From oldest to youngest, we collected hand samples (with an average dimension of 15 × 12 × 8 cm) from the Entrada Sandstone, the Wanakah Formation, and the Tidwell Member of the Morrison Formation (Figure 3). Hand samples were georeferenced using hand-held Global Positioning System units (GPS). The geographic orientation of the samples was determined using a magnetic compass, after correction of the magnetic declination (11°E). Hand samples were extracted using a rock hammer to obtain fresh surfaces and to avoid supergene alteration overprints. The sandstones exhibit light brown to pale tan colors in the Tidwell Member of the Morrison Formation (Figures 4A), brown, pink or tan colors in the Wanakah Formation (Figures 4B) and relatively bright red to orange colors in the Entrada Sandstone (Figures 4C). These colors attest to the presence of iron oxides and hydroxides cement in the bulk sample. Polished petrographic thin-sections were prepared for selected, representative specimens to assess the mineralogy, morphology and grain size of the constituent mineral phases. The thin-sections were examined using a Hitachi S-3400N scanning electron microscope (SEM) in backscatter (BSE) mode at Montana State University Billings. Additional analyses with an energy dispersive spectroscopy (EDS) Ametek EDAX system provided semi-quantitative spot chemical analyses. Before imaging, thin-sections were coated with a 45 Å-thick layer of gold/palladium to prevent charging. The operation conditions of 15–20 kV accelerating voltage and a working distance of 10.0 μm yielded the best results.

We cut several serial 2 cm-thick slabs from the hand samples to produce 2 cm sample cubes. The 2 cm AMS cubes were cut, side-by-side, from these slabs to yield about 15 cubic specimen per one hand sample. These oriented cubes were measured using the AGICO KLY-4S Kappabridge susceptibility meter at Southern Illinois University Carbondale. Measurements were performed at room temperature with an applied magnetic field of 300 A/m and frequency of 875 Hz. The AMS measurements were processed using the AGICO SUFAR program while the acquired data were plotted using the AGICO Anisoft 4.2 program. Directional data obtained from the projection of the three principal magnetic susceptibility axes on equal-area lower hemisphere stereonets, including their 95% confidence ellipses, were calculated using the Jelinek statistics (Jelínek, 1978; Supplementary Material).

The AMS is defined by a symmetric, second-rank tensor consisting of six independent elements represented by an ellipsoid with three mutually perpendicular axes. The three axes represent the three principal magnetic susceptibilities namely: K ₁ (maximum), K ₂ (intermediate), and K ₃ (minimum). The AMS parameters used in this study include the mean magnetic susceptibility K m =   ( K 1   +   K 2 +   K 3   3 ) , indicative of the concentration of magnetic minerals present in the hand sample; the degree of magnetic anisotropy P’ where a 1 = l n ( K 1 K m ) , a 2 = l n ( K 2 K m ) , and a 3 = l n ( K 3 K m )   , which describes the strength of the magnetic fabric, and in general is a proxy for the type of magnetic fabric in the rocks; and the shape factor T = 2 ( ( ln ⁡ K 2 − ln ⁡ K 3 ) / ( ln ⁡ K 1 − ln ⁡ K 3 ) ) − 1 , which represents the shape of the ellipsoid of the magnetic fabric (oblate or prolate), with T ranging from −1 to +1, prolate (with T ≤ 0) and oblate (with T ≥ 0).

We performed both magnetic hysteresis experiments and Isothermal Remanent Magnetization (IRM) experiments at room temperature, on rock fragments to identify the nature of the AMS carriers. Magnetic hysteresis was measured on 11 specimens using the MicroMag 3900-04 vibrating sample magnetometer (VSM; Princeton Measurements Corporation) at Southern Illinois University Carbondale. IRM experiments were performed on two representative specimens with contrasting colors (one specimen from each locality). A field strength of up to 1.5 T was applied to saturate the specimens magnetically. For comparison, the induced magnetization in both specimens was normalized after correcting for high-field slope, in this case, 70% of the maximum applied field.

Due to the assumed presence of hematite in the hand samples, we chose thermal demagnetization of the NRM to constrain the different magnetic phases present in each specimen. Thermal demagnetization of the NRM was performed at the CIMaN-ALP laboratory of paleomagnetism (Peveragno, Italy) using a Superconducting Rock Magnetometer 2G Enterprise 755 hosted in a shielded room. Forty-eight representative AMS cubes (16 per formation) were selected for thermal demagnetization of the NRM. The specimens were subjected to a stepwise thermal demagnetization in 15 steps from room temperature to 600°C, with additional four steps up to 700°C for specimens showing higher unblocking temperatures.

Previous work reports on the petrographic characteristics of the Entrada Sandstone, Wanakah Formation and Tidwell Member of the Morrison Formation (Ejembi et al., in review). Based on Folk (1980) sandstone classification, the Entrada Sandstone consists dominantly of sublitharenites, while the Wanakah Formation and Tidwell Member range from feldspathic litharenite to litharenite. The Entrada Sandstone in the two localities is predominantly pinkish to reddish brown, cross-bedded, and is composed of fine to very fine-grained, sub-rounded to rounded framework grains. The Wanakah Formation consists of distinct, thin beds (<0.15 m) of fine-grained sandstone with interbedded mudrocks that have a characteristic red and green color. The Tidwell Member consists of a regionally extensive, dark gray to tan, marker sandstone bed with overlying beds that are tabular, coarse-grained, and are interbedded with mudrocks. These sandstone beds vary in thicknesses (0.7–1.5 m), with a maximum measured thickness of ∼1.5 m at Escalante Canyon.

The quartz and feldspar grains in these units are strikingly free of any transgranular or intergranular deformation fractures and inherited or cognate plastic deformation microstructures such as undulose extinction. Both optical microscopy and scanning electron microscopy reveal very pristine detrital grains assembled in undeformed, fabric-less assemblages. Cross-bedding structures, when present, result from grain size variation. Figure 5, for example, shows representative SEM-BSE images of polished thin sections along with corresponding maps of major elements from the sandstone specimens using x-ray energy-dispersive systems (EDS) analysis. The maps of iron distribution in the matrix reveal no specific coating along grain margins. The EDS chemical analyses help identify quartz, alkali feldspar, calcium carbonates, and iron oxides/hydroxides. However, magnetite and titanomagnetite detrital grains were observed in some of the specimen (based on the Fe, O and Ti elemental maps, e.g., EC-2Jmt and EC-3Jw (Figure 5). This lack of ferromagnetic grains is not unusual in terrestrial sandstones where these minerals are quickly degraded and precipitated as early hematite grain coatings (e.g., Chan et al., 2007).

The hysteresis behaviors show negligible contributions of diamagnetic minerals, minor contributions of paramagnetic minerals, and the dominance of ferromagnetic phases (Table 1 and Figure 6). The magnetic hysteresis parameters, saturation magnetization (M_s), saturation remanent magnetization (M_r), and magnetic coercivity (H_c) are calculated after correction of the high field slope, which arises primarily from the sum of diamagnetic (quartz and carbonates) and paramagnetic (clay and other detrital grains) and antiferromagnetic contributions. The coercivity of remanence (H_cr) was determined through back-field experiments. Two fundamental behaviors are observed: a single phase with a low to moderate remanent coercivity phase (e.g., EC-1Jmt, H_cr ≈ 49.5 mT; Figures 6A); and a two-phase mixture with high- and moderate-coercivity phases resulting in a wasp-waisted shaped hysteresis loop (e.g., EC-1Jes, H_cr ≈ 176 mT; Figures 6B). Most specimens reach magnetic saturation at fields above 0.6 T. A few specimens show a negative high-field slope (e.g., EC-1Jes) indicative of a very small diamagnetic contribution. As the contribution of the high-coercivity phase is highly variable, the M_r/M_s and H_cr/H_c ratios result from a mixture of several single-phase properties and therefore cannot be used as indicators of magnetic granulometry. The relatively low values of the high field slope (Table 2) indicate that paramagnetic phases cannot account for the observed AMS.

Two types of IRM acquisition behavior for specimens with contrasting lithofacies (i.e., color, texture, and porosity) are shown in Figure 7. The IRM plots show distinct approaches to magnetic saturation up to 1.5 T. These differences correlate with the color of the specimen: red sandstones are more magnetically coercitive than tan sandstones.

The AMS scalar parameters (K _m , T, and P’) are plotted in binary diagrams in Figure 8 and provided in Table 2. The specimens display consistent magnetic properties. P’ shows a negative correlation with K _m , both at the scale of the specimen and across specimens. K _m ranges from ∼8.14 to 31.0 × 10⁻⁶ SI (Figures 8A). There is no correlation between the rock color and AMS parameters. P’ ranges from 1.02 to 1.27, clearly outside the range of values attributed to paramagnetic only assemblages, but most specimens have P’ < 1.05 (Figures 8B). T varies substantially within each specimen but overall shows mainly oblate symmetry (Figures 8C). Although the degree of anisotropy in the specimens is strongly influenced by the weak magnetic susceptibility, it does not provide a reliable estimate of the strain at magnetic susceptibilities close to zero.

The stereonets of the three principal magnetic susceptibility axes (K₁, K₂, and K₃) obtained from the sandstone specimens from the two study sites are shown in Figure 9. The 95% confidence ellipses around averages per specimen are derived from the Jelinek statistics (Jelínek, 1978). The AMS directional parameters are rather consistent at the hand sample scale and between specimens. The specimens display a strong and consistent—but varying degree of steepness of the—magnetic foliation (planar fabric) defined by the K ₁–K ₂ plane (Figure 9), which dips to the SE. The double arrow shows the dip angle of the magnetic foliation. The magnetic lineation (K ₁) is moderately well-defined and show discrete clusters. The AMS planar fabric of these rocks is clearly oblique with respect to the sub-horizontal bedding plane observed on the outcrop where each hand sample was collected from.

Stepwise thermal demagnetization of the NRM is shown for six out of forty-eight representative specimens in Figure 10. The NRM intensities range from 3.69 × 10⁻⁵ to 9.91 × 10⁻³ A/m, with an average of 1.86 × 10⁻³ A/m. Characteristic remanent magnetizations (ChRMs) were successfully isolated in 39 (over 48) specimens, both with normal (16 specimens) and reversed polarities (23 specimens). In most cases, the experiments show a stable single-component behavior, with a relatively straight ChRM toward the origin. Few specimens show evidence of a second directional components, isolated at lower temperatures (e.g., BP-1Jmt), including some with normal and reverse components. The ChRMs are scattered, and do not pass the reversal test (McFadden and McElhinny, 1990). The weakly magnetized specimens (which were independent of the formation or lithology) that lack stable demagnetization behaviors could not be interpreted. In most specimens from the Tidwell Member and the Wanakah Formation, the main demagnetization occurs below 150°C while the remainder is generally removed below 300°C (e.g., Figures 10A–D). Most specimens from the Entrada Sandstone also display demagnetization below 150°C, but some retain part of their magnetization up to 500°C–680°C (e.g., Figures 10F). The stable demagnetization directions vary substantially in inclination (Figure 10) with an average ∼55° for the <150°C stable component, while the variations in magnetic declination are also high.

Previous studies have shown that the origin of magnetic fabrics in sedimentary rocks is controlled by sedimentary vs. tectonic processes (e.g., Tarling and Hrouda, 1993; Li et al., 2014). The sedimentary origin of AMS lies in the depositional process by which particles settling in a slurry acquire a shape preferred orientation parallel to the depositional plane, which is close to horizontal in most cases. The compaction of sediments generally strengthens the planar component of the syn-depositional fabric because it results from the application of a vertical stress (gravity).

When bedding is horizontal, there is either no lineation of the maximum susceptibility, or a very weak lineation. Regardless of the magnetic carriers, the resulting AMS tends to be strongly planar and weakly linear, i.e., oblate fabrics (e.g., Schieber and Ellwood, 1993; Tarling and Hrouda, 1993; Saint-Bezar et al., 2002; Aubourg et al., 2004; Aubourg et al., 2010; Parés et al., 2007; Robion et al., 2007; Dall'Olio et al., 2013). Sedimentary magnetic fabrics in sedimentary rocks are characterized by a tight clustering of the K ₃ axis along the bedding pole, with the K ₁ and K ₂ axes dispersed within the bedding plane (Figures 11A). In rare cases, eolian sedimentary rocks may acquire a linear fabric parallel to the wind direction (e.g., Lagroix and Banerjee, 2002). For materials deposited under a strong current, a linear fabric parallel to the current may be imparted, although such fabrics are typically reported in shales and rarely in sandstones (e.g., Ellwood and Howard, 1981; Schieber and Ellwood, 1993; Tarling and Hrouda, 1993), and the amplitude of the imbrication angle of deposited sediments is very small compared to the strongly oblique magnetic fabric observed in our specimens (Figures 11B).

Tectonic overprints of a sedimentary fabric can be detected with AMS even for anisotropies as low as 1% (1.01), e.g., Burmeister et al. (2009). In clay-bearing rocks, such overprints result in a reorientation of platy minerals perpendicular to the instantaneous shortening axis (i.e., the LPS) due to tectonic strain. In sandstones, tectonic overprints are typically identified through specific microstructures including transgranular or intra-granular fractures, recrystallization rims, lobate grain boundaries and outcrop-scale structures such as stylolites or deformation bands (e.g., Saint-Bezar et al., 2002).

In the case of the Uncompahgre Uplift rocks investigated in this study, detailed petrographic observations indicate a lack of plastic deformation, beyond rare detrital quartz grain that exhibits pre-existing undulose extinction. The tectonic uplift that occurred in this region resulted in broad-scale tilting of beds, in places up to 10°. But at our sampling localities, the bedding is close to horizontal (<3 dip). Based on lack of deformation at the scale of the outcrop, the specimen and that of the thin-section, a tectonic origin for the observed AMS fabric can be ruled out. Further, the generally steep attitude (i.e., dip) of the magnetic foliation (∼50°), if it were of tectonic origin, would require an attitude of the shortening axis inconsistent with regional tectonics.

The AMS fabrics of all the specimens are systematically strongly oblique (28–71 dip) with respect to the subhorizontal bedding. Considering this pronounced paradoxical obliquity and the remarkably consistent dip direction to the South East between hand samples from the same site and across the two sampling sites, a sedimentary origin for the AMS can be ruled out.

To better understand the origin of these magnetic fabrics we need to consider the magnetic minerals present in these rocks. At the outcrop scale, the Entrada Sandstone hand samples commonly show a red color (Figures 4C) indicative of the presence of relatively fine hematite. Most specimen also contain relatively large, multi-domain (up to 200 µm) magnetite grains that we interpret as detrital grains based on their rounded and smooth morphology as well as their location in between quartz clasts (Figure 5). The hysteresis behavior of most specimens shows a strong contribution from a high magnetic coercivity phase that cannot be magnetite because this mineral typically saturates at the field of 0.2 T or less (e.g., Dunlop and Özdemir, 1997). Hematite and goethite are, in a diagenetic environment, by far the most likely phases responsible for this behavior. The IRM acquisition curves (Figure 7) confirm the presence of at least one high coercivity phase. Finally, the thermal demagnetization behavior of these rocks shows that in most cases, goethite dominates the NRM, except in the Entrada Formation where hematite contributes more. In general, the magnetic phase carrying the NRM can be—but is not necessarily the same as—the phase carrying the AMS. In summary, most specimens contain a mixture of detrital magnetite (Fe₃O₄), and its typical alteration products; hematite (Fe₂O₃) and goethite FeO(OH). While the absolute percentage of magnetite may be relatively small, this phase contributes substantially to AMS due to its high intrinsic magnetic susceptibility.

While the respective contributions of these three minerals to the AMS of our specimens remain quantitatively unconstrained, their intrinsic magnetic anisotropy may shed light on the origin of the AMS. Magnetite primarily has a magnetostatic anisotropy (i.e., shape anisotropy) that tends to dominate magnetic fabrics because it has a high magnetic susceptibility (Dunlop and Özdemir, 1997). Hematite, in low fields, has mainly a magnetocrystalline anisotropy that mimics its planar hexagonal shape (Rochette et al., 1992). Goethite grains are elongated along their crystallographic c-axis, which is also the K₃ axis and gives goethite an inverse AMS fabric (e.g., Rochette et al., 1992).

Considering the magnetic assemblage observed in our specimens, two alternative explanations can be proposed:

In these studies, hot fluids precipitate iron oxides and hydroxides in pore spaces of a porous rock and the resulting AMS effectively reflects pore space shape anisotropy.

The first hypothesis cannot explain oblique AMS fabrics because mixtures of normal and inverse AMS fabrics systematically result in perpendicular AMS fabrics (Ferré, 2002). It also seems unlikely because the thermal demagnetization behaviors and magnetic properties of the specimens indicate that the proportion of magnetite, hematite, and goethite varies between hand samples (Figure 10 and Tables 1, 2).

The second hypothesis is consistent with most observations including the variations in magnetic mineralogy across formation depending on the availability of iron in the environment and possible variations in rock permeability. The fundamental mechanism driving fluids through these formations could be the tectonic uplift of the Uncompahgre mountains which would have caused a regionally consistent hydraulic head to the West of the sampling localities (Figures 11B). Although rock fractures are visible at outcrop scale (Figure 4), we did not identify any linkage to presence of systematic fracture or pore networks at the microscopic scale (Figure 5) and assessing the potential role of fracture networks in contributing to the regional fluid flow through these formation is beyond the scope of this work. Today, the paleomagnetic data showing a reverse period of magnetization at least in the Entrada Formation (e.g., EC-1Jes) supports a Cretaceous precipitation event associated with the Uncompahgre Uplift, although this process is highly unlikely because iron is a very mobile element and subsequent erosion created by the uplift and Quaternary downcutting of the Gunnison River almost certainly created a late-stage fluid flow event to the west.

Magnetic fabrics have previously been used both to characterize fluid flow through porous rocks (e.g., Sizaret et al., 2003; Just et al., 2004; Essalhi et al., 2011) and to quantify pore anisotropy (e.g., Pfleiderer and Halls, 1990; Pfleiderer and Halls, 1993; Benson et al., 2003). Crystallographic analyses of magnetite grains and whether its preferred grain orientation align with pore spaces in the bulk rock (if it exists) are outside the scope of this work. Here in this study of the Uncompahgre Uplift, contrarily to many similar studies of sedimentary rocks using magnetic fabrics, we show that the AMS does not inform on depositional processes, compaction or paleocurrents. Instead, the AMS reflects a regionally consistent post-depositional, tectonically-driven fluid flow owing to cycling via redox reactions of a highly mobile element (Fe) over geologic history in porous and permeable media.