Petrology and multimineral fingerprinting of modern sand generated from a dissected magmatic arc (Lhasa River, Tibet)
Published:December 28, 2018
- PDF LinkChapter PDF
Eduardo Garzanti, Mara Limonta, Giovanni Vezzoli, Wei An, Jiangang Wang, Xiumian Hu, 2018. "Petrology and multimineral fingerprinting of modern sand generated from a dissected magmatic arc (Lhasa River, Tibet)", Tectonics, Sedimentary Basins, and Provenance: A Celebration of the Career of William R. Dickinson, Raymond V. Ingersoll, Timothy F. Lawton, Stephan A. Graham
Download citation file:
High-resolution sand petrography and heavy mineral analyses help to frame U-Pb age and Hf isotope data from zircon grains, integrated in turn with geochemical data from detrital apatite, rutile, garnet, and monazite, and with Raman spectroscopy data from detrital amphibole, pyroxene, and epidote-group minerals. This multitechnique approach, including stream-profile analysis, was used to characterize components of the sediment flux and define erosion patterns across the Lhasa block, a complex continental arc terrane caught in the Himalayan collision. Litho-feldspatho-quartzose detrital modes and hornblende-dominated heavy mineral assemblages suggest that the majority (four fifths) of the sand bed load in the Lhasa River catchment is derived from erosion of granitoid batholiths. Gravel composition, however, is markedly different and dominated by volcanic pebbles in the trunk river, as in all of its four major tributaries, testifying to an order-of-magnitude difference in apparent erosion rates between granitoid batholiths and arc lavas. This marked contrast, partly explained by wide exposures of granitoid rocks in the rugged Nyainqêntanglha Range characterized by active incision, is notably amplified by the high sand-generation potential of granitoid rocks, which, in contrast to dense joint blocks of andesitic lavas, tend to disintegrate to sandy grus upon weathering. Sedimentary strata, making up a good half of exposed rocks, are also underrepresented in sand bed load, suggesting selective mechanical breakdown of nondurable shale/slate grains. This exposes a serious bias affecting estimates based on sand only, and it highlights the necessity for taking into account the entire size spectrum from mud to gravel in order to improve the accuracy of sediment budgets. Provenance analysis should involve multiple methods applied to multiple minerals, rather than be based solely on a single rare mineral, even if it is exceptionally laden with potential provenance information, such as zircon.
We here divide arc-derived suites into those eroded from undissected arcs, in which nearly continuous volcanic cover is present, and those from dissected arcs, in which cogenetic plutons are widely exposed from erosional unroofing.
—Dickinson and Suczek (1979, p. 2175)
The mineralogical composition of sediments provides the key to reconstruct the erosional evolution of a source area in time (Dickinson, 1985). The hardest challenge is posed by orogenic settings, which vary notably depending on the nature of converging plates and their geometry of convergence. Orogenic sediments thus cover a wide range of compositions, resulting from mixed detritus from different sources, including upper-plate magmatic-arc remnants trapped in the collision zone, lower-plate passive-continental-margin strata, and underlying basement accreted in the growing accretionary prism, as well as rock assemblages not specifically considered in classic provenance models, such as ophiolitic complexes or neometamorphic axial belts (Garzanti et al., 2007). The erosional evolution of a collision orogen such as the Himalaya results from the integral of erosive histories recorded in each of these tectonic domains, which may be distinguished ideally and traced individually by quantitative provenance analysis of the sedimentary record.
The purpose of this article is to document petrographic, mineralogical, geochemical, and geochronological signatures of sediments shed today by a complex continental-arc terrane incorporated within a large collision orogen. Abundant data are available from undissected intraoceanic or continental arcs, chiefly exposing a calc-alkaline suite of volcanic products and thus representing perhaps the simplest sources of detritus of all (e.g., Stewart, 1978; Klein et al., 1980; Prasad and Hesse, 1982; Baltuck et al., 1985; Packer and Ingersoll, 1986; Lundberg, 1991; Fujioka and Saito, 1992; Marsaglia, 1992; Marsaglia and Ingersoll, 1992; Garzanti et al., 2002; Noda, 2005; Derkachev and Nikolaeva, 2007; Limonta et al., 2015; Marsaglia et al., 2016). Less exhaustive is the existing information on transitional suites (e.g., Stewart, 1977; Gergen and Ingersoll, 1986; Marsaglia et al., 1992) and on more complex arc terranes in advanced stages of dissection, when diverse source rocks including metamorphic wall rocks and sedimentary strata may be exposed (e.g., Bachman and Leggett, 1982; Marsaglia et al., 1995; Ingersoll and Eastmond, 2007; Ingersoll, 2012).
This is why we chose to study modern sand derived from the Lhasa block, the remnant of the active Asian margin involved as part of the upper plate in the continental collision with the lower-plate passive margin of India since the middle Paleocene (Garzanti et al., 1987; Hu et al., 2015). We focus on sediment carried by the Lhasa River, the largest left-bank (northern) Tibetan tributary of the Yarlung Tsangpo (upper tract of the huge Brahmaputra fluvial system) draining entirely within the Lhasa block (Fig. 1).
We emphasize the importance of an integrated, multimethod and multimineral approach to provenance analysis (e.g., Mange and Morton, 2007; von Eynatten and Dunkl, 2012; Smyth et al., 2014). The last decade has seen a rapid increase in provenance studies based principally or even exclusively on U-Pb dating of individual zircon grains, which are preserved widely in sand-stones owing to their relative durability. Although the isotopic fingerprinting of detrital zircon may contribute substantially to enlighten provenance puzzles, relying on zircon alone is simply not enough. Beside the well-known fertility problem and other sources of bias (e.g., Moecher and Samson, 2006; Dickinson, 2008; Malusà et al., 2013, 2016), the average zircon content in sediments is only 1 grain out of ~5000, corresponding to the estimated average Zr concentration in the upper continental crust of 190–193 ppm (Taylor and McLennan, 1985; Rudnick and Gao, 2003). Focusing on zircon only thus inevitably entails missing information on the remaining 99.98% of the sample (Fig. 2).
In this study, high-resolution sand petrography and heavy mineral analyses were integrated with U-Pb age and Hf isotope data on zircon grains, geochemical data on detrital apatite, rutile, garnet, and monazite, and Raman spectroscopy analyses of detrital amphibole, pyroxene, and epidote-group minerals. Such a multidisciplinary approach, including stream-profile analysis, allowed us to characterize the various components of the sediment flux and trace erosion patterns across this continental-arc terrane caught in the Himalayan collision.
The Lhasa block, drained by the Lhasa River, is an E-W–trending tectonic domain distinguished by different sedimentary successions into northern, central, and southern terranes. Cretaceous, Jurassic, and minor Triassic cover strata characterize the northern Lhasa terrane. In the central Lhasa terrane, very low-grade Permian–Carboniferous metasedimentary rocks and an Upper Jurassic to Lower Cretaceous volcano-sedimentary succession, including felsic lavas and volcaniclastic rocks, are intruded by Mesozoic batholiths (Zhu et al., 2011). Granitoid rocks of the Gangdese batholith are widespread in the southern Lhasa terrane, which includes Upper Triassic to Cretaceous sedimentary strata, Jurassic–Lower Cretaceous volcanic rocks, and the widespread uppermost Cretaceous to Eocene Linzizong andesites to rhyolitic ignimbrites (Wen et al., 2008; Ji et al., 2009). Magmatic rocks emplaced between 220 and 40 Ma in the central and southern Lhasa terranes have distinct zircon U-Pb age spectra and Hf isotopes. Zircons with negative εHf(t) values—parts per 104 deviation in initial Hf isotope ratios between zircon samples and the chondritic reservoir at the time of zircon crystallization—predominate in the central Lhasa terrane (Chu et al., 2006; Zhu et al., 2009), whereas positive εHf(t) values predominate in the southern Lhasa terrane (Chu et al., 2006; Ji et al., 2009; Zhu et al., 2015).
The Lhasa block, the last microcontinent accreted to the southern margin of Asia before collision with India (Dewey et al., 1988), is welded by the Bangong-Nujiang suture in the north to the Qiangtang block (Marcoux et al., 1987) and by the Yarlung-Zangbo suture in the south to the Tethys Himalaya (Gansser, 1980; Hébert et al., 2012), representing the northern passive margin of India (Sciunnach and Garzanti, 2012). Collision between the Qiangtang and Lhasa blocks took place between Middle Jurassic and Early Cretaceous time (Kapp et al., 2007; Ma et al., 2017), whereas continental collision with India began in the middle Paleocene (Hu et al., 2016a). After accretion to Asia, the Lhasa block underwent a prolonged erosional evolution, which started at 113–110 Ma, when detritus from the Gangdese arc began to feed the Xigaze forearc basin (An et al., 2014; Wang et al., 2017), and intensified during Paleocene collision (Orme et al., 2015; Hu et al., 2016b) and subsequent stages of Himalayan orogenic growth (DeCelles et al., 2011; Wang et al., 2013).
The Kyi or Lhasa River (drainage area 32,470 km2, length 450 km, average altitude 4500 m above sea level [a.s.l.]) includes four main branches: the Miggi, Dam, Maizho, and Duilong Rivers (Fig. 1). The longest is the Miggi (Rezhen Tsangpo; in Tibetan: tsangpo = big river; chu = water, small river; tso = lake), sourced in the Maidika region (Jiali County), above 5000 m a.s.l. in the southeastern part of the central Tibetan Plateau, south of the Tanggula Mountains and east of the Nam Tso (Fig. 3). The Dam (Phongdolha) and Yangbajain (upper tract of the Duilong) tributaries are sourced from the Nyainqêntanglha Range in the west, reaching 7162 m a.s.l. in altitude. Two major dams built for hydropower and flood regulation exist on the river, the Pondo hydropower station (2011) at the confluence between the Miggi and the Dam Rivers, and the Zhikong hydropower station (2007) at the confluence with the Drigung tributary 65 km downstream. Farther downstream the trunk river is joined by the Maizho River, draining lower-relief areas in the east, and it passes through Lhasa City, receives the Duilong River, and eventually joins the Yarlung Tsangpo at 3590 m a.s.l. in Quxu County (Fig. 1).
The catchment has a semiarid, relatively mild plateau climate, with warm summers and temperatures mainly above freezing, even in the coldest months (average annual temperature 9 °C; 3 °C in winter and 15 °C in summer). Windstorms are frequent in late winter to spring. The annual precipitation of 450–500 mm is concentrated (85%) between June and September. Annual water discharge ranges from 4 to 10 km3. Melting of snow and glaciers starts in May and contributes 20%–30% of the water, but most is supplied by summer-monsoon rainfall between July and August. Floods may occur between July and September, whereas the river has low water and sometimes freezes in winter. Average suspended sediment concentration is 0.1 g/L. Pollution is minor from sewage and minimal from pesticides and fertilizers.
Geology and Geomorphology
Within the Lhasa River catchment, granitoid rocks are estimated to constitute 28% of exposure areas, and volcanic rocks only constitute 19% (largely represented by Linzizong lavas and ignimbrites), with the majority of the catchment being covered by sedimentary rocks (53%; Zhang et al., 2012, p. 1462; R. Guo, 2017, personal commun.). Paleozoic strata are widely exposed in the central part of the basin drained by the Dam River, whereas Jurassic–Cretaceous strata crop out in the southern part drained by the Maizho and Duilong Rivers, and Jurassic strata crop out in the northern part drained by the upper Miggi River (Fig. 1). Granitoid rocks are best exposed in the rugged Nyainqêntanglha Range in the western part of the catchment, representing the western flank of the Yadong-Gulu rift, the most prominent graben system in Tibet (Armijo et al., 1986). The central part of the rift is bounded by a low-angle detachment that has rapidly exhumed in the footwall mylonitic gneisses and granitoids as young as late Miocene during the Pliocene–Pleistocene. Areas of maximum extension are characterized by active basin incision and intrabasin topographic highs (Kapp et al., 2005, 2008). Several peaks in the Nyainqêntanglha Range reach far above 6000 m or even 7000 m in elevation, whereas to the east of the graben, only a few reach above 5500 m.
A stream-profile analysis of main Lhasa River tributaries (Fig. 3) shows a relative homogeneity in their morphology. Numerical morphometric descriptors include the concavity index θ, which expresses the rate of change of channel gradient with drainage area, and the channel-steepness index ks, a measure of bedrock-channel response to differential rock uplift, if other controls such as rock type, climate, flood hydrology, or sediment flux are negligible or sufficiently well constrained (Whipple and Tucker, 1999; Kirby et al., 2003). For most of the analyzed stream segments, the concavity index is within a narrow range of values slightly lower than the norm (θ = 0.32–0.37), as is typical of short, steep streams characterized by rapid sediment transport (Whipple, 2004). A slightly negative value was obtained for the Miggi River, owing to knickpoints associated with faults and/or marked changes in substrate properties. Channel steepness is also remarkably homogeneous in a narrow range for most stream segments (ks = 105–118), whereas it is lower for the uppermost Dam, higher for the Miggi, and highest for the Nyainqêntanglha River draining the steepest slopes of the entire Lhasa River catchment.
SAMPLING AND ANALYTICAL METHODS
Nineteen samples of fine-grained to coarse-grained sand were obtained from active river bars of the Lhasa River and its tributaries from June 2013 to August 2017 (information on sampling sites is provided in Data Repository Table DRA11 and Google Earth map Lhasariver.kmz). We also analyzed eight samples collected from the Yarlung Tsangpo upstream and downstream of the Lhasa River confluence, and five samples from its three major tributaries draining the Lhasa block (Raka, Xiang, and Nyemo Rivers; Fig. 4). The complete petrographic, mineralogical, geochemical, and geochronological data sets and full methodological information are provided in Appendix A of the Data Repository material.
A quartered aliquot of each bulk sand sample was impregnated with araldite, cut into a standard thin section stained with alizarine red to distinguish dolomite and calcite, and analyzed by counting 400–450 points under the microscope (Gazzi-Dickinson method; Ingersoll et al., 1984). Average rank of metamorphic rock fragments was expressed by the metamorphic indices MI or MI*, ranging respectively from 0 (detritus from sedimentary and volcanic rocks) or 100 (detritus from very low-grade metamorphic rocks) to 500 (detritus from high-grade metamorphic rocks; Garzanti and Vezzoli, 2003). Sand classification was based on the main components quartz (Q), feldspars (F), and lithic fragments (L), considered if >10% QFL (e.g., a sand is named litho-feldspatho-quartzose if Q > F > L > 10% QFL; Garzanti, 2016). Median grain size was determined in thin section by ranking and visual comparison with standards of f/4 classes prepared by sieving in our laboratory.
Heavy mineral analyses were carried out on a quartered aliquot of the 32–500 µm class obtained by dry-sieving. The >500 µm class, where the heavy fraction is generally represented mainly by rock fragments, was excluded because the presence of detrital grains with great size differences within a single concentrate makes mounting and identification difficult (Mange and Maurer, 1992). The coarse silt class (32–63 µm) was instead included to be sure not to lose the finest tail of the size distribution, where the densest minerals such as zircon or monazite are concentrated (Garzanti et al., 2008).
Heavy minerals were separated by centrifuging in sodium polytungstate (density ~2.90 g/cm3) and recovered by partial freezing with liquid nitrogen. On grain mounts, 200–250 transparent heavy mineral grains were point counted at a suitable regular spacing under the petrographic microscope to obtain real volume percentages (Galehouse, 1971). Grains of uncertain identification were checked by Raman spectroscopy. The sum of zircon, tourmaline, and rutile over total transparent heavy minerals (ZTR index of Hubert, 1962) estimates the durability of the assemblage (i.e., extent of recycling; Garzanti, 2017). The hornblende color index (HCI) varies from 0 to 100 and estimates formation temperatures of metamorphic and igneous rocks (Andò et al., 2014). Heavy mineral concentration was calculated as the volume percentage of total (HMC) and transparent (tHMC) heavy minerals (Garzanti and Andò, 2007a). Heavy mineral suites ranged from very poor (HMC < 0.5) to poor (0.5 ≤ HMC < 1) and moderately poor (1 ≤ HMC < 2) to moderately rich (2 ≤ HMC < 5) and rich (5 ≤ HMC < 10). Detrital components are listed in order of abundance throughout the text.
From the large (~3 kg) sand sample S4676 collected on a longitudinal bar within the Lhasa River, 12 km upstream of the Yarlung Tsangpo confluence (29°22ʹ44ʺN, 90°51ʹ38ʺE; star in Fig. 1), zircon, apatite, garnet, rutile, and monazite grains were separated by elutriation methods and selected by handpicking. Geochemical analyses of apatite were carried out at State Key Laboratory of Lithosphere Evolution, Institute of Geology of Geophysics, Chinese Academy of Sciences. Analyses of zircon, garnet, and rutile were carried out at the State Key Laboratory of Mineral Deposits Research, Nanjing University. U-Pb ages and Hf isotopes of zircon grains were determined by laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS), whereas the chemical composition of garnet and rutile was measured with a JEOLJXA-8100M electron microprobe (methods described in Jackson et al., 2004; Hu et al., 2010).
Raman spectroscopy analyses were carried out with a Raman Renishaw inVia on 386 transparent heavy mineral grains in the Department of Earth and Environmental Sciences, University of Milano-Bicocca, Italy. After calibration using the 520.6 cm–1 Raman band of silicon, spectra were obtained by focusing the 532 nm laser beam on the grain surface for 40–80 s, depending on signal intensity. Amphiboles, pyroxenes, and epidote-group minerals were identified by their diagnostic peaks in the medium-frequency to high-frequency regions around 666 cm–1, 855 cm–1, and 1000 cm–1, and in the OH region between 3100 and 3700 cm–1 (Tribaudino et al., 2012; Andò and Garzanti, 2014; Bersani et al., 2014).
Forward Compositional Modeling
The relative amounts of bed-load sand contributed to the Lhasa River by its different tributaries, and to the Yarlung Tsangpo by the Lhasa River, were assessed by forward mixing models based on integrated bulk petrography and heavy mineral data (Garzanti et al., 2012; mathematical approach illustrated in detail in Data Repository Appendix A [see footnote 1]). Replicate samples are critical to improve on the accuracy of calculations and to evaluate the variability of sand composition through time associated with short-term variability of erosion patterns in the catchment. In this study we thus analyzed three samples collected by different operators in different years in the terminal tract of the Lhasa River between the Duilong and Yarlung Tsangpo confluences (Fig. 1), four samples collected in the Yarlung Tsangpo between Xigaze and the Lhasa River confluence (Fig. 1), and four samples collected in the Yarlung Tsangpo between the Lhasa River confluence and Gonggar County (Fig. 1).
Potential modifications of sediment transport caused by the recent closure of the Pondo and Zhikong dams were checked by comparing the composition of trunk-river sands collected upstream and downstream of both dams (Fig. 1). Downstream samples were poorer in quartz and more closely resembled sediments contributed by tributaries locally, revealing efficient sequestration of bed load in reservoirs. Marked enrichment in dense minerals shortly downstream of the Zhikong Dam pointed to extensive erosion of previously deposited sediments within the channel, a process inferred to restore the original trunk-river sand composition in many rivers worldwide (e.g., Garzanti et al., 2015; Vezzoli et al., 2016).
Because climate in southern Tibet is semiarid, the effects of chemical weathering on sand composition were considered to be negligible. Hydraulic-sorting control was checked, but for all samples the grain density—calculated as the weighted average of the density of all detrital components (source rock density index of Garzanti and Andò, 2007a)—was between 2.64 and 2.71 g/cm3, which is well within the expected range for sediments derived from magmatic arcs (table 1inGarzanti et al., 2009). The major source of uncertainty by far in the calculation of sediment yields and erosion rates is from the poor assessment of sediment fluxes, owing to the lack of measured data from both Yarlung Tsangpo and Lhasa River.
MODERN SAND FROM THE LHASA BLOCK
In this section we illustrate the key petrographic and heavy mineral signatures of river sand samples derived from diverse sectors of the Lhasa block (Figs. 5 and 6; complete datasets provided in Tables DRA2 and DRA3 [see footnote 1]). A comparison with detrital modes of river sand derived from the Ladakh batholith, representing the western equivalent of the Gangdese batholith in the northwestern Himalaya (Gansser, 1980), is shown in Table 1. Petrographic data on modern sands from the Lhasa River, Yarlung Tsangpo, and some of its tributaries were previously reported in Garzanti et al. (2004), also containing heavy mineral data, and Zhang et al. (2012), focusing instead on U-Pb age and Hf isotopic signature of detrital zircons.
The Nyainqêntanglha, Nu, Gurum, and Yangbajain Rivers drain the Nyainqêntanglha Range, where Cretaceous to Cenozoic granitoids are exposed together with their wall rocks, including orthogneisses and Paleozoic metasedimentary rocks (Fig. 1). These rivers carry quartzo-feldspathic to feldspatho-quartzose plutoniclastic sand with plagioclase ≥ K-feldspar, a few high-to very high-rank metamorphic rock fragments including sillimanite, and micas (biotite > muscovite; Figs. 5A and 6). Poor to moderately rich, hornblende-dominated heavy-mineral assemblages include garnet, apatite, epidote, zircon, clinopyroxene, and titanite. The Womaguo right-bank tributary of the Nyainqêntanglha River, draining very low-grade Paleozoic metasedimentary rocks (Fig. 1), carries quartzo-lithic sand dominated by shale/slate and siltstone/metasiltstone with some metarhyolite/metadacite lithic grains (Figs. 5B and 6). The very poor heavy mineral assemblage includes amphibole, pyroxene, epidote, and garnet. The Pengbo right-bank tributary of the Lhasa River, chiefly draining Mesozoic strata capped by the Linzizong volcanic rocks (Fig. 1), carries quartzo-feldspatho-lithic sand with shale, slate, volcanic, and very low-rank metavolcanic grains. The moderately rich heavy mineral assemblage includes amphibole, epidote, garnet, and clinopyroxene.
The Duilong, Miggi, and Dam Rivers, draining granitoid batholiths and their Paleozoic to Cretaceous volcano-sedimentary covers in diverse proportions, carry litho-feldspatho-quartzose sand with sedimentary and very low- to high-rank metasedimentary lithics. Duilong sand contains microlitic to felsitic volcanic rock fragments derived from Linzizong arc lavas and ignimbrites, and a moderately poor heavy mineral suite with amphibole and subordinate epidote, garnet, and clinopyroxene. A similar suite characterizes Miggi sand, whereas the Dam suite is moderately rich and hornblende-dominated. Sand of the Maizho River is quartzo-feldspatho-lithic, poorer in quartz and K-feldspar, and richer in volcanic and metamorphic detritus (Figs. 5C and 6). The poor heavy mineral suite includes mainly amphibole with epidote and garnet. Lhasa River sand downstream of the Duilong confluence is litho-feldspatho-quartzose with microlitic and felsitic volcanic, shale/slate/phyllite, siltstone/metasiltstone, metadacite, chloritoschist, and rarer higher-rank metamorphic rock fragments (Fig. 5D). The moderately rich, amphibole-dominated suite contains epidote, and minor garnet, zircon, apatite, and clinopyroxene.
Gravel composition is markedly different, and dominated by volcanic pebbles in the trunk river (Fig. 4B) as in all of its four major branches.
The Nyemo River (Fig. 1) carries feldspatho-quartzose sand with a rich hornblende-dominated suite derived mostly from the Gangdese batholith. In contrast, gravel is dominated by pebbles derived from Linzizong volcanic rocks, with only subordinate granitoid clasts (Fig. 4D). The Raka Tsangpo west of Xigaze and the Xiang River east of Xigaze carry sands with less feldspar and more sedimentary to very low-rank metasedimentary and volcanic to metavolcanic rock fragments. Heavy mineral suites are similarly hornblende-dominated but only moderately rich, and they include epidote and clinopyroxene, and minor apatite, zircon, tourmaline, titanite, and garnet.
In this section, we summarize the geochemical and geochronological signatures of detrital minerals separated from the large sand sample S4676 collected in the terminal tract of the Lhasa River (Fig. 4A). Our aim is to define a modern-sand reference useful to constrain provenance diagnoses for ancient Himalayan sandstone derived from the active margin of Asia since the Cretaceous (e.g., Garzanti et al., 1996; Najman et al., 2008, 2010; DeCelles et al., 2011; Hu et al., 2016b), and a standard of comparison for mineralogical studies of sediment and sedimentary rock derived from comparable dissected-arc sources worldwide. Therefore, we focused on all chemically resistant detrital species, which are more likely to survive diagenesis and to be preserved in the stratigraphic record. Tourmaline and Cr-spinel were not extracted in sufficient amounts to be investigated.
U-Pb Zircon Ages and Hf Isotopes
We obtained 107 concordant U-Pb ages from zircons in sample S4676 (71% Cenozoic, 14% Mesozoic, and the remaining 15% scattered between 455 and 1784 Ma; Figs. 7A and 7B). The youngest ages were 7.5 ± 0.3 Ma, 18.9 ± 0.9 Ma, and 20.9 ± 0.6 Ma, with discordances of 3.6%, 4.8%, and 2.8%, respectively. The youngest peak was at 21.1 Ma, and the main peak was at 47.5 Ma (data set provided in Table DRA4 [see footnote 1]). Among zircon grains of Jurassic to Cenozoic age, 33 displayed high 176Hf/177Hf and positive εHf(t), and 22 displayed low 176Hf/177Hf and negative εHf(t). U-Pb geochronology data and Hf isotope signatures of zircon grains are provided in Tables DRA4 and DRA5 [see footnote 1]).
Rocks exposed in the Lhasa block are characterized by distinctive zircon age patterns and Hf isotopes, but Mesozoic sediments show mixed features (Fig. 7; Leier et al., 2007). Detrital zircons in Lhasa River sand mainly yielded U-Pb ages between 40 and 100 Ma combined with positive εHf(t) values, indicating that they are mainly derived from Upper Cretaceous to Eocene magmatic rocks of the Gangdese batholith (Fig. 7F). The less abundant zircon grains with Early Cretaceous to Jurassic ages and negative εHf(t) values are derived instead from magmatic rocks of the central Lhasa terrane. The 16 ages older than 400 Ma indicate recycling of Paleozoic to Mesozoic sedimentary and volcaniclastic strata. Few detrital zircons yielded ages and isotopic features comparable to those of northern Lhasa magmatic rocks (Fig. 7).
The age spectrum obtained by Zhang et al. (2012) on a Lhasa River sand collected ~50 km upstream of the confluence with the Yarlung Tsangpo included a higher percentage of zircons yielding Early Cretaceous to Jurassic ages and negative εHf(t) values, indicating greater contribution from central Lhasa than for sample S4676 (Figs. 7A, 7B, and 7F).
The 60 apatite grains analyzed from sample S4676 displayed significant variations in rare earth element (REE) patterns, which allowed us to identify two groups, a dominant one with light (L) REE enrichment and positive or mildly negative Eu anomalies, and a subordinate one with middle (M) REE enrichment and a markedly negative Eu anomaly (Fig. 8). LREE enrichment reflects low heavy (H) REE content of the magma from which the apatite crystallized, whereas low LREE content points to crystallization of accessory phases such as monazite or allanite. The dominant apatite grains of group 1, displaying steeper right-inclined REE pattern (Fig. 8A), compare with apatite crystals in metaluminous rocks, including the least-fractionated Gangdese mafic plutons and postcollisional adakites. Apatite grains of group 2, with a flat pattern and negative Eu anomaly, resemble those of crystals in fractionated peraluminous I-type and S-type granites exposed in the Gangdese batholith and central Lhasa terrane (Chu et al., 2009).
Most apatite grains in Lhasa River sand have compositions similar to those from the Gangdese batholith (Figs. 8C, 8D, and 8E), and display negative εNd(0) values and 147Sm/144Nd ratios around 0.1, suggesting a depleted mantle source (Fig. 8B). Trace-element data and Sm/Nd isotope signatures of apatite grains are provided in Tables DRA6 and DRA7 [see footnote 1]).
The 61 rutile grains analyzed from sample S4676 showed a great variability in trace-element concentrations. Nb ranged from <70 to 7810 ppm, Fe ranged from 8 to 15,882 ppm, V ranged from 3298 to 8405 ppm, Cr ranged from <20 to 7909 ppm, Zr ranged from <20 to 3388 ppm, and Al ranged from <5 to 153 ppm (dataset provided in Table DRA8 [see footnote 1]). Most grains were rich in Nb, indicating felsic rather than metapelitic source rocks (Meinhold, 2010). Only a few were Cr-rich, suggesting a mafic source (Fig. 9A). Formation temperatures calculated after Watson et al. (2006) range between 500 °C and 900 °C, with 37 grains crystallized at lower temperatures and only eight crystallized at temperatures >750 °C (Fig. 9B). The U-Pb age distribution of detrital rutile matches closely the U-Pb age distribution of detrital zircon (R. Guo, 2017, personal commun.).
The 74 garnet grains analyzed from sample S4676 were characterized by diverse compositions. SiO2 varied from 35.1 to 37.3 wt%, Al2O3 varied from 20.6 to 21.9 wt%, FeO varied from 17.3 to 37.0 wt%, MnO varied from 5.1 to 25.6 wt%, CaO varied from 0.1 to 9.1 wt%, MgO was ≤1.2 wt%, and Cr2O3 was ≤0.01 wt% (dataset provided in Table DRA9 [see footnote 1]). Almandine and spessartine, the dominant end members, were present in similar amounts. Virtually all detrital garnets in Lhasa River sand have geochemical affinity with those derived from granitoid rocks (Krippner et al., 2014). They plot (Fig. 10A) in the field of large granitic intrusive bodies, leucosomes, and migmatites in the diagram of Suggate and Hall (2014), and in the field (Fig. 10B) of intermediate-felsic igneous rocks in the discrimination diagram of Mange and Morton (2007). Two grains rich in grossular molecule plot in the field of calc-silicates, skarns, and rodingites (Fig. 10A).
The 50 monazite grains analyzed from sample S4676 displayed a wide range of compositions (dataset provided in Table DRA10 [see footnote 1]). P2O5 varied from 22.6 to 30.9 wt%, SiO2 varied from 0.2 to 4.9 wt%, La2O3 varied from 10.8 to 15.8 wt%, Ce2O3 varied from 22.9 to 32.0 wt%, Nd2O3 varied from 8.1 to 12.1 wt%, and ThO2 varied from 3.7 to 22.7 wt%. The relative high Th content, marked LREE enrichment, and negative Eu anomaly suggest dominant provenance from granitoid rocks (Schandl and Gorton, 2004; Williams et al., 2007).
Amphibole, Pyroxene, and Epidote-Group Minerals
Among the 146 amphibole grains analyzed with Raman spectroscopy from sample S4676, hornblende was much more common (44%) than actinolite (8%); the remaining amphibole grains (47%) yielded spectra with intermediate character between hornblende and Fe-actinolite. Blue-green to green amphibole is dominant in Lhasa River sand (92 ± 4 of detrital amphiboles), as throughout the catchment (Table 1). Green-brown to brown hornblende represents up to 11% of amphibole grains in Maizho sand, which also contains oxy-hornblende.
The 68 pyroxene grains analyzed with Raman spectroscopy from sample S4676 were diopside (51%) and augite (47%) in similar amounts; one pigeonite grain was identified, whereas neither hypersthene nor enstatite was detected. Among other samples, only the Womaguo tributary contains common brown augitic clinopyroxene. Hypersthene was sporadically recorded throughout the catchment; one enstatite grain was observed in one trunk-river sample.
The 172 epidote-group minerals analyzed with Raman spectroscopy from sample S4676 were 65% epidote and 35% clinozoisite; zoisite grains were not observed. Green pistacitic epidote was dominant. Most samples yielded some yellow-brown allanite grains, most frequent in the Duilong River and in its tributaries draining the Nyainqêntanglha Range. Diagnostic Raman peaks for each analyzed grain are indicated in Table DRA11 [see footnote 1].
In the classic model of Dickinson and Suczek (1979), the composition of sediment derived from a magmatic arc is a function of the degree of unroofing (i.e., of the relative amount of exposed plutonic roots), which tends to increase progressively during erosion of volcanic covers after the cessation of volcanic activity and/or during tectonic uplift. An undissected arc exposing lavas and pyroclastic products exclusively generates feldspatho-lithic volcaniclastic detritus with pyroxene-rich heavy mineral suites. Basalt-derived sand is dominated by dark glassy or lathwork lithic fragments, augitic clinopyroxene, and commonly olivine. Sand derived from porphyric andesite includes more plagioclase, microlitic lithic fragments, oxy-hornblende or brown kaersutitic amphibole, and biotite. Sand derived from more felsic products (e.g., dacite) contains common to dominant hypersthene, and sand derived from rhyodacite to rhyolite contains abundant felsitic to vitric pumiceous lithics and significant quartz, commonly displaying hexagonal (quartz β) euhedral outline, straight extinction, and glass-filled embayments indicating magmatic resorption (Dickinson, 1970; Garzanti et al., 2013). The progressive dissection of the tonalitic-granodioritic core of the arc massif generates quartzo-feldspatho-lithic to quartzo-litho-feldspathic sand with mixed volcano-plutonic signature. Eventually, when and where volcanic covers are largely removed, lithoquartzo-feldspathic to quartzo-feldspathic sand includes quartz, plagioclase, and K-feldspar in similar proportions, and heavy mineral suites dominated by blue-green hornblende (Marsaglia and Ingersoll, 1992; Garzanti and Andò, 2007b). Such a stratigraphic evolution of detrital modes has long been documented in great detail in the Great Valley Group of California, where it reflects the progressive dissection of the Sierra Nevada arc massif concurrent with northward migration of the Mendocino triple junction and propagation of the right-lateral San Andreas transform system (Ojakangas, 1968; Dickinson and Rich, 1972; Mansfield, 1979; Ingersoll, 1983, 2012; Ingersoll and Eastmond, 2007). A similar evolution is documented in triple-junction and strike-slip tectonic settings worldwide (Marsaglia and Ingersoll, 1992; Marsaglia, 2004). The same compositional trends are displayed by detrital modes in the Transhimalayan forearc-basin succession, which records the rapid unroofing of the Ladakh and Gangdese batholiths during the initial stages of the India-Asia collision (Garzanti and van Haver, 1988; Henderson et al., 2010; Hu et al., 2016b).
Sand derived from cordilleras (DeCelles and Hertel 1989; Johnsson et al., 1991) and wide continental arc-terranes characterized by a long polyphase accretion history such as the Lhasa block (Zhu et al., 2011) may display a notably more varied composition, including lithic-rich, sedimentary to low-rank metasedimentary and paleovolcanic to low-rank metavolcanic detritus mixed in various proportions with the neovolcanic and plutonic components shed from the arc massif (Fig. 11). The similarity in detrital modes between Lhasa River sand and Quaternary to modern sands shed by magmatic arcs all along the western continental margin of North and Central America (Table 2) is one more confirmation of the validity of the arc-unroofing model (Dickinson, 1985). This model represents a valuable reference for provenance interpretation of ancient sandstone suites derived from west-facing continental-margin arc-trench systems generated by eastward/northeastward subduction of oceanic lithosphere (Dickinson, 1978, 1985, 1988; Garzanti et al., 2007).
PROVENANCE BUDGETS, SEDIMENT GENERATION, AND EROSION RATES
Forward mixing calculations suggest that most bed-load sand in the Lhasa River is generated from erosion of granitoid batholiths (81% ± 2%), which is fully consistent with indications provided by the geochemical and geochronological fingerprints of detrital zircon (Fig. 7), apatite (Fig. 8), rutile (Fig. 9), and garnet (Fig. 10). The rest is derived from Paleozoic to Paleocene sedimentary/very low-grade metasedimentary (16% ± 3%) and volcanic/metavolcanic (3% ± 1%) cover strata (Table 3). Field observations indicate that the contribution from the Linzizong and other volcanic units, although minor for sand, is major for gravel (up to 80%), testifying to an order-of-magnitude difference in sand-generation potential between arc lavas and granitoid batholiths. A high ratio of volcanic to plutonic clasts was also observed in most conglomerates of the Great Valley Group of California, which was explained by “the tendency for granitic rocks to disintegrate to sandy grus upon weathering, whereas dense joint blocks of aphanitic volcanics are persistent” (Dickinson and Rich, 1972, p. 3011–3012). This explanation is fully consistent with our field observations.
The Miggi branch is estimated to supply up to one third of total Lhasa River bed-load sand, the Maizho, the Dam, and the Duilong supply up to a fifth each, and the rest is supplied by the Drigung, Pengbo, and other lesser tributaries (Table 3). The Lhasa River is calculated to supply as much as half of the sediment flux of the Yarlung Tsangpo downstream of their confluence, an assessment that suffers from significant uncertainty because of the overall similar mineralogy of bed-load sand in the two rivers (Fig. 11). Such an estimate would imply that rates of sand production are up to 3–4 times higher in the Lhasa River catchment than in the much larger Yarlung Tsangpo catchment upstream (Table 3). If data on sand-sized sediments can be extrapolated to bulk-sediment production, and if an annual sediment load of 30 million tons is assumed for the Yarlung Tsangpo downstream of the Lhasa River confluence based on reapportioning of total Brahmaputra sediment fluxes (Garzanti et al., 2004; Enkelmann et al., 2011; Darby et al., 2015), then indicative erosion rates of 0.17 mm/yr and 0.05 mm/yr are indicated for the Lhasa River catchment and for the Yarlung Tsangpo catchment upstream of the Lhasa River confluence, respectively (Table 3). Given the different proportion of exposure areas for granitoid, volcanic, and sedimentary rocks in the Lhasa River catchment, calculated as 28:19:53 (Zhang et al., 2012; R. Guo, 2017, personal commun.), sand production would be one full order of magnitude higher in granitoids than in sedimentary rocks, and twice in sediment rocks than in lavas. The corresponding erosion rates based on the assumptions made above would range from as high as 0.5 mm/yr for granitoid batholiths to as low as 0.05 mm/yr for sedimentary rocks and 0.02 mm/yr for volcanic rocks.
This very large difference, partly explained by wide exposures of granitoid rocks in the rugged Nyainqêntanglha Range characterized by active incision (Fig. 3), is notably amplified by the high sand-generation potential of granitoid rocks (Palomares and Arribas, 1993). Sand-generation potential appears to be an order of magnitude lower for both volcanic and sedimentary rocks for opposite mechanical reasons. Lavas are tough to break and largely represented by gravel in mountain areas, whereas tuffs and shale/slate grains (Fig. 5B) are easily comminuted in fine particles largely carried as suspended load. This exposes a serious bias affecting sediment budgets based on data on sand-sized sediments only (e.g., Resentini et al., 2017), and it highlights the necessity to take into account the entire size spectrum from mud to gravel in order to improve the accuracy of the estimates.
Even less accurate are sediment budgets based on single sand-sized minerals, particularly if they are rare components of the sand framework (Vezzoli et al., 2016), which is invariably the case for zircon (Fig. 2). Let us compare the zircon age spectrum of our Lhasa River sample S4676 with those obtained on samples collected in the terminal tract of the Lhasa River (AY09-08-08) and in the Yarlung Tsangpo upstream (Quxu sample AY09-08-08) and downstream of the Lhasa River confluence (Gonggar sample AY06-28-06; figs. 3E, 4B, and 4CinZhang et al., 2012). We note that the variability is greater between the two Lhasa River samples, which should ideally be identical and yet show percentages of Mesozoic–Cenozoic grains from 62% to 85% (Fig. 7A), than between the two Yarlung Tsangpo samples, which would theoretically show a difference related to the zircon input from the Lhasa River and yet show virtually identical percentages of Mesozoic–Cenozoic grains (68.6% and 69.1%; figs. 4B and 4CinZhang et al., 2012). The close similarity of the cumulative probability plots of zircon ages displayed by the Yarlung Tsangpo samples collected at Quxu and Gonggar, and the lack in the Gonggar sample of the age and Hf isotopic signatures of the northern Lhasa terrane characterizing many Lhasa River zircons led Zhang et al. (2012, p. 1465) to conclude “that the northern Lhasa detrital-zircon signatures are completely diluted in the Yalu River at the Gonggar site.” A robust zircon budget can hardly be calculated from these data, principally because the 50 Ma age peak characteristic of the southern Gangdese arc is dominant not only in Lhasa River sand but in Yarlung Tsangpo sand as well (Fig. 7A). Pre-Mesozoic zircons diagnostic of sediment supply from right-bank Yarlung Tsangpo tributaries draining the Himalaya (e.g., 500 Ma peak; DeCelles et al., 2004; Gehrels et al., 2011; Carrapa et al., 2017) show up only as a small peak in Yarlung Tsangpo samples (Zhang et al., 2012, p. 1457). Close inspection of figure 8A in Zhang et al. (2012) indicates that the 500 Ma peak decreases in size by almost half between Quxu (where it represents ~13% of zircon grains) and Gonggar (where it represents ~8% only), which is consistent with major supply from the Lhasa River.
Robust zircon provenance budgets are equally difficult to construct within the Lhasa River catchment, owing to the considerable variability between our S4676 sample and sample AY09-08-08 of Zhang et al. (2012). Considering that the Miggi branch drains northern and central Lhasa in similar proportions, that the Duilong branch drains central and southern Lhasa in similar proportions, and that the Dam and Maizho branches drain almost entirely within central and southern Lhasa, respectively, based on our data set, we would expect equivalent sand contribution from central and southern Lhasa, with subordinate additional supply from northern Lhasa (Table 3). The calculated proportions do not change much if we consider detrital supply from granitoid rocks only. According to the available data, zircon fertility does not seem to be markedly different in granitoid rocks within the northern, central, and southern terranes of the Lhasa block (Fig. 7H). Zircon grains in Lhasa River sand should thus show the age and isotopic signatures of central and southern Lhasa batholiths in similar proportions. This is definitely more the case for sample AY09-08-08 of Zhang et al. (2012), whereas the results from our sample S4676 are biased in favor of the southern Lhasa Gangdese batholith for reasons we are unable to ascertain (Figs. 7A, 7B, and 7F).
Following the pioneering work of Bill Dickinson and coworkers, who defined the paradigm of provenance analysis in plate-tectonic terms and dedicated specific attention to arc terranes and their erosional evolution in space and time, sedimentary-petrology studies have concentrated on relatively simple detrital sources such as undissected intraoceanic or continental arcs. Less well studied are more complex arc terranes in their advanced stage of dissection, when granitoid batholiths together with a spectrum of metamorphic wall rocks and their sedimentary cover strata are exposed. The present study helps to fill the gap by focusing on modern sand derived from the Lhasa block—the remnant of the Asian active margin that collided with the passive margin of India in the Paleocene—and specifically on sand from the Lhasa River, the largest left-bank Tibetan tributary of the Yarlung Tsangpo/Brahmaputra.
Although Paleozoic and Mesozoic sedimentary rocks and Jurassic to Eocene volcanic rocks represent the majority of exposure area in the Lhasa River catchment, litho-feldspatho-quartzose detrital modes with hornblende-dominated heavy mineral assemblages indicate that the majority (four fifths) of sand bed load is derived from erosion of granitoid batholiths, widely exposed in the highly elevated, rapidly eroded Nyainqêntanglha Range. This conclusion is corroborated by U-Pb age spectra and Hf isotopic fingerprints of detrital zircon and by the geochemical signatures of detrital apatite, rutile, garnet, and monazite, all of which consistently point to dominant provenance from granitoid batholiths. Geochronological and geochemical fingerprints of detrital zircons indicate that most are shed from the Gangdese batholith rather than from central and northern Lhasa magmatic rocks. Only a few zircon grains are recycled from Paleozoic to Mesozoic strata.
Integrated petrographic, mineralogical, geochemical, and geochronological data sets of modern sands shed by the Lhasa block, a complex continental-arc terrane incorporated within the Himalayan collision orogen, define a modern-sand reference useful for constraining provenance of ancient Himalayan sandstones derived from the active margin of Asia since the Cretaceous, and a standard of comparison for mineralogical studies of sediment and sedimentary rock derived from comparable dissected-arc sources worldwide.
The path leading to this study began back in 1984, fostered by the insight and encouragement offered generously by Bill Dickinson to a stray student in Calabria, and owes much to the advice received from Ray Ingersoll in more recent years, also in the form of numerous kindly supportive reviews. Our work benefited from assistance by Gaoyuan Sun and Juan Li in the field and additional samples provided by Jérôme Lavé, Christian France-Lanord, Henri Munack, Jan Blöthe, Peter Clift, and Alberto Resentini. Fundamental help and advice with Raman spectroscopy analyses were provided by Sergio Andò. This study was supported financially by the National Natural Science Funds for Distinguished Young Scholar to X. Hu (41525007) and by Progetto di Rilevante Interesse Nazionale, Ministero dell’Istruzione, dell’Università e della Ricerca (PRIN-MIUR) to E. Garzanti (2015EC9PJ5). Very careful constructive reviews by Kathleen Marsaglia, Barbara Carrapa, and Raymond Ingersoll are very gratefully acknowledged.
Supplementary data associated with this paper can be found as item 2018350 at www.geosociety.org/datarepository/2018/, or on request from email@example.com or Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301-9140, USA. These include information on sampling sites (Table DR1) and the complete data sets for bulk sand petrography (Table DR2), heavy minerals (Table DR3), detrital zircon U-Pb geochronology (Table DR4), Hf isotope signatures of zircon grains (Table DR5), trace-element data on detrital apatite (Table DR6), Sm and Nd isotope signatures of apatite grains (Table DR7), and geochemistry of detrital rutile (Table DR8), garnet (Table DR9), and monazite (Table DR10). Raman-spectroscopy data on detrital amphibole, pyroxene, and epidote-group minerals are provided in Table DR11. Table captions and methodological details, including the description of the approach followed in the calculation of provenance budgets and data sources for zircon ages in the Lhasa block, are given in Appendix DRA. The Google Earth map of sampling sites, Lhasariver.kmz, is also provided.
Figures & Tables
Tectonics, Sedimentary Basins, and Provenance: A Celebration of the Career of William R. Dickinson
- absolute age
- andesitic composition
- clastic sediments
- Far East
- garnet group
- heavy minerals
- igneous rocks
- isotope ratios
- Lhasa China
- plate collision
- plate tectonics
- Raman spectra
- stable isotopes
- Tibetan Plateau
- volcanic belts
- Xizang China
- zircon group
- detrital zircon
- Nyainqentanglha Range
- Lhasa River