Natural fracture networks exert a first-order control on the exploitation of resources such as aquifers, hydrocarbons, and geothermal reservoirs, and on environmental issues like underground gas storage and waste disposal. Fractures and the mechanical stratigraphy of layered sequences have been intensively studied to unravel the relationships between bed thickness and fracture spacing, but less attention has been paid to intrabed fracturing patterns due to the intrinsic local variability of sedimentary processes and products. Among sedimentary rocks, turbidites show great lateral and vertical variability of textural characteristics and depositional facies, which are expected to strongly influence the location and density of fractures. To better understand the contribution of stratigraphic, sedimentologic, and petrophysical properties on fracture patterns, we performed a high-resolution study on a selected stratigraphic interval of jointed foredeep turbidites in the Miocene Marnoso-Arenacea Formation (Northern Apennines, Italy). Cumulative statistical relations of field and laboratory structural, sedimentologic, and petrophysical data significantly improved when analyzed at the sedimentary facies scale. In particular, for facies recording different cross-flow (i.e., longitudinal to the paleocurrents) depositional conditions within the parent turbidity currents, we observed three-dimensional anisotropies of rock hardness (i.e., uniaxial compression) that were positively correlated with normalized fracture intensities, indicating a primary sedimentary control on fracture distribution. This type of intrabed joint distribution has crucial practical implications for the lateral prediction and evaluation of mesoscale fracture patterns in turbidite sequences.
Fracture networks impact subsurface fluid flow, including downward percolation and upward migration of groundwater, geothermal fluids, and hydrocarbons, thus playing a crucial role in environmental and energy resource issues (see, e.g., Nelson, 2001; Singhal and Gupta, 2010). Such significant societal impacts motivate the study of the factors determining fluid flow at the various scales, and in particular at the micro- and mesoscale (e.g., fracture aperture and connectivity), which are not fully covered by drilling and geophysical means, and thus, the controls on fracture nucleation, slip, propagation, mechanical closing, and sealing by mineral precipitates. Fracture-related fluid flow is critical for specific lithologies characterized by low-permeability or porosity such as massive carbonate sequences, basement rocks, and igneous intrusions, and it is equally important for dual porosity-permeability systems developed in bedded sedimentary rocks, where the bulk fluid retention-transmission is controlled by the superposition of a secondary fracture network onto a primary matrix volume, either of which can show specific porosity-permeability (see, e.g., Spence et al., 2014, and reference therein).
Increasing numbers of studies show that fracturing in layered sedimentary rocks characterized by the vertical alternation of different lithological units, both siliciclastic and carbonate, is controlled by many factors including: (1) local to regional stress regime, (2) driving deformation mechanism (e.g., structural bending vs. hydrofracturing), (3) hydrogeological architecture, (4) the intrinsic mechanical properties of lithological units (e.g., friction coefficient, rock strength, viscosity, cohesion, etc.) and their coupling at interfaces, (5) the local to regional tectonic and lithostratigraphic setting, and (6) diagenetic processes (e.g., dissolution, cementation, compaction, and replacement; see, e.g., Pollard and Aydin, 1988; Narr and Suppe, 1991; Gross, 1993; Becker and Gross, 1996; Bai et al., 2000; Underwood et al., 2003; Barbier et al., 2012; Tan et al., 2014; Hooker and Katz, 2015; Ogata et al., 2014a, 2014b; Tavani et al., 2015, and reference therein). In particular, the rheological properties and thickness of each sedimentary bed, and the nature of the bounding surfaces (i.e., layer interfaces) define the mechanical stratigraphy of a given sedimentary succession, while the intra- and interbed fracture network geometry defines the fracture stratigraphy (Shackleton et al., 2005; Laubach et al., 2009; Savage et al., 2010; Barbier et al., 2012; Hooker et al., 2013; Gale et al., 2014). In particular, Laubach et al. (2009) pointed out the possible incremental differentiation between mechanical and fracture stratigraphy due to the evolution of rock properties during diagenesis and deformation.
Studies on well-bedded sedimentary sequences show the importance of well-developed, systematic mechanical stratigraphy in the first-order definition of fracture patterns, due to the regular alternation of relatively “rheologically homogeneous” beds separated by weaker (soft) interlayers (Ladeira and Price, 1981; Corbett et al., 1987; Narr and Suppe, 1991; Ruf et al., 1998; Wennberg et al., 2006; Tavani et al., 2008; Strijker et al., 2012). At smaller scale, studies focusing on internally heterogeneous shallow-marine carbonates and siliciclastic eolian sandstones indicated a “facies” rather than “bedding” control on fracture patterns, highlighting how the (primary) sedimentary structures deeply influence second-order fracturing (Ortega et al., 2010; Storti et al., 2011; Barbier et al., 2012; Deng et al., 2015).
In order to investigate this “facies versus bedding control,” turbidite sandstones are the best candidates, since they show a great variability of depositional structure, grain-size distribution, packing, sorting, and mineralogy, and nonetheless maintain tabularity and constant thickness over long distances (Mutti, 1992; Muzzi Magalhaes and Tinterri, 2010).
Here, we present results from a study integrating stratigraphic and sedimentologic logging, structural analysis, and in situ measurements of rock hardness (i.e., uniaxial compressional strength) tensors, coupled with mercury-intrusion porosimetry and petrographic and microstructural studies supported by optical scanning electron microscope (SEM) and image analysis. We carefully logged a selected stratigraphic interval of the Miocene Marnoso-Arenacea Formation of the Italian Northern Apennines, which provides a world-class example of a typical basin plain turbidite succession (see, e.g., Ricci Lucchi, 1986; Tinterri et al., 2012). In particular, the studied interval is characterized by high facies variability due to spatial confinement of depocenters in the Langhian inner foredeep, affected by the growth of thrust-related folds during deposition (Lucente and Pini, 2003; Tinterri and Tagliaferri, 2015). This facies-based approach minimizes the possible bias induced by changes in the depositional environment and structural position and maximizes the accounted range of lithological variability. Facies are described and interpreted as a result of the along- and cross-current (i.e., downslope and along-slope) parent flow evolution (see, e.g., Mutti, 1992).
Miocene to Pliocene, NE-verging, thrust-related growth anticlines affecting the ∼2000-m-thick Marnoso-Arenacea Formation are well exposed in the external foothills of the Northern Apennines (e.g., De Donatis and Mazzoli, 1994). The investigated area is located in the NW-plunging periclinal sector of the Palazzuolo anticline (Diaterna anticline in Landuzzi, 2005), which is bounded to the SW by the Mount Nero thrust and to the NW by the main thrust zone of the Mesozoic Ligurian allochthonous nappe (i.e., Sillaro Line; Landuzzi, 1994, 2005; Fig. 1A herein). The exposed folded succession includes Langhian to Tortonian turbidites recording the progressive, structurally controlled and axially fed sedimentary infill of the inner sector of the foredeep (e.g., Tinterri and Tagliaferri, 2015; see Figs. 1B and 1C herein). The anticlinal geometry is characterized by a gently inclined back limb, a wide crestal zone, and a steep to overturned forelimb cut by the NE-directed Mount Castellaccio thrust zone (e.g., Landuzzi, 2005; Fig. 1D herein).
The investigated succession is exposed along the Santerno River, which allows observations and data sampling on clean rock surfaces (Fig. 2). This interval, characterized by a high local sandstone-mudstone ratio (S/M = 2.8), is located ∼90 m below the regional, basinwide Contessa key bed (e.g., Ricci Lucchi and Valmori, 1980) and overlies a mudstone-dominated interval. The latter caps, in turn, a tens-of-meters-thick mass transport deposit (i.e., Acquadalto mass transport deposit; see Fig. 2A), which represents another regional stratigraphic marker (Muzzi Magalhaes and Tinterri, 2010; Tinterri and Tagliaferri, 2015). The analyzed 6.6-m-thick interval consists of four thick to very thick (from 80 to 160 cm) medium- to coarse-grained sandstone beds, and two thin, fine- to medium-grained sandstone beds, separated by marly and clayey silty mudstones. Overall, beds show a NNE-SSW strike and gently dip toward the W, with NW-SE–oriented paleocurrents recording a southeastward transport direction (stereonet in Fig. 2A), in accordance with the regional paleogeographic framework (see Fig. 1A).
The studied beds differ from the classical “Bouma sequence” because they are “composite” units made up of two or more internal subdivisions (i.e., laminasets; see following). The different sedimentary divisions reflect complex depositional processes, ascribed to the interplay between a lower dense (laminar) and an upper diluted (turbulent) flow (Tinterri et al., 2003; Talling et al., 2004; Tinterri and Muzzi Magalhaes, 2011). In particular, the thicker beds can be classified as “hybrid turbidites,” “slurry,” “debrite,” or “sandwich” beds. Such strata are interpreted to result from highly erosive sediment flows experiencing complex along- and across-current transformations in flow characteristics (e.g., Amy and Talling, 2006), which can be expected from seafloor morphologies creating frontal and lateral confinement (e.g., Muzzi Magalhaes and Tinterri, 2010). Each bed is thus composed of one or more facies developed at the lamina and laminaset scale (sensu Campbell, 1967). Different facies types, separated by either well-defined or transitional surfaces, are composed of thicker and coarser beds, whereas thinner and finer ones are characterized by relatively the uniform and homogeneous distribution of internal sedimentary structures (Figs. 2B and 2C). This type of internal subdivision clearly influences fracture propagation, creating a marked transversal, intrinsic, composite anisotropy (e.g., Griera et al., 2013; Gomez-Rivas et al., 2015), which favors intrabed multilayer fracturing (e.g., Bons et al., 2012), as detailed in the following sections.
The turbidite sandstones of the Langhian–Serravallian Marnoso-Arenacea Formation appear ubiquitously well cemented throughout the basin. In detail, the original porosity is rarely preserved, even in coarser-grained beds, having been destroyed by the widespread intergranular precipitation of microcrystalline calcite cement and by the enhanced compaction of clay minerals in the matrix (Fontana et al., 1986; Cantisani et al., 2013).
The overall fracture network affecting the Marnoso-Arenacea Formation in this region consists of two systematic, bed-perpendicular joint sets (Sani, 1990; Landuzzi, 1994), with the more persistent, master joint set striking WNW-ESE and the abutting cross joint set striking NNE-SSW (Gross, 1993; Figs. 3A and 3B). Such joints are typically confined within facies layers (i.e., facies-bound) rather than beds (Fig. 3C). In the studied outcrop, master joints are dominant (i.e., pure dilatational sidewall displacement) and show smooth bounding surfaces, sometimes with faint plumose patterns. Thin remnants of calcite coatings on sidewalls of joints, often removed by later dissolution due to meteoric water percolation, suggest vein reopening.
Layer-parallel calcite shear veins and fractures with dip-slip slickenfibers occur in the marly layers immediately below the erosive base of sandstone facies, which is usually characterized by incipient conjugate thrusts emanating from the layer-parallel shear vein (Figs. 3D and 3E). Shear veins display pervasive internal banding provided by multiple welded slip planes, indicating crack-and-seal opening during layer-parallel shortening/extension and possibly flexural slip folding (Sani, 1990). As highlighted by preceding studies on zircon and apatite fission tracks, backed up by clay mineralogy (i.e., vitrinite reflectance and illite-smectite ratio), the maximum burial depth varied from 2.5 to 4 km in response to the differential tectonic loading exerted by the overriding Ligurian Nappe (e.g., Reutter et al., 1983; Zattin et al., 2002).
In summary, the described structural association records a polyphase deformation history that spans from late Miocene fold-and-thrust tectonics to the subsequent Pliocene–Pleistocene uplift and unroofing, when the bulk of jointing is thought to have occurred (see, e.g., Landuzzi, 1994, 2005).
The sedimentologic approach used in this paper is based on the process-oriented “facies tract” concept: an association of genetically related facies types making up a turbidite system, which reflects the evolution of turbidity currents within each considered system (Lowe, 1982; Mutti, 1992). Consequently, an ideal facies tract can be considered as a bed deposited by a single sediment gravity flow undergoing along- (i.e., from proximal to distal) and cross-current (i.e., from flow axis to margins) transformations during its basinward motions, due to the influence of the physiography of the depositional setting (see, e.g., Tinterri and Tagliaferri, 2015, and references therein). In these terms, the concept of gravity flow facies can be applied to a specific depositional division within a bed (i.e., lamina or laminaset; sensu Campbell, 1967), with each one characterized by specific grain-size distributions and sedimentary structures. It follows that, within a bed, the vertical facies sequence represents temporal changes in the flow conditions at a fixed location, while the horizontal facies association records how the flow conditions changed spatially through progressive flow transformations (Tinterri et al., 2003).
Optical microscopy observations, conducted on oriented thin sections in direct and polarized transmitted light by a digital camera–equipped Zeiss AxioPlan 2® microscope, were used to quantify grain size and sorting and classify the corresponding sedimentary facies of each bed. The relative abundances of textural components, clasts, matrix (including interstitial micro-pores), and cement were quantitatively determined by automated image analysis using the Zeiss AxioVision® software on manually line-drawn backscattered SEM images, acquired from selected thin sections representative of each identified facies (i.e., 21 samples; at least four secondary electron backscatter images, 600 × 600 μm wide, for each sample).
Fracture Data Collection
Master joint attributes such as linear density (P10 as standardly termed in oil-industry applications; Dershowitz and Einstein, 1988; Dershowitz and Herda, 1992), spacing, and height, along with spatial distribution, were collected along 21 linear scan lines, covering each recognized facies. Three-dimensional exposures allowed us to maximize data collection by adapting the standard method, carefully orienting scanlines perpendicular to the master joint trend in order to measure the actual fracture spacing without any possible bias (see, e.g., Zeeb et al., 2013, and reference therein). The very few joints shorter than half of the correspondent mechanical layer were not considered. This modified method actually minimizes the size bias, censoring bias, orientation bias, and truncation bias (e.g., Zhang and Einstein, 1998; Zeeb et al., 2013) without any further data correction or processing.
Schmidt Hammer Rock Hardness
To quantify in situ uniaxial compression strength, we systematically collected Q rebound coefficients (Winkler and Corbett, 2014; Winkler et al., 2014) using a SilverSchmidt® integrated electronic Schmidt hammer (PROCEQ Instruments; N-type; impact energy 2.207 Nm [1.63 ft-lb]):
According to the standard procedure (ISRM, 1981; Aydin and Basu, 2005; Aydin, 2009; Yagiz, 2009; Barton, 2014), we collected at least 90 rebound data points for each investigated mechanical unit (i.e., sedimentary facies). Before acquiring the data, we carefully polished the selected areas with a corundum abrasive tool to remove any possible superficial alteration, ensuring uniform contact between the rock and the hammer tip. At least 30 data points were acquired in three spatial directions (x, y, z): parallel to the bedding dip, parallel to bedding strike, and perpendicular to bedding, respectively. Q values were systematically measured at a distance of at least 10 cm from layer-parallel shear veins to prevent the possible influence of abrupt changes on mechanical properties (Ruf et al., 1998; Cooke and Underwood, 2001; Larsen et al., 2010). Raw data are provided in the GSA Data Repository (DR1).1
To avoid measurement-related bias, data analysis was performed on the higher 50% interval of the total data set (Katz et al., 2000; Barton, 2014). Given the resultant normal distributions, the mean Q values calculated in three dimensions and the cumulative data acquired in each sedimentary facies are hereafter directly used as proxies to estimate the relative (not absolute) rock strength.
A PoreMaster 33® porosimeter (Quantachrome Instruments, Boynton Beach, Florida, USA) was used to acquire total porosity and pore-size distribution by mercury intrusion. In total, 21 representative samples from all mechanical units (i.e., facies) were measured. Before measurement, samples were dried at 40 °C for 24 h, and then ∼1.5–2 g of material were analyzed. Rock density, obtained prior to mercury porosimetry, was measured by an Ultrapic© Elium picnometer (Quantachrome Instruments). The following parameters were used for data acquisition: (1) The sample cell was 1.0 × 3.0 cm; (2) the pressure range was 3.45–227,527 kPa (0.5–33,000 psi); (3) the pore size range was 0.0064–950 μm; (4) the contact angle of mercury was 140°; and (5) the surface tension of mercury was 0.48 N/m (480 dyn/cm). The obtained mercury intrusion and extrusion curves were interpreted to estimate pore-size distributions using the Washburn equation
where the applied hydraulic pressure P is related to the cross-sectional radius R of pore-throats accessible by the pressured mercury, together with two material-related thermodynamic parameters: surface tension of mercury γ and its contact angle θ with the sample material involved (Washburn, 1921; León y León, 1998). Raw data are provided in the GSA Data Repository (DR2; see footnote 1).
Joint Attributes Analysis
Statistical spacing distributions for each mechanical facies were also tested to evaluate their best-fitting regressions to the theoretical distribution classically used in the literature: (1) exponential, (2) normal, (3) gamma, and (4) log-normal (see, e.g., Dekking et al., 2005; Delignette-Muller and Dutang, 2015). For a robust approximation of the best-fitted distributions, median p-values used to quantify goodness-of-fit were obtained through the application of three different standard tests: the Kolmogorov-Smirnov, Cramér-Von Mises, and Andreson-Darling tests (see Dekking et al., 2005; Delignette-Muller and Dutang, 2015). These tests show that since the normal and log-normal distributions are particular cases of the gamma distribution, all the facies can be described in terms of the shape parameter, which is a function of the skewness of the gamma distribution. Tan et al. (2014) introduced the joint saturation ratio (JSR), which is related to the shape parameter of the closely approximated theoretical gamma distribution by an empirically derived function, as a thickness-independent indicator of the state and type of fracturing (i.e., natural hydrofracturing vs. flexural bending-related fracturing).
Bed and Facies Types: Sedimentologic Significance and Attributes
In order to better correlate single turbidite beds and the related intrabed facies with mesostructural data, in situ rock hardness measurements, and the samples used for microstructural and petrophysical analyses, we adopted the same progressive (top to bottom) alpha-numeric classification and color code throughout text and figures (see following).
The four principal beds evaluated in this study are numbered 1–4 from top to base of the outcrop/section (Fig. 4). These beds range in thickness between 85 and 160 cm and are partitioned into at least three stacked facies: (1) basal, well-sorted, medium-grained massive sandstone (F8-type facies of Mutti, 1992; division “a” of Muzzi Magalhaes and Tinterri, 2010), (2) an intermediate slurry-debrite, i.e., a poorly sorted to unsorted muddy sandstone characterized by liquefaction/fluidization features and mudstone clasts (division “b” of Muzzi Magalhaes and Tinterri, 2010), and (3) an upper layer made up of well- to moderately sorted fine sandstone with plane-parallel, oblique, and convolute laminations (F9-type facies of Mutti, 1992; division “c” of Muzzi Magalhaes and Tinterri, 2010; see Fig. 2B herein).
The mudstone intervals, representing the fine-grained “tails” of the parent turbulent flows (i.e., hemiturbidites; Muzzi Magalhaes and Tinterri, 2010), have a maximum thickness of 60 cm. These are characterized by faint laminations and dark- and light-gray color banding due to the presence of very thin (up to 1 cm) siltstone layers and alternating marl-dominated and clay-dominated lithologies, sometimes appearing pervasively bioturbated. Two 5–10-cm-thick, fine-grained sandstone beds are interpreted as F9 facies of Mutti (1992). They are characterized by well-developed laminations, with evidence of hummocky-type structures and biconvex ripples (Muzzi Magalhaes and Tinterri, 2010), and they can be defined as “single-facies” beds (see Fig. 2C).
In addition to the internal facies subdivision, bed 3 shows a second-order bipartition, with lower, coarser-grained, massive- to crudely laminated and convoluted laminasets with mudstone clasts (F8 facies of Mutti, 1992), and an upper finer-grained part characterized by plane-parallel to undulated and convolute laminae (F9 facies of Mutti, 1992). In the cross-flow evolutionary facies scheme proposed by Tinterri and Tagliaferri (2015) and Tinterri et al. (2016), this kind of bed represents the sedimentary record of the axial part of the parent flow, which develops the slurry-debrite hybrid bed counterparts outlined above (i.e., beds 1 and 2) at the flow margins (Fig. 5). In this scheme, bed 4 represents a singularity, being characterized by mixed, intermediate-facies characteristics, likely recording the transition between the slurry-debrite hybrid type and the more typical F9-type beds at the flow margins (Tinterri et al., 2016).
We identified six main representative facies types based on micro- to mesoscale, stratigraphic-sedimentologic characteristics (Figs. 6A and 6B): (1) massive clay-rich mudstones-siltstones (facies A); (2) massive, bioturbated, marl-rich mudstones-siltstones (facies B); (3) well-sorted, fine-grained, laminated sandstones (facies C; F9-type facies of Mutti, 1992); (4) poorly sorted, fine-grained, laminated sandstones (facies D); (5) unsorted to poorly sorted, mud-rich, medium- to coarse-grained sandstones with fluidization/liquefaction features and mudstone clasts (facies E; slurry-debrite interval, according to Muzzi Magalhaes and Tinterri, 2010); and (6) massive- to crudely laminated, medium- to coarse-grained sandstones (facies F; F8-type facies of Mutti, 1992). The latter three cases represent the main upper, intermediate, and lower component facies of multifacies beds (i.e., tripartite), respectively. The boundary surfaces between different facies are usually transitional, shifting laterally to sharper surfaces and varying along with the internal sedimentary characteristics (see Fig. 4). The base of the sandstones is always well defined and sharp, sometimes with irregular surfaces (e.g., sole marks, erosional scours, load casts) testifying to the highly erosive capacity of the turbidity currents.
At the microscale, these facies show distinctive clast-matrix textures and fabric ranging from massive, matrix-sustained to laminated clast–sustained textures, depending on the dominant grain size (see Fig. 6). In thin sections, primary pores appear largely cemented by micritic calcite and filled by matrix (depending on the lithologies), with some primary porosity preserved along laminae and secondary porosity possibly due to microcracks observed along grain boundaries, and possibly synchronous with joint formation (see Fig. 6B).
Basic textural components such as clasts, cement, and matrix (here intended as the intergranular fine-grained material including related micropores) were quantified in terms of areal percent through automated image analysis of SEM backscatter images (see “Facies Characterization” section) and represent complementary diagnostic features for the identification of each facies type (Figs. 6B and 6C).
In a ternary diagram of textural components, the different facies are grouped in specific clusters (see Fig. 6C): The more sorted sandstone facies (i.e., facies C, D, and F) differ mainly on the basis of their clast/cement ratio, while the unsorted, mud-rich facies E shows a relatively higher spread due to the different content of fine-grained intergranular matrix plus micropores. The mudstone facies A is almost entirely made up of fine-grained matrix with dispersed bioclasts (see Fig. 6B), and thus it represents an outlier. The facies B data point falls close to the facies C and D clusters, suggesting an affinity to the F9-type beds.
Petrophysical Properties and Fracture Attributes
A summary of the petrophysical and fracture properties is provided in the correlated logs of Figure 7, and the specific results are discussed next.
The mean uniaxial strength log indicates that Q values range from ∼38 to ∼74. Higher values were obtained from facies F (66 and 58.5 Q), and comparable ones were obtained from facies D (57.7 and 69 Q) in beds 1 and 4, respectively. The two values from facies C are consistent (53 and 59 Q), whereas data from facies E are more scattered, from ∼60 to ∼40. Facies A and B typically provided mean Q values lower than ∼60. Cumulative data analysis by facies type confirms the highest Q values from facies F, the high scattering of data from facies D, and the lower Q values from facies A, B, and E, with the latter showing high scattering as well (Fig. 8A). The log of Q values separated for the three measurement directions with respect to bedding does not show a consistent behavior of Q values spreading in the different facies, apart from facies C (see Fig. 7). Data from facies F are very similar for the x, y, and z directions in bed 4 and become more differentiated moving from bed 1 to bed 3 and then bed 2. Data from facies D are scattered in bed 4 and more clustered in the remaining strata. A variable behavior in different beds is observed also for facies E and B.
Q values measured on three mutually perpendicular surfaces were used to obtain the shape of rock hardness tensors and to analyze it using an L-F diagram, where L and F are the shape factors of an ellipsoid defined by the aspect ratios between the maximum-intermediate and intermediate-minimum tensor axes, respectively. The resultant diagram is used to illustrate the directional anisotropy in a geometric form (Jelinek, 1981). For this purpose, the maximum, intermediate, and minimum principal axes of the Q ellipsoid were labeled as Q1, Q2, and Q3, respectively. Facies D, E, and F were selected for this analysis because of the lack of sharp mechanical bounding interfaces (Cooke and Underwood, 2001; Larsen et al., 2010), such as bedding surfaces and layer-parallel shear veins, which occur in facies A, B, and C. The resultant Q ellipsoids differ on an intrafacies basis, with the dominant prolate geometries characterized by Q1 orthogonal to bedding planes (i.e., vertical), and equidimensional to oblate ones characterized by Q1 parallel to the bedding dip (Fig. 9). This evidence indicates that higher anisotropy of rock strength occurs in the unsorted to poorly sorted facies (facies D and E) compared to the well-sorted ones (facies F). In this framework, Q anisotropy mirrors the cross-flow evolutionary depositional trend recorded by the internal facies variability. Similar trends arise for facies E and F, with an opposite pattern for facies D.
Porosity and Pore Size
The porosity log trend follows that of the mean uniaxial strength (see Fig. 7). The lower values are provided by facies C with a mean porosity of 0.61%, followed by facies F, whereas facies D has porosity values comparable to F in beds 2 and 3, and higher values in beds 1 and 4. Facies E has systematically higher porosity (ranging from 5.5% to 10.3%), while facies A and B show a variable behavior (ranging from 1.9% to 7.8%). Analysis of cumulative porosity data by facies shows high values and very high scattering of data from facies B, as well as from the other mud-rich facies E (see Fig. 8B), possibly due to reorientation of planar minerals during compaction. Noticeably, the porosity values refer to the total volume of unconnected micropores. Thus, porosity does not directly influence the overall permeability, which is relatively low for the investigated lithologies (less than 10%).
The pore size log is characterized by contrasting values for the component facies in beds 1 and 2 (0.065–0.107 μm for facies D, 0.121–0.634 μm for facies E, and 0.485–2.49 μm for facies F), whereas similar and more homogeneous values were obtained from beds 3 and 4 (0.13–0.175 μm for facies D, 0.224–0.32 μm for facies E, and 0.149–0.205 μm for facies F; see Fig. 7). The corresponding cumulative median pore size plot indicates that smaller pores occur in facies A and B, facies D is characterized by pore size of ∼0.1 μm, facies E has slightly larger pores of ∼0.2 μm, and facies C and F have the larger pore sizes, ranging ∼0.3–1 μm (see Fig. 8C). Analysis of mean Q values plotted versus the area of pores plus intergranular matrix, obtained by SEM images (see “Facies Characterization” section), shows a nonlinear inverse correlation, particularly when an outlier data point from facies B is removed from the best fit (see Fig. 8D). The correlation is very good for data from facies E and one facies F data point. Three other data points follow the general trend, while two of them, from facies D and F, respectively, do not match the best-fit line.
Cumulative data analysis indicates that the number of master joints in the studied outcrop overwhelms that of cross-joints, and that ∼72% of them are confined within the correspondent mechanical facies (see Figs. 2B and 2C). In single facies beds, facies-bound joints are also bed-bound, whereas the subordinate through-going joints randomly stop at different heights (i.e., vertical length in the order of meters), especially in the thickest amalgamated bedsets. The corresponding log shows significant variations, with the higher amount of through-going joints (27%–76% for facies D, 50%–100% for facies E, and 40%–90% for facies F) occurring in beds 3 and 4 (see Fig. 7). This behavior is influenced by both layer thickness and the sharp versus smooth type of facies transitions (i.e., intrabed boundaries), and it is particularly evident for multifacies beds, where the percentages of facies-bound joints are relatively low if compared to the single-facies ones (Fig. 10A). Through-going joints terminate against layer-parallel calcite veins occurring close to the basal sandstone facies (see “Geologic Setting”), which provide effective mechanical discontinuities.
The log of one-dimensional (1-D) linear joint density (P10) shows higher values in the two single-facies beds occurring just below bed 1, being bounded by layer-parallel shear veins (see Fig. 7). Clay-rich layers of facies A and B have a high joint frequency as well. Facies D and F have comparable joint P10 values, which become significantly lower in facies E. As a general feature, beds 1 and 2 have lower joint P10 values than the overlying strata. Joint spacing (S) and the associated vertical persistency or height (H) values have very similar trends in the corresponding logs, which show that longer joints in vertical cross sections are more widely spaced than shorter ones (see Fig. 7). The coefficients of variability (Cv) of both S and H have similar trends as well, testifying to their overall even distribution, and supporting the direct correlation between vertical persistency and spacing. This result is further confirmed by the analysis of median joint spacing obtained from all scanlines (for a discussion on best-fitted distributions, see “Facies Variability and Joint Density” section) plotted versus the corresponding H values (Fig. 10B). Apart from an outlier data point from facies E, cumulative best fitting of the whole data set provides a very good result, which progressively improves as the single facies are considered. The facies having three or more data points provide different best-fit trends that show a strong correlation between the corresponding H and S values, with the only exception of facies D.
Facies Variability and Joint Density
The log of normalized spacing values by the H/S ratio shows very similar values of ∼1 in bed 4, followed by significant scattering upward up to bed 1, where data approach values close to 2 (Fig. 7). The data appear to be influenced by stratigraphic position rather than facies type, although when single facies types are considered, their first-order contribution to H/S ratio becomes clear. Analysis of cumulative H/S ratios by facies type shows that mudstone facies B has far higher values than the sandstones, among which H/S values increase from facies D to F (Fig. 10C). Facies A and C show H/S values in accordance with those obtained for mudstones and sandstones, respectively, despite the two available data points. Normalized joint spacing has a weak inverse correlation with the areal content of intergranular matrix and related micropores, provided that an outlier data point from facies A, which represents a peculiar matrix-dominated end member (see earlier herein), is not included in the best fit (Fig. 10D). Unlike mean Q values (see Fig. 8D), data from facies E are more scattered and significantly contribute to the weaker best fit.
We also calculated the classically used standard parameters such as the fracture spacing ratio (FSR; Becker and Gross, 1996), the fracture spacing index (FSI; Bai and Pollard, 2000), and the joint-saturation ratio (JSR; Tan et al., 2014; see Fig. 7). As expected for a layered succession, FSR, defined as the mechanical layer thickness to median joint spacing ratio calculated for each scanline, has a pattern that roughly mimics the H/S trend. This testifies to the dominantly facies-bound nature of joints. The relationship is weakened in particular by the spreading of the facies E data points, which is due to their faint and transitional bounding surfaces. The same happens for the mudstone facies A and B, in which the recognition of single beds and mechanical intervals is usually internally complicated by early diagenetic discontinuities (see earlier herein) and faint surfaces bounding juxtaposed layers with low rheological contrast.
The single FSR values can be visualized in a mechanical layer thickness–median joint spacing scatter plot, in which the slope of the best-fitted linear regression curve defines the FSI. As such, the FSI indicates the average state of fracturing of the entire layered succession (Narr and Suppe, 1991). In the data presented here, the cumulative analysis provides a low FSI value (= 0.68), with an inconsistent statistical regression, due to the very high scatter of mud-rich facies A, B, and E data points (Fig. 11A). After excluding such facies and an outlier point from facies D, representing a singularity (i.e., scan-line 18), statistical relationships significantly improve (Fig. 11B), providing a FSI value (= 1.35) that falls within the range of typical joint-saturated bedded sedimentary successions (Narr and Suppe, 1991; Gross, 1993; Bai and Pollard, 2000; Davis et al., 2011).
The statistical tests conducted for our spacing data suggest that the gamma distribution is the most representative. Subordinately, also the normal and log-normal distributions are sometimes represented, although with minimal (two digits) p-value differences, whereas the exponential distribution is always rejected. This outcome is in line with other published data sets worldwide (see, e.g., Tan et al., 2014, and reference therein). The results of the statistical analyses are provided in the GSA Data Repository (DR3; see footnote 1).
All the facies, with the exception of facies E, fall in the field of flexural bending, with higher median JSR values for facies A, B, and C (Fig. 11C). The large spread of facies D is due to the inclusion of the outlier data point 18, which displays overlapping facies D, E, and C characteristics.
H/S ratio values plotted versus the biaxial Q anisotropy, calculated as the difference between values measured along bedding dip and strike, respectively, show high dispersion (Fig. 12). However, when sedimentary facies are considered separately, good best fits are observed despite the different trends. Such facies trends are similar to those on the L-F diagram (see Fig. 9), showing that H/S ratio values decrease along with increasing dip-parallel Q component (and decreasing strike-parallel Q component) for facies E and F. Facies D instead shows again an opposite pattern.
This research points illustrates field and laboratory evidence for the systematic control exerted by facies distribution and architecture on fracture frequency, both at the intra- and interbed scales. This agrees with the well-demonstrated sedimentary control at the bed scale (e.g., Narr and Suppe, 1991; Gross, 1993; Bai et al., 2000) and at the facies scale (e.g., Ortega et al., 2010; Barbier et al., 2012; Deng et al., 2015), indicating that fracturing in sedimentary rocks can be strongly influenced by vertical and lateral variations in grain size and/or microfabric, imprinted during deposition. Accordingly, in well-stratified sedimentary successions, beds are not always the basic mechanical unit.
The pattern observed in the L-F diagram (see Fig. 9) for the basal and intermediate intervals of multifacies beds (facies F and E, respectively) reflects a progressive switching of the mechanical anisotropy from bedding strike– to bedding dip–parallel Q1 along with the cross-flow evolution of the parent turbidity currents. The opposite pattern observed for the upper interval (facies D) might be due to the “ponding” effect usually observed for turbidites deposited in structurally confined depocenters (i.e., contained-reflected turbidites; Muzzi Magalhaes and Tinterri, 2010). Nearly opposite paleocurrent directions developed in the upper fine-grained interval (facies D) due to perturbation of the turbulent cloud, causing complex interactions and interference patterns of the traction-plus-fallout structures, which in turn are expected to create the resultant composite, transversal anisotropies in mechanical properties. The dominance of bedding-orthogonal Q1 in clay-rich and laminated facies (i.e., facies E and D) is likely due to enhanced vertical compaction. This intrafacies, coplanar (x-y) mechanical anisotropy, related to the intrinsic sedimentologic variability, is responsible for the resultant joint patterns.
Our results suggest that for a given degree of flexural bending around an axis approximately parallel to the direction of the paleocurrents, structurally confined turbidites develop master joint patterns with a linear density controlled by vertical and lateral facies variations. Changes in sedimentary facies attributes are the record of the cross-flow position with respect to the flow direction of the parent turbidity current, along with a switching of the mechanical tensor from flow-perpendicular orientation at the axis to flow-parallel orientation at the margins of the turbidity current. This mechanical-sedimentological variation applies to the massive, coarse-grained sandstone facies characterizing the basal interval of multifacies beds (i.e., facies F), and to a smaller extent to the intermediate “debrite” interval (i.e., facies E), which records a decreasing fracture density from axial- to marginal-flow beds (Fig. 13). The opposite mechanical-fracturing trend is observed for the upper laminated, fine-grained sandstones (i.e., facies D). Similar trends are also expected in the vertical direction, along with comparable sedimentological up-section variations in structurally confined turbidite successions recording lateral depositional onlap at larger scales. A mechanical explanation for preferential synfolding fracturing in coarse sandstone facies (facies F) can be found in their higher uniaxial compressional strength, which likely favors stress channeling (e.g., Mandl, 2000) and tensile failure.
The aim of this study was to investigate the contribution of primary depositional (i.e., flow-related) bed architecture to joint formation in turbidites, with an emphasis on the micro- to mesoscale control of the intrabed sedimentary facies.
The results of our high-resolution, field- and laboratory-based work demonstrate that in beds characterized by different sedimentary facies (i.e., laminaset associations; Campbell, 1967), massive basal sandstone intervals (facies F) appear more cemented and fractured than those showing well-developed lamination and segregation of the grain-size populations (facies D). Conversely, mud-rich sandstones related to sedimentary processes that cause homogenization and lithologic mixing of fine- and coarse-grained populations (facies E) are less prone to nucleate and propagate fractures. This suggests that clay minerals were able to impede the complete diffusion of cement-precipitating fluids in this facies, as indicated by its higher porosity and the larger amount of micropore/matrix. In addition, the abundance of clay minerals makes these lithologies less elastic and able to accommodate low strain by plastic deformation.
The cross-flow evolution of these types of structurally confined turbidites is recorded by multifacies beds, characterized by a coplanar rotation of the mechanical anisotropy tensor from flow-perpendicular to flow-parallel orientation in the massive, coarse-grained basal sandstones (facies F), with an opposite pattern in the laminated, fine-grained upper sandstones (facies D). At the same boundary conditions, with the application of flexural bending around an axis parallel to the paleocurrent direction, jointing of coarse-grained sandstone beds (facies F) is expected to be higher in the axial part and lower at the margins of the paleoflow, with an opposite pattern expected for fine-grained sandstone beds (facies D).
This study highlights the actual complexity associated with intrabed fracturing, suggesting that specific fracture trends in jointed turbidites can only be constrained by combining detailed sedimentological, structural, and petrophysical studies. In this framework, mesoscale sedimentary control on fracture patterns in turbidites is expected to have a strong impact on high-resolution permeability predictions in the subsurface. This study also notes that current models used for fractured siliciclastic layered successions not considering internal facies subdivisions can be oversimplified in terms of both fracture frequency and orientation, regardless the size and robustness of the parent database, and predictive laws empirically derived from analysis of such large heterogeneous data sets may provide results that are too simplistic.
Our findings, showing that fracture patterns in turbidites are imprinted at the facies scale, have important and timely socioeconomic implications for exploitation of hydrocarbon reservoirs and aquifers in fractured rocks, utilization of unconventional resources and geothermal fields, and the geotechnical characterization of mines, quarries, waste disposal sites, and underground geologic storage.
This work was funded by Petrobras, Petróleo Brasileiro S.A. (FRASI Project; grants to F. Storti). Helpful support from P. Muzzi Magalhaes (Petrobras) is greatly appreciated. L. Clemenzi, A. Artoni, A. Borsani, A. Piazza, and A. Tagliaferri are gratefully acknowledged for their help during field work and for the stimulating discussions. Suggestions, comments, and criticism from S.E. Laubach, J. Hooker, and C.J. Landry on an early version of the manuscript greatly improved its final quality. We greatly appreciated the constructive revision by E. Gomez-Rivas, A. Braathen, and an anonymous reviewer. We deeply thank Petrobras for releasing this material for publication.
↵† Present address: Faculty of Earth and Life Sciences, VU University, 1081 HV, Amsterdam, The Netherlands; .
↵1GSA Data Repository item 2016252, supplementary material consisting of porosity and pore-size distribution data set (DR1), three-dimensional rock strength (Q rebound units) data set (DR2), and statistical test results (DR3), is available at http://www.geosociety.org/pubs/ft2016.htm or by request to .
Science Editor: Bradley S. Singer
Associate Editor: Bernhard Grasemann
- Received 19 February 2016.
- Revision received 7 July 2016.
- Accepted 3 August 2016.
- © 2016 Geological Society of America