls, which provide new insights for the prediction and production of fault‐controlled dolomitized hydrocarbon reservoirs. A very thick succession (>1600 m) of Aptian–Albian shallow‐marine carbonates of the Benassal Formation that crop out in the Benicàssim area (Maestrat Basin, eastern Spain) is partly replaced by dolomite, resulting in dolostone geometries ranging from massive patches to stratabound bodies. Detailed mapping, systematic logging and correlation were carried out to characterize the structural, sedimentary and sequence stratigraphic framework of the area and to constrain the principal controls on the full‐range of dolostone geometries. The results show that carbonate sediments accumulated in a half graben stacked in three transgressive–regressive sequences. Large‐scale massive dolostone patches (with up to kilometre extension) formed near large‐scale faults indicating that they acted as entry points for warm dolomitizing fluids into the basin. These dolostone patches laterally pass to large stratabound bodies that extend for long distances (at least 7 km) away from feeding faults, forming a continuum. The presence of a regional unconformity and a clastic fine‐grain low‐permeability unit (Escucha Formation) on top of the Benassal Formation likely constrained the dolomitization fluids to an up to 580 m thick interval below the base of the Escucha Formation. Thus, only limestones within this interval, corresponding to the two uppermost transgressive–regressive sequences, were dolomitized. There is a clear relationship between the stratigraphic framework and the preferred replaced beds. Dolomitization preferentially affected sediments deposited in inner to middle ramp settings with predominant wackestone to packstone textures. Such facies are laterally most abundant in the east of the study area (i.e. basinward) and vertically in layers around the maximum flooding zone of the top sequence, which is preferentially affected by dolomitization.


INTRODUCTION
Carbonate reservoirs are estimated to contain 60% of the world's remaining oil reserves (Roehl & Choquette, 2012). They typically exhibit a heterogeneous distribution of petrophysical properties, and consequently are challenging to explore, develop and produce. The main difficulties are associated with the fact that the distribution of porosity and permeability, and thus reservoir quality, in carbonate rocks is strongly affected by a variety of post-depositional diagenetic processes such as compaction, dissolution, cementation and mineral replacement (e.g. Tucker, 2009;James & Jones, 2015) that often occur at different times in the burial history resulting in a complex paragenetic evolution. The lateral and vertical extension of diagenetic products (cements, dissolution vugs, stylolites, etc.) are very difficult to characterize in three dimensions using subsurface data, especially since drill cores only represent a very small fraction of the reservoir (e.g. Howell et al., 2014). Dolomitization (i.e. the replacement of calcite with dolomite) is one of the most important diagenetic processes affecting limestone successions. It occurs in a wide range of environments and modifies the pore space geometry and volume as well as permeability (e.g. Allan & Wiggins, 1993;Warren, 2000;Machel, 2004). Dolomitization is a widely studied diagenetic process for understanding controls on fluid flow in carbonates (e.g. Qing & Mountjoy, 1992, 1994Whitaker et al., 2004) because, unlike other diagenetic processes such as cementation or dissolution, the effects of dolomitization are easy to visualize in many cases on outcrops and in drill core due to the colour and texture change associated with the mineral change (e.g. Humphrey et al., 2019). Outcrop studies on dolomitization are thus essential for understanding how diagenesis impacts reservoir quality, and to unravel the key controls on diagenetic fluid flow producing rock alterations (e.g. Mansurbeg et al., 2016).
There are several types of dolomitization (for comprehensive reviews see Hardie, 1987;Warren, 2000;Machel, 2004;Whitaker et al., 2004), among which fault-controlled high-temperature dolomitization (HTD) is particularly relevant (see review by Davies & Smith, 2006) both from a scientific and economic point of view. Hightemperature dolostone normally forms when warm fluids are transported up faults and invade the host limestones, resulting in the replacement of calcite with dolomite as well as in the formation of dolomite cement. The flow of high-temperature dolomitizing fluids in limestone successions can result in a variety of replacement geometries, from massive patches next to faults (e.g. Wilson et al., 2007;Sharp et al., 2010;Dewit et al., 2012), to stratabound geobodies that extend for long distances away from them (e.g. Sharp et al., 2010;Lapponi et al., 2011;Mart ın-Mart ın et al., 2013Dewit et al., 2014;Gomez-Rivas et al., 2014;Gasparrini et al., 2017;Hollis et al., 2017).
The majority of published studies dealing with fault-controlled dolostones focus on the stratigraphic, petrographic and geochemical characterization of outcrop analogues, with the aim of providing an overall picture on how the dolomitization process is controlled by depositional facies, diagenesis and structures. Recent representative studies include the analysis of outcrops in the Zagros belt in Iran (e.g. Sharp et al., 2010;Lapponi et al., 2011), the Vasque-Cantabrian Basin in Spain (e.g. L opez-Horgue et al., 2010;Shah et al., 2010;Dewit et al., 2012Dewit et al., , 2014Gasparrini et al., 2017), the Iberian Chain in Spain (e.g. Nadal, 2001;Mart ın-Mart ın et al., 2013Corbella et al., 2014;Gomez-Rivas et al., 2014), the Oman mountains (e.g. Vandeginste et al., 2013;Beckert et al., 2015), North America (e.g. Azmy et al., 2008;Diehl et al., 2010;Lavoie et al., 2010;Slater & Smith, 2012;Morrow, 2014; see review by Davies & Smith, 2006; and references therein), the Alps Jacquemyn et al., 2015), the Apulia Carbonate Platform (Rustichelli et al., 2017) and the Suez Rift (Hollis et al., 2017), among many others. These studies highlight a series of aspects that impact fault-controlled dolomitization, including the characteristics of fault zones that act as entry points for dolomitization fluids, the permeability and composition of host limestones (associated with depositional facies and pre-dolomitization diagenesis), and the properties of fracture networks that can facilitate or prevent dolomitization by acting as either conduits or barriers for dolomitization fluid flow.
Despite the recent advances on the understanding of high-temperature dolomitization, there are still fundamental open questions on how this process operates and controls the volume, geometry and distribution of the resulting diagenetic geobodies. Key questions to be addressed include: (i) what is the spatial relationship between faults and dolostone geobodies at multiple scales; (ii) which, if any, sedimentary facies and stratigraphic sequences are preferentially replaced; (iii) to what extent do pre-dolomitization diagenetic products control replacement; and (iv) what is the role of low-permeability sedimentary units as seals for dolomitizing fluids?
These questions were tackled through the study of the high-temperature dolostones of the Benic assim outcrop analogue (Maestrat Basin, eastern Spain), with the overarching aim of providing an improved framework for the prediction of high-temperature dolomitization geometries in shallow-marine carbonates. In this area, a very thick pile (>1600 m) of Lower Cretaceous shallow-marine carbonates of the Benassal Formation are well-exposed. The limestones are partially dolomitized in close association with regional faults. The Benic assim area has been the focus of several recent studies (e.g. Mart ın-Mart ın et al., 2013Corbella et al., 2014;Gomez-Rivas et al., 2014), which focused on dolostones with stratabound distribution, and dealt with the analysis of depositional facies, diagenetic evolution and source of the dolomitization fluids. These studies, however, did not tackle controls on variable dolomitization geometries across different fault blocks and in relation to the sedimentary facies and stratigraphic sequences.

Structure, transgressive-regressive sequences and lithofacies
The Benic assim area is located in the southeastern part of the Maestrat Basin (eastern Spain) (Fig. 1). This rift basin developed during the Late Jurassic-Early Cretaceous syn-rift cycle of the intraplate Iberian Rift System (Salas & Casas, 1993;Salas et al., 2001). The Maestrat Basin was subsequently inverted during the late Eocene-early Oligocene Alpine Orogeny (Guimer a et al., 2004), and then experienced extension during the late Oligocene-Miocene to form the present-day western margin of the Valencia Trough (Roca & Guimer a, 1992).
The study area is characterized by two systems of extensional faults that intersect one another: (i) the NNE-SSW striking and southeast-dipping Benic assim Fault system; and (ii) the east-west striking and south-dipping Campello and Juvell us Fault system (Mart ın-Mart ın et al., 2013) (Fig. 1). According to the latter authors, the Benic assim and Campello faults are inherited from the Hercynian Orogeny and were reactivated as extensional faults during the Mesozoic. During the Late Jurassic and Early Cretaceous rifting more than 2 km of syn-rift deposits were accommodated in the Benic assim area. A significant proportion of these sediments constitute the Late Aptian-Early Albian Benassal Formation, which represents a limestone succession dominated by shallow-marine carbonates with a present-day thickness of more than 1600 m (Mart ın-Mart ın et al., 2013).
The Benassal Formation was deposited on a carbonate ramp very rich in orbitolinids, rudists, corals and algae with scarce siliciclastic input (Tom as et al., 2008;Mart ın-Mart ın et al., 2013). Based on the analysis of the sedimentary succession cropping out in the Orpesa Range fault block (see Fig. 1), the latter authors defined 12 lithofacies types (LFT; Table 1) and three transgressive-regressive (T-R) sequences (Fig. 2). The transgressive systems tract (TST) of sequence I is characterized by basinal marls, whereas the regressive systems tract (RST) is marked by peloidal and orbitolinid grainstones, and peloidal and coral facies. The latter deposits are topped by a thick unit of rudist floatstones that represent the most regressive deposits of sequence I. The latter unit shows an erosional surface with palaeokarst at its top, revealing subaerial exposure (sequence boundary A in Fig. 1). The TST of sequence II initiates with a few metres of sandy limestones displaying tidal bundles and cross-bedding. Above, most of the TST is characterized by orbitolinid-rich and spicule-rich muddy facies. The RST of sequence II is distinguished by coral-bearing and rudistbearing limestones stacked in a progressive pattern, that alternate with hummocky crossstratified grainstones and peloidal grainstones. The RST ends with a several tens of metres thick unit made up of rudist and Chondrodonta bivalves with floatstone to boundstone textures, the top of which represent the maximum regressive surface (MRS) of the T-R sequence II. The TST of sequence III is mainly formed by nodular bedded bioclastic limestones containing muddrapes and orbitolinids that are overlain by up to a hundred metres of dolostones with several limestone stringers dominantly composed of mud-dominated facies. One of these stringers corresponds to spicule-rich wackestones at the top of which the maximum flooding zone (MFZ) is placed. The bottom of the RST is partially dolomitized and passes upward to ooidal-rich,  bioclastic-rich and peloidal-rich facies. The top of sequence III is faulted and covered, and thus the boundary with the Escucha Formation was not recognized in the Orpesa Range. Within the Maestrat Basin the top of the Benassal Formation is bounded by a regional subaerial unconformity represented by a palaeokarst that is overlain by up to 90 m of fine-grained tidal claystones and sandstones of the Escucha Formation (Salas et al., 2001).

Fault-controlled dolostones
The Benassal Formation is partly dolomitized throughout the Benic assim area. This replacement, as well as locally associated galena-sphalerite  Fig. 1) indicate that the dolomitization of the Benassal Formation host limestones took place after a major stage of mechanical and chemical compaction, and thus during burial. The latter authors described four types of dolomite: (i) most of the replacive mass is formed by planars to non-planar dolomite crystals with characteristic cloudy appearance and mimetic fabric ( Fig. 3A to C); (ii) mimetic dolomite appears frequently recrystallized in close association with fractures and permeable areas, forming planar-s to non-planar replacive dolomite crystals with cloudy cores and clear rims (Fig. 3D); (iii) planar-e crystals that formed after precipitation of clear dolomite rim overgrowths, replacive cores (i.e. overdolomitization) (Fig. 3E); and (iv) saddle dolomite filling fractures, The dolomitization fluids are interpreted as hydrothermal because their estimated temperatures (exceeding 80°C and at least up to 110°C) are considerably higher than host rock temperatures according to the geothermal gradient. These would have ranged between ca 30°C and 60°C from the uppermost to the lowermost replaced limestones, respectively . This study also revealed that the dolomitization fluids were likely derived from infiltrated seawater that exchanged with basin fluids and basement rocks, resulting in up to 23% eq. wt. NaCl salinity. Geochemical and mass-balance calculations suggest that interstitial fluids alone could not account for the volume of dolomite, and that only a pervasive fluid circulation mechanism based on thermal convection could have provided a sufficient volume of fluids to produce the replacement reaction during the Late Cretaceous . Two-dimensional (2D) fluid flow and reactive transport simulations presented in Gomez-Rivas et al. (2010) and Corbella et al. (2014) suggest that the fluid flow system that produced the large-scale stratabound dolostone geobodies in this area had a strong lateral component and probably lasted for about a million years. There is no evidence from field and petrographic observations that the fluid that caused replacive dolomitization circulated at high pressure.

Methods
Field mapping and logging were used to constrain: (i) the depositional lithofacies of the Benassal Formation and their distribution; and (ii) the distribution of dolostones and their association with faults, lithofacies and stratigraphic sequences (Figs 1 and 4 to 7). The vertical and lateral distribution of lithofacies was based on 19 logs (with a total logged thickness of about 6000 m) that cover the three major faulted blocks of the study area (Orpesa Range, Ferradura and Cingle del Morral). Thirteen of these logs are reported here for first time (logs 6 to 14 and 16 to 19) and complement those published by Mart ın-Mart ın et al. (2013,2017) in the Orpesa Range (logs 1 to 5) and Ferradura (log  15) blocks, respectively (Figs 4 to 7). Correlation between logs and faulted blocks was based on photographic panels, key surfaces, bedding traces and characteristic facies. The correlation is partially calibrated by ammonite specimens (see Mart ın-Mart ın et al., 2013 andGarcia et al., 2014). The datum for the stratigraphic correlation between faulted blocks is placed at the base of the Escucha Formation following Salas et al. (2001).
Fifty rock samples including limestones and dolostones were collected during field work for thin section description to determine rock textures and components, and thus for depositional lithofacies classification. The classification of limestone textures follows Dunham (1962) and Embry & Klovan (1971). A total of 15 lithofacies were considered including those defined by Mart ın-Mart ın et al. (2013) (LFT 1 to 12) and three additional types defined in this study (LFT 13 to 15) (see summary in Table 1; Figs 8 and 9).
The Benassal Formation is arranged into three T-R sequences following the sequence stratigraphy framework defined in the study area by  The geometries of dolostones are recognized based on their dimensions and spatial association with faults and bedding. The areal extent of the dolostone bodies was assessed using conventional field mapping and virtual outcrop models (VOs). Dolostone bodies of eleven selected locations were ground checked with emphasis on the transition between different geometries, and their potential relationship with depositional facies, T-R sequences, fault systems, fractures and stylolites. Light Detection and Ranging (LiDAR) and an unmanned aerial vehicle (UAV) were used to collect geospatially constrained field data in order to build high-resolution VOs of three selected outcrops (e.g. Buckley et al., 2008Buckley et al., , 2010Buckley et al., , 2013). The LiDAR model was a RIEGL VZ-2000 (RIEGL Laser Measurement Systems GmbH, Horn Austria) equipped with a Nikon D100 camera (Nikon, Tokyo, Japan) and the UAV was a DJI Phantom 3 Advanced (DJI, Shenzhen, China). Five local high-resolution models were used to compare in detail dolostone geobodies geometries in different outcrops.

Sequence stratigraphy and lithofacies evolution
The characteristics of lithofacies evolution and sequence stratigraphy framework significantly vary across the three study fault blocks (Orpesa Range, Ferradura and Cingle del Morral). Accordingly, they are summarized for each fault block below (see Table 1 and Fig. 2). The new section (log 6) appears faulted at its bottom and comprises 185 m of dolostones replacing both mud-dominated and grain-dominated facies. Accordingly, limestone stringers within dolostones, which range in thickness between 10 m and 30 m, are composed of bioclastic wackestones (LFT11), rudist floatstones (LFT7), coral limestones (LFT6) and peloidal grainstones (LFT3). The top of the section is covered by Quaternary deposits, and thus the boundary with the Escucha Formation is not exposed in the area. Based on field observations, however, the covered succession is assumed to be in the range of several tens to a few hundred metres (Fig. 5).

Ferradura
The Ferradura area represents the footwall block of the Benic assim Fault and the hanging wall block of the Campello Fault (Figs 1 and 6). The thickness of the Benassal Formation measured in this area is 1215 m (logs 12 to 18), although the total thickness is assumed to be one to several hundred metres higher as the bottom and top of the Benassal Formation are covered by vegetation and Quaternary deposits (Fig. 6).   Transgressive-regressive sequence I is 289 m thick and is well-exposed along the road CV-148 (log 12 ; Figs 1 and 4C). The TST is characterized by ochre-coloured marls with echinoids and terebratulida brachiopods (LFT1) with several metre-thick intercalations of peloidal and orbitolinid grainstones (LFT2) rich in glauconite and quartz grains together with minor mica (Figs 6, 8A and 8B). The MFZ of sequence I is placed at the top of the first and thickest marly unit. Above, the RST is topped by a thick bed-set of rudist floatstones (LFT7) containing minor nerineid gastropods. The rudists appear mostly fragmented, locally forming bouquets in life position, and are dominated by requieniids (Toucasia sp.). The rudist floatstones, which are covered and faulted towards the top, are the most regressive deposits of sequence I, and thus the MRS and the boundary between sequences I and II is placed at the top (Figs 4 and 6).
Transgressive-regressive sequence II is about 548 m thick (logs 13 and 14; Figs 4C and 6). The TST is characterized by the predominance of muddy facies dominated by orbitolinid-rich facies (LFT4) that pass upward to several tens of metre-thick bed-sets of bioclastic wackestones to packstones (LFT11). Intercalated between the latter facies appears a bed-set composed of very fine-grained packstones with sponge spicules, calcispheres and chert nodules (LFT9) (Fig. 8C). The MFZ is placed at the top of the spicule-rich facies, which are interpreted as the deepest water deposits of sequence II. Above, the lower part of the RST comprises peloidal (LFT3), bioclastic (LFT 11) and a several tens of metresthick unit of foraminifera and green algae packstone and grainstone (LFT13) facies that appear dolomitized and covered by Quaternary deposits (Figs 6 and 8C). The RST of sequence II ends with an ca 80 m thick unit of massive and several metre-thick beds of rudist and Chondrodonta floatstone to boundstones (LFT7). The rudist bivalves are dominated by large requieniids (Toucasia sp.) and radiolitids specimens. The MRS of sequence I is placed at the top of this rudist bed set (Fig. 6).
Sequence III, which is 469 m thick, starts with a TST that shows a 115 m thick unit of dolostones replacing mud-dominated and graindominated facies as indicated by several stringers made up of coral, bioclastic and peloidal limestones (logs 14 to 16; Figs 4C and 6). The dolostones are topped by orbitolinid rudstones (LFT4) with abundant chert nodules, which are considered the most transgressive deposits and at the top of which is placed the MFZ. Above, the RST is dominated by a set of several metrethick beds made of medium to coarse-sized peloidal grainstones (LFT3) (Logs 16 to 18; Figs 4C and 8E). The top of the T-R sequence III, and thus the top boundary of the Benassal Formation, is covered, and not recognized in the Ferradura fault block. However, the presence in the surroundings of loose and fragmented beds of ochre-coloured and quartz-rich sandstones typical of the Escucha Formation suggests that the boundary between both formations is nearby (Fig. 1).

Cingle del Morral
The Cingle del Morral is located at the footwall block of the Campello Fault and the Benassal Formation is only 674 m thick here (log 19; Figs 1 and 7). This thickness represents 456 m measured in outcrop and 234 m more of cover calculated assuming a constant bed dip until the top boundary with the Escucha Formation is reached (Figs 4D and 7). The bottom boundary of the Benassal Formation with the underlying Villarroya de los Pinares Formation is characterized by a sharp lithological change from wellbedded and cross-laminated ooidal grainstones below to marly facies (Fig. 7).
Sequence I, which is 232 m thick, shows a TST dominated by marly packstones to rudstones rich in coarse-sized skeletal components, echinoderms and coral fragments (LFT14) that gradually pass to ochre-coloured marls with echinoids, terebratulid brachiopods and belemnites (LFT1) (Figs 7 and 8F to H). These transgressive marls include the ammonite species Epicheloniceras sp. and Caseyella sp. indicating a Late Aptian age. The MFZ is placed at the top of the marly unit, which represents the most transgressive deposit of sequence I. Above the marls, the TST is characterized by a ca 20 m thick set of centimetre to decimetre limestone beds exhibiting poorly sorted packstone to rudstone textures rich in echinoids and sponges (LFT14). The latter limestones show erosive bases, cross-bedding and thin intercalations of grey marls, and form coarsening and thickening up cycles. Together with the echinoderms and sponges, these limestones are abundant in mud pebbles and mud drapes, terebratulid and rhynchonellid brachiopods, corals, oysters, pectinids, gastropods, belemnites, ammonites and other unidentified components. These limestones pass progressively upward to grey marls containing coral colonies in life position, sponges and echinoids (LFT15), a facies equivalent to those described in detail by Bover-Arnal et al. (2012) (Fig. 8I to J). The coral colonies, which reach up to one metre in height, are domal and tabular shaped and appear as individuals or growing one on top of the other. The corals show abundant lithophaga bivalve borings and their tops frequently are the substrate for sponges, oysters and other undifferentiated organisms. Towards the top of this unit the corals considerably increase in abundance with respect to the marls forming patches of several metres (i.e. bioherms) between the fine sediment (Fig. 8J).
The RST ends with a 20 m thick bed set of bioclastic wackestone to packstones with abundant coral fragments (LFT11) topped by 8 m of rudist limestones exhibiting floatstone to boundstone textures (LFT7). The rudist bivalves are dominated by large requieniids (Toucasia sp.) and radiolitid specimens, occasionally forming bouquets in life position. Together with rudists, Chondrodonta bivalves are locally abundant, forming the main skeletal component of the limestones. The latter floatstones, which represent the most regressive sediments of sequence I, show an erosive surface at their top where the MRS is placed (Fig. 6).
Sequences II and II are not defined, because the upper part of the succession is mostly covered. Sequence I, however, is overlain by a 20 cm thick conglomerate made of centimetresized quartz pebbles that represents the sole evidence of siliciclastic sediments in the Cingle del Morral section (Fig. 8J). This siliciclastic deposit is considered time-equivalent to those of LFT8 that overlie sequence I in the Orpesa Range (log 3; Fig. 4). The rest of the succession is characterized by miliolid and orbitolinid packstones with abundant green algae and aggregate grains (LFT13), and bioclastic (LFT11) and peloidal (LFT3) facies. The latter are laterally dolomitized at the outcrop scale, forming thin dolostone beds between the host limestones. Following the dolostones, the succession is characterized by several bed-sets of limestones with coral fragments and, less frequently, in life position (LFT6) that pass upward to rudist floatstone (LFT7). These bed-sets are rich in ferruginized grains and are locally capped by mud cracks (Fig. 8L). The uppermost part of the outcropping section is mostly replaced by dolostones that show intercalation of limestone stringers made up of peloidal (LFT3) and rudist (LFT7) facies. The top of the section, and thus the top of the Benassal Formation, is covered although the overlying Escucha Formation crops out further to the north. There, the Escucha Formation is made up of cross-laminated quartzrich sandstones and grey marls.

Dolostone geometry and distribution
Dolostones extensively replace limestones of the Benassal Formation across the Benic assim halfgraben and are especially abundant in the hanging wall blocks of the Benic assim and Campello faults, particularly south of their intersection (Fig. 1). Two main types of dolostone geobodies are distinguished in the Benic assim area based on their dimensions and spatial association with faults and bedding: (i) fault-controlled massive dolostone patches (Figs 9E, 9F and 10); and (ii) stratabound dolostones (Figs 1, 10 and 11). Both types are described below.
1 Patches of massive, non-stratabound dolostones crop out close to faults at various scales, from metre-scale to kilometre-scale faults ( Fig. 10A and B). In such cases dolomitization fronts cross-cut bedding. The massive dolostone patches have widths ranging from 300 to 1600 m and thicknesses varying between 20 m to at least 200 m. Patches of massive dolostone with a pipe-like shape (sensu Sharp et al., 2010) follow fault traces upward and appear rooted in stratabound dolostone bodies.
2 Stratabound dolostones are defined as dolostone bodies replacing specific beds (Figs 10A and 11). They are typically kilometres wide and are a few metres to hundreds of metres thick. Massive patches are typically connected with stratabound dolostone geobodies away from the feeding faults, in a continuous way so that the stratabound bodies are terminations of the massive ones (Fig. 10).
The two main types of dolostone geobodies described here are continuous and present the same petrographic characteristics and diagenetic evolution (see Mart ın-Mart ın et al., 2015. Dolostones are laterally continuous along the Orpesa Range fault block, from north to south. The largest massive patch in the area, which is spatially associated with the Campello Fault and appears in its hanging wall, is at least 200 m thick and 1400 to 1600 m wide and contains a few limestone stringers (Figs 1, 10A and 10B). This large patch passes laterally to large stratabound dolostone geobodies that act as its termination. The largest stratabound body has a maximum thickness of 150 m and extends for more than 7 km away from the Campello Fault, replacing host rocks of the upper part of the Benassal Formation (Figs 1, 10A and 12). Also in the Orpesa Range fault block, but away from the Campello Fault, some kilometre long faults with a similar strike (for example, the Altos del Señor Fault and Fault D, which strike NW-SE to WNW-ESE) are surrounded by massive dolostone patches (hundreds of metres wide and tens of metres thick) (Fig. 1). A ca 500 m wide massive dolostone patch cropping out in the hanging wall of the Altos del Señor Fault passes laterally to a stratabound dolostone geobody that is 70 m thick and at least 1700 m long (Fig. 5). Stratigraphically, these beds represent the lowermost dolostones in the whole study area. The equivalent beds are not replaced in the north part of the Orpesa Range fault block (i.e. close to the Campello Fault). The bottom boundary of this body is about 550 m below the base of the Escucha Formation (Figs 1, 10A and 12). The stratigraphically uppermost dolostones lie just about 130 m below the base of the Escucha Formation in this fault block. Dolostones in the Orpesa Range fault block replace sediments of T-R sequence II and, in a larger extent, of sequence III (Fig. 5). In the north, where the depocentre is located, only the T-R sequence III is affected by replacement, with a slight preference for sediments deposited in the RST. However, in the southern part of this fault block, the replacement affects sediments of sequence III and the RST of sequence II.
Dolostones in the Ferradura fault block replace limestones corresponding to the middle and upper parts of the Benassal Formation (Figs 1, 6 and 12). The lowermost dolostones in this area crop out about 2200 m away from the Campello Fault and replace limestones at a depth ranging from 750 to 600 m below the top of the Benassal Formation. These sediments correspond to the regressive part of sequence II. The lower part of these dolostones is formed by a few dolomitized beds with individual thicknesses of a few metres. They are in contact with spicule wackestones to packstones (LFT9) that contain silica nodules (Fig. 8B). The upper part of these dolostones is affected by several faults and presents a thickness of about 80 m. Limestone stringers within these dolostone bodies correspond to foraminifera and green algae grainstones (LFT13) (Fig. 6). Dolostones replacing the upper part of the Fig. 12. Correlation of the Benassal Formation across the three fault blocks showing the distribution of dolostones in the stratigraphic framework. The sequences are coloured in blue (sequence I), green (sequence II) and purple (sequence III). The grey coloured zone refers to the vertical dolomitization interval, which represents the lowest and highest occurrences of dolostones within the Benassal Formation. Note that the datum corresponds to the base of the Escucha Formation (top of the Benassal Formation) to highlight the effect of faulting during deposition. Foram, foraminifera; HCS, hummocky cross-stratification; Orbitol, orbitolinid; and Pel, peloidal. Ferradura block are dominated by a thick stratabound dolostone body (115 m thick, 550 m long) that terminates at a fault and contains abundant limestone stringers (Figs 6 and 11B). This lies about 1500 m south of the Campello Fault and corresponds to the transgressive part of sequence III. A small massive dolostone patch (20 m thick) located about 800 m away from the Campello Fault appears stratigraphically higher than the aforementioned stratabound geobody (see patch close to logs 17 and 18 in Figs 1 and 6). This thin massive dolostone patch passes laterally to highly-cemented peloidal grainstone (LFT3) and belongs to the regressive part of sequence III. Limestones cropping out below the base of the Escucha Formation are formed by LFT3 and are only partly dolomitized here. All of the uppermost dolostone bodies in this fault block are bounded at their west side by Fault C, which is a NNW-SSE trending normal fault with a length of 1800 m and an offset of tens of metres. The corresponding beds of these dolostones are not replaced in the footwall block of Fault C except for a small patch (log 14; Figs 6 and 11B). Dolostones in the Ferradura fault block are restricted to a vertical interval whose bottom and top are about 750 m and 70 m below the base of the Escucha Formation (Figs 8E and 12).
Dolostones in the Cingle del Morral fault block replace limestones of the middle part of the Benassal Formation and present a stratabound geometry, extending at least 1000 m away from the Campello fault. They are restricted to a vertical interval ranging from 100 to 400 m below the base of the Escucha Formation (Fig. 12). Dolostones replace several thin limestone beds (decimetre to 3 m thick) of foraminifera and green algae wackestones to grainstones (LFT13) in the lower part of the body, and a relatively thicker unit (50 m) above (Figs 7 and  8L). These dolostones correspond to sequence II.

Dolomitization fronts
Dolomitization fronts are mostly sharp (Fig. 9C to H) but they very occasionally are marked by a transition zone, ranging from 0.5 to 1.0 m, of pure limestone to pure dolostone ( Fig. 9I and J). Dolostone bodies are observed to pinch out at various scales, from tens of metres to centimetres ( Fig. 9C and D). However, most dolomitization fronts are laterally bounded by faults at a range of scales, from major faults with lengths of hundreds of metres to outcrop-scale faults and fractures with lengths of metres ( Fig. 9E to G). Vertically, many dolomitization fronts correspond to bedding planes or bed-parallel stylolites ( Fig. 9G and H). Vertical dolomitization fronts can be composed of a combination of faults, fractures and stylolites in certain cases (Fig. 9C, D, F and G).

Structure of the Benic assim half-graben
Correlation of sedimentary logs of the exposed Benassal Formation in the Benic assim area reveals significant variations in sediment thickness across the three main blocks defined by the Campello and Benic assim major faults (Figs 5, 6, 7, 12 and  13). The most important structure at the time of deposition was the west-east trending Campello Fault, as indicated by the >1600 m thick succession observed in its hanging wall (Orpesa Range block) compared with 700 m at its footwall (Cingle del Morral block) (Figs 7 and 13). This northsouth structural framework is consistent with a half-graben basin with a depocentre immediately south of the Campello Fault. Along the hanging wall of the Campello Fault, the Benassal Formation progressively thins from >1600 m in the north to ca 1250 m in the south, likely resulting in a regional onlap in that direction (Figs 5 and 12).
The east-west structure of the study area was influenced by the north-south trending Benic assim Fault that records a difference of 300 m in the thickness of the Benassal Formation between the hanging wall (Orpesa Range block) and the footwall (Ferradura block) (Figs 5, 6 and 12). The variation in thickness between faulted blocks confirms that the Benic assim Fault was also active during the deposition of the Benassal Formation. This east-west framework indicates that the carbonate platform was tilted towards the east, and thus the basin depocentre was located in the north-east of the study area.
The analysis of the Benassal Formation lithofacies and their stacking pattern throughout the study area reflects a dominant shallow-marine carbonate platform with very scarce terrigenous influx that evolved from basinal to inner platform environments with time (Figs 12 and 14). The depositional system was interpreted by Mart ın-Mart ın et al. (2013) as a ramp-type carbonate platform based on: (i) the excellent lateral continuity of facies belts, which indicates a carbonate ramp with a minimum width in the range of kilometres; (ii) a very similar stacking patterns of lithofacies across fault blocks; and (iii) the lack of step-like structures.

Controls on the geometries of dolostone geobodies
The Benic assim half-graben hosts a series of large-scale dolostone geobodies that constitute an excellent example of well-exposed massive diagenetic alteration. These dolostone bodies crop out in an area that is at least 3.2 km long and 9.0 km wide (Fig. 1). The arrangement, petrographic and geochemical characteristics suggest that these dolostones are high-temperature and closely related to the Campello and Benic assim faults, and were therefore interpreted as fault-controlled (see Geological setting; Gomez-Rivas et al., 2014;Mart ın-Mart ın et al., 2015. Two types of dolostone geometries have been recognized, showing a clear spatial association between them: (i) fault-controlled massive patches, including dolostone pipes; and (ii) stratabound dolostones. Field data suggest that faultcontrolled massive dolostone patches laterally evolve to stratabound bodies, forming a continuum equivalent to the classical Christmas tree pattern at multiple scales (Fig. 10A), but with a very long, lateral stratabound component. Dolostone pipes are rooted in the thickest stratabound dolostones and are spatially relatively close to the massive patches. Such association of geometries is common in most cases of fault-controlled dolomitization in a way that these geometries are typically interpreted as being genetically related to the same dolomitization process (e.g. Davies & Smith, 2006;Wilson et al., 2007;Sharp et al., 2010;Lapponi et al., 2011). However, there are also cases in which such geometries have been formed during different events (Hollis et al., 2017). In either case, the relative extension of fault-controlled massive patches is a key factor for hydrocarbon exploration because good reservoir quality can typically be found close to the transition zone from massive patches to stratabound dolostones (e.g. Wilson et al., 2007).

Structural controls on replacement
Different faults in the study area are interpreted to have acted as feeders or barriers for dolomitization fluids at various scales. Based on results from this and previous studies, faults spatially associated with massive dolostone patches and dolostone pipes are interpreted to have acted as entry points for dolomitization fluids into the basin, in agreement with other studies of fault-controlled dolomitization (e.g. Davies & Smith, 2006;Wilson et al., 2007;Sharp et al., 2010;Lapponi et al., 2011;among many others). Feeding faults in the study area are some of the syn-rift structures, chiefly the Campello Fault and a series of largescale faults with a similar strike (such as the Altos del Señor Fault and Fault D) (see Fig. 1). According to field, petrographic and geochemical data from the area, these faults were likely connected in a way that the dolomitizing fluid came from underlying basin and basement rock units and then was distributed by these structures. Despite the different detachment level of the two dominant fault systems (Antol ın-Tom as et al., 2007), the intersections of both systems very likely allowed their connectivity to facilitate fluid circulation. Moreover, geochemical and mass balance calculations , together with reactive transport models  indicate that a pervasive fluid flow system driven by thermal convection, and controlled by these fault systems, could have delivered enough volume of dolomitization fluid and at the required temperature to account for the volume of dolostone in the Benic assim half graben. These studies interpreted that the dolomitization fluid was composed of evolved seawater that interacted with underlying sedimentary cover and basement metamorphic rocks. Therefore, seawater drawdown and thermal convection would not have occurred at the same time. The large volume and high reactivity of warm fluids invading the rock pores next to fault zones, and especially at their intersection, resulted in the formation of massive dolostone patches without a significant facies control. Dolostone pipes are a special case of massive replacement around faults, since they appear rooted in the largest stratabound dolostone bodies. Their geometry and arrangement reveal that they formed when warm fluids that were focused along specific beds reached a fault and flowed upward. Such structures were also described in detail by Sharp et al. (2010) in the Zagros Mountains.
A large volume of massive dolostone that passes laterally to a series of large-scale stratabound dolostone geobodies is associated with the Campello Fault. This huge volume of high-temperature dolostone, which forms a continuum from the fault in the north to a distance of 7 km to the south suggests that the Campello Fault was the master feeding fault (i.e. main entry point) for Mg-rich dolomitization fluids in the Benic assim area. This interpretation is also supported by the fact that this fault also acted as the main conduit for the delivery of mineralizing fluids forming Mississippi Valley Type (MVT) deposits in its vicinity, and also by the presence of more Fig. 14. Sketches of the half-graben structure showing major depositional environments (inner, middle and outer ramp) and the distribution of dolostones across the study area. Dolostones indicated with a questions mark in sequence II are inferred due to lack of exposure. See Fig. 1 for section location. radiogenic burial carbonate cements occurring in the fault surroundings Mart ın-Mart ın et al., 2015).
The role and importance of the master feeding faults on the distribution of dolostones have extensively been reported in studies of faultcontrolled dolomitization (see Davies & Smith, 2006;Wilson et al., 2007;L opez-Horgue et al., 2010;Hollis et al., 2017). In this study, the lateral dolomitization extension along the hanging wall of the master feeding fault (Campello) ranges from 790 to 1600 m for massive dolostone patches and is at least 7000 m long for stratabound dolostones (Fig. 14).
Faults that acted as barriers for dolomitization fluids can also be identified because they coincide with the termination of dolostone bodies. A series of structures ranging from outcrop fractures to large-scale faults bounding metre-thick limestone stringers or hundreds of metre-thick dolostone clusters are clearly recognized (Figs 1,9,10 and 11). Fault C is one of the typical examples of a low-offset (with a displacement on the order of a few metres) and large-scale (with a length of at least 2.4 km) fault that acted as a barrier for dolomitization fluids. These structures were probably not cemented when dolomitization took place, since only two pre-dolomitization marine cementation phases are found prior to replacement (Mart ın-Mart ın et al., 2015). Therefore, these structures caused the diversion of the dolomitization fluids either due to the presence of fault gouge or simply because they acted as planes of discontinuity for fluid flow. Similar cases of faults acting as dolomitization stoppers are rarely reported in case studies (e.g. Rustichelli et al., 2017), despite the fact that faults can act as barriers for fluids (Knipe et al., 1998;Bense & Person, 2006;Pei et al., 2015). The presence of faults with similar orientation, dimensions and probably age, but that present opposite transport behaviour (i.e. acting as either conduits or barriers for dolomitizing fluids), illustrates how challenging it is to predict fault permeability in the subsurface.
Finally, another structural control at the outcrop and micro scale are bedding-parallel stylolites. These structures, which pre-date replacement Depositional controls on replacement Dolostones in the Benic assim area only occur in the Benassal Formation and are vertically restricted to a zone below the base of the Escucha Formation. This dolomitization zone ranges from ca 400 to 650 m in thickness (Fig. 12), even though the Benassal Formation is >1600 m thick. The preferred dolomitization of the uppermost part of the Benassal Formation has been interpreted to be related to the presence of a regional unconformity, which corresponds to the top of sequence III, and the overlying Escucha Formation, which likely acted as a top seal for dolomitization fluids Mart ın-Mart ın et al., 2015). Despite the similar characteristics (composition, texture, facies, etc.) of limestones of the whole Lower Cretaceous succession, the diagenetic fluids only replaced the upper part of the Benassal Formation sediments below the regional unconformity (Figs 1 and 2).
There is also a 70 to 130 m thick limestone zone just below the regional unconformity at the base of the Escucha Formation which is not affected by dolomitization (Fig. 12). This configuration is very different from the frequently reported shale seals in fault-controlled dolomitization systems, where limestones just below the unconformity and sealing shale unit normally appear extensively dolomitized (see reviews by Davies & Smith, 2006, and many other case studies like Azmy & Conliffe, 2010;Saller & Dickson, 2011;Freiburg et al., 2012;Slater & Smith, 2012;Hendry et al., 2015). The presence of a non-dolomitized zone below an unconformity and clastic seal is probably related to the composition and original porosity and permeability of such a zone. In the study area, the uppermost sediments of the Benassal Formation were affected by subaerial exposure during the transition from a shallowmarine carbonate platform (Benassal Formation) to a tidal flat dominated by clastic sedimentation (base of the Escucha Formation; Querol et al., 1992;Villanueva-Amadoz et al., 2010). This scenario is different from that described in the aforementioned carbonate systems that are overlain by shales associated with transgression and drowning which did not experience subaerial exposure and/or meteoric water cementation. Subaerial exposure in carbonates is commonly associated with significant porosity creation (karst) that may facilitate subsequent dolomitization (e.g. Sharp et al., 2010;Poros et al., 2012;Swennen et al., 2012). However, they have also been reported to result in the formation of seals for dolomitizing fluids (e.g. Lavoie & Chi, 2006), likely similar to the situation in this study. In this case, a vertical dolomitization interval associated with a top seal in the Benassal Formation can be defined by subtracting the non-dolomitized zone (70 to 130 m beneath the base of the Escucha Formation) from the preferred circulation zone of dolomitization fluids (400 to 650 m beneath it; Fig. 12). What sediments are replaced within this interval depends on other factors, such as structures (see above), and facies distributions, components, texture and pre-dolomitization diagenesis (which are discussed below).
The stratigraphic framework of the Benic assim area is mainly controlled by the activity of the two large-scale syn-rift faults (the Campello and Benic assim faults). Therefore, these structures determined the thickness of the succession in each part of the basin and, thus, which part of the Benassal Formation was located within the vertical dolomitization interval at each location. Consequently, these syn-rift faults strongly influenced the potential distribution of dolostones in the sequence stratigraphic framework (Fig. 12). The Benassal Formation is very thick in the north of the Orpesa Range (i.e. at the half graben depocentre), with the vertical dolomitization interval only affecting T-R sequence III (Figs 5 and 12). In contrast, the Benassal Formation is relatively thinner in both the southern side of the Orpesa Range and in the Ferradura fault block, where the vertical dolomitization interval covers sediments of sequence III and the upper part of sequence II (Fig. 12). Finally, the Benassal Formation is much thinner in the Cingle del Morral fault block, where the vertical dolomitization interval affects the three T-R sequences (Figs 7 and 12). However, only sediments of the uppermost two sequences were affected by dolomitization.
The majority of dolomitized sediments in the study area were deposited in inner to middle ramp settings (Fig. 14). This observation is supported by the following facts: (i) muddy facies were deposited in outer ramp to basinal settings at the bottom of the Benassal Formation, while grainy facies were deposited in the inner ramp towards the top of the Benassal Formation, and are generally not replaced with dolomite; (ii) dolostones (both massive patches and stratabound dolostones) are most abundant in the eastern side of the study area, where middle ramp facies are prevalent; (iii) dolostones tend to replace the regressive sediments of T-R sequences when the sequence is mostly associated with an outer to middle ramp setting (such as in cases of T-R sequence II); and (iv) dolostones preferentially replace sediments around the maximum flooding zone of a T-R sequence when they represent middle to inner ramp settings (such as in case of T-R sequence III). Non-replaced limestones between or within dolostone bodies are formed by either muddy facies (for example, spicule wackestones) with suture and sharp peak and wave-like stylolites (sensu Koehn et al., 2016) or grainy facies that are highly-cemented including pre-dolomitization marine cements (Fig. 9K and L;Mart ın-Mart ın et al., 2017). Both rock types were likely either impermeable to dolomitization fluids or their components were less reactive and were consequently not replaced. Reactive transport simulations indicate that a permeability contrast of at least two orders of magnitude between adjacent beds was required in order to cause preferential dolomitization of specific beds . Additionally, Mart ın-Mart ın et al. (2015) revealed that replacement of calcite with dolomite started in the matrix and then affected other components. This is probably due to the high reactive surface area of micrite. Therefore, relatively grainy (i.e. initially permeable) facies containing some micrite (for example, wackestones and packstones) have a stronger potential for dolomitization.
The depositional setting and distribution of lithofacies are key controls on the extent of dolomitized geobodies, and normally determine the units that can be preferentially dolomitized (e.g. Barale et al., 2012;Hips et al., 2016). Previous studies indicate that fault-controlled dolomite can replace sediments deposited in a variety of environments, from relatively shallow-water settings rich in grains (e.g. Barale et al., 2012;Hendry et al., 2015) to basinal environments dominated by relatively muddy facies (e.g. Smith, 2006;Hips et al., 2016). However, the preferred dolomitization of wackestones to packstones seems very common in cases of fault-controlled replacement (e.g. Wilson et al., 2007;Beavington-Penney et al., 2008;Mart ın-Mart ın et al., 2015). This is also the case here, where the combination of initial permeability of these sediment textures together with the presence of a certain volume of micrite (with high reactive surface area) favour their replacement.

CONCLUSIONS
The results of this study reveal a series of controls on the geometry and distribution of hightemperature dolostone geobodies in extensional settings, based on the systematic study of the Benic assim outcrop analogue. The main conclusions from this study are: 1 The strong syn-sedimentary activity of the west-east trending Campello Fault during the deposition of the Benassal Formation controlled the development of a half-graben basin and determined the depositional framework in the study area. As a result, the Benassal Formation is much thinner and presents shallower facies in the footwall block of the Campello Fault compared with its hanging wall.
2 The sedimentary analysis of the Benassal Formation throughout the study area reflects a depositional system with the proximal area in the west and south, and the basin depocentre towards the north-east. The carbonate succession represents a general sedimentary evolution from basinal to inner-platform depositional settings dominated by shallow-marine environments. It is characterized by the lack of deposits representing platform barriers or steps, and the scarce influx of terrigenous sediments.
3 Two end-member dolostone geometries (fault-controlled massive patches and stratabound) have been recognized in the study area. They show a clear spatial association between the two end-members, since massive patches laterally extend to stratabound bodies. Moreover, dolostone pipes are rooted in the thickest stratabound dolostones and follow the fault plane upward.
4 The distribution of dolostones is controlled (in order of importance) by the master feeding faults (i.e. entry points for warm dolomitization fluids), the sequence stratigraphic framework and the lithofacies. Specifically: (i) the hanging wall of the Campello and Benic assim faults (corresponding to the basin depocentre) was the preferred place for dolomitization, which is massive next to the faults; (ii) a clastic unit (Escucha Formation) above a regional unconformity, which corresponds to the top of the uppermost stratigraphic sequence, acted as a seal for reactive fluids and constrained the vertical interval in which sediments could be replaced; (iii) the thickness of transgressive and regressive sequences determined what sediments were affected by replacement in each fault block; and (iv) the vast majority of dolomitized sediments were deposited in middle-ramp settings dominated by wackestones and packstones.