ABSTRACT

The Wajid Group of southwestern Saudi Arabia consists of a dominantly sandy succession of Cambrian to Permian age that spans several discrete phases in the tectonic evolution of the Arabian Peninsula. The principal aim of this study was to determine whether successive changes in the tectonic setting are reflected in changes in provenance-related mineralogy. Because of the relatively limited compositional range of the Wajid sandstones, heavy-mineral assemblages have been used as the primary tool for assessing changes in provenance signature. A comparison of heavy-mineral and petrological data has, however, also been carried out.

Variation in the relative abundances of zircon, rutile, monazite, tourmaline and apatite has revealed significant changes in provenance signature between the Dibsiyah (Cambrian–Ordovician), Sanamah (Ordovician–Silurian), Khusayyayn (Devonian–Carboniferous) and Juwayl (Carboniferous–Permian) sandstones. Since previous studies have established that northward-flowing rivers deposited the fluvial sandstones of the Wajid Group, it appears that the source area lay to the south. In the absence of data from the region to the south, it is not possible to identify specific source areas. It is clear, however, that the successive changes in provenance signature must reflect exposure of new source rocks through progressive denudation, changes in the pattern of tectonic uplift or changes in the drainage system. It is also possible that some of the observed mineral variation is related to lateral influx of sands through long-shore drift during times of high sea level.

Two distinct mineral compositions occur within the Dibsiyah sandstones, indicating that a major change in provenance took place during deposition of the Upper Dibsiyah sands. The boundary between the Dibsiyah and Sanamah formations is sharply defined, although the overall composition of the Sanamah sandstones is in many respects similar to that of the Dibsiyah sandstones. There is a relatively small difference in composition between the Sanamah sandstones and the associated diamictites. A major change in provenance is indicated at the base of the Khusayyayn Formation, with an increase in the proportion of monazite and staurolite. This change in composition persists into the Juwayl Formation although the greater variability displayed by the Juwayl heavy-mineral assemblages indicates contribution from several sources. Heavy-mineral assemblages in the Juwayl sandstones are comparable to those of the Unayzah C and B sandstones of central Saudi Arabia, but differences suggest mixing between a southern (Juwayl) and western (Shield) source for the Unayzah sandstones.

Compositionally, Wajid sandstones range from quartz arenite to arkose. Comparison of the petrographic and heavy-mineral data is hampered by the different grain-size ranges studied. However, it would appear that samples with similar heavy-mineral provenance character do not necessarily possess similar feldspar percentages, even when the latter are corrected for in-situ kaolinization. The data set is too small to establish an explanation for this apparent discrepancy.

INTRODUCTION

The Wajid Group comprises a succession of Palaeozoic sandstones that crop out in southwestern Saudi Arabia, along the southeastern flank of the Arabian Shield (Figure 1). Its northern limit is defined by the southern flank of the Central Arabian Arch. The Wajid strata extend eastwards into the subsurface and are believed to be in lateral continuity with the Palaeozoic succession of the Rub’ Al-Khali Basin.

In April 2004, Saudi Aramco sponsored a reconnaissance field trip to the Wajid area, led by Saudi Aramco geologists Mike Hulver, Kent Norton, and Abdelmuttaleb Al-Qahtani. The opportunity was taken during this field trip to conduct preliminary sampling for petrography and heavy-mineral analysis.

The Wajid Group provides the principal surface exposure of Palaeozoic rocks in southwestern Arabia and is thus key to understanding the early Phanerozoic tectonic and palaeogeographic evolution of the region. Four of the tectonostratigraphic megasequences (AP2–AP5) of Sharland et al. (2001) are represented, as well as two periods of Gondwana glaciation (latest Ordovician and Permian–Carboniferous). The aim of this paper is to document the evolution of the Wajid heavy-mineral assemblages with time, and in particular to assess the relative roles of tectonism, provenance and climate in determining the composition of the assemblages.

STRATIGRAPHY

The term Wajid Sandstone was introduced by Powers et al. (1966) to comprise the clastic succession that lies between the limestones of the Khuff Formation and the Proterozoic basement between latitudes 17°30'N and 20°30'N and longitudes 43°30'E and 46°30'E, in southwestern Saudi Arabia (Figure 1). Powers (1968) formally defined the Wajid Sandstone Formation and proposed a composite type section comprising outcrops between Jabal Wajid (19°25'N) and Bani Ruhaiya (19°50'N). Subsequently, Kellogg et al. (1986) defined four members within the Wajid Sandstone Formation of the Wadi Tathlith Quadrangle. In ascending sequence, these are the Dibsiyah Member, the Sanamah Member, the Khusayyayn Member and the Juwayl Member. Although Evans et al. (1991) recommended abandonment of these terms in favour of those established for central and northern Saudi Arabia, Stump and van der Eem (1995) retained them (in modified form) on the grounds that they better reflect the distinctive character of the Wajid succession. The revisions that Stump and van der Eem (1995) proposed were to elevate the Wajid Sandstone to group status and the component members to formation status. They also recognised the Qalibah Formation (Qusaiba Member) as an additional component of the Wajid Group. The scheme of Kellogg et al. (1986), as modified by Stump and van der Eem (1995), is followed in this paper.

Scarcity of fossils has hampered precise dating of the Wajid Group succession, but the group is now considered to range from Mid-Cambrian to Permian in age (Stump and van der Eem, 1995). The stratigraphic framework is summarised in Figure 2.

The present study is restricted to sandstones exposed in the central and northern part of the Wajid Group outcrop (Figure 1). A brief description of the four sandstone formations is given below, in ascending order.

The Dibsiyah Formation (c. 170 m) rests unconformably on the Proterozoic basement. Stump and van der Eem (1995) recognised two informal divisions (‘lower’ and ‘upper’) within the Dibsiyah Formation. The basal part of the ‘lower’ Dibsiyah Formation consists of a palaeosol overlain by quartzite pebble-bearing litharenitic and sublitharentic sandstones of highly variable thickness. The remainder of the ‘lower’ Dibsiyah Formation consists of mature, cross-bedded quartz arenitic sandstones. The ‘lower’ Dibsiyah sandstones were interpreted as having been deposited by a northward-prograding braided stream system. The ‘upper’ Dibsiyah consists of cross-bedded quartz arenites. The trace-fossil Tigillites (= Skolithos) is locally abundant, indicating deposition in intertidal environments. The Dibsiyah Formation was given a tentative Mid-Cambrian to ?Early Ordovician age by Stump and van der Eem (1995). It is equivalent, at least in part, to the Saq Sandstone of central and northern Saudi Arabia and is part of a vast sequence of quartz-rich sandstones deposited across northern Gondwana, from North Africa to Arabia. Considerable effort has gone into trying to understand the source of this “continent-wide braided stream system” with a consistently south to north transport direction (Avigard et al., 2003, 2005; Kolodner et al., 2006).

The Sanamah Formation (0 to c. 140 m) occupies channels incised into the Dibsiyah Formation. It consists of a basal conglomerate of mixed pebble type, overlain by structureless, variably pebbly arkosic and subarkosic sandstones with rare diamictites. The bulk of the formation was interpreted as glacio-fluvial in origin by Stump and van der Eem (1995) who argued for a fluvial origin for the ‘tillites’ (diamictites) and ‘striated pavements’ described by some previous authors (McClure et al., 1988; Vaslet, 1990; Evans et al., 1991). The presence of bioturbated sandstones at the top of the formation in the type section (Stump and van der Eem, 1995) indicates an upward transition from a fluvial to marine environment. No biostratigraphically significant taxa have been recovered from the Sanamah Formation, but a Late Ordovician to early Silurian age has been inferred by Stump and van der Eem (1995). It is, therefore, partly equivalent to the Sarah and Zarqa formations, which also bear evidence of Gondwana glaciation in central and northern Saudi Arabia (Evans et al., 1991).

The Qalibah Formation (Early Silurian) rests unconformably on the Khusayyayn Formation. In the Wajid area it is composed mostly of mudstone (Qusaiba Member) and no sandstone samples were available for this study.

The Khusayyayn Formation (minimum c. 200 m) rests disconformably on either the Sanamah Formation or the Dibsiyah Formation. It consists of rather uniform, medium- to coarse-grained quartzose sandstones, pebbly in the basal part. The sediments were interpreted by Stump and van der Eem (1995) as of marginal-marine to upper-shoreface facies in the lower part and continental (fluvial, lacustrine, aeolian) in the upper part. An age range of Early Devonian to ?mid-Carboniferous is suggested for the Khusayyayn Formation by Stump and van der Eem (1995). According to Stump et al. (1995, p. 14, their figure 2), the Khusayyayn consists of a lower division of Devonian age and an upper division of Carboniferous (Visean) age. It is equivalent, at least in part, to the Jauf, Tawil, and Jubah formations of central Saudi Arabia. Lower Devonian fish fossils have been reported from the South Qara Al Madarah roadcut area (Forey et al., 1992).

The Juwayl Formation (0–150 m) occupies two channels cut into the Khusayyayn Formation (locally into the Sanamah Formation). It consists of coarse to fine, partly pebbly sandstones. The pebbles are mostly of quartz, but relatively large clasts (up to boulder size), in the lower part of the unit, consist of sandstone derived from the channel walls. Sediments with diamictite fabric are locally present, but make up only a small proportion of the overall succession. These have been interpreted as tillites by some authors (Helal, 1965, 1966; McClure, 1978, 1980; McClure and Young, 1981; McClure et al., 1988), but are regarded as of non-glacial, fluvial origin by Hadley and Schmidt (1974, 1975) and Stump and van der Eem (1995). The Juwayl Formation outcrops have produced no biostratigraphically significant taxa, but dating of supposed equivalent subsurface successions suggests a late Carboniferous (Stephanian) to Early Permian age (McClure, 1980; Al-Husseini, 2004). It is partly time-equivalent to the Unayzah C and perhaps lower Unayzah B units of central Saudi Arabia (Melvin and Sprague, 2006).

The Juwayl clastics are overlain by the Ruhaiya Limestone in the Bani Ruhaiya area near the eastern limits of the Juwayl outcrop belt. Although originally mapped as Khuff, subsequent integration of outcrop and shallow borehole data in that area suggests that the Ruhaiya Limestone may be older than the middle to upper Permian Khuff (see discussion in McClure, 1980 and Sharland et al., 2001, p. 172). McClure (1980) gives an Artinskian age for the Ruhaiya Limestone (20°00'N, 44°48'E, his figure 1) and a Sakmarian age for the immediately underlying clastics (Juwayl Formation of this paper). Norton and Hulver (2006) recently reported possible aeolian sandstones beneath the Ruhaiya Limestone in outcrops nearby, and they believe the limestone to belong to the Khuff Formation. They further argue that the aeolian sandstone is younger than the Juwayl Formation, probably equivalent to the Unayzah A (post-Artinskian) of central Saudi Arabia.

Samples from all four Wajid sandstone formations were analysed in this study, together with a single sample of present-day wadi silts. Details of individual samples are presented in Table 1, with further qualification for specific samples as follows.

The Dibsiyah Formation samples W02 and W03 were collected from a level believed to be only 15–30 m above the base of the Dibsiyah Formation, but the presence of Tigillites burrows (marine indicators) in sample W02 would appear to preclude assignment to the ‘lower’ Dibsiyah Formation.

The Sanamah Formation samples include three samples that have been identified as diamictites (W08, W11 and W20). Sample W08 is very poorly sorted pebbly sandstone with tabular clasts of mudstone and siltstone up to one metre long. Sample W11 is very poorly sorted pebbly sandstone with rounded quartz pebbles up to 2 cm in length and siltstone/mudstone rip-up clasts up to 10 cm in length. Sample W20 is very poorly sorted conglomeratic sandstone with tabular slabs of (?Dibsiyah) sandstone/siltstone up to one metre long and quartz pebbles set in a sandy matrix.

Two of the Khusayyayn Formation samples (W54A and W68) are from localities where the formation overlies basement. Sample W54A rests on a regolith of weathered granite and consists of small pebbles and granules of quartz, weathered feldspar and tabular clasts of fine sandstone/siltstone up to 5 cm long. Sample W68 is from the siltstone unit reported by Stump and van der Eem (1995, p. 434 and their figure 8: section f) as having yielded late Early Devonian fossil fish faunas.

The Juwayl diamictite sample W34 consists of granules and small pebbles of sandstone and quartz set in a sandy matrix. Samples W44 and W45 are from the uppermost part of the formation at Bani Ruhaiya, near the contact with the overlying Ruhaiya Limestone.

PETROLOGY

Thin-section petrographic analysis was carried out on all of the Wajid sandstone samples studied for heavy-mineral analysis. Primary features are summarised below:

The Dibsiyah sandstones are quartz arenites when kaolinite is recalculated as feldspar (Figure 3a). Monocrystalline: total quartz ratios are high (> 95%) and there are small amounts (< 5%) of biotite and muscovite in most samples. The sandstones typically are matrix-free or have very little detrital matrix. Grain size ranges from fine to very coarse, and some sandstones are bimodal. Trace amounts of K-feldspar are present, but diagenetic kaolinite content ranges up to 15%, suggesting that more feldspar may have been present prior to alteration. Assuming that kaolinite is an alteration product of K-feldspar, restoration of the framework mineralogy (QFL) is shown in Figure 3a. There are no obvious geographic differences in framework grain mineralogy.

The Sanamah sandstones vary in composition from quartz arenites to arkoses when kaolinite content is re-calculated as feldspar (Figure 3b). Most are subarkoses. Grain size ranges from very fine sandstone/siltstone to pebbly sandstone. There is no clear geographic trend in composition where a sufficient number of samples exist to test variability. Sanamah sandstones often contain significant detrital matrix (2–20%). Carbonate cement is typically low except in samples that may be cemented due to recent caliche formation (e.g. W06, W07). Monocrystalline/total quartz ratios (55–98%) are much lower than in the Dibsiyah sandstones and K-feldspar content is higher, commonly 5–18%. Diagenetic kaolinite ranges from 0–12%.

Khusayyayn sandstones are generally fairly clean with little detrital matrix; they range from fine- to very coarse-grained. The Khusayyayn sandstones show a significant variation in restored framework grain composition, from quartz arenites to arkoses (Figure 3c). Like the underlying Sanamah sandstones, most are subarkoses. There is no clear geographic trend in composition and samples at a single outcrop can vary from arkose to quartz arenite over short stratigraphic intervals, e.g. at Khashm Habuna where fluctuations in feldspar content from bed to bed are obvious to the naked eye. At Khashm Habuna, quartz arenites seem to be more common high in the section, whereas arkoses are more common lower in the section. At Jabal al Khashab, all samples are quartz arenites. Although the exact stratigraphic position of these samples is unknown, they are from the easternmost exposures of the Devonian section in the central Wajid area, and presumably would also be high in the Khusayyayn succession. There may therefore be an upward stratigraphic decrease in feldspar within the Khusayyayn Formation, although this cannot be confirmed without additional sampling.

Only 3 of 20 samples contained 5% or more carbonate cement. Kaolinite ranges from 0–10%. Hematite cement is abundant in some samples (up to 20%) but is likely to be due to surface weathering. Monocrystalline:total quartz ratios are mostly > 85% except for one sample from a coarse conglomerate (W64), where the ratio is only 59%. K-feldspar ranges from 0–20% of the bulk rock, even before recalculation of the kaolinite as feldspar.

Juwayl sandstones range from fine- to coarse-grained and pebbly. They are mostly quartz arenites and show the least mineralogical variation of all the Wajid sandstones sampled (Figure 3d). Only two samples (W42 and W43) from near Bani Ruhaiya falls outside the quartz arenite field, and this only after recalculating kaolinite as feldspar. The low feldspar content of the Juwayl contrasts markedly with the arkosic and subarkosic Khusayyayn sandstones that unconformably underlie it.

Juwayl sandstones are generally matrix-free and have little carbonate cement. K-feldspar is below 3% and kaolinite is not abundant – less than 2% in 8 of the 11 Juwayl samples examined. Monocrystalline: total quartz ratios are mostly high (> 90%).

Discussion

Standard QFL plots of the Wajid sandstones are shown in Figure 4. Figure 4a shows the percentage of feldspar actually identifiable in thin section, while Figure 4b shows restored feldspar percentages based on the assumption that associated authigenic kaolinite is derived from the decomposition of former feldspar grains.

Lithic grains are extremely scarce or lacking in the majority of samples. The highest percentage is in the basal Khusayyayn conglomerate W54A, with significant percentages otherwise being restricted to Dibsiyah and Sanamah samples (including all three Sanamah diamictite samples).

Feldspar percentages (%feldspar) show a much greater range than lithic percentages (%lithics) and thus have greater potential for comparison with the heavy-mineral provenance data. Feldspar percentages are, however, known to be strongly influenced by grain size because of the greater availability of feldspar grains of fine sand grade than coarse sand grade. This is illustrated in Figure 5, in which measured %feldspar (Figure 5a) and restored %feldspar (Figure 5b) for the Wajid sandstone samples are plotted against their respective mean grain size. It is apparent from both plots that there is a gross trend of increasing %feldspar with decreasing grain size, with the highest values for each grain size interval being displayed by Sanamah and Khusayyayn samples.

In Figure 5a (measured %feldspar), the majority of the samples display low to very low %feldspar values, whereas in Figure 5b (restored %feldspar) several samples show a shift towards the high-value Sanamah and Khusayyayn samples. However, most of the restored %feldspar values fall short of the trend defined by the highest-value samples. Either these samples never possessed such high %feldspar values or the restored percent feldspar figures are underestimated, i.e. not all of the removed feldspar is represented by kaolinite cement.

The mineralogical and textural maturity of all Wajid sandstones (very low lithics, high quartz content, and moderate to low K-feldspar), coupled with the maturity of their heavy-mineral suites, suggest derivation from a cratonic source with significant reworking and/or weathering. Reworking of older sediments is also evident in the common presence of recycled quartz overgrowths (Figure 6). This indicates reworking of sediments that were previously buried deeply enough to enter the quartz cementation window, generally associated with temperatures in excess of 60–70°C.

HEAVY-MINERAL ANALYSIS

Although it is widely accepted that heavy minerals are valuable indicators of provenance, it is also apparent that the composition of an individual heavy-mineral assemblage is dependent on many other factors (Morton, 1985; Morton and Hallsworth, 1994, 1999), including chemical weathering (leading to removal of minerals susceptible to oxidation or acid dissolution), grain-size availability (minerals that typically occur as small crystals will be under-represented in coarse sands), transport (leading to hydraulic separation of minerals with differing densities and shape) and diagenesis (leading to removal of minerals susceptible to dissolution by the circulation of warm pore waters).

These processes can lead to detrital assemblages that are very different in character to those of their host rocks, with some minerals being absent and others present in very different proportions. Unless highly distinctive minerals are present that can be related to a specific source rock, this modification in the range and proportion of minerals can lead to considerable difficulty in interpreting provenance signatures. Thus the processes of meteoric and diagenetic dissolution of unstable minerals can result in the development of assemblages composed of relatively few ultrastable minerals (typically zircon, rutile, tourmaline and monazite) from what were originally very different heavy-mineral assemblages. Conversely, assemblages derived from a single source rock can display substantial variation in composition because of the effects of grain size availability and hydraulic sorting.

Heavy-Mineral Indices

Because of the effects described above, comparison of heavy-mineral assemblages using percentage data alone can be highly unreliable. Morton (1985) proposed methods to minimise the effects of hydraulic sorting. One approach was the use of two-component indices, which reflect the relative abundance of minerals with similar hydraulic behaviour (i.e. similar density and shape characteristics). A two-component index may relate to two different minerals or to two varieties of the same mineral. A further constraint proposed by Morton (1985) was that heavy-mineral analysis should be carried out on a single size fraction, thus eliminating the effects of grain-size availability. The very fine sand fraction (64–125 microns) is used because heavy minerals of this size can be obtained from sandstones of all grain sizes and from the sand matrix of conglomerates.

Diagenetic effects are minimised by the use of indices based on heavy minerals resistant to diagenetic alteration. Five such heavy-mineral indices are routinely used in provenance studies: rutile:zircon, garnet:zircon, monazite:zircon, chrome-spinel:zircon, and apatite:tourmaline. These are selected because of the resistance of all the component minerals to diagenetic dissolution. These indices are thus equally applicable to deeply buried sandstones as to sandstones that have undergone minimal burial. For sandstones that have experienced little or no burial dissolution (as indicated by a lack of significant surface etching on unstable minerals), it is possible to devise additional indices based on unstable minerals.

Unstable minerals encountered in the Wajid sandstones include, in order of increasing stability: pyroxene, amphibole, epidote, staurolite, kyanite and garnet. As none of these minerals shows significant surface etching, post-depositional dissolution has no effect on the relative abundance of the unstable mineral components. Two of the more abundant unstable mineral species (hornblende and staurolite) are therefore used as mineral index components in this study. Though not subject to the testing undertaken by Morton (1985), these indices (staurolite:tourmaline and hornblende: tourmaline) are considered to provide a useful indication of the relative proportion of the component minerals in the source rocks.

Other indices used in this study are the pink (purple) zircon index (pZi), defined by the percentage of pink (purple) zircons in the total zircon assemblage and the euhedral zircon index (eZi), defined as the percentage of euhedral zircons within the total zircon assemblage. Formulae for the heavy-mineral indices are given in Table 2. The standard chrome-spinel:zircon index (CZi) is omitted because chrome spinel has not been encountered in the Wajid sandstones.

The fundamental premise of these mineral indices is that they are more or less unaffected by the effects of hydraulic sorting and thus provide a reasonably accurate reflection of the relative abundance of minerals (of the same grain-size fraction) in the source rock. The mineral indices provide the most reliable means of comparing provenance character between different sandstone bodies. They thus provide the most appropriate tool for establishing stratigraphic relationships within and between sandstone successions, which is the primary aim of this paper.

Although all of the mineral indices presented in Table 2 are considered to be largely independent of the effects of hydraulic sorting, they are not all independent of the effects of weathering. In particular, the apatite:tourmaline index (ATi) can be affected by the dissolution of apatite by acidic meteoric groundwater, in either the source area or the depositional area. For this reason, ATi values may not reflect the composition of source rocks where weathering has taken place under humid climatic conditions. The hornblende:tourmaline index (HTi) can also be affected by weathering-related dissolution. These two indices are therefore of limited value as provenance indicators except for sediments that have undergone rapid erosion and deposition or those deposited under arid and/or cold-climatic conditions. They can, however, provide useful information on sediment maturity, since the degree of weathering is dependent not only on climate, but also on relief and rates of erosion, transportation and deposition. Variation in ATi values in particular can provide a record of sea-level variation (Morton and Hallsworth, 1999) and thus be of value in correlation cycles of sedimentation between sections.

A key element in the interpretation of all heavy-mineral assemblages is the degree to which the mineral composition may have been inherited from pre-existing sediments. In particular, a high degree of physical and mineralogical maturity does not necessarily reflect intense weathering and prolonged transport; it can equally well reflect repeated recycling of the component sands. Fuller accounts of the concept and interpretation of heavy-mineral indices is provided by Morton and Hallsworth (1994, 1999) and references therein.

The significance of each of the mineral indices is briefly discussed below.

The rutile:zircon index (RZi) is a highly reliable provenance indicator as both rutile and zircon are highly stable. However, this stability also favours recycling, such that much of the signature may be inherited from earlier sediments. Rutile is a common constituent of regionally metamorphosed pelites and basic igneous rocks, whereas zircon is a widely distributed mineral in granites and other plutonic rocks. Variation in RZi values may thus reflect variation in the proportion of metamorphic to plutonic source rocks or variation in the type of metamorphic source rock.

The monazite:zircon index (MZi) is a reliable provenance indicator as both minerals are highly stable. However, this stability also favours recycling, such that much of the signature may be inherited from earlier sediments. Monazite occurs as a minor constituent of both plutonic and metamorphic rocks.

The garnet:zircon index (GZi) is a reliable indicator of derivation from regionally metamorphosed pelitic rocks. Garnet is generally unaffected by weathering, but dissolves in strongly acidic groundwaters and is susceptible to dissolution under conditions of deep burial (greater than c. 3,000 m). Grains undergoing dissolution are characterised by strongly etched grain outlines. This feature is not seen in the Wajid sandstone assemblages and GZi values are therefore considered to be unaffected by post-depositional dissolution.

The chrome-spinel:zircon index (CZi) is a highly reliable provenance indicator. Chrome spinel is associated with ultrabasic igneous rocks and ophiolites, but has not been observed in the Wajid sandstones.

The chloritoid:tourmaline index (CtTi) is an indicator of derivation from alumina-rich metapelites, which are the principal source of chloritoid; tourmaline is most common in granitic igneous rocks. It should be noted, however, that index values may be influenced to some degree by hydraulic sorting, since chloritoid possesses a platy habit, which is likely to make it more buoyant than its density would otherwise indicate. The intraformational variation displayed by this index may therefore be greater than of other indices. Little is known of the stability of chloritoid. However, it appears to be stable under most weathering conditions and appears to be more stable than garnet under conditions of deep burial (Morton and Hallsworth, 1994, p. 245).

The apatite:tourmaline index (ATi) is a potential provenance indicator, but is also strongly influenced by weathering, especially under conditions of low relief and humid weathering. Apatite is a ubiquitous and abundant component of heavy-mineral assemblages in both igneous and metamorphic source rocks. Low or very low ATi values are therefore indicative of a significant degree of weathering of the primary heavy-mineral assemblages or reworking of low-apatite sands. They can also result from leaching of the sandstones at outcrop. This is unlikely to be a significant factor in present-day weathering in the Wajid area, but could possibly have been significant in earlier periods of higher rainfall.

The staurolite:tourmaline index (STi) is a good indicator of derivation from low-grade meta-sedimentary rocks. Staurolite is common in schists, phyllites, and gneisses. It appears to be relatively resistant to weathering, but is susceptible to dissolution under conditions of deep burial (greater than c. 2,000 m). Significant dissolution is indicated by the development of strongly serrated grain outlines. This feature is not seen in the Wajid sandstone assemblages and STi values are therefore considered to be unaffected by post-depositional dissolution.

The hornblende:tourmaline index (HTi) is a good indicator of derivation from basic igneous and basic metamorphic source rocks. However, hornblende, along with other ferromagnesian minerals, is susceptible to decomposition when exposed to oxidising meteoric waters and is also subject to dissolution under conditions of deep burial (greater than c. 600 m). Significant dissolution is indicated by the development of strongly serrated grain terminations. This feature is not seen in the Wajid sandstone assemblages and HTi values are therefore considered to be unaffected by post-depositional dissolution.

The pink zircon index (pZi) indicates the percentage of pink (to purple) zircons within the entire zircon assemblage. It is considered to be a reliable indicator of provenance, but with the proviso that pink zircons are likely to have been supplied from more than one source. The more deeply coloured pink zircons are commonly well-rounded and are likely to be derived from pre-existing sediments. The paler pink zircons range from well-rounded grains of probable sedimentary origin to euhedral zircons that are clearly derived from local igneous or metamorphic basement.

The euhedral zircon index (eZi) indicates the percentage of zircons with euhedral crystal form. For this purpose, the term euhedral is applied to all grains that show well-defined pyramid terminations, including those that display some abrasion of the interfacial angles. Most of the euhedral zircons thus defined are likely to be of first-cycle origin. Variation in the euhedral zircon index may reflect differing degrees of abrasion (physical maturity) or differing provenance (e.g. distinguishing between sands derived from igneous and mature sedimentary source rocks).

WAJID GROUP HEAVY MINERALS

The heavy-mineral percentage data (Table 3) reveals the full range of detrital non-opaque heavy-mineral species encountered in this study: apatite, calcic amphibole, chloritoid, clinopyroxene, epidote, kyanite, monazite, orthopyroxene, rutile, staurolite, titanite (sphene), tourmaline and zircon. Non-opaque detrital minerals that were not included in the modal analysis include carbonate minerals, anatase and barite (which can be confused with their authigenic counterparts) and mica and chlorite (which, because of their flaky habit, are subject to extreme hydraulic sorting). Most of the minerals listed above are found in the majority of the samples studied, but two are of very restricted occurrence. Orthopyroxene is present (as a very minor constituent) only in the recent wadi sediment (W61), while chloritoid is present in three samples, two Juwayl sandstone samples (W34, W40) and a Khusayyayn conglomerate sample W54A that rests directly on weathered granitic basement. The abundance of chloritoid in the latter sample (13.5%), reflected in the extremely high CtTi value (87.9), is a highly unusual feature and points to derivation from an alumina-rich metapelite source rock.

The presence of unstable minerals such as pyroxene, calcic amphibole and epidote, principally in the Juwayl Formation, indicates that for part of the time at least the sands were not subjected to intense meteoric weathering. The absence of surface etching of the unstable mineral grains also indicates that the Wajid sandstones have never been buried more deeply that a few hundred metres. The minerals display marked variation in their degree of rounding. The garnet, staurolite and epidote grains are sharply angular, indicating that they represent first-cycle influx from a basement source. The amphibole grains commonly show more rounded outlines, perhaps reflecting their lower hardness. The tourmaline grains show a wide range in roundness, with both highly rounded and sharply angular grains commonly present in the same sample. This indicates that there are at least two populations of tourmaline within the Wajid sandstones. Zircon and rutile are mostly subrounded to rounded and are thus mostly likely to be recycled. The soft minerals apatite and monazite are typically well rounded.

Heavy-mineral index values are presented in Table 4. It should be noted that not all samples have yielded valid counts for all indices. For example, the assemblage in the present-day wadi silt (W61) is so dominated by ferromagnesian minerals that the zircon count was insufficient for reliable calculation of the pZi, eZi and MZi indices.

Stratigraphic Mineral Plots

The lack of stratigraphic markers and the reconnaissance nature of the field trip precluded detailed stratigraphic correlation among different outcrops. The relative stratigraphic position of samples from different localities is therefore uncertain. Nevertheless, we have attempted to reconstruct a hypothetical stratigraphic sequence for the samples (Figure 7) by using regional dip and global positioning system (GPS) elevations taken at each locality to project the samples into a single sequence, realising the limitations of the approach. This stratigraphic sequence provides a useful framework for viewing the data and serves as a working hypothesis for future testing.

Although most samples project into the hypothetical sequence within their currently recognised formation, in some cases samples from one formation project into another (e.g. a Dibsiyah sample projecting into the Sanamah). This could be due to one or a combination of several factors: GPS elevation errors, local deviations from regional dip, faulting, and significant local topography on unconformity surfaces (e.g. Sanamah valleys cut into Dibsiyah, and Juwayl channels cut into Khusayyayn).

In these cases, the sample in question is moved upwards or downwards to place it into the correct formation. The four samples where this had to be done are Dibsiyah samples W22 and W23, which project in above the Dibsiyah-Sanamah contact, and Khusayyayn samples W36, and W48, which project in above the Khusayyayn-Juwayl contact. These samples have been adjusted downwards into their respective formations in the stratigraphic plots. Samples whose relative positions have been projected in from different localities are connected by dashed lines in Figure 7. Samples taken at a single outcrop where stratigraphic relations are certain are connected by solid lines. All samples are plotted with equal spacing in Figure 7 for simplicity, and because of the large uncertainty in estimating actual spacing for samples from different outcrops.

Two of the Khusayyayn samples (W54A and W68) are plotted separately on Figure 7 as they rest directly on basement and cannot therefore be fitted meaningfully into the more complete sedimentary sequences developed elsewhere in the Wajid region. The present-day wadi silt sample (W61) is included for comparison.

Looking at the succession as a whole, it is apparent that a major change in mineral character takes place across the Sanamah–Khusayyayn boundary, with an upward increase in average MZi, pZi and STi values, coupled with highly variable HTi and eZi values. A significant change is also apparent across the Dibsiyah–Sanamah boundary, with a sharp decrease in RZi and eZi values. Mineralogical changes across the Khusayyayn–Juwayl boundary are less pronounced, being limited to a minor decrease in RZi values and a very slight increase in ATi values. There is also some indication of stratigraphic mineral trends within each formation, although because of the limited nature of the data set and the uncertainty concerning the stratigraphic succession, any inferences must be regarded as tentative.

The four Dibsiyah sandstone samples are all from the ‘upper’ Dibsiyah Formation, as indicated by the presence of the marine trace-fossil Tigillites in the lowermost sampled sandstone. So far as can be determined, W2 and W3, which are estimated to be no more than c. 30 and 40 m above the basement, respectively, are from near the base of the ‘upper’ Dibsiyah Formation, whereas W22 and W23, which are exposed beneath Sanamah diamictite, are believed to be near the top of the Dibsiyah Formation. The major upward increase in RZi values between samples W03 and W23 thus appears to take place within the ‘upper’ Dibsiyah Formation. This is accompanied by an increase in pZi and HTi values and a decrease in STi values. The eZi index is relatively high in all Dibsiyah samples and shows a steady upward increase toward the Sanamah contact.

The Dibsiyah-Sanamah contact at East Madarah is marked by a dramatic change in heavy-mineral indices. The systematic upward increase in eZi observed in the Dibsiyah ends abruptly with a fall to low values in the lower Sanamah. The remainder of the Sanamah section shows a steady upward decrease in eZi values. RZi and pZi values also drop sharply across the formation boundary, after which there is an overall upward trend of increasing RZi values. It may be noted that while the diamictite samples follow this trend, their RZi values are lower than those of the associated sandstones. The Sanamah succession also displays a slight upward increase in MZi and STi values and a slight decrease in HTi values. The diamictite samples possess slightly higher STi values than the associated sandstones. The two stratigraphically highest Sanamah samples at North Madarah show an increase in GZi values, representing the first appearance of garnet in the Wajid sandstone succession.

The boundary between the Sanamah and Khusayyayn formations at North Madarah is marked by an increase in MZi, pZi, STi and eZi values. Index values within the Khusayyayn are somewhat variable, with sample W48 having a high HTi value and a very low RZi value. Khusayyayn samples W54A and W68 (plotted beneath the stratigraphic plot) both rest unconformably on basement, so that their age relative to the Sanamah–Khusayyayn boundary is not known. Both samples display very low RZi values and W68 a very low MZi value. W54A is also notable for its high GZi and extremely high ATi values. These features, together with the exceptionally high CtTi value, are attributed to derivation from nearby basement rocks.

At Jabal Khurb Al Aswad, the boundary between the Khusayyayn and Juwayl formations is primarily marked by an upward decrease in RZi values. Index values within the Juwayl Formation are highly variable, indicating significant changes in provenance with time. The high HTi values in the middle samples is particularly notable, as are the higher than average ATi values.

Mineral Index Cross-Plots

Mineral index cross-plots are presented in Figure 8. This mode of presentation reveals provenance-related compositional groupings within each formation and allows easy visual comparison between formations. RZi values provide the basis for all six cross-plots because RZi is one of the most reliable provenance indicators and has yielded valid counts for all of the samples examined. In the following discussion, the RZi/MZi plot (Figure 8a) is regarded as the primary measure of provenance variation, as it shows the greatest discrimination between the successive formations. For comparative purposes, the groupings identified on the RZi/MZi plot are identified also on the RZi/pZi (Figure 8b) and RZi/STi (Figure 8c) plots. Four such groupings have been recognised within the Juwayl sandstones. These are numbered 1–4, reflecting their position within the stratigraphic reconstruction of Figure 7.

Dibsiyah Formation

All four samples are characterised by extremely low MZi values (Figure 8a). Significant variation in RZi and STi values is displayed, however, with W2 and W3 possessing low RZi and high STi values and W22 and W23 possessing high RZi and low STi values (Figure 8c). W2 and W3 also show lower pZi values than W22 and W23. All the Dibsiyah samples display extremely low GZi and ATi values.

Sanamah Formation

Two distinct compositional groups can be identified within the Sanamah Formation. Five of the six sandstone samples (W12 to W16) possess relatively high RZi and MZi values, whereas the three diamictite samples (W8, W11 and W20) possess relatively low RZi and MZi values (Figure 8a). The two groups also show a slight difference in STi values (Figure 8c). The sandstone sample W10 displays RZi and STi values similar to those of the main group of sandstones, but differs in its extremely low MZi value.

Khusayyayn Formation

Five of the Khusayyayn sandstone samples (including a clast (W39) from within a Juwayl channel fill) possess moderate RZi values. They form a compositional group in which MZi values decrease with increasing RZi values. Two samples (W54A and W68), both with very low RZi and MZi values, plot outside this field. Both rest directly on basement. All of the Khusayyayn samples possess moderate to high pZi and STi values. W54A differs from the remainder in possessing a high GZi value and an extremely high ATi value. It also contains abundant chloritoid, reflected in an extremely high CtTi value (see Table 4).

Juwayl Formation

The Juwayl sandstones fall into four distinct compositional groups on the RZi/MZi plot. These groups also display different pZi and eZi values. All of the Juwayl sandstones display high STi values and very low to extremely low GZi values. Most samples possess extremely low-to-low ATi values but one sample (W41), which represents the matrix of a conglomeratic channel fill, has an ATi value of 19.4. Apatite is also present in sample W40, which possesses similar RZi, MZi and STi values, although the ATi index is much lower (4.0).

The Juwayl samples show a wider range in composition than the underlying formations. Sample numbers are limited, but it would appear that four separate compositional groups can be identified on the RZi/MZi, RZi/pZi and RZi/STi plots (types 1–4, Figures 8a–8c). These groups are associated with different stratigraphic and geographic sample locations.

Type 1 comprises samples W34 and W35 (Figure 7), which are from a locality where the Juwayl diamictite rests directly on sandstones of the Devonian Khusayyayn Formation. W34 is from the sandy matrix of the diamictite; W35 is from a cross-bedded fluvial sandstone that directly overlies the diamictite.

Type 2 comprises samples W40 and W41 (Figure 7), which are from a location where a Juwayl channel cuts down into sandstones of the Khusayyayn Formation. W40 is from the channel wall and W41 from the sandy matrix of the conglomeratic channel fill. They plot well away from all other Wajid sandstone samples on the RZi/MZi, RZi/pZi and RZi/STi plots.

Type 3 comprises samples W37 and W38 (Figure 7), which are from the channel fill and channel wall, respectively, of an intraformational conglomeratic channel in the south of the study area. These samples are distinguished from the other Juwayl samples by their very low MZi, pZi and eZi values (Figures 8a, 8b and 8f).

Type 4 comprises samples W42 and W43. These samples are believed to represent the highest part of the Juwayl Formation analysed for heavy minerals (Figure 7), although it should be noted that their distance below the base of the Ruhaiya Limestone at nearby Bani Ruhaiya is uncertain. Samples W42 and W43 are characterised by high MZi values and are quite distinct from the other Juwayl samples.

Present-day Wadi Silt

The present-day wadi silt sample yielded reliable counts for only three indices. It is characterised by a very high RZi value and by extremely high GZi and ATi values. None of the Wajid sandstone samples possesses a comparable provenance signature. Although the Khusayyayn sample W54A, interpreted as having been derived direct from local basement rocks, possesses similarly high GZi and ATi values, it has very different RZi values. Clearly the two samples are derived from different source rocks.

INTERPRETATION OF THE WAJID HEAVY-MINERAL DATA

Almost all of the Wajid sandstone samples display zero to very low ATi values (see Figures 7 and 8e). These include samples from the Sanamah and Juwayl formations, which are known to have been deposited during glacial periods. Under such climatic conditions, chemical weathering of upland source rocks will have been minimal and apatite more or less unaffected by acidic meteoric leaching. The scarcity of apatite in these sandstones must therefore reflect either a scarcity of apatite in the source rocks or an effect of Quaternary weathering. For the Juwayl sandstones at least this appears to be an original feature, since comparable low-ATi sandstones (unpublished data) are present in time-equivalent subsurface sections of central Saudi Arabia. The abundance of apatite in the Khusayyayn sample W54A also indicates that the extremely low ATi values in other samples are an original feature, since there is no reason to suppose that this sample, believed to have been derived from a nearby basement source, should have undergone any less Quaternary weathering than other outcrop samples.

Variation in ATi values may therefore reflect the extent to which sands were derived directly from unweathered basement rocks. The only samples that appear to have been derived largely from the local basement are the basal Khusayyayn conglomerate sample W54A and the recent wadi silt sample W61, with ATi values of 95.1 and 95.8, respectively. A minor contribution from unweathered basement rocks is perhaps indicated by the elevated ATi values (4.2–19.4) in three of the Juwayl samples (W38, W40 and W41). The remainder of the Juwayl samples, together with all of the Dibsiyah, Sanamah, and Khusayyayn samples display extremely low ATi values, indicating derivation from pre-existing sediments or highly weathered basement rocks. A sedimentary source is favoured by the presence of re-worked quartz overgrowths in some samples (see above).

Amphibole and garnet are both susceptible to meteoric leaching, but less so than apatite. The extremely low GZi and HTi values in many of the Wajid sandstones, including those of the cold-climate Sanamah and Juwayl formations, are thus likely to be inherited and not the result of weathering in the Wajid area. In the present study, high GZi values have been encountered in only two samples: the basal Khusayyayn conglomerate (W54A) and the present-day wadi silt (W61). In both instances, the association with high ATi values indicates derivation from nearby unweathered basement rocks. The inherited low GZi values may reflect intense weathering at the time of deposition of the source sandstones. Alternatively, deep burial of the source sandstones may have led to the dissolution of garnet during burial diagenesis. High HTi values are restricted to a few Khusayyayn and Juwayl samples. They are also interpreted as reflecting a contribution from nearby basement rocks.

Since the limited occurrence or absence of etching features on even the most unstable stable minerals precludes any significant diagenetic removal of minerals within the Wajid sandstones, it thus seems likely that the present range in composition of the Wajid heavy-mineral assemblages is a true reflection of the depositional composition of the sands. The following discussion of the stratigraphic evolution of the assemblages and their provenance is based on this premise.

The cross-plots of the provenance-sensitive mineral indices RZi, MZi, STi and GZi (Figure 8) and the stratigraphic plots (Figure 7) indicate significant changes in the composition of the Wajid sandstone source rocks with time. Given the highly consistent northward transport directions revealed by palaeocurrent data in the fluvial facies (e.g. Dabbagh and Rogers, 1983) it seems unlikely that any fundamental change in palaeogeographic setting took place during deposition of the Wajid sandstone. It is therefore possible that all of the sandstones were derived from a single-source region, with compositional changes resulting from the exposure of new source rocks through progressive denudation, through changes in the catchment area or through differential uplift of individual tectonic elements. Compositions are also likely to have been affected by changes in climate.

It is possible that not all the sandstones were derived from a southerly source, however. Another factor that should be taken into account is that the ‘upper’ Dibsiyah succession and parts of the Khusayyayn succession are believed to have been deposited in nearshore marine environments (see Stump and van der Eem, 1995). Sandstones within these intervals could have been brought into the basin from other source areas through the process of long-shore drift. Significant bed-to-bed variations in feldspar content at the outcrop scale in the Khusayyayn sandstones (see below) may be the result of such mixing. Local fluvial influx from basement highs may also have taken place. The existence of such highs within the Wajid area is indicated by the occurrence of Khusayyayn sediments resting directly on basement at some localities.

Taking the Wajid succession as a whole, it is apparent that the four sandstone-dominated formations represent discrete phases of sand deposition, reflected in successive change in bulk sand composition. Where compositional overlap does exist between sandstones of different formations, this can generally be attributed to reworking or derivation direct from the same basement rocks. Successive changes in mineral composition are discussed below.

Dibsiyah Formation

Sample pairs W2–W3 and W22–W23 are interpreted to be from the lower and upper parts, respectively, of the upper Dibsiyah. If they are representative of these levels, then a major change in provenance must have taken place during deposition of the ‘upper’ Dibsiyah sands. The relative increases in RZi and HTi suggest that this change involves an increased contribution of basic metamorphic and/or igneous rocks. The apparent steady upward increase in the eZi index (Figure 7) suggests an increase contribution of first-cycle sediments along with this increase in basic material. Additional sampling will be required to test these ideas.

Sanamah Formation

The upward increase in eZi observed in the Dibsiyah succession is terminated by a sharp downward shift across the Dibsiyah-Sanamah contact. Apparently, the sand supply that had earlier contributed euhedral zircons to the Dibsiyah was shut off (by glaciation?) or swamped by an increased amount of reworked sediment.

Sanamah diamictite samples plot close to the Dibsiyah samples W2 and W3. Since the diamictites include slabs (up to a metre across) of Dibsiyah sandstone and siltstone, it is possible that their relatively low RZi and MZi values reflect a substantial content of reworked Dibsiyah sand. It should be noted, however, that the diamictite W20 overlies Dibsiyah sandstones (samples W22 and W23), that possess high RZi values (Figure 7) and cannot therefore have contributed significantly to the diamictite. An alternative explanation for the low RZi values in the diamictites is that they contain sand derived from basement granites, which are typically low in rutile.

In the stratigraphic plots, the diamictite samples project into the hypothetical stratigraphic sequence at different levels, but in each case RZi values are lower in diamictites than in associated non-diamictite samples. It appears that the glacial diamictites were tapping a slightly different source than the non-diamictite samples. In both diamictite and non-diamictite samples RZi values tend to increase upwards (Figure 7). MZi values show a general upward increase, but there are reversals in the trend at some levels (Figure 7).

The Sanamah assemblages may thus be seen as comprising glacio-fluvial sandstones that were derived from an increasingly monazite-rich, possibly distant, source and diamictite matrix sands that were derived from a different source, possibly older Wajid sandstones.

Khusayyayn Formation

W54A is a conglomerate that sits directly on a granitic regolith, indicating that Sanamah and Dibsiyah sediments have been removed or were never deposited at this location. Contribution from the underlying granite could explain the extremely low RZi values (lower, even, than the lowest Dibsiyah RZi value), but the very high GZi value (Figure 3d) and the extremely high chloritoid:tourmaline index (KTi) value (see Table 4) are suggestive of a metamorphic source for the bulk of the sand. The extremely high ATi value (comparable to that of the present-day wadi silt sample W61) indicates a lack of acidic leaching of both the source rock and the present-day sandstone. The W68 sandstone, which rests directly on granitic basement, lacks the distinctive metamorphic minerals present in W54A, and it seems likely that its extremely low RZi and MZi values reflect derivation from the underlying granite.

The main compositional group of Khusayyayn sandstones displays a distinct trend of decreasing MZi values with increasing RZi values (Figure 8a). This is best explained by mixing of sand derived from high-MZi source rocks (represented by sample W48) with sand derived from low-MZi source rocks (represented by sample W36). Sample W36 is from a level immediately below the Juwayl Formation and represents the youngest part of the Khusayyayn Formation sampled. Therefore, the low MZi source appears most prevalent or least diluted in the upper Khusayyayn section. W48 is from about 30 km north of W36 and from a level interpreted to be lower in the section (Figure 7). Therefore both the high MZi source and the low MZi source, which are responsible for the mixing trend observed in Figure 8a were seemingly able to contribute somewhat independently at times and together at other times. This seems more compatible with separate drainages rather than an unroofing of source terrane with time.

A large sandstone clast (W39) collected from a Juwayl channel fill at Jabal Fard al Ban plots close to and within the main compositional fields for the Khusayyayn sandstones (Figures 8a8c) and away from the enclosing Juwayl sandstones (W40 and W41: group 3). This suggests that the clast was derived from the Khusayyayn Formation rather than from a pre-existing Juwayl channel-fill deposit.

Juwayl Formation

Juwayl heavy-mineral indices show large variations and no clear long-term stratigraphic trend. However, in terms of the four types of sandstone described previously, several general comments can be made.

The stratigraphically lowest Juwayl sandstones, Type 1 (W34 and W35), have much lower RZi and much higher MZi indices than the directly underlying Khusayyayn sandstones (W36) at the same locality. These samples plot between the two groups of Khusayyayn samples on Figure 8, indicating that they could have been generated by reworking and mixing of Khusayyan sands.

Type 2 (W40 and W41), stratigraphically above Type 1, plot well away from other Juwayl sandstones in terms of MZi and STi (Figure 8a, c). They also have somewhat higher eZi values (Figure 8f) than underlying Juwayl samples. These characteristics indicate a distinctive new source consisting in part at least of aluminous metapelites and first cycle granitic sediments.

Type 3 (W37 and W38) lie stratigraphically above Type 2 sandstones, but display much lower MZi, pZi, and eZi values (Figure 8a, b, f). They are closer in composition to the low-RZi sandstones of the Khusayyayn, Sanamah and Dibsiyah formations than to the other Juwayl sandstones. Since Aramco unpublished geological maps (Hulver, 2004) suggest that the Juwayl in this area cuts down into the Sanamah, it is possible that the Type 3 sandstones are reworked from the Sanamah and other pre-Permian formations.

The distinctively high MZi values of Type 4 sandstones (W42 and W43) represent a different stratigraphic interval than the Type 3 sandstones, which are believed to be lower in the Juwayl succession.

The RZi/MZi, RZi/pZi and RZi/STi plots of Figures 8a8c demonstrate that there is little compositional overlap between the four Juwayl sandstone compositional types and the underlying Khusayyayn, Sanamah and Dibsiyah sandstones, mostly due to lower RZi and higher STi values in the Juwayl. Although a limited amount of reworking of older sands is likely, it seems that sand types 4 and 2 represent an influx of sands derived from newly exposed source rock types. Sand type 4 has the highest MZi values of any formation sampled, whereas sand type 2 has the highest STi, pZi and eZi values of any formation and among the highest HTi values.

As discussed above, the low ATi values displayed by most Juwayl sandstones (Figures 7 and 8e) is likely to be an original feature and the slightly elevated values in the middle part of the Juwayl Formation are thus likely to reflect a change in provenance.

The RZi/HTi plot (Figure 9a) shows that most samples plot within a restricted compositional range, displaying a trend of decreasing HTi values with increasing RZi values. Samples that lie outside this field include the younger, high-RZi Dibsiyah sandstones, a single Khusayyayn sandstone, three Juwayl sandstones and the present-day wadi silt. Counts were insufficient to obtain an HTi value for the distinctive W68 Khusayyayn sandstone. The younger Dibsiyah sandstones do not show anomalously high HTi values, but lie outside the main compositional group on account of their exceptionally high RZi values, which are comparable to that of the present-day wadi silt. These three samples also display similar STi values (see Figure 8c). It is therefore possible that the Dibsiyah sands were derived from source rocks similar to those that supplied the wadi silt, but with the HTi (and ATi values: Figure 9b) greatly reduced by weathering.

The anomalously high HTi values of the Khusayyayn sample W48 and the three Juwayl samples are probably related to provenance since they both are associated with elevated eZi values. Since the high HTi values in the Juwayl Formation are associated with the different sandstone types (2 and 3) as defined on the RZi/MZi plot (Figure 8a), it is likely that the influx of hornblende represents a contribution from an independent source, probably located on nearby basement.

The eZi values (Figures 7 and 8f) display significant stratigraphic variation. Both the low-RZi and high-RZi Dibsiyah sandstones display relatively high eZi values, whereas the Sanamah sandstones and all but one of the Khusayyayn sandstones (W48) display low eZi values. Most of the Juwayl sandstones possess moderate eZi values. Because the variation in eZi values is largely related to variation in the associated provenance indices, it probably reflects eZi variation within the source rocks. An element of transport-related abrasion cannot be ruled out, however, especially in those assemblages derived from far distant sources. Provenance control is most clearly exemplified by the recent wadi sand sample W61, which possesses a very low eZi value, despite being a mineralogically immature, first-cycle sediment. In this instance, the low eZi value is presumed to have been inherited from a mature metasedimentary source rock. The relatively high-eZi Dibsiyah assemblages are therefore interpreted as derived from a source of euhedral zircons that was no longer available during Sanamah and most of Khusayyayn sedimentation.

Although the heavy-mineral provenance data, being derived from a single-size fraction, cannot be correlated numerically with the petrographic data, qualitative comparison of the two data sets can be made. It might be expected that samples that show closely similar heavy-mineral provenance signatures would also show similar feldspar contents. However, in the case of the five Sanamah sandstones that plot closely together on the RZi/MZi plot (Figure 8c), this is clearly not so. Four of these sandstones are of comparable fine sand grade (125–250 μm) and yet display restored feldspar percentages that range between 11 and 40 (Figure 5).

It is difficult to understand how such large differences could be attributed to differences in the diagenetic processes, since the sandstones are of similar grain size and from the same fluvial channel-fill outcrop. It would appear, therefore, that the sandstones differed in their primary K-feldspar content. This could reflect derivation from separate high-feldspar and low-feldspar sources that had similar heavy-mineral suites. Alternatively, the sands could have been derived from a single suite of source rocks in which feldspar had already undergone varying degrees of feldspar removal, either though weathering or, in the case of sedimentary source rocks, burial diagenesis. In view of the very close similarity in heavy-mineral composition, the second option seems the more probable.

The greater variation in both heavy-mineral composition and grain size makes it difficult to assess the relationship between heavy-mineral provenance signatures and feldspar content in the remainder of the Wajid samples.

Implications for Palaeogeographic Evolution

While palaeocurrent data clearly indicate that the successive Wajid fluvial sands were derived from a southerly direction (Dabbagh and Rogers, 1983; Stump and van der Eem, 1995), the source of the sands is poorly constrained. Several authors have commented on the high mineralogical and textural maturity of much of the Wajid sandstones (e.g. Dabbagh and Rogers, 1983, p. 53, 55; Stump and van der Eem, 1995, p. 425), which is a feature of the equivalent strata in central and northern Saudi Arabia (Powers et al., 1966; Vaslet, 1990, p. 37, 57). Previous studies on the heavy-mineral assemblages in Saudi Arabian Palaeozoic sandstones have also pointed to the dominance of the highly stable minerals zircon, rutile and tourmaline (Powers et al., 1966, p. 24–25; Vaslet, 1990, p. 38, 58; Hussain et al., 2004). Even the diamictites (interpreted by some as tillites) possess a high degree of mineral maturity (Vaslet, 1990, p. 82–83).

The mature mineral assemblages can be interpreted as indicating derivation from source rocks dominated by pre-existing sediments (e.g. Hussain et al., 2004). The presence of subordinate amounts of fresh and often angular grains of amphibole, epidote, garnet, kyanite, pyroxene and staurolite could reflect additional contribution from nearby metamorphic basement outcrops. An alternative explanation is that such assemblages could have been derived from metamorphosed, sand-rich sediments in which the sand component possesses a moderate to high degree of physical and chemical maturity. An example of such a lithology occurs in a road cut at 17°36.222 N, 043°33.735 E, southeast of Zhahran al Janub, where metamorphosed quartzose sandstone (now quartz-mica schist) immediately underlies sandstones of the Khusayyayn Formation. Derivation from pre-existing sediments or metasediments is supported by the occurrence of reworked quartz overgrowths in Wajid sandstones (Figure 6).

The consistent northward palaeoflow directions recorded from successive fluvial sandstones within the Wajid Group (e.g. Dabbagh and Rogers, 1983; Stump and van der Eem, 1995) points to a stable structural setting, with uplift to the south and subsidence to the north. Palaeocurrent data from the pre-Permian formations clearly indicate that the streams were flowing towards what is now the exposed eastern margin of the Arabian Shield.

Although Dabbagh and Rogers (1983, their figure 3) do not differentiate between the different formations, it is evident that their palaeocurrent data are largely derived from the Khusayyayn Formation, which constitutes the majority of the Wajid Group outcrop. These indicate that the average transport direction was slightly to the east of north. Sandstones assigned by Hussain et al. (2004, their figure 2) to the Dibsiyah and ?Khusayyayn formations in outliers to the southwest of the Wajid Plateau also display an average palaeocurrent direction to the east of north.

So far as can be determined, therefore, sands were supplied from the south during both Cambrian and Devonian sedimentation. The very marked change in provenance represented by the first influx of monazite-rich sands in the Devonian Khusayyayn Formation, must therefore reflect the exposure of new source rocks to the south. This could have resulted from the progressive denudation of the source area (unroofing), through a change in the pattern of differential uplift within the source area, or through a pronounced contraction or extension of the drainage system.

Heavy-mineral analysis of Cambrian and Ordovician sandstones in Hawtah field, some 600 km NNE of the Wajid area, has revealed that sandstones assigned to the Saq and Sarah formations display significantly lower RZi values than their Wajid counterparts, the Dibsiyah and Sanamah formations (Figure 10). The Sarah sandstones also show much lower MZi values than the approximately equivalent Sanamah sandstones. This is interpreted as reflecting the operation of independent, northward-directed sand dispersal systems (Figure 11a) draining different regions within the southern hinterland. No data are available for Khusayyayn-equivalent sandstones in the Hawtah area.

Few palaeocurrent data are available for the Juwayl sandstones. The most specific reference to palaeocurrent directions is that of Evans et al. (1991, p. 951), who indicate a northeastward direction of transport in stacked channel sequences at Jabal Khurb al Ahmar, in the eastern (N–S trending) valley fill. Stump and van der Eem (1995, p. 435) give a less precise direction for the basal channel-fill sediments of the Juwayl Formation at Bani Khurb (also in the eastern valley), describing the transport direction as more northerly than for the ‘upper’ Khusayyayn Formation, which was itself described (p. 434) as having a more easterly transport direction than the northerly or northeasterly direction of the ‘lower’ Khusayyayn. Transport to the northeast would thus appear to be indicated in both localities (Figure 11b).

Northeastward transport of the Juwayl sands is supported by the presence of sandstones with relatively high MZi values at the base of the Permian-Carboniferous succession along the southeastern margin of the Central Arabian Arch (Nuayyim, Hawtah, Hazmiyah, Ghinah and Usaylah fields: Figure 12). These sandstones belong to the Unayzah B and C units of Melvin and Sprague (2006). Although these sandstones are comparable to the Juwayl sandstones in their trend of increasing MZi values with increasing RZi values, their compositional range is more restricted. This may in part reflect increasing homogenisation of the sands with increasing distance of transport. Such a process cannot, however, account for the lower average MZi and RZi values compared with the Juwayl sandstones. Either the Unayzah B and C sandstones were not derived from precisely the same source as the Juwayl outcrop sandstones or the southerly-derived sands underwent mixing with low-MZi and high RZi sands. It is therefore possible that high-MZi Juwayl-type sands underwent mixing with low-MZi sands introduced by the southward to eastward directed fluvial drainage system that became established along the southeastern margin of the Arabian Shield at this time (McGillivray and Husseini, 1992, p. 1484) (Figure 13). The introduction of low-MZi sands from other locations along the Arabian Shield margin may account for the anomalously high RZi samples in some of the Unayzah B and C sandstones. It should nevertheless be stressed that, in the absence of information from intermediate shallow subsurface sections, the correlation between the Juwayl sandstones and the Unayzah B sandstones is not definitively established.

Even without a precise correlation into the subsurface, it is apparent that palaeocurrent data (available only for the eastern valley fill) favour northeastward transport of the high-MZi sands along the southeastern flank of the central Arabian Arch, as shown in Figure 11b. In this context, the orientation of the Juwayl valleys, especially the western, NNW–SSE trending valley, appears anomalous. Without further examination in the field, particularly concerning the precise location and stratigraphic level of the palaeocurrent data, it is not possible to explain this discrepancy.

Despite the uncertainty concerning the orientation of the Juwayl valleys, the palaeocurrent and mineral data provide a strong indication that the mid-Carboniferous uplift of the Central Arabian Arch (McGillivray and Husseini, 1992, p. 1480; Konert et al., 2001, p. 427; Al-Husseini, 2004, p. 54) was sufficiently well-established to cause eastward deflection of the regional northward flow direction.

The precise location of the Wajid sandstone source areas is uncertain. Some authors favour a far distant source, in central Africa (e.g. Stump and van der Eem, 1995, p. 428), whereas Hussain et al. (2004), who carried out a heavy-mineral analysis on Wajid sandstones to the southwest of the study area, concluded that they might have been derived from the Neoproterozoic terranes and infra-Cambrian cover rocks in southernmost Arabia. The relative angularity of many of the less stable mineral components (amphibole, garnet, kyanite and staurolite) encountered in the present study indicates that some of the sandstone components were derived from basement sources that were not far distant. Any metamorphic source must, however, have been of different composition to the source of the present day wadi sediment sample W61 or the basement-derived conglomerate sample W54A, since both possess very much higher GZi and ATi values (Figure 8d and 8e) and W61 a lower STi value (Figure 8c).

The absence of monazite in the Wajid sandstones studied by Hussain et al. (2004) suggests that representatives of the Khusayyayn Formation are not present and that both the Red Unit and the overlying Gray Unit can be assigned to the Dibsiyah Formation. The influx of monazite-rich sands in Devonian times (Khusayyayn Formation) and their persistence into the Permian (Juwayl Formation) is indicative of a substantial change in the nature of the source rocks. This could have resulted either from differential uplift within the pre-existing source regions or from an extension of the drainage system to the south, into the area of present-day Ethiopia and Somalia.

Correlation Potential

The marked difference in mineral composition shown by the bulk of samples in each of the four Wajid sandstone formations provides a potential means for identification and correlation of these formations both at outcrop and into the adjacent subsurface. Such direct correlation can, however, be achieved only within a single-sand dispersal system, since provenance signatures are likely to differ significantly between dispersal systems. In such cases, interregional heavy-mineral correlation must rely on signatures that are likely to have transcended individual dispersal systems, such as eZi and ATi values, which may reflect regional ‘events’, such as tectonic rejuvenation of source areas, sea-level change and climate change. Regional tectonic events can also be identified by an upward change in provenance signatures along formation boundaries, even where the provenance signatures show regional differences. For example, the tectonic restructuring and sea-level fall that accompanied the onset of glaciation in latest Ordovician times (e.g. Konert et al., 2001, p. 421) is marked by a change in provenance signatures in both the Wajid area (base Sanamah Formation) and Hawtah area (base Sarah Formation), although the sandstones in the two areas were evidently not derived from the same source area (Figure 10).

The four provenance groups identified in this study equate with four of the major depositional packages or ‘Super Groups’ (II–V) identified by Stump and van der Eem (1995) and Stump et al. (1995). These authors regarded the boundaries of each Super Group as representing major periods of diastrophism associated with interregional unconformities and fundamental changes in sedimentation that reflected a common geological history within across the northern Gondwanaland (Stump et al., 1995, p. 5). They are therefore likely to be represented by significant stratigraphic changes in the composition of the heavy-mineral assemblages, irrespective of geographic location and provenance.

CONCLUSIONS

Heavy-mineral analysis has revealed that significant changes in provenance took place during deposition of the Wajid sandstones. Since there are strong indications that the principal source area lay to the south, changes in the composition of the source rocks are ascribed to a combination of progressive unroofing, changes in the pattern of tectonic uplift and changes in the configuration of the drainage system.

The stratigraphic changes in heavy-mineral composition mostly represent successive, unconformity-bounded, tectono-sedimentary packages, represented by the Dibsiyah, Sanamah, Khusayyayn and Juwayl formations. A marked contrast in heavy-mineral assemblages has also been identified within the upper Dibsiyah sandstones. In this instance, however, there is no field evidence for a major discontinuity. This uncertainty could probably be resolved by the acquisition samples from the Dibsiyah Formation.

Comparison of the heavy-mineral assemblages from the Wajid sandstones and those from deep subsurface sandstones in the Central Saudi Arabia fields area to the northeast reveals distinct compositional differences over much of the succession. The sandstones of the central Saudi Arabia fields area were evidently derived from different source rocks. Only in the Permian succession is there evidence for sandstones of Wajid (Juwayl) composition extending into the Central Saudi Arabia fields area. This may reflect uplift of the Central Arabian Arch prior to the onset of Juwayl sedimentation, which would have led to diversion of the northward-flowing rivers to the northeast.

Comparison of heavy-mineral and petrographic provenance data reveals a strong correlation in some formations but a poor one in others. The reason for the poor correlation is uncertain, although it is possible that the feldspar percentage values are affected by in-situ or inherited diagenetic overprints.

Although the stratigraphic changes in provenance signature recorded from the Wajid sandstone succession are likely to be specific to southwest Saudi Arabia, the correlation with regionally recognised tectonostratigraphic megasequences means that it is likely that comparable changes in mineral composition will be displayed in successions elsewhere in the region. Heavy-mineral assemblages thus have the potential for subsurface identification and correlation of Palaeozoic sandstone formations throughout the Arabian region.

ACKNOWLEDGEMENTS

Dr. Mike Hulver of Saudi Aramco, Dr. Graham Lott of the British Geological Survey and two anonymous reviewers are thanked for their thorough and constructive reviews of an earlier version of this paper. The authors wish to thank Saudi Aramco for permission to publish this paper. R.W.O’B. Knox publishes with the approval of the Director, British Geological Survey. The assistance of Nestor Niño Buhay in adding finishing touches to the graphics is much appreciated.

ABOUT THE AUTHORS

Robert W.O’B. Knox is an Independent Consultant specializing in the application of heavy mineral analysis to the correlation and delineation of reservoir sandstones in the Paleozoic and Mesozoic successions of the Arabian Peninsula. He was formerly with the British Geological Survey, where he gained over 25 years experience on the stratigraphy and sedimentology of UK and overseas petroleum basins.

robertknox@rknox.fsnet.co.uk

Stephen G. Franks is a Senior Geological Consultant in the Exploration and Petroleum Engineering Advanced Research Center of Saudi Aramco. His primary research is directed toward pre-drill reservoir quality prediction in sandstones. Steve has spent more than 30 years in the oil and gas industry. He has a BSc degree in Geology from Millsaps College, an MSc in Geology from the University of Mississippi, and a PhD in Geology from Case Western Reserve University. He was with Atlantic Richfield (ARCO) for 26 years, spending his first 10 years and last 8 years in R&D, separated by an 8-year period in exploration. He was appointed as both a Senior Exploration Advisor and Senior Research Advisor to ARCO management. In 1999 he formed RockFluid Systems, Inc., a consulting firm specializing in pore-level reservoir characterization and formation water geochemistry. Steve joined Saudi Aramco in 2001 to develop reservoir quality prediction models for the deep sandstone gas reservoirs of Saudi Arabia. In addition to reservoir quality prediction, he has conducted studies of deep formation water chemistry, modern sabkha pore waters, and investigated early diagenesis of modern desert sediments using stable isotope geochemistry and optical luminescence age dating. Stephen is a member of the AAPG, SEPM, and Dhahran Geoscience Society.

stephen.franks@aramco.com

Joshua D. Cocker spent a rewarding career specializing in granitoid rock petrogenesis, alteration of ophiolite sequences and sandstone reservoir quality. He received his BSc (Honors) and PhD (1975) from the University of Tasmania, Australia. After teaching for seven years at the University of Alberta and Oregon State University, Josh joined Mobil Oil as a Research Scientist focusing on the origin and properties of authigenic clays. He was seconded to Saudi Aramco in 1989 and focused on clay diagenesis, sandstone reservoir quality and stratigraphic correlation in sandstone sequences. Josh has retired to Tasmania to develop habitat for kangaroos, to travel and to dabble in things geological.

roo@roowithaview.com