Structural control on ﬂuid ﬂow and shallow diagenesis: insights from calcite cementation along deformation bands in porous sandstones

. Porous sandstones are important reservoirs for ge-oﬂuids. Interaction therein between deformation and cementation during diagenesis is critical since both processes can strongly reduce rock porosity and permeability, deteriorating reservoir quality. Deformation bands and fault-related diagenetic bodies, here called “structural and diagenetic heterogeneities”, affect ﬂuid ﬂow at a range of scales and potentially lead to reservoir compartmentalization, inﬂuencing ﬂow buffering and sealing during the production of geoﬂuids. We present two ﬁeld-based studies from Loiano (northern Apennines, Italy) and Bollène (Provence, France) that elucidate the structural control exerted by deformation bands on ﬂuid ﬂow and diagenesis recorded by calcite nodules associated with the bands. We relied on careful in situ observations through geo-photography, string mapping, and unmanned aerial vehicle (UAV) photography integrated with optical, scanning electron and cathodoluminescence microscopy, and stable isotope ( (cid:14) 13 C and (cid:14) 18 O) analysis of nodules cement. In both case studies, one or more sets of deformation bands precede and control selective cement precipitation. Cement texture, cathodoluminescence patterns, and their isotopic composition suggest precipitation from meteoric ﬂuids. In Loiano, deformation bands acted as low-permeability bafﬂes to ﬂuid ﬂow and promoted selective cement precipitation. In Bollène, clusters of deformation bands restricted ﬂuid ﬂow and focused diagenesis to parallel-to-band compartments. Our work shows that deformation bands control ﬂow patterns within a porous sandstone reservoir and this, in turn, affects how diagenetic heterogeneities are distributed within the porous rocks. This information is invaluable to assess the uncertainties in reservoir petrophysical properties, especially where structural and diagenetic heterogeneities are below seismic resolution.


Introduction
Porous rocks, such as sandstone and carbonate, are important reservoirs for geofluids. Structural and diagenetic processes commonly affect the petrophysical properties and reservoir quality in these rocks. The importance of the interaction between deformation and structures, fluid flow, and diagenetic processes has been emphasized only during the last 2 decades (e.g., see the recently coined term structural diagenesis; Laubach et al., 2010;Mozley and Goodwin, 1995;Eichhubl et al., 2009;Balsamo et al., 2012;Philit et al., 2015;Antonellini et al., 2017Antonellini et al., , 2020Del Sole et al., 2020). If deformation influences diagenesis and vice versa, a feedback can eventually develop between these two processes. Early diagenesis influences the mechanical properties of rocks  and, in turn, their mechanical stratigraphy (Laubach et al., 2009;La Bruna et al., 2020). "Structural and diagenetic heterogeneities" (referred to as SDHs from now on) can determine the textural characteristics (e.g., grain size, grain shape, relative proportion of grains, and matrix material) as well as the petrophysical and mechanical properties of the rock volume hosting them Aydin, 2000;Faulkner et al., 2010;Bense et al., 2013;Pei et al., 2015;Del Sole et al., 2020). Cement precipitation in granular porous siliciclastic rocks leads to porosity loss, reduction in permeability (Tenthorey et al., 1998;Morad et al., 2010), and, in turn, overall reservoir quality deterioration (Ehrenberg, 1990;Morad et al., 2010). Carbonate cement is commonly concentrated within a few specific horizons or nodules with various shapes and arrangements (Kantorowicz et al., 1987;Bjørkum and Walderhaug, 1990;Mozley and Davis, 1996), making porosity and permeability prediction more challenging (Davis et al., 2006;Morad et al., 2010). Furthermore, cement increases the mechanical strength of the host rock (Dvorkin et al., 1991;Bernabé et al., 1992;Boutt et al., 2014), influencing fault-zone architecture and potential fault reactivation (Dewhurst and Jones, 2003;Flodin et al., 2003;Wilson et al., 2006;Williams et al., 2016;Philit et al., 2019;Pizzati et al., 2019).
Cement has been found in association with DBs. Localization of cement along these structural features may significantly enhance porosity and permeability reduction caused by mechanical crushing and reorganization of grains, thus increasing their sealing or buffering potential (Edwards et al., 1993;Leveille et al., 1997;Fisher and Knipe, 1998;Parnell et al., 2004;Del Sole et al., 2020). The occurrence, distribution, and petrophysical properties of cement along DBs therefore need to be properly characterized and implemented into reservoir quality modeling to predict porosity, permeability, and their heterogeneity (e.g., Morad et al., 2010).
Models of calcite cementation, in particular, are fundamental for predicting sandstone and fault-rock properties such as porosity, permeability, compressibility, and seismic attributes. Diagenetic processes related to fluid flow mechanisms and evolution within DBs are not fully constrained; in particular, how DBs steer the origin and distribution of calcite cement remains poorly understood. Different processes account for enhanced fluid flow within DBs, such as unsaturated flow relative to the host rock in arid to semiarid vadose zones (Sigda et al., 1999;Sigda and Wilson, 2003;Wilson et al., 2003) and transient dilation in the early stage of DB formation (e.g., Main et al., 2000). Also, these mechanisms have been employed to explain the occurrence of cement and other processes (e.g., cementation, hydrocarbon inclusion entrapment, removal of iron oxide coatings) in and around the band (Fowles and Burley, 1994;Labaume and Moretti, 2001;Ogilvie and Glover, 2001;Parnell et al., 2004;Parry et al., 2004;Sample et al., 2006;Wilson et al., 2006;Cavailhes et al., 2009;Balsamo et al., 2012;Lommatzsch et al., 2015). These mechanisms, however, appear to be limited to specific conditions (e.g., cement precipitation in the early stage of DB formation or in vadose environments) assuming that DBs behave as fluid "conduits" in order to explain the occurrence of cement or other authigenic products within these structures. Nevertheless, a significant number of studies on DBs show that in most cases they are baffle or seals to fluid flow (see Ballas et al., 2015, for a review). Much less attention has been paid to fluid flow and diagenetic mechanisms leading to (post-DB formation) selective cementation in association with low-permeability baffle DBs (e.g., Philit et al., 2015), and models of cement precipitation in these DBs are limited to quartz cement and are mostly experimental Williams et al., 2015). Different mechanisms, then, need to be invoked to explain the occurrence of (carbonate) cement in association with DBs in a broader set of conditions.
The aim of our work is to elucidate the influence of DBs on fluid flow and their role in fostering diagenesis and localizing diagenetic products in porous sandstones. The novelty of our work is that by using a multiscalar and cross-disciplinary approach integrating structural and diagenetic analysis, we assess the control exerted by DBs on flow pattern, the origin of diagenetic heterogeneities, and spatial distribution by means of the systematic characterization of the occurrence, as well as the spatial and microstructural relationship between DBs and cement nodules in two porous sandstone reservoir analogs. We examine two field sites in Italy and France where calcite cement nodules are spatially associated with DBs. The comparison between the two locations with different geological settings makes it possible to derive general conclusions that can be extended to other cases in which DBs and diagenetic processes interact. Our study also allows for the evaluation of the impact of both structural and structural-related diagenetic heterogeneities on present-day fluid circulation and on subsequent deformation.
2 Geological framework 2.1 Loiano field site, northern Apennines (Italy) The Loiano study area is in the northern Apennines (Emilia-Romagna region, Italy), 20 km to the south of the city of Bologna (Fig. 1a). The northern Apennines are an orogenic wedge formed in response to the upper Cretaceous-Eocene closure of the Ligurian-Piedmont ocean (Marroni et al., 2017) and the subsequent Oligocene-Miocene convergence and collision between the Adriatic Promontory and the Sardinia-Corsica Block of African and European origin, respectively (Vai and Martini, 2001). Our work focused on the Loiano Sandstones of the Epiligurian Successions (Fig. 1ac), the middle Eocene to middle Miocene siliciclastic infill of thrust-top, piggyback basins discordant to the underlying Ligurian units, which migrated passively to the NE during the Apennines orogeny atop the entre orogenic wedge (Vai and Martini, 2001). The 300-1000 m thick, late Lutetian-Bartonian Loiano Sandstones are a fan delta to proximal turbidite deposit (Papani, 1998). They are medium-to coarsegrained, poorly consolidated, immature arkosic sandstones and conglomerates deposited in a relatively small lenticular basin (a few tens of kilometers in width and length; Fig. 1a, c). They are composed of 49 %-60 % quartz and 39 %-48 % feldspar, the rest being rock fragments, detrital carbonate clasts, and minor accessories (Del Sole and Antonellini, 2019).

Bollène field site, Southeast Basin (Provence,
France) The Bollène site is in the Southeast Basin of Provence (France), 15 km to the north of the city of Orange (Fig. 2a). The Southeast Basin is a triangular region between the Massif Central to the northwest, the Alps to the east, and the Mediterranean Sea to the south. It is a Mesozoic cratonic basin on the edge of the Alpine orogen, approximately 200 km long and 100 to 150 km wide. Three main tectonic episodes affected the region (Arthaud and Séguret, 1981;Roure et al., 1992;Séranne et al., 1995;Champion et al., 2000): SSW-NNE Pyrenean contraction from the Paleocene to Oligocene, NW-SE Gulf of Lion extension from the Oligocene to early Miocene (rifting), and, lastly, SW-NE Alpine contraction from the Miocene to Quaternary (Fig. 2a). The site of Bollène is exposed in a quarry (Fig. 2c) located in Turonian sand (low-cohesion sandstone) between 10 and 200 m thick and is situated north of the E-W Mondragon anticline (Fig. 2b, c). The Turonian sands at the Bollène quarry are laminated and fine-to coarse-grained with modal and bimodal grain size distributions; they formed in deltaic and eolian environments (Ferry, 1997). The host sands are not ce-mented. They are composed of 88 % to 92 % quartz, the rest being feldspar. The median grain diameter (D 50 ) is 0.31 mm, i.e., medium sand. Their porosities range from 20 % to 43 %, and the precise value at study site is 22 % (Ballas et al., 2014).

Outcrop analysis
The geometry and distribution of DBs and nodules were documented by detailed field mapping at different scales for both sites. At the Loiano site, a map (370 m 2 ) at the 1 : 25 scale (1 cm 4 m) was made by standard topographic compass and tape mapping (Fig. 3). The Bollène quarry site pavement was mapped using a DJI PHANTOM ™ drone. Photographs were taken at different heights above the ground surface and were then used to build a 3D mesh and extract high-resolution orthophotos using Agisoft PhotoScan Metashape software (© Agisoft LLC). The high-resolution orthophoto mosaic (1 px 1-1.5 mm) was used for the detailed mapping of DBs and nodules. Furthermore, DBs and nodule patterns, as well as their characteristics and spatial relationships, were documented in the field on high-resolution photographs (15 megapixels) both in Loiano (Figs. 4 and 5) and Bollène (Figs. 7 and 8). Oriented samples were collected for thin section preparation, microstructural, and stable isotopes analysis. The orientation of DBs was measured at each site and plotted in lower-hemisphere equal-area stereograms, rose diagrams, and frequency histograms (Figs. 1d and 2d) using the Daisy3 software (Salvini, 2004).

Microstructural analysis
Polished thin sections of host sandstones, DBs, and nodules were analyzed by standard petrographic microscopy, cold cathodoluminescence, and backscattered electron imagery using a JEOL JSM-5400 and an FEI Quanta FEG 200 environmental scanning electron microscope (SEM). These microscopy techniques were used to examine the textural characteristics (e.g., grain size, shape, arrangements, contact relationships) and microstructures of host rock and DBs, as well as the cement distribution and texture (e.g., cement type and degree of cementation, cement crystal size and shape) (Figs. 9-11). In particular, cold cathodoluminescence (CL) analysis of carbonate cement in nodules was conducted with a CITL cold cathodoluminescence 8200 Mk5-1 system (operated at 14-15 kV beam energy and 250 µA beam current) equipped with a standard petrographic microscope (Olympus BH41). CL was used to describe the cement crystal properties (texture, fabric, luminescence) and the micron-scale spatial distribution and textural relationship among the cements, framework detrital grains, and fractures (sensu lato). This information is used to (i) understand the interrelation between deformation, fluid flow, and diagenesis (e.g., cement precipi- tation); (ii) assess the relative timing of each process; (iii) describe porosity evolution with time; and (iv) understand the mechanisms and the geochemical environment of cement precipitation when coupled with other tools (e.g., stable isotopes analyses). The CL features (color, brightness) of the carbonate minerals are controlled primarily by the relative abundances of Mn 2+ , REEs, and Fe 2+ . These differences, in turn, reflect specific physiochemical conditions of forma-tion waters during mineral growth, including fluid chemistry (salinity), pH and Eh, temperature, pressure, ion activity, and biological activity (e.g., Marshall, 1988;Barnaby and Rimstidt, 1989;Machel, 2000;Hiatt and Pufahl, 2014).  Wibberley et al., 2007;Saillet and Wibberley, 2010;Ballas et al., 2012Ballas et al., , 2013Ballas et al., , 2014Soliva et al., 2013;Philit et al., 2018). Panels (a) and (b) as well as the stratigraphic column in (c) are modified from Philit et al. (2015Philit et al. ( , 2018. The geological map in (c) is modified from Ballas et al. (2012). (d) Lower-hemisphere equal-area projection indicates the orientation of the different sets of DBs (64 data points) at the study site. DBs associated with carbonate nodules are highlighted in red. Dotted lines indicate the main attitude of tabular carbonate nodules. DBs azimuth (strike; N ± 90 • ) frequency rose diagram and dip angle ( • ) plotted against frequency. Best-fit Gaussian curves superimposed on the corresponding data histograms (frequency distributions). Gaussian peaks and related standard deviations (±SD) are indicated for each population. The numbers in square brackets [n] are the same as used in Fig. 6b to rank different sets of DBs.

Stable isotope characterization
Stable carbon and oxygen isotope data from cements from within carbonate nodules were used to constrain the geochemical environment of precipitation and possible source of fluids. Powder samples for bulk rock carbon and oxygen stable isotopes analysis were ground with a dental drill from unweathered or altered sections of the nodules. A total of 46 sites were sampled from nodules in Loiano (n = 30; Fig. 12a) and Bollène (n = 16; Fig. 12b). Powders samples were ana- lyzed with a Thermo Finnigan DELTA plus XP mass spectrometer coupled with a Thermo Finnigan Gas Bench II gas preparation and introduction system. δ 13 C and δ 18 O refer to the international standard VPDB (Vienna Pee Dee Belemnite). Isotope determination analytical precision was 0.10 ‰ and 0.15 ‰ VPDB for carbon and oxygen, respectively. The uncertainty was ca. 0.15 ‰ for carbon and ca. 0.20 ‰ for the oxygen isotopes.
4 Deformation bands and cement: field observations

Loiano
At the study site, bedding strikes NW-SE and dips at an average of 38 • to the NE (Fig. 1d). Deformation bands cluster into three different trends striking 340 • (NNW-SSE), 26 • (NNE-SSW to NE-SW), and 81 • (ENE-WSW) (Fig. 1d). All DBs dip moderately to steeply, mostly in the west and south quadrants (Fig. 1d). Deformation bands commonly occur with a positive relief and appear as whitish linear traces with minor undulations forming eye and ramp structures, whereby they branch and merge (Figs. 3 and 4). Already at the outcrop they exhibit a significant reduction in grain size and porosity in comparison to the surrounding host rocks (Fig. 4f). Deformation bands occur both as single features and as clusters or zones of bands, i.e., in narrow zones with variable thickness (0.8-60 cm) with subparallel DBs (up to 40). Single DBs accommodate minor offsets from a few millimeters up to 40 mm, whereas clusters can accommodate offsets up to 0.5 m (Fig. 4b, d). Deformation bands display a variety of apparent normal and strikeslip offsets (Fig. 3). Different sets of DBs show ambiguous and conflicting crosscutting relationships. Field observations indicate that the NNW-SSE and NNE-SSW sets have mutual crosscutting relationships typical of faults forming synchronously (Fig. 3).
The peculiar characteristic of the Loiano Sandstones is the occurrence of spatially heterogeneous carbonate cement in the form of isolated or multiple spheroids or irregular nodules and continuous tabular nodules (Figs. 4 and 5). The nodules weather out in positive relief because they are more resistant to weathering than the weakly cemented host rock. Isolated nodules range in diameter (major horizontal axis) from 0.2 to 3 m (Fig. 4c, d), whereas tabular concretions have a thickness ranging from 0.10 to 0.8 m and a long axis ranging from 3 up to 15 m in length (Fig. 5a). Generally, the nodule shape in Loiano is similar to that of an oblate spheroid. There is no evidence of spherical nodules or prolate spheroids. The volume of the carbonate nodules ranges from 0.001 m 3 to > 10 m 3 . Nodules form about 20 % of the exposed outcrop volume. Two types of nodules can be distinguished depending on whether they are associated with DBs and/or clusters (i.e., DB-parallel nodules; Figs. 3, 4, and 5a-c) or with bedding planes (i.e., bedding-parallel nodules; Figs. 3 and 5d, e). The former represent roughly 75 % of the total nodules in the study area and are the main target of this work. The association between DBs and nodules occurs in the form of (i) parallelism and spatial overlap between DBs and nodules and (ii) confinement of the nodules by the DBs. In all cases, nodules are oriented with the major axis (elongation direction) parallel to the DBs and the minor axis (i.e., thickness) perpendicular to them. Deformationband-parallel nodules are isolated ellipsoids (Fig. 4b, d) or, alternatively, continuous tabular objects (Figs. 4a and 5a). Nodules may be located along the DB (or zone of bands) trace (Figs. 4a,d and 5a,b), placed in between, and confined by DBs (Figs. 4b, e, and 5c), or they may be asymmetrically placed on one side of the DBs (Figs. 4c and 9e). In some cases, nodules lie at the intersection of different DB planes (Fig. 3). Some DBs are not spatially associated with nodules (Figs. 3 and 4b). Among the multiple sets of DBs, those mostly associated with carbonate nodules are the NNW-SSE and the NNE-SSW ones. As a result, most nodules are elongated along these two structural directions (  sets are rarely associated with carbonate nodules. Nodules are never cut across by the DBs. Bedding-parallel nodules are either isolated (Fig. 3) and multiple but laterally discontinuous (Figs. 3 and 5d, e) or laterally continuous layers with a tabular geometry ( Fig. 3; e.g., "nodular beds" in Del Sole et al., 2020). Nodular beds are continuous pervasively cemented layers that extend along the bedding plane for several meters (up to 15 m in length) with a nearly constant thickness of ca. 35-50 cm. Nodules along bedding planes are more rounded gentle boundaries (Fig. 5d, e) than those associated with DBs, which are instead more tabular and exhibit angular and sharp boundaries (Figs. 4 and 5a-c). In some cases, nodule geometry and elongation direction follow both bedding surfaces and DBs (Fig. 3). Nodules, despite being ubiquitous in the sandstone, are mostly observed within coarse levels with a grain size equal to or larger than medium sands (0.25-0.5 mm). We did not observe any nodules in sedimentary rocks with a grain size finer than sand (siltstone and clay). Bedding-parallel nodules are commonly located in sandstone levels confined between clay-silty levels or fine-grained sand levels (Figs. 3 and 5d, e).
A set of joints and veins ( Fig. 5a, b) was found exclusively within the carbonate nodules. They postdate DBs and nodules and do not propagate into the surrounding host sandstone.

Bollène
At Bollène, DBs occur as belonging to three different trends oriented (  The bedding is emphasized mostly by clay and silt horizons, sporadic well-defined thin levels of gravel, and the alignment of bedding-parallel nodules. (d) Black arrows point to multiple but laterally discontinuous bedding-parallel nodules. Here, a single DB (white line) crosses a bedding-parallel nodule without causing any offset. The position of (d) is indicated in Fig. 3. (e) Photomosaic showing a series of laterally discontinuous nodules just below, above, or in between several continuous impermeable clay-rich levels. The deformation pattern changes depending on the host rock properties (e.g., sorting degree, porosity φ, grain size; see Supplement S1 for details). Cataclastic deformation is accompanied by clay smear (see inset) whereby DBs cut thin dark-colored clay levels.
8a, c) and just a few to NE; a second one is characterized by dominant dextral strike-slip kinematic bands with higher dip angles (70-90 • ) and a NNW-SSE trend (Figs. 6b and 7f). Trend (ii) can be divided into two sets; one set is characterized by dominantly left-lateral and minor right-lateral sub-vertical strike-slip conjugate bands striking NE-SW to ENE-WSW (set 2 in Fig. 6b; Fig. 7a-c); a second set is instead characterized by a set of conjugate DBs with moderate dips (∼ 60 • ), NE-SW orientation (set 3 in Fig. 6b), and undetermined kinematics (likely normal-sense). Trend (iii) is com- posed of ESE-WNW conjugate bands, with reverse kinematics and low dip angles (30-40 • ). In the field, DBs appear as whitish linear traces with minor undulations and characteristic eye structures whereby they branch and merge (Figs. 6b and 7). In most cases, DBs weather out in positive relief. Frequently, DBs occur in narrow zones (a few millimeters up to 5-15 cm in thickness). Field observations indicate that ESE-WNW DBs are crosscut by NE-SW strike-slip DBs (Fig. 6b). The latter also crosscut the NW-SE/NNW-SSE set ( Fig. 7a-c, f). Bedding is oriented NW-SE and dips gently (< 10 • ) to the S (Fig. 2b, d). Bedding is difficult to recognize on the floor of the quarry because of its low dip and the massive texture of the rock (Fig. 6b). The Turonian Sandstones outcrop on the quarry floor. There are two lithotypes. The first one is represented by massive porous sands with DBs and localized carbonate cementation (see description below). The second one is characterized by a massive calcrete level with tabular geometry (see Supplement S2 for details).
The Turonian Sandstones in the Bollène quarry are characterized by a spatially heterogeneous cementation (Fig. 6b). These diagenetic heterogeneities occur as spherical and tabular nodules (Figs. 7 and 8). Spherical nodules are arranged as isolated bodies within the surrounding host rock (Figs. 7c, g and 8a, d, e) or aggregated in tabular clusters (Fig. 7a, d). Nodules weather out in positive relief. Spherical nodules range in diameter from a few millimeters (0.004-0.005 m) to a few tens of centimeters (0.2 m), whereas tabular ones have a thickness ranging from a few centimeters to 0.1 m and a long axis up to 5 m in length (Figs. 6b and 7a-c). Assessment of the nodule lateral extension is hampered by the presence of vegetation and debris cover, whereas subsurface extension cannot be measured because of the limited vertical exposures of the outcrops. Hence, the values reported here are minimum values. In general, the nodule shape may be approximated by a sphere for which length, width, and thickness are "equal" and by an oblate spheroid for which length and width are larger than the nodule thickness. Carbonate nodule volume ranges from 10 −7 (small spherical nodules) to > 2.5 m 3 (tabular nodules assuming length and width of 5 m and thickness of 0.1 m). In Bollène, the nodules are all spatially and geometrically associated with DBs (i.e., DB-parallel nodules; Figs. 6b and 7). This association occurs in the form of (1) parallelism between DBs and nodules, (2) geometric congruence between the DB trend and Figure 7. Calcite cement occurs isolated (a, c, g) or in clusters (a, d) of (S) spherical nodules and continuous (t) tabular nodules (a-c). Nodules are arranged in compartments parallel to clusters of NW-SE normal bands (a-c) and to NNW-SSE dextral strike-slip bands (f, g). NE-SW/ENE-WSW strike-slip bands displace both the NW-SE bands and their associated nodules (a-c, f). (e) Spherical nodule (about 5 cm in diameter) in spatial superposition with a NW-SE DB that does not displace the cement. The lens cover in (a, f) has a 5.5 cm diameter. The figures (a-d, f, g) are in map view. The position of (a, f) is indicated in Fig. 6b. Hammer length in (b) is 30.5 cm.  Eichhubl et al., 2010;Soliva et al., 2013) endorses the normal kinematic of these bands (see Supplement S1 for details). "Up" refers to the topography. Close-up of spherical (d-e) and tabular (f) nodules from the field. Pen and marker length in (a, f) is about 13.5 cm. The lens cover in (d) has a 5.5 cm diameter. the nodule (or nodules cluster) shape, and (3) confinement of nodules in parallel-to-band compartments. In all these cases, tabular nodules and clusters of spherical nodules are oriented with the major axis (elongation direction) parallel to the DBs and the minor axis (i.e., thickness) perpendicular to them (Figs. 7a and 8a). Unlike what we have seen in Loiano, in Bollène nodules are in compartments among DBs. Carbonate nodules are associated with the NW-SE/NNW-SSE DB set (Figs. 6b and 7a, f). Although this set is conjugate with bands dipping to the SW and to the NE, the tabular cement bodies dip only to the SW (Figs. 2d and 8a-c). No nodules are cut by the NW-SE bands. The NE-SW/ENE-WSW strike-slip bands cut through the NW-SE/NNW-SSE bands and the associated NW-SE-trending carbonate nodules (Figs. 7a-c, f). There is clear evidence of these crosscutting relationships both at the outcrop and at the microscale (see Sect. 5.2). For this reason, we focus on the NW-SE DBs and nodules in the remaining part of this study.
5 Deformation bands and cement: textural and microstructural characteristics

Loiano
Host rock total porosity (minus cement φ) is between 20 % and 26 % ( Fig. 9; Del Sole and Antonellini, 2019). Porosity is predominantly intergranular, whereas intragranular (e.g., pores within bioclasts) and "oversize" pores are due to dissolution of detrital grains (Figs. 9 and 10). Deformation band total porosity (minus cement φ) is lower by an order of magnitude (below 5 %) than the host rock porosity (Fig. 9d). In the nodules, the host rock porosity is almost completely filled by cement (Fig. 10) so that the remnant porosity (voids) is low (down to 1.3 %) (Fig. 9a, c). The presence of cement within the DBs enhances the porosity reduction caused by grain crushing and compaction (Fig. 9d). The microstructure of the DBs is characterized by reduced grain size, porosity, and pore size compared to the host rock ( Fig. 10i-l). Within the DBs a few coarse grains are surrounded by a fine-grained matrix.
Despite the different effects of mechanical and chemical compaction as well as minor authigenic alterations (refer to Supplement S1 for details), the major diagenetic components of the Loiano Sandstone are calcite cements. These cements fill mainly intergranular, and to a lesser extent intragranular (intraskeletal), pore spaces and intragranular fractures, and they encase the framework grains and all other diagenetic features. Bedding-parallel nodules ( Fig. 10a-h) are characterized by a mosaic texture of blocky sparite to poikilotopic bright-orange to orange CL calcite cement. Crystal size is typically 40-100 µm and up to 300 µm (Fig. 10g, h). This cement phase is the most widespread one and it is uniform almost everywhere in terms of texture and CL pattern, if not for some minor dark CL subzones (Fig. 10g, h). A minor calcite cement phase is associated with detrital carbonates (bioclasts) and it shows a bright-orange CL (Fig. 10cd). It occurs as pore lining formed by elongate and sharp rhombohedral calcite (dogtooth) that outlines the outer rim of the bioclasts (Fig. 10c) and drusy mosaic calcite that fills intraskeletal pores, outlines bioclasts, and fills intragranular fractures in bioclasts (Fig. 10d). This subordinate phase was observed only in bedding-parallel nodules. Pore-lining cement around bioclasts is present only where there was pore space (now filled), whereas it is absent where other grains are in contact with the bioclast. All cement phases described above (intergranular, intraskeletal, and pore lining) encase compacted grains, and cements are undeformed, still preserving the original shape. Pore-filling cement in DB-parallel nodules (Fig. 10i-n) shows a similar texture and CL pattern as that described for the main intragranular cement phase in the bedding-parallel nodules (Fig. 10a-h). The main features that differentiate DB-parallel nodules from beddingparallel ones are the finer crystal size of calcite within DBs ( Fig. 10i-l) and the absence of a bright-orange CL cement phase described in association with bioclasts in beddingparallel nodules (Fig. 10c, d). Some bright-orange CL cement was observed only in detrital form (crushed) within the DB (Fig. 10k). The pore filling in DBs is fine-grained sparite; no evidence was found of crushed calcite crystals belonging to the dominant calcite phase. The finer fraction within the DBs is a matrix made up of comminuted angular and finegrained clasts (flakes) of feldspar and to a smaller degree quartz, encased by the cement (Fig. 10j-l). Similarly, the cement fills the microfractures that cut through coarser grains and it encase the fine-grained clasts that are present within these fractures. Although these microfractures are frequent in the host rock sectors in proximity to the DB (Fig. 10i), they were also observed in bedding-parallel nodules (Fig. 10a, b). Host rock volumes within DB-parallel nodules are still characterized by blocky sparite cement with some minor dark CL growth subzones (Fig. 10i, m, n), similar to what was observed in bedding-parallel nodules.  To evaluate any sign of dissolution in nodules, we carefully checked cement crystal morphologies adjacent to poorly cemented or non-cemented host rock sectors at the edges of nodules. Here, cement crystal boundaries are regular and sharp (Fig. 10m, n).

Bollène
The host sands at Bollène are weakly cemented, with the exception of localized carbonate cementation described above. Host rock grains are mostly rounded and lack a fabric (Fig. 11a, f). Here, we describe the microstructure of NW-SE/NNW-SSE normal-sense and strike-slip bands, as well as NE-SW/ENE-WSE strike-slip bands sets. The most recognizable features that characterize both DB sets are the reduction of grain size and porosity, as well as a tighter packing relative to the host rock (Fig. 11). NE-SW strike-slip bands ( Fig. 11i-l) have a higher degree of grain comminution, porosity reduction, and tighter packing when compared to NW-SE bands (Figs. 8c and 11g-h). Most grains within the bands are fractured and angular. Despite the strong comminution, a few rounded large survivor quartz grains are preserved in the DB matrix (Fig. 11g, h, k). Fine angular grains that are mostly comminuted feldspar fragments and secondary quartz and minor oxides make up the matrix. We also observed fine particles of crushed calcite cement among the matrix grains within NE-SW bands (Fig. 11k, l). In some cases, the grains in the host rock areas in proximity to the DB are encased by relatively undeformed carbonate cement (Fig. 11i, j). Some grains in the host rock are corroded and partially replaced or coated by calcite cement (Fig. 11c).
The main cement in spherical and tabular nodules is a poikilotopic calcite that infills intergranular pores (Fig. 11ae). Most of the cement is non-luminescent (dark luminescence) under CL (Fig. 11c, d), but a few crystals show partial overgrowths with a bright-orange CL color (Fig. 11e). When the crystal has a heterogeneous CL pattern, the nonluminescing zones are mainly in the crystal core, whereas the luminescing subzones are mostly at the crystal edges (Fig. 11e). A very thin film (up to ca. 10 µm thick or less) of bright-orange CL calcite cement commonly coats the detrital grains (Fig. 11c), and it is also visible under natural light (Fig. 11b). In the nodules, some of the intragranular microfractures at contact points are filled by cement (Fig. 11a); a few are not (Figs. 11b). The cements described above (pore filling and grain coating) are relatively undeformed (i.e., no microfractures, no twin lamellae) and still preserve the original shape, except where the NE-SE/ENE-WSW strike-slip bands crosscut the cement nodules. At the crosscutting site, indeed, and more specifically in the host rock sectors in proximity to the NE-SW/ENE-WSW bands, we observe intragranular fractures at contact points and the onset of cement comminution between quartz clasts (Fig. 11i, j). Fine particles of crushed detrital calcite cement are found among the cataclastic matrix grains within NE-SW/ENE-WSW strike- slip bands where they interact with nodules (Fig. 11k, l). At the microscale, no preferential or significant calcite cementation was observed in association with the NW-SE bands. The association between cements and these latter bands was observed only at the mesoscale (see Sect. 4.2).
6 Cement stable isotope geochemistry 6.1 Loiano Cement from the nodules of the Loiano samples has δ 13 C values between −7.68 ‰ and −1.47 ‰ (VPDB) and δ 18 O values between −4.42 ‰ and −1.35 ‰ (VPDB) (Fig. 12a). The cement from DB-related nodules is characterized by isotope compositions between −5.41 ‰ and −1.47 ‰ (VPDB) for δ 13 C and between −4.42 ‰ and −1.40 ‰ for δ 18 O (VPDB). The cement from bedding-parallel nodules has isotope compositions between −7.68 ‰ and −5.94 ‰ (VPDB) for δ 13 C and between −2.09 ‰ and −1.35 ‰ (VPDB) for δ 18 O. Both cement groups (DB-parallel and beddingparallel nodules) have a relatively narrow range of oxygen isotopic composition featuring a nearly vertical alignment in the δ 18 O-δ 13 C cross-plot. DB-parallel nodules show a slightly wider span of δ 18 O composition when compared to bedding-parallel nodules. However, carbon isotopic composition shows a wide range of variability when considering both the total isotopic composition data and the cement group data.

Bollène
Stable isotope analysis of the Bollène samples also defines two groups of data in the δ 18 O-δ 13 C space (Fig. 12b). The cement group referring to the DB-related nodules has δ 13 C values between −7.73 m and −4.68 ‰ (VPDB) and δ 18 O values between −7.70 ‰ and −5.88 ‰ (VPDB). The other group is from cement sampled in a calcrete level observed within the same Turonian Sandstone a few meters above the studied outcrop (see Sect. 4.2 and the Supplement S2 for details), and it is characterized by isotope compositions between −2.54 ‰ and −2.39 ‰ (VPDB) for δ 13 C and between −6.58 ‰ and −6.32 ‰ (VPDB) for δ 18 O. Both cement groups have a relatively similar δ 18 O signature and a relatively narrow range of δ 18 O composition varying only between −7.70 ‰ and −5.88 ‰ (VPDB). In a similar way, cement sampled from the calcrete has a narrow range of δ 13 C composition and shows heavier δ 13 C values in the data set. However, the δ 13 C composition of DB-related nodules has a wider variability range and is the most depleted in the data set.

Discussion
In the following, we compare the two field sites, highlighting their similarities and differences concerning the interaction between deformation, fluid flow, and diagenesis. We discuss the influence of DBs on fluid flow and their role in enhancing diagenesis and localizing diagenetic products (nodules). Finally, we propose an explanation for the geochemical environment within which fluids were sourced and precipitated the nodule cement. We then explore the implications of SDHs for subsurface fluid flow and reservoir characterization.

Cement distribution and its relationship with deformation bands
The distinctive feature of the Loiano Sandstones is a spatially heterogeneous cementation in the form of nodules. Field evidence indicates that DB formation predates calcite cementation. All nodules are spatially related to DBs (Figs. 3 and 5ac) except for those that are situated along bedding planes (∼ 25 % of the total nodules; Figs. 3 and 5d, e). In contrast, not all DBs are associated with nodules. A clear correspondence always exists between the shape and elongation direction of the nodules and the DBs direction. This pattern is also observed from aerial photographs (Del Sole and Antonellini, 2019). Localized cement along these structural features is itself an indication that deformation preceded cement precipitation (e.g., Eichhubl et al., 2004). If the sandstones were completely cemented at first and then completely removed except from the DBs, the cement in the DB matrix would most likely have been preserved in orange-reddish hues (oxidation residues), but nothing like that was observed. Moreover, cement morphologies adjacent to the porosity at nodule edges imply that cement dissolution has not occurred (Fig. 10m, n), thus excluding the possibility that nodules, both those parallel to bedding and those parallel to DBs, are relicts from an overall dissolution process. No DBs crosscut the cement, at least for those sets that are spatially related to nodules (NNW-SSE to NE-SW), indicating that cementation postdates DB development. In support of this, the precipitation of cement and (the consequent) lower porosity would favor the formation of joints over DBs in the sandstone (Flodin et al., 2003;Aydin et al., 2006; Fig. 5a, b). The presence of pore-filling cement would increase the strength of the sandstone (Del Sole et al., 2020), preventing rotation and sliding of particles, and increase rock cohesion (Bernabé et al., 1992) and grain contact area, thus yielding a uniform contact stress distribution and higher stiffness (Dvorkin et al., 1991). Extensive cement, then, would inhibit DB development.
Results from microstructural observations show that intergranular cement in the nodules encloses the grains within both host rock and DBs, and it overprints burial-related mechanical and chemical compaction features (Fig. 10a-h). This evidence suggests that the formation of authigenic cements occurred after significant compaction (Cibin et al., 1993;Milliken et al., 1998). Estimated burial depths for the top of the Loiano Sandstones are 800-1000 m (Cibin et al., 1993) and 700-1200 m (McBride et al., 1995). Transgranular microfractures at grain contacts are due to stress concentration at contact points and they are interpreted as load-bearing structures within the granular framework (e.g., Eichhubl et al., 2010;Soliva et al., 2013). In DBparallel nodule samples the cement that fills the transgranular fractures is in continuity (i.e., same textural and CL characteristics) with the pore-filling cement outside the fractures. The presence of undeformed cement within structural-related features such as microfractures and crushed grains (Fig. 10il), both within and outside the DBs, proves that cement precipitation occurred after (at least after the early stages of) deformation.
The bands are the main controlling factor on the location, geometry, and elongation direction of DB-parallel nodules. The occurrence and location of bedding-parallel nodules are instead controlled by grain size and contrast in grain size within the host rock. Although bedding-parallel nodules are found in all sands, they are more common within coarse-grained levels (≥ medium sands; i.e., size range: 0.25-0.5 mm). There are no nodules in sediments below the sand range or in layers with permeability below 100 mD . Moreover, bedding-parallel nodules are often restricted to the sand level in contact with clay-silty levels (Fig. 5d, e) above and/or below. Hence, grain size and permeability variations are the most important factors control-ling diagenesis and nodule formation in Loiano. The grain size and permeability variations as dominant controls on nodule development in porous media is also reported by other authors (Mozley and Davis, 1996;Hall et al., 2004;Davis et al., 2006;Cavazza et al., 2009;Balsamo et al., 2012). In general, bedding-parallel nodules show a more rounded morphology when compared to DB-related nodules. The former nodule type owes its smooth morphology to a homogeneous and isotropic weathering; the sharp and squared shapes of DB-parallel nodules is probably due to the anisotropy introduced by the DBs in the host rock that influences the cementation. The interplay between band strength and erosion may also have an influence on nodule shape.
In the Bollène quarry, all calcite nodules occur in association with the DBs, in particular with the NW-SE/NNW-SSE set (Fig. 6b). At this site, we observe complex relationships among multiple deformational and diagenetic events. Timing of bands and nodules is inferred from crosscutting relationships. There is no evidence of low-angle ESE-WNW reverse-sense DBs crosscutting the cement nodules, whereas NE-SW-to ENE-WSW-trending strike-slip DBs offset the reverse-sense bands, the NW-SE bands, and the NW-SE-trending cement nodules (Figs. 6b and 7). The localization and parallelism between DBs and cement are similar at the two field sites, with the exception that NW-SEtrending nodules and DBs in Bollène are not superposed. Here, DBs are always overprinted by cement but the spatial overlap between DBs and nodules (Fig. 7e) is unusual. Nodules occur in compartments that are spatially confined by DB zones. Tabular nodules and clusters of spherical nodules are oriented with the major axis parallel to the NW-SE DBs (Figs. 6b and 7a). The NW-SE bands do not crosscut the cement; therefore, calcite cementation occurred between the NW-SE band formation (Pyrenean contraction or Oligocene-Miocene extension?) and the NE-SW strike-slip bands (Miocene-Quaternary age Alpine shortening?). Please refer to Supplement S1 for details on how DBs relate to the tectonics of the area. Microstructural observations show that the dominant phase of intergranular calcite cement encloses the grains within the nodules, and it overprints only a proportion of the transgranular microfractures at grain contact points. All microfractures in the nodules are filled by a cement that is in continuity (same texture and CL characteristics) with the pore-filling cement outside the grain. Unfilled microfractures (Fig. 11b) were not connected to the pore network, and they were potentially quickly isolated by the calcite mineral growing in the pore space. It is less likely that they formed after cement precipitation; otherwise, the cement would have been broken.

Role of deformation bands in fluid flow and diagenesis
The localized diagenesis observed in the form of nodules at Loiano and Bollène provides evidence for the effect of struc-tural heterogeneities, such as DBs, on fluid flow in porous sandstones (Eichhubl et al., 2004Balsamo et al., 2012;Philit et al., 2015;Del Sole and Antonellini, 2019;Pizzati et al., 2019;Del Sole et al., 2020). The petrophysical properties (porosity, permeability, capillary entry pressure) of DBs influence fluid flow and localize diagenesis and cement precipitation. Cataclastic DBs increase flow tortuosity in reservoirs and produce capillary barriers that severely baffle the flow at the reservoir scale and limit cross-flow between host rock compartments (Harper and Mofta, 1985;Edwards et al., 1993;Lewis and Couples, 1993;Leveille et al., 1997;Gibson, 1998;Antonellini et al., 1999Antonellini et al., , 2014Sternlof et al., 2004;Rotevatn and Fossen, 2011;Ballas et al., 2012;Medici et al., 2019;Romano et al., 2020). Smaller pores within bands result in higher capillary forces than in the host rock. This may cause higher water saturation within the bands with respect to the host rock (Tueckmantel et al., 2012;Liu and Sun, 2020). A higher degree of flow tortuosity (reduction in pore interconnectivity) and lower porosity and permeability within the bands may increase the fluid retention time regardless of the water saturation conditions (Antonellini et al., 1999;Sigda and Wilson, 2003;Wilson et al., 2006). Recently, Romano et al. (2020) documented with single and multiphase core flooding experiments that cataclastic bands can strongly influence the fluid velocity field. Other authors (Taylor and Pollard, 2000;Eichhubl et al., 2004) recognized that a slower rate of solute transport relative to the fluid within the bands causes the formation and local perturbation of diagenetic alteration fronts. In light of these considerations and the temporal and spatial relationships between bands and cements obtained from field and microstructural observations, we discuss a model for selective cement precipitation associated with DBs. In our model we assume a reservoir in saturated conditions (see also Sect. 7.3).
A marked grain surface roughening and reduction of grain size, porosity, and pore size characterize the DBs presented in this work. In Loiano, the combined effect of cataclasis and compaction in the DBs causes porosity reduction by 1 order of magnitude, permeability reduction by 3 orders of magnitude, and advective velocity reduction by 2 orders of magnitude with respect to the host rock (Del Sole and Antonellini, 2019; Supplement S3). Similarly, DBs in the Bollène quarry have lower permeability (up to 3 orders of magnitude) and porosity (up to 50 %) (Ballas et al., 2014; see also Supplement S3) when compared to the host rock. A lower permeability, a higher degree of tortuosity (i.e., lower pore size), and reduced section area available for flow (i.e., lower porosity) in the DBs compared to the host rock may cause a flow "slowdown". In Loiano, the slowdown effect would be more pronounced when considering the normal-to-DB flow than the parallel-to-DB one given that normal-to-DB permeabilities are lower (1 order of magnitude in average) when compared to parallel-to-DB ones . Cata-clasis has competing effects on advective flow velocity; it causes (i) an increase in flow velocity linked to the porosity reduction and (ii) a decrease in the hydraulic conductivity (if the hydraulic gradient does not change). The decrease in hydraulic conductivity (3 orders of magnitude; Del Sole and Antonellini, 2019) dominates over the flow velocity increase caused by porosity reduction (1 order of magnitude; Del Sole and Antonellini, 2019). As a result, there is a net decrease in advective flow velocity in the DBs. A reduction in flow velocity (i.e., slower flow path in the DB with respect to the host rock) might increase the residence time of the fluid migrating through the reactive material (see next paragraph).
A first mechanism responsible for cement nucleation in association with DBs would be the presence of highly reactive crushed and pervasive fractured siliciclastic grains within the cataclastic DBs (e.g., Lander et al., 2009;Williams et al., 2015). The comminuted material of the DBs has a large amount of reactive surface area (nucleation spots) and very tiny pore spaces among the crushed grains. With these conditions, cement precipitation requires less free energy to occur (Wollast, 1971;Berner, 1980), whereas greater cement abundances (e.g., Walderhaug, 2000) and faster rates of cement emplacement (Lander et al., 2008;Williams et al., 2015) are promoted. Despite the fact that the role of fracturing in promoting cement precipitation in sandstones has been essentially explored for quartz cement, we think that this mechanism can be applied to calcite cement too. There is plenty of evidence of calcite precipitation over a silica substrate (e.g., Stockmann et al., 2014), but it would require either more time or a higher degree of supersaturation (e.g., Noiriel et al., 2016) to occur (i.e., to lower the energy barrier for nucleation) when compared to a carbonate substratum. This mechanism has already been proposed to explain the presence of (quartz and calcite) cement within the band pore space and contrast in the degree of cementation between the bands and the surrounding rock Knipe et al., 1997;Fisher and Knipe, 1998;Milliken et al., 2005;Philit et al., 2015;Del Sole and Antonellini, 2019;Pizzati et al., 2019). In this work we show that DBs are a preferred site of cement precipitation. Within the DB there are more nucleation spots for cement nucleation and smaller pores that lead to fast pore clogging. DB pores close faster than host rock ones. Once the cement begins to precipitate, the fresh carbonate substrate (cement) could further enhance precipitation (e.g., Noiriel et al., 2016). This mechanism may be relevant for Loiano where the calcite cement fills small pore spaces among fresh quartz and feldspar surfaces created during fracturing. This process can explain why in most cases DBs are more cemented than the surrounding host rock. On the contrary, this process was less relevant in the Bollène quarry where the bands are not cemented by carbonate and the cementation is localized in compartments between zones of bands rather than within them. The arrangement of nodules in Bollène indicates that the low-permeability DBs hin-dered the cross-flow, restricted the fluid flow, and focused the diagenesis to parallel-to-bands compartments.
A second mechanism could have worked in combination with the presence of more reactive fine-grained comminution products to promote cementation in the DBs in Loiano. According to their experiments on an analog fault gouge, Whitworth et al. (1999) suggested a membrane behavior for faults in sandstone during cross-fault flow and solute-sieving-aided calcite precipitation. A membrane effect and solute sieving by faults may locally increase the concentrations of components needed for calcite cementation (e.g.. Ca and bicarbonate) on the high-pressure side of the membrane and induce precipitation. The DBs could have acted as a semipermeable membrane in baffling chemically reactive flow and favor cement precipitation. This process may explain a higher concentration of cement along the DBs in nodules and the asymmetric distribution of cement on one side of DBs (upstream side; Figs. 4c, e and 9a, e). An analogous mechanism was proposed by other authors to explain the occurrence of the preferred and asymmetric distribution of the authigenic alterations (carbonate and clay cements, Eichhubl, 2001;hematite bleaching, Eichhubl et al., 2004) on the upstream side of DBs in sandstones.
Other factors that may have locally favored (the initiation of) calcite cement precipitation are the growth of cement on detrital grains (Loiano, Fig. 10e; Bollène, Fig. 11c) and the presence of broken detrital carbonate clasts (e.g., shell fragments) that act as a "seed" (cement nucleation sites) (e.g., Bjørkum and Walderhaug, 1990). The latter case was observed in Loiano, mainly in bedding-parallel nodules (Fig. 9c, d). The mechanisms discussed above explain how and why cement precipitation would occur within the band and in its proximity, as observed on-site. Our field observations confirm the theoretical and flow simulations as well as the analog experiments, which demonstrated that DBs can negatively affect the fluid flow in porous sandstones (e.g., Rotevatn and Fossen, 2011;Antonellini et al., 2014;Romano et al., 2020) and enhance cement precipitation (e.g., Lander et al., 2009;Williams et al., 2015).

Structural diagenesis scenario for carbonate nodule formation
We integrate petrographic observations and the stable isotope characterization of cements here with the mesoscale spatial organization and microscale textural relationships between nodules and DBs to discuss the geochemical conditions and potential fluid sources that controlled the formation of carbonate nodules in the studied areas. In Loiano, the first calcite cement to precipitate was the intraskeletal and porelining cement associated with bioclasts in bedding-parallel nodules ( Fig. 9c-d). The cement fabric and textures, circumgranular dogtooth and void-filling drusy mosaic, suggest a phreatic environment (Longman, 1980;Moore, 1989;Adams and Diamond, 2017). Drusy calcite spars can result from re-placement of aragonite in bioclasts in meteoric environments (Flügel, 2013). The second, more pervasive phase of cementation is documented by the intragranular cement observed in all the nodules. The mosaic of blocky sparite with coarse crystals and homogeneous distribution also point to phreatic conditions (Longman, 1980;Flügel, 2013;Adams and Diamond, 2017). The intergranular cement pattern is analogous in DB-parallel nodules and bedding-parallel nodules, meaning they probably formed in a similar phreatic environment.
Oxygen isotope data suggest a meteoric environment (Fig. 12a). According to the compilation made by Nelson and Smith (1996), the moderately depleted δ 18 O and δ 13 C values we found in both types of nodules (Fig. 12a) are grouped in the field of "meteoric cements" (Nelson and Smith, 1996) and are in support of precipitation under meteoric conditions.  Giustini et al., 2016). The isotopic signal of the cementing fluids may have been buffered by interaction with the host rock (dissolution of detrital marine shells in the sandstone framework; e.g., Fig. 10c-d), by mixing with local formation water (McBride et al., 1995), or, for instance, by more regional factors (e.g., atmospheric temperature, humidity, precipitation, seasonality of precipitation and recharge, vegetation; Cortecci et al., 2008;Jasechko, 2019). In this view, further investigations would be needed to properly constrain the cementing fluid conditions (e.g., clumped isotopes, fluid inclusions). The δ 13 C composition of cement (Fig. 12a), however, seems to reflect soil weathering processes as the primary source of bicarbonate dissolved in the waters (Hudson, 1977;Nelson and Smith, 1996) and supports precipitation from meteoric fluids. DB-parallel and bedding-parallel nodules show a similar composition for δ 18 O; however, bedding-parallel nodules have more depleted δ 13 C values. This might reflect a higher contribution of organic carbon from soil-derived CO 2 (Hudson, 1977), possibly indicating that bedding-parallel nodules formed in shallower conditions with respect to DB-parallel nodules. Another explanation is that the two types of nodules were formed by different episodes of water inflow with different (external) environmental conditions. The difference in isotopic composition between these two types of nodules indeed suggests different cement precipitation timing and water compositions as proposed by McBride et al. (1995) and Milliken et al. (1998). Phreatic meteoric conditions for nodule formation point to a shallow diagenesis, and it is consistent with the shallow burial depths estimated for the Loiano Sandstones (see Sect. 7.1).
Cementation patterns can be used to infer the paleo-fluid flow direction at the time of calcite precipitation (Mozley and Goodwin, 1995;Mozley and Davis, 1996;Cavazza et al., 2009;Eichhubl et al., 2009;Balsamo et al., 2012). The different spatial arrangements between DBs and nodules in Loiano make the paleo-fluid flow direction reconstruction challenging. The asymmetric distribution of cement in some nodules associated with DBs can be explained by lateral fluid circulation (Fig. 13a), and cement would accumulate on the upstream side of the DBs (Figs. 4c,9,and 13a). In other cases, cement is roughly symmetrical with respect to the bands, or it is placed where conjugate bands intersect (Figs. 4b, 5c, and 13a). The most likely interpretation is that both lateral flow under saturated conditions and "direct" meteoric infiltration from the surface, with percolation through the rock, contributed to the formation of nodules in Loiano (Fig. 13a).
Calcite (i.e., diffusive supply of Ca 2+ and HCO 3− ) is possibly derived from the infiltration of CaCO 3 -saturated meteoric fluids carrying soil-derived CO 2 (Hudson, 1977;Nelson and Smith, 1996), and/or it is locally derived from detrital carbonate grains in the sandstone layers or from intra-formational shale beds and calcite-rich clays layers (McBride et al., 1995;Milliken et al., 1998). In both scenarios calcite precipitates in correspondence to zones of DBs (DB-parallel nodules) and close to low-conductivity layers (bedding-parallel nodules). McBride et al. (1995) suggest that calcite precipitation along faults (DBs) in Loiano was induced by the mixing of locally derived formation water with meteoric water introduced along the faults or, alternatively, by a loss (exsolution) of CO 2 along the fault zones. These mechanisms, however, imply that DBs were fluid conductive. This hypothesis is at odds with our measurements of the DB hydraulic behavior (Del Sole and Antonellini, 2019;Del Sole et al., 2020). More likely, carbonate DB cementation resulted from CO 2 -saturated groundwater (Fig. 13a). We cannot exclude the possibility of a role played by normal faults in the area (Picotti and Pazzaglia, 2008;Picotti et al., 2009;Fig. 2a) that might have steered regional subsurface fluid circulation. These faults could have cut through top and/or bottom seals and driven fluid migration from above and/or underneath aquifers (Fig. 13a). Episodic fault activity can also favor (episodic) horizontal fluid migration along layering at the time of faulting, possibly explaining the occurrence of nodules (Fig. 13a) and their different isotopic signature (Fig. 12a). From our observations, we can say that the selective cementation process in the Loiano Sandstones depends on "regional" hydrological factors (e.g., topographic gradient, bedding, faults?) locally coupled to the presence of DBs.
In the Bollène quarry, the relative timing of DB formation and cementation in the Turonian Sandstones is complex to unravel. Carbonate cementation occurred between distinct deformation phases with multiple DBs forming (see Sect. 7.1). The dominant dark cathodoluminescence pattern and homogeneously distributed poikilotopic spar texture could suggest an oxidizing (high pO 2 ) meteoric phreatic environment (Longman, 1980;Moore, 1989;Flügel, 2013;Hiatt and Pufhal, 2014). Oxygen isotope data also support a meteoric source for the fluids (Fig. 12a). The range of moderately depleted δ 18 O and δ 13 C values of nodules in Bollène is consistent with a meteoric environment in a continental setting (Nelson and Smith, 1996). The δ 18 O values of nodule cement correspond to parent fluids with an oxygen isotope composition (δ 18 O fluid ) varying between −7.85 ‰ and −5.98 ‰ VSMOW (see Supplement S4 for details on the back calculation of δ 18 O of parent fluids). These values are consistent with the expected δ 18 O of present-day meteoric fluids characterizing this area (e.g., Genty et al., 2014; Jasechko, 2019), thus supporting precipitation from meteoric fluids. Maximum burial depth of the Turonian Sandstone was estimated through stratigraphic constraints to be 400±100 m Soliva et al., 2013). These data support the shallow conditions for nodule diagenesis in Bollène. A phreatic environment is more probable given that in vadose conditions we should have observed meniscus cements and because massive calcrete such as observed in the study area generally forms in a groundwater environment (e.g., Alonso-Zarza, 2003). The δ 18 O values of cement from the calcrete layer correspond to parent solutions with δ 18 O varying from −6.70 ‰ to −6.44 ‰ VSMOW (see Supplement S4), suggesting a meteoric source of fluids. In the vadose zone, DBs would also enhance unsaturated flow relative to the host rock (Sigda et al., 1999;Wilson et al., 2006;Cavailhes et al., 2009;Balsamo et al., 2012).
Field evidence suggests that clusters of low-permeability DBs in Bollène impeded cross-fault flow since no cement was found in superposition with the DBs. The presence of nodules between the DB clusters indicates that the DBs forced the fluid flow and localized the diagenesis in parallelto-band compartments. This evidence and the fact that nodules are homogenous along their elongation direction discredit the hypothesis of lateral flow. The cement could have been originated from (i) downward fluid flow directly from infiltration of meteoric waters or (ii) upward flow of basinal fluid (pressurized aquifer) along fractures and fault pathways in the carbonate rocks (Fig. 13b). In both cases, the water flow was potentially driven by the vertical continuity of DB clusters that have acted as propagation features of faults in overlying (i) or underlying (ii) series and aquifers (Fig. 13b). This scenario might explain why the cement is found only in association with the NW-SE DBs. In both cases (i and ii), the constituent necessary for the precipitation of cement in nodules (i.e., Ca and bicarbonate) would come from the surrounding carbonates. Above the Turonian Sandstones there are several carbonate layers in the upper Turonian and Santonian interval ( Fig. 2c; Ferry, 1997), whereas below there are carbonates belonging to the Jurassic and Cretaceous series (Fig. 2b, c;Debrand-Passard et al., 1984). In the first case (i), continental meteoric waters saturated with meteoric carbon dioxide have dissolved the necessary con- stituents along their path through the rock succession toward the high-porosity Turonian Sandstones. The water percolation through the soil favored fluid acidification. Similar depleted δ 18 O values between nodules and cement from the calcrete level (Fig. 12b) support the (i) hypothesis, and they may have originated from a similar surficial cement source from downward water flow in association with variations of bicarbonate concentration and/or pH in the water table. In the second (ii) hypothesis, nodule cement resulted from CO 2 exsolution during the upward flow of basinal brine or CO 2saturated groundwater in a pressurized aquifer.
7.4 Implications for subsurface fluid flow, reservoir characterization, and resource development Models for calcite cementation are of fundamental importance for predicting sandstone and fault-rock properties such as porosity, permeability, compressibility, and seismic attributes. In Loiano, DBs have acted as fluid flow baffles. First, they buffered the fluid flow and localized cement precipita-tion, acting as areas of preferential cementation in otherwise excellent-porosity sandstones. The resulting diagenetic products enhance porosity and permeability reduction caused by cataclasis, further affecting subsequent fluid circulation. The presence of structural-related cement in the form of concretions (i) strengthens the rock volume, (ii) degrades porosity and permeability, thereby increasing the buffering effect or sealing capacity of DBs, and (iii) imparts mechanical and petrophysical anisotropy to the host rock . We think that it is important to consider the possibility of concretions to form in association with faults within siliciclastic reservoirs, especially where these structures (DBs) are below seismic resolution (e.g., Del Sole et al., 2020). It is also critical to understand SDH spatial organization, extension, continuity, density, hydraulic role in terms of fluid flow circulation, and mechanical influence on the host rock. This information should be included in a robust reservoir characterization, and, in general, it is beneficial during geofluid exploration and energy appraisal, resource development strategies (groundwater, geothermal, hydrocarbon), well production, reservoir simulation modeling, geomechanical evaluation of a drilling site, and other environmental and industrial operations (e.g., waste fluid disposal; groundwater contaminants; geologic CO 2 sequestration; enhanced oil recovery -EOR). The incorporation of this information into aquifer or reservoir (flow) models requires implicit representation of the SDH network and the upscaling of its structural and petrophysical properties (e.g., Fachri et al., 2013;Antonellini et al., 2014). When the cementation is heterogeneous, such as in the examples presented in our work, it could be difficult to model and predict, especially when data are spatially discontinuous (e.g., wells). In these cases, outcrop-based studies allow for continuous and more reliable reconstructions of cement distribution. The characterization of the SDH network distribution (e.g., Del Sole et al., 2020) allows us to predict where (i.e., location and volumes) and how (i.e., spatial organization) the reservoir compartments are arranged and how the fluid circulation can be affected. The kind of study that we present here might be helpful to extract those statistical parameters necessary to implement reservoir studies that account for heterogeneity in petrophysical properties and their association with seismic and sub-seismic structural heterogeneities.

Conclusions
In this contribution, we present two examples of structural control exerted by DBs on fluid flow and diagenesis recorded by calcite nodules strictly associated with DBs. The objective of this research was to constrain the role of DBs in affecting the flow pattern and in localizing cement precipitation in porous sandstones, as well as to elucidate the mechanisms involved in these processes. The major results of our study can be summarized as follows.
1. At both study sites, one or more sets of DBs precede and control selective calcite cement precipitation in the form of nodules. The later localization of cementation along these structural features results in a complex and spatially heterogenous cementation pattern (SDHs).
2. Selective cementation of nodules associated with DBs indicates interaction between deformation structures, fluid flow, and chemical processes. The volumetrically significant presence of cement (10 %-25 % of the exposed outcrop volume) indicates that fluid flow and mass transport have been strongly affected by the presence of low-permeability DBs.
3. Two main processes are discussed to explain selective carbonate cementation associated with DBs in Loiano.
(i) The high concentration of nucleation sites on finegrained comminution products with increased reactive surface area of the pore-grain interface and small pore throats in the DB trigger cement precipitation and fast pore clogging with respect to the host rock. (ii) Solute sieving across the DB (membrane effect) promotes Ca and bicarbonate concentration increase on the upstream side.
4. In Bollène no clear superposition among bands and cement was observed. Here, the clusters of bands acted as hydraulic barriers to cross-flow, thus compartmentalizing fluid circulation and localizing diagenesis in volumes arranged parallel to the bands.
5. In both areas, cement textures, cathodoluminescence patterns, and their isotopic signature suggest that the cement in the nodules precipitated in a phreatic environment from fluids of meteoric origin.
6. In a framework of late-stage diagenesis (post-DB formation) and saturated conditions (phreatic environment), the processes commonly employed to explain focused fluid flow and preferential cement precipitation associated with DBs, such as "transient dilation" and "capillary suction" (see Sect. 1), appear not to be pertinent. In Bollène and Loiano the DBs buffered and compartmentalized fluid flow and localized diagenesis.
7. Further analyses, such as flow simulations and cement precipitation modeling, are deemed necessary to further explore microscale fluid flow and diagenetic mechanisms that drove preferential calcite cement precipitation along DBs in the studied porous sandstones.
8. DBs control flow pattern and affect how diagenetic heterogeneities are distributed within a porous sandstone. The association of diagenetic cementation with DBs further increases the flow-buffering potential of these structural features. It also creates SDHs that impart a mechanical and petrophysical anisotropy to the host rock volume and can seriously affect the subsurface fluid circulation in porous sandstones. These features should be considered during reservoir characterization, especially where SDHs are below seismic resolution.
Data availability. All the data produced and used to write the paper are contained in it and in the Supplement. More detailed information will be made available on request by contacting the corresponding author.
Author contributions. MA and LDS conceived the paper. LDS collected and processed field and laboratory data, provided their interpretations, drew the figures, wrote the paper, and did the revisions. RS and GB contributed to fieldwork at the Bollène study site and to cathodoluminescence data interpretation. MA, FB, and GV participated in fieldwork at the Loiano study site. All authors actively participated in discussing the results and drawing the conclusions, as well as critically revising the paper.
Competing interests. The authors declare that they have no conflict of interest.
Special issue statement. This article is part of the special issue "Faults, fractures, and fluid flow in the shallow crust". It is not associated with a conference.
Acknowledgements. Leonardo Del Sole kindly acknowledges Natalia A. Vergara Sassarini for fruitful discussion concerning the interpretation of cathodoluminescence imaging data and Mattia Pizzati for technical support during sampling for stable isotope analysis. The Laboratoire Géosciences Montpellier is acknowledged for hosting Leonardo Del Sole as a visiting PhD student in the period between April and July 2019, during which the cathodoluminescence analysis and the fieldwork in the Bollène quarry were carried out. The authors also wish to thank Paola Iacumin, Enricomaria Selmo, and Antonietta Di Matteo for stable isotope analysis in the SCVSA Department (University of Parma). Constructive criticism and comments by James P. Evans and Geoffrey C. Rawling greatly improved our paper. This research is part of a PhD project of the first author. Leonardo Del Sole dedicates this work to the loving memory of Antonio Del Sole.
Review statement. This paper was edited by Randolph Williams and reviewed by Geoff Rawling and James Evans.