Terrace formation in the upper Bengal basin since the Middle Pleistocene: Brahmaputra fan delta construction during multiple highstands

EAGE Basin Research (2018) 30 (Suppl. 1), 550–567, doi: 10.1111/bre.12236 Terrace formation in the upper Bengal basin since the Middle Pleistocene: Brahmaputra fan delta construction during multiple highstands Jennifer L. Pickering,* ,1 Steven L. Goodbred Jr,* Jeremiah C. Beam,† John C. Ayers,* Aaron K. Covey,* Haresh M. Rajapara‡,§ and Ashok K. Singhvi§ *Department of Earth and Environmental Science, Vanderbilt University, Nashville, TN, USA †Department of Chemistry, Vanderbilt University, Nashville, TN, USA ‡Department of Physics, Electronics and Space Science, Gujarat University, Gujarat, India §Physical Research Laboratory, Navarangpura, Ahmedabad, Gujarat, India ABSTRACT Floodplains, paleosols, and antecedent landforms near the apex of the Brahmaputra fan delta in north-central Bangladesh preserve cycles of fluvial sediment deposition, erosion and weathering. Together these landforms and their associated deposits comprise morphostratigraphic units that define the river’s history and have influenced its channel position and avulsion behaviour through the Late Quaternary. Previously, temporal differentiation within these units has not been sufficient to decipher their sequence of deposition, an important step in understanding the spatial pattern of migration of the Brahmaputra River. Holocene units in this region are fairly well established by radiocarbon dating of in situ organic material, but pre-Holocene units are considered Pleistoceneaged if organic material is dated >48 000 year BP (the limit of radiocarbon dating) or the sediments are positioned beneath a prominent paleosol, interpreted as a buried soil horizon that developed during a previous sea level lowstand. In such cases, these morphostratigraphic units have been broadly interpreted as Pleistocene without knowing their absolute depositional ages or relative evolutionary chronology. Here we use detailed sediment analysis to better differentiate morphostratigraphic units at the Brahmaputra’s avulsion node, establishing the sequence of deposition and subsequent weathering of these bodies. We then test this relative chronology by luminescence dating of the sands beneath these landform surfaces. This work provides the first absolute depositional age constraints of terrace sediments for the Middle to Late Pleistocene Brahmaputra River and upper Bengal basin. The luminescence ages are complemented by detailed compositional trends in the terrace deposits, including clay mineralogy and the degree of weathering. Together, these newly dated and carefully described morphostratigraphic units reflect eustasy-driven cycles of terrace development by way of highstand floodplain deposition and subsequent lowstand exposure and weathering, along with active tectonic deformation. Defining this Late Quaternary history of terrace development and position of the Brahmaputra River is a first step toward an integrated understanding of basin and delta evolution over multiple glacioeustatic cycles and tectonically relevant timescales. INTRODUCTION The use of an integrated systems or ‘source-to-sink’ model for linking sedimentary processes with geomorphic landforms has risen to prominence in continental margin studies over the last two decades (e.g. Nittrouer & Driscoll, 1999; Goodbred, 2003; Allen, 2008; Sømme et al., 2009; Davidson et al., 2011; Romans & Graham, 2013). This approach has established floodplains and terraces as Correspondence: Jennifer L. Pickering, 3333 Highway 6 South, R-1004A, Houston, Texas 77082, USA. E-mail: Jennifer.Pickering@ Shell.com 1 Present Address: Shell International Exploration and Production, Inc., Houston, TX 77079, USA 550 important depocentres in the midstream sediment routing system (e.g. Granet et al., 2010; Wittmann et al., 2011; Clift & Giosan, 2013). In addition, morphostratigraphic units such as terraces are, by definition (Frye & Willman, 1962), the sediment bodies that link environmental process (e.g. floodplain deposition followed by channel abandonment) to preserved stratigraphy (e.g. a buried paleosol atop fluvial sediments). Floodplain stability timescales increase with catchment size (M
etivier et al., 1999; Sømme et al., 2009), so floodplain-turned-terrace sediments in large deltas have the potential to record relatively long depositional histories. Broadly, toward an understanding of the processes governing stratigraphic preservation and landform © 2017 The Authors Basin Research © 2017 John Wiley & Sons Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists Pleistocene terraces of the Brahmaputra development on river behaviour and delta evolution, we consider the terrace record of the Brahmaputra River in the upper Bengal basin. This sediment routing system is among the world’s largest (in terms of catchment, delta, as well as fan size) and contains a prominent set of alluvial terraces in the midstream subaerial delta system. Considering the size of the catchment, these terraces should be relatively stable, and thus have been subject to multiple cycles of erosion and overprinting. The resulting composite terrace morphology leaves an incomplete and overprinted chronostratigraphic record of terrace evolution for the upper Brahmaputra delta system. Despite a general understanding of delta evolution since the Last Glacial Maximum (LGM) (Goodbred & Kuehl, 2000) and increasingly detailed reconstructions of Holocene depositional patterns (Goodbred et al., 2014; Pickering et al., 2014; Sincavage et al., in press), the dynamics of Brahmaputra deposition prior to the LGM remain unknown. Likewise the origins of outcropping Pleistocene-aged landforms within the delta, whether depositional-erosional features created by major rivers or tectonically uplifted surfaces, have not yet been resolved (see Rashid et al., 2006 and references therein). In the current study, detailed analyses of fine-grained paleofloodplain deposits are used to decipher the origin and weathering history of these terraces. Further, we expand upon a published borehole transect from the Brahmaputra avulsion belt near the apex of the fan delta in north-central Bangladesh (Pickering et al., 2014) and present the chronology of formation of pre-Holocene terraces of the Brahmaputra fluvial system, linking for the first time Brahmaputra stratigraphy to specific marine isotope stages. Study area Where the Brahmaputra River exits the Assam valley, it abruptly turns southward and flows between the Shillong Plateau and the Tista alluvial fan, where it becomes the ‘Jamuna’ River. This location marks the apex of the Brahmaputra fan delta (Fig. 1), a braided, distributive system characterised by relatively steep, sand-dominated Fig. 1. Digital elevation model (DEM) of the upper Bengal basin, showing relevant regional highlands and physiographic features. The area in the white box is shown in Fig. 2, and the Indian subcontinent with geopolitical borders is inset for reference. © 2017 The Authors Basin Research © 2017 John Wiley & Sons Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists 551 J. L. Pickering et al. Fig. 2. Above: Boxed area from Figure 1. Paths of Brahmaputra-Jamuna (B-J) and Old Brahmaputra (OB) river courses are indicated. Three terraces, the focus of this paper, are labelled (Bogra, Jamulpur, and Sherpur). Dauki thrust fault is shown, and lower elevation depression that runs parallel to the fault (shown in green to blue, south of fault) is the Dauki foredeep. Borehole locations are shown by white circles. Below: Borehole transect shown above with locations of corresponding terraces and modern BrahmaputraJamuna channel indicated. Lithology is generalised in this figure to metre-scale resolution, so thin fine-grained units, shown in blue, are not depicted. Orange dashed line indicates Holocene-Pleistocene contact determined by lithology and radiocarbon ages (Pickering et al., 2014). morphology constructed through bank migration, channel avulsion (switching), and distal overbank flooding (Wilson & Goodbred, 2015). Throughout the Holocene, the trunk stream has occupied two main pathways, the western Brahmaputra-Jamuna course and the eastern Old Brahmaputra course (Fig. 2), with some relatively minor offtakes and splays occurring along each river path (Pickering et al., 2014; Sincavage et al., in press). At the apex, the lateral extent of the Brahmaputra fan delta is limited by high topography (Fig. 1). The surrounding highlands are pre-Holocene surfaces, ranging from a few to tens of metres higher than the channel belts. The western side is confined by the Tista fan in the north and the Bogra Terrace in the south, and the eastern side is 552 confined by the Shillong Plateau. In addition, the Jamulpur and Sherpur terraces outcrop within the fan delta (Fig. 2). The Jamulpur Terrace, the northernmost extent of the much larger Madhupur Terrace, bifurcates the Brahmaputra trunk stream and the Old Brahmaputra distributary. Since Morgan & Mcintire (1959) mapped some of these terraces decades ago, most researchers accept that they were deposited during the Pleistocene; however, reliable ages have yet to be reported in publicly available literature. Indeed, only a handful of pre-Holocene ages have been reported at all, typically radiocarbon ages obtained from organic material buried beneath Holocene alluvium (Japan International Cooperation Agency, 1976; Umitsu, © 2017 The Authors Basin Research © 2017 John Wiley & Sons Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists Pleistocene terraces of the Brahmaputra 1987; Stanley & Hait, 2000; McArthur et al., 2008; Pate, 2008; Sarkar et al., 2009). terraces establish periods of aggradation during relative base level highstands, followed by long-term exposure during falling base level. Chronostratigraphic framework Utilising sediment lithology and geochemical provenance analyses of sediments from a transect comprising 41 boreholes spaced ~3 km apart, sampled at 1.5 m depth intervals up to 94 m deep. Pickering et al. (2014) recognised two principal sedimentary facies associations: the Brahmaputra Valley facies association, which is widespread across the transect, and the Dauki Foredeep facies association, which is confined to the northeastern-most margin of the transect near the Dauki fault (Fig. 2). These facies associations were subdivided into lithofacies based on grain size and unit thickness, comprising braidbelt sands, overbank muds, and basinal muds, as well as a coarse-grained, extrabasinally sourced unit (not considered herein). Radiocarbon ages obtained from buried organic material were then used to refine Holocene and Pleistocene chronostratigraphic units for each facies. Thus, the chronostratigraphic framework established by Pickering et al. (2014) includes Holocene and Pleistocene Braidbelt Sands, Holocene and Pleistocene Overbank Muds, and Holocene and Pleistocene Basinal Muds. Although the Holocene-Pleistocene stratigraphic contact was identified, and a detailed history of fluvial deposition and abandonment was interpreted for Holocene sediments, the depositional sequence of Pleistocene stratigraphy was not resolvable by the chronostratigraphic method used in that study. In an effort to establish a chronology for the Pleistocene terrace deposits, we have further analysed sediments from each of the 4 fine-grained chrono-lithofacies (referred to as facies by Pickering et al., 2014; : Table 1): the Holocene Overbank Muds (HOM), the Pleistocene Overbank Muds (POM), the Holocene Basinal Muds (HBM), and the Pleistocene Basinal Muds (PBM). The Brahmaputra Valley facies association was recognised on the basis of surface morphology and lateral distribution, resulting in morphostratigraphic units that define different fluvial terraces; here we have refined morphostratigraphic units of the POM chrono-lithofacies to include the Bogra Terrace (BT), Jamulpur Terrace (JT), and Sherpur Terrace (ST). A summary of the classification scheme used herein, with new results for each unit is presented in Table 1. In this follow up paper, we present analyses that compare the degree of weathering and paleosol development between the fine-grained terraces and foredeep sediments. The timing of burial of sediments from each morphostratigraphic unit has also been determined by measuring the luminescence signals of stimulated feldspar grains; these absolute ages, while sparse, are the first dates of Pleistocene stratigraphy in the upper Bengal basin and Brahmaputra fan delta. The resulting burial ages confirm the relative chronology of terrace formation first established by pedogenesis and weathering interpretations; taken together, the burial and exposure histories of these Depositional model The stratigraphic architecture of the Late Quaternary Brahmaputra delta in northern Bangladesh is characterised by relatively homogeneous fluvial sands (i.e. well sorted with consistent lithologic composition) that occur in packages tens of metres thick and rarely separated by thin muds deposited during overbank flooding (Pickering et al., 2014). Both coarse- and fine-grained sediments exhibit varying degrees of weathering, indicating that the history of deposition at different locations has been intermittent through time. This evidence for weathering includes horizons of oxidised, brown-tinted sands within the generally grey-coloured reduced sediment, as well as mottling indicative of oxidation-reduction reactions between the soil and groundwater, and occasionally the preservation of paleosols recognised by stiff, impermeable mud. In some of these paleosols, stiffness decreases with depth, while in other paleosols, the stiffness of the sediment is maintained down column. Recognition and interpretation of these weathering horizons in conjunction with radiocarbon dating has led to the demarcation of the Holocene-Pleistocene contact in the upper ~60 m of stratigraphy. Vertically juxtaposed Pleistocene and Holocene sequences are similar across the Bengal delta (Goodbred & Kuehl, 2000; Pate et al., 2009; Pickering et al., 2014). We therefore consider the mode of sediment deposition during the modern (i.e. the Marine Isotope Stage [MIS] 1 sea level highstand) to be generally analogous with deposition during previous highstands. Based on observations of the modern system, sand is principally deposited as channel bars in the active braidbelt, and mud is deposited on the adjacent floodplains through overbank flooding (Fig. 3a). Because the braidbelt is prone to lateral migration, sand deposition is widespread and floodplain mud preservation is infrequent due to reworking at the mobile riverbanks. Although sediment deposition is focused within and adjacent to the active braidbelt, channel avulsions lead to infilling of low-lying topography, enabling deposition across the active fan delta surface (Reitz et al., 2015). The result is a sand-dominated stratigraphic architecture with overbank mud preserved primarily as thin, localised units (Fig. 3b). Active highstand deposition in the Bengal delta reverses during falling sea level, with abandonment of active floodplains, as incision decouples the valley system from the adjacent floodplain (Umitsu, 1993; Goodbred & Kuehl, 2000; McArthur et al., 2008). Sand deposition is confined to channel bars, and floods capable of aggrading the abandoned floodplains have decreasing likelihood as incision increases the terrace relief (Gibling et al., 2005) (Fig. 3b). Lateral migration and channel avulsion is increasingly limited as the valley incises, and the resulting © 2017 The Authors Basin Research © 2017 John Wiley & Sons Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists 553 J. L. Pickering et al. Chronolithofacies Facies associations Table 1. Chrono- and morpho-stratigraphic framework for fine-grained sediments of borehole transect with lithology and sediment analyses results Morphostratigraphic units Dauki Foredeep facies association Pleistocene Overbank Muds (POM) Brahmaputra Valley facies association (fine-grained fraction) Holocene Overbank Muds (HOM) Lithology Unit Thickness Paleosol development (based on stiffness) Microtexture Mineralogy Poorly sorted coarse silt; typically, blue-gray or dark graybrown 1—5 m None to weak uniformly euhedral with relatively large crystals (up to ~50 µm) chlorite, illite, kaolinite, and smectite anhedral, except one sample directly below paleosol uniformly illite, except paleosol sample is uniformly smectite Pleistocene Bogra Terrace (PBT) Very poorly sorted coarse silt, friable; yellow-gray to yellowbrown; oxidized below 2-3 m depth 3—8 m Moderate to strong at 2—5 m depth; moderate at 9 m depth (adjacent to main terrace); very strong at 41—47 m depth Pleistocene Jamulpur Terrace (PJT) Poorly sorted coarse blue-gray mottled silt; hydromorphic below 2-3 m depth 5—11 m Moderate to strong at 2—13 m depth anhedral with few euhedral crystals; euhedral crystals have a granular habit uniformly illite 10 m Moderate at 8—18 m depth anhedral with few euhedral crystals; euhedral crystals have a granular habit uniformly illite up to 20 m None to weak mixed euhedral-anhedral; euhedral crystals are <20 µm dominantly kaolinite with some illite up to 55 m wide variability, but up to 20 m of very strong development at NE margin anhedral with few euhedral crystals have a platy habit dominantly kaolinite with illite and chlorite Pleistocene Sherpur Terrace (PST) Holocene Basinal Muds (HBM) Pleistocene Basinal Muds (PBM) Poorly sorted coarse silt of variable color; from top: yellowgray above dark gray above blue-gray above red-brown above yellow-brown; oxidized below 9 m depth Poorly sorted coarse silt; highly variable color, including yellowgray, red-brown, blue-gray, black, green, brown-gray, and tan silts Poorly sorted coarse silt; highly variable color, including gray, red-brown, blue-gray, dark gray, green-gray, and brown silts long-term exposure of the abandoned highstand floodplains facilitates weathering and soil formation to develop a distinct terrace surface. METHODS The shallow stratigraphy of the upper Brahmaputra delta plain was defined by analyses of 980 sediment samples collected by borehole drilling (Pickering et al., 2014). These analyses included particle size and bulk geochemical compositions (major oxides given by weight percent) using a Malvern 2000E laser-diffraction particle-size analyser and an Oxford Instruments MDX 1080+ X-ray fluorescence (XRF) mass spectrometer, respectively. Weathering proxies of fine-grained sediments Of the samples analysed for grain size, 180 samples were fine-grained (muds), defined by mean grain size diameters <62.5 lm. Using data collected by Pickering et al. (2014), we explore the element ratio of K/Si relative to Al/Si for the fine-grained sediments, as a proxy for chemical weathering after Lupker et al. (2013). We also note the presence of paleosol horizons in the Pleistocene stratigraphy, and 554 we compare the degree of stiffness as well as the thickness of these paleosols, which we consider to reflect development of the soil profile, an indication of the weathering history of the deposit. For this paper, we performed additional analyses on a subset of 17 of the 180 previously analysed muds (sample locations shown in Fig. 4a); this subset includes samples from each of the 4 chrono-lithofacies, as well as the 3 distinct morphostratigraphic units (i.e. terraces) from the POM facies. In order to identify textural and compositional differences across spatial and temporal boundaries, we imaged these 17 sediment samples using a Scanning Electron Microscope (SEM) and determined clay mineralogy using X-ray diffraction (XRD). Samples were initially preserved as saturated cohesive mud plugs (as they were extruded while drilling) and were stamped onto glass slides. The transferred sediment was then carbon-coated, and slides were imaged using a Tescan Vega 3 LMU SEM with a standard view field of 150 lm at 20 kV accelerating voltage. Textural differences between the Holocene-aged and Pleistocene-aged samples were observed, yielding a qualitative comparison of euhedral vs. anhedral overall crystal morphology. Mud plugs were then subsampled for XRD scans, which entailed air-drying and powdering 25 g of sediment using a mortar and pestle. Non-oriented powder scans were obtained on a Scintag © 2017 The Authors Basin Research © 2017 John Wiley & Sons Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists Pleistocene terraces of the Brahmaputra Fig. 3. Conceptual depositional model for Bengal system illustrating relationships between grain size, discharge, and depositional environments. (a) Extent of water (and sediment) transported by the modern river is shown on the land surface during low to normal flow (above) and during a monsoon flood (below); while sand and mud deposition is occurring at both times, the images illustrate sand (bar) and mud (overbank) deposition. (Moderate-resolution Imaging Spectroradiometer [MODIS] images courtesy of the Land Processes Distributed Active Archive Center [LP DAAC] managed by the NASA Earth Science Data and Information System [ESDIS] project.) (b) Generalised locations for sand, mud, and no deposition on land surface; depth dimension shows sediment packages preserved in relation to changing sea level. Weathering of exposed sediment is also indicated. Modified extensively after McArthur et al. (2008). X1 h/ h automated powder XRD with a Cu target, a Peltier-cooled solid-state detector, and a plastic sample holder. Each sample underwent a set of three analyses: (1) the air-dried sample was analysed for smectite + illite and kaolinite + chlorite clay minerals; (2) the sample was glycolated for 12 h at 60°C to expand the smectite layers for analysis; (3) the sample was heated at 550°C to break down illite and analyse chlorite. Estimates of the relative abundance of clay minerals were determined for illite, smectite, chlorite, and kaolinite by measuring the mineral peak area and dividing by the mineral intensity factor, after Biscaye (1965). Presence or absence of goethite was also noted, and the ratio of quartz to clay was estimated by the use of a zinc oxide internal standard and comparison of peak areas to a standard curve of known quartz concentrations. © 2017 The Authors Basin Research © 2017 John Wiley & Sons Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists 555 J. L. Pickering et al. Fig. 4. (a) Interpreted cross section of transect shown in Fig. 2 based on stratigraphy presented by Pickering et al. (2014); updated with distribution and relative development of paleosols (hatched lines), as well as XRD & SEM sample locations (red stars). (b) Detailed borehole stratigraphy for all boreholes at each terrace. Locations of XRD & SEM results are shown for reference to cross section (a). Locations sampled for burial age dating (sampled depths shown by blue dots) with IRSL age results are also shown. Burial age dating of coarse-grained sediments To obtain first order age constraints for terrace formation along this transect, we measured the optically stimulated luminescence (OSL) ages of the sands directly beneath the mud caps of the three terraces along the braidplain (Fig. 4a). These sands were sampled roughly 30 cm below the first occurrence of sand beneath the mud caps of each terrace (Fig. 4b). At the depth to be sampled (ranging from 8.5 to 19 m), the drill string was removed and a split spoon sampler containing an opaque PVC pipe was attached. The drill string and sampler were then lowered and the sample was obtained; under cover of a dark, opaque tarp, the sample was removed and wrapped for transport. The coarse sand fraction of quartz and 556 K-feldspar was extracted from the bulk sample using standard protocols (Wintle, 1997; Aitken, 1998). Monolayer feldspar grain aliquots (~3 mm diameter) were mounted and stimulated by an array of infrared light emitting diodes (IR-LEDs) at 870  40 nm and measured using a Risø TL/OSL DA-20 automated reader (BøtterJensen et al., 2000, 2003) with a 90Sr / 90Y beta source with a strength of 0.113 Gy s 1. The detection optics comprised a Schott BG-39 and a Corning 7–59 filter combination, allowing transmission in the blue-violet band (370–440 nm) with a peak at ~410 nm (Aitken, 1998). Because the infrared stimulated luminescence (IRSL) signal of feldspar minerals is subject to fading, a post infrared infrared stimulated luminescence (pIR-IRSL) single aliquot regeneration protocol was employed (Table 2) after Buylaert et al., 2009 and Thiel et al. (2011). The intensity © 2017 The Authors Basin Research © 2017 John Wiley & Sons Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists Pleistocene terraces of the Brahmaputra of the initial 6 s was used as the signal and the final 60 s (of a total of 150 s read out) was used for the background correction. A 50 Gy test dose was used to normalise the natural and regenerated luminescence signals. For all samples, selection criteria for the max permissible value were: 5% recuperation ratio, 10% recycling ratio, 10% test dose error and 5% measurement error. The laboratory generated dose response curve was fitted to a single saturating exponential. The dose rate was measured using a High Purity Germanium (HPGe) detector between ranges of 2.3–3.6 Gy ka 1. Details of the dose rate and De are given with the results in Table 3. To correct for fading of the pIR-IRSL signal, samples were irradiated at 200 Gy in the laboratory, and the signal decrease over time was measured following Auclair et al. (2003) and Biswas et al. (2011). The fading correction by Huntley & Lamothe (2001) was employed and the g value was in the range of 2–3% per decade for all samples. A mean value of 6 aliquots was used to correct the De. Burial age corresponding to both the corrected and uncorrected De is provided, and it is clear that use of either would not affect the interpretations (Table 3). CHEMICAL ALTERATION OF FINEGRAINED SEDIMENTS For the chronostratigraphic framework established by Pickering et al. (2014), radiocarbon ages were used to define stratal packages interpreted as channel occupations and abandonments, enabling reconstruction of the Holocene avulsion history. However, ages of organic material in the Pleistocene units were beyond the resolution of radiocarbon dating, so while it was determined that those sediments were deposited >48 ka, neither the precise ages of the Pleistocene morphostratigraphic units (terraces) nor their depositional ages relative to each other were determined. In the present study we attempt to better establish age control of those Pleistocene morphostratigraphic units. Table 2. Sequence of steps for pIR-IRSL SAR protocol modified after Thiel et al. (2011). A plot of the ratio of Lx/Tx with given dose Di and interpolation of the normalised intensity of signal from samples as received (natural) on this plot, provides De. Step Treatment 1 2 Dose, Di Preheat, 320°C, 30 s 3 4 5 6 IR Stimulation, 100 s at 50°C IR Stimulation, 150 s at 290°C Give Test Dose, DT Preheat, 320°C, 30 s 7 8 9 10 IR Stimulation, 100 s at 50°C IR Stimulation, 150 s at 290°C IR Stimulation, 300 s at 325°C Return to step 1 Observed Remove thermally unstable signal Towards this, chemical weathering for each unit was assessed by tracking the ratio of mobile K to immobile Si in sediments over time. For modern Himalayan rivers, K/Si vs. Al/Si is expressed by a linear relationship, for which the slope decreases (K is lost) as chemical weathering occurs (Lupker et al., 2013). In Fig. 5, Lupker et al.’s values for bank sediments from the modern Himalayan rivers and the Brahmaputra in Bangladesh are shown for reference, along with each of the four chrono-lithofacies (Table 1) from this study. A progressive decrease in slope occurs from modern Himalayan frontal riverbanks (0.283) to modern Brahmaputra riverbanks in Bangladesh (0.204), indicative of K depletion with time as K-feldspar and other primary K-bearing minerals such as muscovite weather to form K-poor clay mineral phases such as kaolinite. The same weathering trend is apparent from the decrease in K in buried Holocene Brahmaputra Overbank Muds (0.104) to Pleistocene Brahmaputra Overbank Muds (0.027). Likewise, the Dauki Foredeep facies shows a similar trend, for which the slope of the Pleistocene Basinal Muds is decreased relative to the Holocene Basinal Muds, indicative of chemical weathering. Sediments from the Brahmaputra Valley and Dauki Foredeep facies are plotted separately because they have different provenances, as Pickering et al. (2014) demonstrated with Sr geochemistry. That K is decreasing relative to Al in both the Brahmaputra Valley facies and the Dauki Foredeep facies is noteworthy as it illustrates weathering of Pleistocene sediments relative to Holocene sediments regardless of geomorphic provinces. PALEOSOLS, CLAY MINERALOGY, AND MICROTEXTURAL MORPHOLOGY OF FINE-GRAINED LITHOFACIES Using the chronostratigraphic framework summarised above, we describe the generalised lithology of each of the fine-grained facies from the Brahmaputra Valley and Dauki Foredeep facies associations, as well as the character of any paleosols within each facies or morphostratigraphic unit. In addition, we present new analysis results of 17 fine-grained samples, representing each facies and morphostratigraphic unit, yielding a more detailed finegrained stratigraphy of the Brahmaputra braidbelt in the upper delta plain. Analyses of clay mineralogy (Fig. 6), as well as the microtextural morphology (Fig. 7), show characteristic variations among facies; these results are given below with contextual grain size information (from Pickering et al., 2014) and summarised in Table 1. Lx Remove thermally unstable signal Brahmaputra valley facies association Holocene overbank mud facies Tx Hot Optical Wash The Holocene Overbank Mud (HOM) facies consists of buried, thinly bedded, dark-coloured silts interspersed throughout the Holocene Braidbelt Sands and interpreted © 2017 The Authors Basin Research © 2017 John Wiley & Sons Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists 557 J. L. Pickering et al. Table 3. Equivalent dose, De, and ages corrected for fading using Huntley & Lamothe (2001) Sample name pIRSL De*,† (Gy) Fading g-Value* (%/decade) BH-3 (JT) BH-5 (ST) BH-8 (BT) 488  26 371  20 341  17 2.9  1.4 2.1  1.1 2.4  1.3 Dose rate data 238 U (ppm) 2.5  0.1 2.3  0.1 3.2  0.2 232 Th (ppm) 17.5  1.4 13.2  1.1 16.3  1.3 Potassium (wt %) Dose Rate (Gy ka 1)‡,§ Age without fading correction (ka) Age after fading correction (ka)¶ 2.2  0.1 2.2  0.1 2.3  0.1 3.33  0.25 3.08  0.23 3.53  0.23 146  14 121  11 97  9 165  16 132  12 107  10 *For each sample, 6 aliquots were measured for De and fading. Stated De and g-Values are mean values of measurements of all the aliquots. The error is the Standard Error (SE) on mean. †Residual pIR-IR signal equivalent to 25 Gy was subtracted from the total pIR-IRSL De. ‡Water content for all samples was 20  5%. §Contribution from cosmic radiation was 150  15 lGy a 1. This may be marginally overestimated in view of depth but will not affect the results. ¶Fading corrections were applied after Huntley & Lamothe (2001). Fig. 5. Compositional variability of K/ Si in Himalayan system and Brahmaputra River in Bangladesh from Lupker et al. (2013), with 4 chrono-lithofacies from this study. as preserved floodplain deposition (Pickering et al., 2014). The locations of these deposits within the stratigraphy are indicated in Fig. 4a. These strata are highly deformable, and do not exhibit signs of chemical alteration due to weathering. The average clay content of this facies is ~71%, and clay mineralogy is a suite of inherited minerals that includes chlorite, illite, kaolinite, and smectite phases, with no dominant mineral trends (Fig. 6a); this mixed composition is similar to clay mineral distributions found in Holocene fluvial overbank deposits throughout the delta (Heroy et al., 2003; Tennant, 2005; Aftabuzzaman et al., 2013). Under magnification, these silts display a uniformly euhedral mineral habit with relatively large crystals (up to ~50 lm) (Fig. 7). Sample 05559 contains noticeably smaller crystals (generally <20 lm); despite the relatively small crystal size, the 558 euhedral habit and mixed mineralogy of sample 05559 are consistent with the other four HOM samples. Pleistocene overbank mud facies Unlike the HOM facies, the Pleistocene Overbank Mud (POM) facies is currently exposed at the surface, forming mud caps for each of the three local terraces (Bogra, Jamuna, and Sherpur), coarsening downward to Pleistocene Braidbelt Sands (Pickering et al., 2014). In general, POM deposits are oxidised and consist of stiffer muds than their Holocene counterparts. The POM facies is subdivided into three morphostratigraphic units informally named for the location that coincides with the exposed land surface of the unit: (1) Bogra Terrace, forming the western boundary of the © 2017 The Authors Basin Research © 2017 John Wiley & Sons Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists Pleistocene terraces of the Brahmaputra Fig. 6. X-ray diffraction (XRD) scans of non-oriented powders for 17 samples labelled for comparison with Figs. 3 and 6. (a) Samples representing the Holocene Overbank Mud (HOM) facies; (b) Samples representing the Pleistocene Overbank Mud (POM) facies; (c) Samples representing the Dauki Foredeep facies association, including the Holocene Basinal Mud (HBM) and Pleistocene Basinal Mud (PBM) facies. Sme = smectite, Ill = illite, Chl = chlorite, Kln = kaolinite, Qtz = quartz, and Gt = goethite. Brahmaputra braidbelt system, (2) Jamulpur Terrace, the interfluve that bifurcates the Brahmaputra-Jamuna and Old Brahmaputra courses, and (3) the Sherpur Terrace, in the eastern portion of the transect, a smaller interfluve that bifurcates two sub-valleys of the Old Brahmaputra course. Sedimentologic results for each of the morphostratigraphic units that make up the POM facies are presented below. Bogra terrace morphostratigraphic unit. The Pleistocene Bogra Terrace (BT) is a morphostratigraphic unit that consists of very poorly sorted, friable, coarse silts that cap brown, leached, fine to medium sands. The capping unit is roughly 3 m thick (Fig. 4b), composed of sediments oxidised to a yellow-grey or yellow-brown colour (Table 1). The shallow stratigraphy comprises a compound paleosol, or multiple paleosols vertically separated by relatively less weathered sediments (Kraus, 1999). Here, the upper unit is a thin, moderately to strongly developed paleosol at 2–5 m depth, and the lower paleosol is much thicker and very strongly developed, from 41–47 m depth. These two silty paleosols are each underlain by a package of leached fluvial sand (Pickering et al., 2014). Immediately to the east of these stacked sequences, © 2017 The Authors Basin Research © 2017 John Wiley & Sons Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists 559 J. L. Pickering et al. Fig. 7. Scanning electron microscope (SEM) images of 17 samples, grouped by facies. Morphostratigraphic unit is additionally labelled for POM facies: BT = Bogra Terrace, JT = Jamulpur Terrace, ST = Sherpur Terrace. Pie charts indicate relative amounts of chlorite, illite, kaolinite, and smectite for reference (see Fig. 5 for detailed XRD results). Horizontal view field is 150 lm for each image. a buried paleosol is recognised at 9 m depth. This paleosol is moderately developed, and it is overlain by sediments of the HOM facies (Fig. 4). The fine-grained fraction of BT sediments averages ~78% clay content, and of the five clay mineral phases for which we measured, only illite and smectite are detectable in BT muds. The only occurrence of measureable smectite was detected in a compact paleosol sample (Fig. 6b: 00809), 560 and although illite is not considered a highly altered mineral, the dominance of smectite in the paleosol is indicative of weathering (Gardiner & Miller, 2004). Likewise, despite the yellow hue of the BT muds, we did not detect goethite (Fig. 6b). BT sediments display anhedral crystal habits, for the most part, with the exception being the silt (00811) that is directly beneath the compact paleosol (00809) (Fig. 7). © 2017 The Authors Basin Research © 2017 John Wiley & Sons Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists Pleistocene terraces of the Brahmaputra Jamulpur terrace morphostratigraphic unit. The Pleistocene Jamulpur Terrace (JT) consists of poorly sorted, coarse silt capping brown fine to medium sand. The finegrained capping unit is 5–11 m thick (Fig. 4b) and composed of mottled blue-grey gleysols characterised by a completely reduced (grey) matrix with interspersed concentrated zones of oxidation, forming rings around clay nodules or root channels (Table 1). At 2–13 m depth, these mottled sediments comprise very hard strata forming a moderately thick, cumulative paleosol. JT sediments average ~70% clay content, and like the BT, these sediments are composed predominantly of illite, indicating clay mineral neoformation. Unlike the BT, JT sediments contain detectable goethite (Fig. 6b). JT sediments have a generally anhedral crystal habit with large, nearly equant grains interspersed throughout the anhedral matrix (Fig. 7). Sherpur terrace morphostratigraphic unit. The Pleistocene Sherpur Terrace (ST) consists of a relatively thick (18 m) silt cap atop reddish-brown very fine to fine sand. Colour of the silts is variable, beginning with yellow-grey at the surface, followed by a dark grey organic layer at 6 m depth. A blue-grey mottled cumulative paleosol marks the top of the Pleistocene surface and begins at 8 m depth, grading into oxidised red to yellow-brown silt at 18 m depth. This 10 m thick paleosol is harder than the overlying fine-grained sediments, but this paleosol exhibits slightly weaker development relative to paleosols of the BT and JT. Like the BT and JT sediments, ST sediments have a generally anhedral crystal habit (Fig. 7) and the clay mineral fraction (~77%) is composed predominantly of illite but contains detectable goethite (Fig. 6b). Dauki foredeep facies association Holocene basinal mud facies The upper ~20 m of strata in the Dauki Foredeep subbasin are Holocene-aged deposits (Pickering et al., 2014). Strata of the Holocene Basinal Mud (HBM) facies are very heterogeneous and include coarse, angular sands as well as finer basinal muds, which are highly deformable, poorly sorted, coarse silts of variable colour, including yellow-grey, red-brown, blue-grey, black, green, browngrey and tan deposits. Under magnification, these silts display a mix of euhedral and anhedral crystal habits, and the euhedral crystals are typically <20 lm (Fig. 7). The average clay content of HBMs is ~66%, which is dominated by kaolinite with some illite (Fig. 6c). Presence of goethite is detectable in these samples. Pleistocene basinal mud facies The Pleistocene Basinal Mud (PBM) facies consists of poorly sorted, coarse silts of various colours, similar to the HBM facies. These are very thick deposits, up to 55 m in the northernmost area of the transect. The top of the Pleistocene unit was demarcated by the presence of a thick (up to 20 m) composite paleosol that is equally or more strongly developed, relative to paleosols of the POM terraces. Under magnification, these muds have a predominantly anhedral crystal habit with few interspersed euhedral crystals, which have a platy habit (Fig. 7). The average clay content of the PBM facies is ~74%, which is, like the HBM facies, primarily kaolinite with smaller percentages of illite and chlorite (Fig. 6c). Goethite is also detectable in half of the samples. RELATIVE WEATHERING OF FINEGRAINED LITHOFACIES In Himalayan systems, smectite and kaolinite are generally considered to be pedogenic clays, suggesting chemical weathering, while feldspar, quartz, illite and chlorite are primary minerals, and goethite is indicative of leaching (Debrabant et al., 1993; Derry & France-Lanord, 1996; Colin et al., 2006). Within the Brahmaputra Valley facies association, the HOM facies has a mixed clay composition that includes chlorite, illite, kaolinite, and smectite; the POM, facies alternatively, comprises largely illite and smectite mineral phases, with some strata containing goethite. The mixed composition, and particularly the presence of illite and chlorite in the HOM facies indicates physical weathering, i.e., disintegration of larger clasts, or only early diagenesis rather than extensive chemical alteration (Segonzac, 1970; Weaver & Pollard, 1973; Gardiner & Miller, 2004). The presence of smectite in the POM facies, however, is indicative of pedogenesis. Likewise, the nearly uniform mineral composition of POM muds in general is noteworthy; we interpret this to be indicative of the intensity of weathering the sediments have undergone, i.e., the weathering process has converged, resulting in transformation toward a single mineral phase (e.g. Duchaufour, 1982; Velde, 1992). These contrasting clay mineral distributions indicate a prolonged period of development in the POM facies compared to the HOM facies, which is very much in agreement with the chronostratigraphic framework defined by Pickering et al. (2014), based on radiocarbon dated organic material from each chronostratigraphic unit. Interestingly, in the Dauki Foredeep facies association, both the HBM facies and the PBM facies are composed dominantly of kaolinite (Fig. 6). The dominance of kaolinite does not vary across the Holocene-Pleistocene contact in the Dauki Foredeep, indicating that sediments of both the HBM facies and the PBM facies have been weathered to a similar degree, despite absolute age control that would suggest otherwise (Pickering et al., 2014). The geomorphic Dauki Foredeep, i.e., the surface expression, is shaped by the Dauki thrust fault (Fig. 2) at the southern margin of the Shillong Plateau. This area is flooded during the monsoon season, and it also accommodates overflow of south-flowing rivers that drain the © 2017 The Authors Basin Research © 2017 John Wiley & Sons Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists 561 J. L. Pickering et al. southwestern plateau. These dueling sediment transport regimes generate the heterogeneous stratigraphy of coarse angular sands interspersed within fine muds found within the HBM and PBM facies. The local gradient, from the relatively high elevation of the southern Shillong Plateau to the low-lying foredeep below, facilitates drainage of Shillong-sourced rivers into the foredeep. Because exposed sediments on the plateau are interpreted to be considerably older alluvium, the finegrained sediments of the Dauki Foredeep seem to be dominantly composed of recycled alluvium that was weathered on the plateau, prior to redeposition in the foredeep. This interpretation is consistent with the dominance of kaolinite and the presence of goethite in basinal muds above and below the Holocene-Pleistocene contact. As a result of this sediment recycling, a mineralogical comparison of the two facies associations cannot be used to decipher relative ages because the Dauki Foredeep muds have inherited their advanced pedogenic mineral phases, as evidenced by the similar mineralogy of Holocene and Pleistocene Dauki Foredeep basinal muds, whereas the Brahmaputra Valley muds have undergone in situ alteration, as evidenced by the different compositions of the Holocene and Pleistocene overbank sediments. Likewise, relative ages of the Pleistocene morphostratigraphic units, i.e., the Bogra, Jamulpur, and Sherpur terraces, cannot be deciphered from compositional data or mineral phase alteration because distinctions in mineralogy and composition are not apparent. Microtextural morphology results yield a similar interpretation. In the Brahmaputra Valley, the HOM facies exhibits consistently euhedral and relatively large grains, compared to the anhedral and relatively finer grained morphology of the POM facies. Based on the mineralogical results discussed above, we propose that the more anhedral mineral habit of the POM facies is indicative of chemical alteration as opposed to physical weathering. We find two exceptions: First, within the euhedral HOM facies, sample 05559 (Fig. 7) appears to be somewhat anhedral, but upon close inspection, the sample merely has relatively small grains compared to the other HOM samples. This is perhaps due to the extreme burial depth relative to the other HOM samples (59 m compared to 2–11 m depth for the other 4 samples); indeed, organic material picked from sample 05559 confirms that the sediment was deposited in the Middle Holocene (6586 cal yr BP; Pickering et al., 2014). Second, sample 00811 of the POM facies, appears to be anomalously euhedral. However, this deposit was located directly beneath sample 00809, which is a compact paleosol composed dominantly of smectite; we suggest that the overlying smectite acted as an impermeable seal that protected the underlying sediment from precipitation and eluviation. Regardless, the differences in microtextural morphology are more apparent between the HOM and POM facies than they are in the HBM and PBM 562 facies, which are both composed of relatively small, anhedral minerals; although the HBM facies appears to have some grains with a more euhedral habit, these are not nearly as large or euhedral as grains analysed from the HOM facies. TERRACE FORMATION OF THE BRAHMAPUTRA VALLEY FACIES ASSOCIATION Although sediment composition, mineralogy, and microtextural morphology are useful for comparing differences between facies associations and time of deposition within the Brahmaputra Valley facies association, they do not yield an obvious distinction within the Pleistocene chronostratigraphic unit, i.e., the relative chronology of terrace deposition. Paleosol development within the terraces, however, is used as a proxy for land surface exposure, which represents the timing of fluvial abandonment and terrace formation. We couple this relative chronology of exposure with absolute burial ages of the sands beneath the fine-grained terrace caps. Together, these chronologies have enabled a first order interpretation of landform evolution in the upper Bengal basin. Sequence of terrace formation Paleosols within each of the Brahmaputra Valley Pleistocene landforms have distinctive thicknesses and degrees of development, relative to each other (Table 1, Fig. 4). These hard mud horizons are interpreted as floodplain paleosols formed due to subaerial exposure during relative sea level lowstands (Fig. 3b); stiff brown and blue-grey clays in particular have been recognised across the Ganges-Brahmaputra delta (Alam et al., 1997; McArthur et al., 2008; Hoque et al., 2012; Pickering et al., 2014; Williams, 2014). By synthesising paleosol development, thickness, and stratal position a relative sequence of terrace formation is proposed. Development of paleosols involves deposition of the overbank sediments, abandonment of the surface resulting from either avulsion of the channel away from the active floodplain or incision of the channel to a lower elevation, and ongoing weathering of the sediments due to exposure at the land surface. As such, the complete history of development may encompass 104 to 105 years. For simplicity, we interpret the paleosol ‘age’ to be the time of exposure, which roughly coincides with the initiation of weathering. The deep weathering profile of the Bogra Terrace comprises two stacked normally graded sequences, each capped by a paleosol. The upper paleosol is relatively thin (3 m) and moderately developed, and the lower paleosol is relatively thick (6 m) and very strongly developed. The relative thickness and development of each of these © 2017 The Authors Basin Research © 2017 John Wiley & Sons Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists Pleistocene terraces of the Brahmaputra paleosols is a reflection of the rate of sediment accumulation and the time of exposure (e.g. Kraus, 2002); we interpret the lower paleosol as a floodplain that accumulated and was exposed to weathering for a longer period of time than the upper paleosol. About 5 km east of this succession, beneath the modern floodplain of the BrahmaputraJamuna, is another paleosol at 9 m depth, which is thin (1 m) and moderately developed. The spatial relationship between these BT paleosols is shown in Fig. 4. From the geometry of the Holocene-Pleistocene contact beneath the BT, we interpret step-wise incision of the channel from the upper paleosol and the more eastern (middle) paleosol, i.e., the higher elevation paleosol exposed at the surface is older than the adjacent middle paleosol (sensu Blum et al., 2013), and the lowest elevation paleosol must be the oldest of the three. Thus the upper 50 m of BT stratigraphy has preserved at least two distinct normally graded sequences, each followed by exposure and soil development, likely with negligible erosion (Kraus, 1999). The adjacent terrace step is also normally graded, but only about 12 m of POM stratigraphy is documented there, because the depth of recovery was limited by an impenetrable gravel layer interpreted as the LGM maximum incision surface by Pickering et al. (2014). Working backward from that maximum incision event (MIS 2), the BT step then might have formed during fluctuating sea level in MIS 3, which would mean the BT surface was exposed prior to that, perhaps at the end of MIS 4 (Fig. 8). The upper normally graded unit, including the Pleistocene Braidbelt Sands, would have been deposited during MIS 5, with most of the aggradation occurring during MIS 5e. The lower normally graded unit then might have been deposited during MIS 7, forming the thick lower paleosol during the relatively extended sea level lowstand of MIS 6. On the eastern side of the Brahmaputra fan delta, the Jamulpur Terrace and the Sherpur Terrace are each characterised by cumulative paleosols, or those that formed as thin, incrementally deposited strata that underwent steady pedogenesis with minimal erosion, typical of floodplains or floodbasins distal to the mainstem (sensu Kraus, 1999). Perhaps these cumulative paleosols reflect a history of relatively short-lived or partial channel occupations in the Old Brahmaputra course, i.e., partial flow remained routed through the mainstem Brahmaputra channel; although this is speculative, the occupation timescale for the Old Brahmaputra was shorter than that of the mainstem Brahmaputra-Jamuna for the Holocene (Pickering et al., 2014). In terms of relative degrees of weathering, the JT paleosol is more strongly developed than the ST paleosol and is exposed at the surface, while the weaker ST paleosol is buried by about 6 m of less weathered silt. On the basis of paleosol development, the surface paleosols of both the BT and the JT appear to have been exposed longer than the buried ST paleosol. However, consideration of the morphologies of the JT and ST surfaces reveals similarlyshaped bifurcating interfluves (Fig. 2), suggesting that they are genetically related and were perhaps approximately coevally exposed (Blum & T€ornqvist, 2000). While the JT and ST paleosols themselves are roughly the same thickness, the total thickness of the fine-grained sediment cap of the ST is much greater. This may be explained by the relative proximity of the JT and ST to the Dauki Foredeep: The ST is closer to the foredeep and thus may be subject to a higher rate of subsidence. Modest accommodation at the ST is readily filled with overbank fine sedimentation, resulting in a thicker mud cap and a buried paleosol that is more weakly developed compared with the thinner but moderately to strongly developed paleosol mud cap of the JT. This is also consistent with the paleosol depositional model, which predicts formation of cumulative paleosols in floodbasin environments (e.g. Kraus, 1999). Taking these arguments together, we interpret the JT and ST paleosols to have similar ages of formation, with the JT paleosol potentially being slightly older. In the absence of multiple paleosols within each unit, it is difficult to assign a specific MIS to either paleosol, but the sand thickness beneath each terrace (JT: >65 m, ST: >35 m) suggests a prolonged period of sea level rise that followed a high magnitude sea level drop, capable of generating the accommodation necessary to be infilled by such a thick sand sequence. We also note here that, while the degree of upper paleosol development is similar for all three terraces, the depth to which the sand is oxidised is extremely deep (>65 m) beneath the JT (Fig. 4b). Absolute timing of sand deposition After establishing the relative timing of paleosol exposure based on paleosol development, thickness, and stratal position, we endeavoured to date the depositional timing of the sands just beneath the terrace mud caps, in order to establish the maximum length of time that the paleosols above those sands may have been exposed to weathering at the surface. We retrieved a sand sample from beneath each terrace mud cap, and while we acknowledge that the chronology would be improved by more ages, the three sand burial ages measured using pIR-IRSL do not contradict our relative age model, established by weathering proxies of the fine-grained sediments. Beneath the Bogra Terrace mud cap, sand from 8.5 m depth was buried 107  10 ka; this deposit is positioned below the uppermost BT paleosol, approximately at equal depth to the middle, adjacent BT paleosol, and above the lowermost BT paleosol. Beneath the Jamulpur Terrace mud cap, sand from 11.5 m depth was buried 165  16 ka; this deposit is positioned just below the JT surface paleosol. Beneath the Sherpur Terrace mud cap, sand from 19 m depth was buried 132  12 ka; this deposit is positioned just below the ST paleosol. These age results are reported in Table 3, shown in stratigraphic context in Fig. 4b, and shown with the global sea level curve in Fig. 8. © 2017 The Authors Basin Research © 2017 John Wiley & Sons Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists 563 J. L. Pickering et al. Since pIR-IRSL age dating for feldspar grains of fluvial sands is a relatively new methodology designed to control age underestimation that resulted from anomalous fading of the luminescence signal in IRSL (Thiel et al., 2011), the ages estimated herein should be considered against chronologies established by complementary methods. Unfortunately, the only other absolute ages of correlative strata in the Bengal basin have resulted from radiocarbon dating, which is not applicable for organic material that is >48 ka. Therefore, we proceed with our interpretation of landform evolution of these terraces, but we caution the reader to be aware that this chronology is based on firsttime depositional age estimates and a proxy-based relative sequence of exposure. These burial ages have facilitated stratigraphic unit correlations across the fan delta that provoke further research. For example, the top of the sand unit beneath the lower BT paleosol is ~30 m deeper than the correlative unit is at the JT location. This is an interesting revelation because, as mentioned briefly in the introduction, the Jamulpur Terrace is the northern extension of the Madhupur Terrace, which has been interpreted as a structurally controlled feature based on possible faulting observed along the western margin (Coleman, 1969). GPS velocities collected over the last 13 years suggest that the Madhupur Terrace is currently vertically stable (Reitz et al., 2015), but further research is needed to fully address the tectonic history of the Madhupur Terrace. Chronology of landform evolution and implications SUMMARY AND CONCLUSIONS The burial age of the sand stratum just beneath the BT surface suggests that aggradation was diminishing around MIS 5d or 5c, which is in good agreement with our proposed surface paleosol age of MIS 4 (Fig. 8). The burial age of the sand beneath the JT surface indicates diminishing aggradation around MIS 6d or 6c, leading up to the Penultimate Glacial Maximum (PGM), at MIS 6a. Weathering of the JT paleosol probably initiated during MIS 6c or 6a, and the extensive sand unit beneath the mud cap must have been incrementally deposited throughout the MIS 7 midstand (and possibly during the MIS 9 highstand as well) leading up to the MIS 6 lowstand. The burial age of the sand stratum just beneath the buried ST paleosol corresponds to the rapid sea level rise after the PGM leading up to the MIS 5e highstand. This unit is then correlative with the upper sand unit of the BT, and the buried ST paleosol age is post MIS 5e, perhaps MIS 5d. This correlation is corroborated by the similar lithologies of the upper BT sand unit and the ST sand unit, including thicknesses of both gravelly strata and oxidised sediments (Fig. 4b). In order to better constrain the history of Pleistocene deposition by the Brahmaputra River in its upper delta plain, sediments from fine-grained units of the Brahmaputra fan delta and the adjacent Dauki Foredeep were analysed, and in both facies associations weathering proxies of sediments corroborated the previously established Holocene-Pleistocene contact. We demonstrated that Dauki Foredeep sediments are recycled alluvium from the adjacent Shillong Plateau, so chemical alteration of that facies association cannot be compared to chemical alteration of the Brahmaputra Valley facies association, for the purpose of interpreting relative ages. Within the Brahmaputra Valley facies association, we were able to establish a relative sequence of terrace formation by the Brahmaputra River using paleosol development and thickness as weathering proxies. We tested this previously established relative chronology by absolute age dating of associated Brahmaputra braidbelt sands buried beneath the terrace mud caps, establishing a more detailed history of delta evolution since MIS 7 and perhaps before. The westernmost Bogra Terrace preserves two normally graded fluvial sequences, deposited during MIS7 Fig. 8. Marine isotope stages (MIS) of the last 350 000 years modified from Railsback et al. (2015) with interpreted periods of deposition and exposure based on weathering proxies of terrace mudcaps and burial ages of underlying sands. Burial ages are shown by blue errors; width of the transparent blue bar shows the error estimated for each age. Dashed lines represent periods of aggradation or incision inferred from the stratigraphy where no absolute ages exist. 564 © 2017 The Authors Basin Research © 2017 John Wiley & Sons Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists Pleistocene terraces of the Brahmaputra and 5e, with exposure after each highstand generating a paleosol. The surface paleosol has been exposed since MIS4, and step-wise incision of the Brahmaputra has also preserved what is interpreted as an MIS 3b paleosol, buried adjacent to the Bogra Terrace surface. Timing of sediment deposition of the Jamulpur Terrace and Sherpur Terrace is less constrained because only a single fluvial sequence is preserved at each location. Nonetheless, the Jamulpur Terrace is interpreted as the oldest of the three features, because the surface paleosol has been exposed since the PGM (MIS6a). The shallowly buried Sherpur Terrace paleosol, formed during MIS 5d or later overlies sands that were buried in MIS 5e, which makes the Sherpur sand unit and the upper Bogra sand unit correlative. These are the first absolute burial age dates for sediments from beneath the Pleistocene contact, measured using a method other than radiocarbon dating, which has allowed us to extend the history of landform evolution in the Brahmaputra fan delta back to the Middle Pleistocene. ACKNOWLEDGMENTS Funding for this research was provided by NSF PIRE Award #0968354. We thank A. Densmore for his thorough and insightful review, editor A. Mulch for his commitment to publish this work, and A. Borcherds for her assistance throughout the publishing process. JLP acknowledges Z. Mahmood and S. Hossain for assistance with fieldwork, C. Lukehart for providing access to the xray diffractometer, and BanglaPIRE collaborators near and far: R. Sincavage, S.H. Akhter, C. Grall, J.L. Grimaud, L. Seeber, M. Steckler, and C. Wilson. AKS thanks the Departments of Science and Technology and Atomic Energy. HMR was supported by the JC Bose fellowship granted to AKS. CONFLICT OF INTEREST No conflict of interest declared. REFERENCES AFTABUZZAMAN, M., KABIR, S., ISLAM, M.K. & ALAM, M.S. (2013) Clay mineralogy of the pleistocene soil horizon in barind tract, Bangladesh. J. Geol. Soc. India, 81(5), 677– 684. AITKEN, M.J. (1998) Introduction to optical dating: the dating of Quaternary sediments by the use of photon-stimulated luminescence. Clarendon Press, Gloucestershire, UK. ALAM, M.S., KEPPENS, E. & PAEPET, R. (1997) The use of oxygen and carbon isotope composition of pedogenic carbonates from Pleistocene palaeosols in NW Bangladesh, as palaeoclimatic indicators. Quatern. Sci. Rev., 16(2), 161–168. ALLEN, P.A. (2008) From landscapes into geological history. Nature, 451(7176), 274–276. AUCLAIR, M., LAMOTHE, M. & HUOT, S. (2003) Measurement of anomalous fading for feldspar IRSL using SAR. Radiat. Meas., 37(4), 487–492. BISCAYE, P.E. (1965) Mineralogy and sedimentation of recent deep-sea clay in the Atlantic Ocean and adjacent seas and oceans. Geol. Soc. Am. Bull., 76(7), 803–832. BISWAS, R.H., MORTHEKAI, P., GARTIA, R.K., CHAWLA, S. & SINGHVI, A.K. (2011) Thermoluminescence of the meteorite interior: a possible tool for the estimation of cosmic ray exposure ages. Earth Planet. Sci. Lett., 304(1), 36–44. € BLUM, M.D. & TORNQVIST , T.E. (2000) Fluvial responses to climate and sea-level change: a review and look forward. Sedimentology, 47(s1), 2–48. BLUM, M., MARTIN, J., MILLIKEN, K. & GARVIN, M. (2013) Paleovalley systems: insights from Quaternary analogs and experiments. Earth Sci. Rev., 116, 128–169. BØTTER-JENSEN, L., BULUR, E., DULLER, G.A.T. & MURRAY, A.S. (2000) Advances in luminescence instrument systems. Radiat. Meas., 32(5), 523–528. BØTTER-JENSEN, L., ANDERSEN, C.E., DULLER, G.A.T. & MURRAY, A.S. (2003) Developments in radiation, stimulation and observation facilities in luminescence measurements. Radiat. Meas., 37(4), 535–541. BUYLAERT, J.P., MURRAY, A.S., THOMSEN, K.J. & JAIN, M. (2009) Testing the potential of an elevated temperature IRSL signal from K-feldspar. Radiat. Meas., 44(5), 560–565. CLIFT, P.D. & GIOSAN, L. (2013) Sediment fluxes and buffering in the post-glacial Indus Basin. Basin Res., 26(3), 369–386. COLEMAN, J. (1969) Brahmaputra River: Channel processes and sedimentation. Sed. Geol., 3, 129–239. COLIN, C., TURPIN, L., BLAMART, D., FRANK, N., KISSEL, C. & DUCHAMP, S. (2006) Evolution of weathering patterns in the Indo-Burman Ranges over the last 280 kyr: effects of sediment provenance on 87 Sr/86 Sr ratios tracer. Geochem. Geophys. Geosyst., 7, Q03007. DAVIDSON, S.K., LELEU, S. & NORTH, C.P. (Eds.). (2011) From river to rock record: The preservation of fluvial sediments and their subsequent interpretation, pp. 327–355. SEPM (Society for Sedimentary Geology), Tulsa, OK, USA. DEBRABANT, P., FAGEL, N., CHAMLEY, H., BOUT, V. & CAULET, J.P. (1993) Neogene to Quaternary clay mineral fluxes in the Central Indian basin. Palaeogeogr. Palaeoclimatol. Palaeoecol., 103(3), 117–131. DERRY, L.A. & FRANCE-LANORD, C. (1996) Neogene Himalayan weathering history and river 87Sr/86Sr: impact on the marine Sr record. Earth Planet. Sci. Lett., 142(1), 59–74. DUCHAUFOUR, R. (1982) Pedology: Pedogenesis and classification. Springer Science & Business Media, London, UK. FRYE, J.C. & WILLMAN, H.B. (1962) Stratigraphic commission: note 27–morphostratigraphic units in pleistocene stratigraphy. AAPG Bull., 46(1), 112–113. GARDINER, D.T. & MILLER, R.W. (2004) Soils in our Environment. 10th edn. Prentice Hall, Upper Saddle River, NJ, USA. GIBLING, M.R., TANDON, S.K., SINHA, R. & JAIN, M. (2005) Discontinuity-bounded alluvial sequences of the southern Gangetic Plains, India: aggradation and degradation in response to monsoonal strength. J. Sediment. Res., 75(3), 369–385. GOODBRED, S.L. (2003) Response of the Ganges dispersal system to climate change: a source-to-sink view since the last interstade. Sed. Geol., 162(1), 83–104. GOODBRED, S.L. & KUEHL, S.A. (2000) The significance of large sediment supply, active tectonism, and eustasy on margin sequence development: late Quaternary stratigraphy and © 2017 The Authors Basin Research © 2017 John Wiley & Sons Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists 565 J. L. Pickering et al. evolution of the Ganges-Brahmaputra delta. Sed. Geol., 133 (3), 227–248. GOODBRED, S.L., YOUNGS, P.M., ULLAH, M.S. & PATE, R.D. (2014) Piecing together holocene stratigraphy of the gangesbrahmaputra-meghna river delta using Sr sediment geochemistry: reconstructions of river-channel behavior and delta evolution. Geol. Soc. Am. Bull., 126(11/12), 1495–1510. GRANET, M., CHABAUX, F., STILLE, P., DOSSETO, A., FRANCELANORD, C. & BLAES, E. (2010) U-series disequilibria in suspended river sediments and implication for sediment transfer time in alluvial plains: the case of the Himalayan rivers. Geochim. Cosmochim. Acta, 74(10), 2851–2865. HEROY, D.C., KUEHL, S.A. & GOODBRED, S.L. (2003) Mineralogy of the ganges and Brahmaputra Rivers: implications for river switching and Late Quaternary climate change. Sed. Geol., 155(3), 343–359. HOQUE, M.A., MCARTHUR, J.M. & SIKDAR, P.K. (2012) The palaeosol model of arsenic pollution of groundwater tested along a 32 km traverse across West Bengal, India. Sci/ Total Environ., 431, 157–165. HUNTLEY, D.J. & LAMOTHE, M. (2001) Ubiquity of anomalous fading in K-feldspars and the measurement and correction for it in optical dating. Can. J. Earth Sci., 38(7), 1093–1106. Japan International Cooperation Agency (1976) Vol. 6: Geology and stone material, Part 1: Geology. In People’s Republic of Bangladesh Jamuna River Bridge Construction Project: Feasibility study report (p. 140). Japan International Cooperation Agency. KRAUS, M.J. (1999) Paleosols in clastic sedimentary rocks: their geologic applications. Earth Sci. Rev., 47(1), 41–70. KRAUS, M.J. (2002) Basin-scale changes in Floodplain Paleosols: implications for interpreting Alluvial Architecture. J. Sediment. Res., 72(4), 500–509. LUPKER, M., FRANCE-LANORD, C., GALY, V., LAVE
, J. & KUDRASS, H.R. (2013) Increasing chemical weathering in the Himalayan system since the Last Glacial Maximum. Earth Planet. Sci. Lett., 365, 243–252. MCARTHUR, J.M., RAVENSCROFT, P., BANERJEE, D.M., MILSOM, J., HUDSON-EDWARDS, K.A., SENGUPTA, S., BRISTOW, C., SARKAR, A., TONKIN, S. & PUROHIT, R. (2008) How paleosols influence groundwater flow and arsenic pollution: a model from the Bengal Basin and its worldwide implication. Water Resour. Res., 44, W11411.
METIVIER , F., GAUDEMER, Y., TAPPONNIER, P. & KLEIN, M. (1999) Mass accumulation rates in Asia during the Cenozoic. Geophys. J. Int., 137(2), 280–318. MORGAN, J.P. & MCINTIRE, W.I.G. (1959) Quaternary geology of the Bengal Basin, East Pakistan and India. Geol. Soc. Am. Bull., 70 (3), 319–342. NITTROUER, C. & DRISCOLL, N. (1999). Source to Sink. MARGINS Newsletter 3, 2–3. PATE, R.D. (2008) Multiple-proxy records of delta evolution and dispersal system behavior: fluvial and coastal borehole evidence from the Bengal Basin, Bangladesh (Masters thesis). Retrieved from Vanderbilt University ETD Repository. PATE, R.D., GOODBRED JR, S.L. & KHAN, S.R. (2009) Delta double-stack: Juxtaposed Holocene and pleistocene sequences from the Bengal basin, Bangladesh. Sediment. Rec., 7 (3), 4–9. PICKERING, J.L., GOODBRED, S.L., REITZ, M.D., HARTZOG, T.R., MONDAL, D.R. & HOSSAIN, M.S. (2014) Late Quaternary sedimentary record and Holocene channel avulsions of the Jamuna and Old Brahmaputra River valleys in the upper Bengal delta plain. Geomorphology, 227, 123–136. 566 RAILSBACK, L.B., GIBBARD, P.L., HEAD, M.J., VOARINTSOA, N.R.G. & TOUCANNE, S. (2015) An optimized scheme of lettered marine isotope substages for the last 1.0 million years, and the climatostratigraphic nature of isotope stages and substages. Quatern. Sci. Rev., 111, 94–106. RASHID, T., MONSUR, M.H. & SUZUKI, S. (2006) A review on the Quaternary characteristics of Pleistocene tracts of Bangladesh. Okayama University Earth Sci. Rep., 13(1), 1–13. REITZ, M.D., PICKERING, J.L., GOODBRED, S.L., PAOLA, C., STECKLER, M.S., SEEBER, L. & AKHTER, S.H. (2015) Effects of tectonic deformation and sea level on river path selection: theory and application to the Ganges-Brahmaputra-Meghna River Delta. J. Geophys. Res., series F, 120(4), 671–689. ROMANS, B.W. & GRAHAM, S.A. (2013) A deep-time perspective of Land-Ocean linkages in the sedimentary record. Annu. Rev. Mar. Sci., 5(1), 69–94. SARKAR, A., SENGUPTA, S., MCARTHUR, J.M., RAVENSCROFT, P., BERA, M.K., BHUSHAN, R., SAMANTA, A. & GRAWAL, S. (2009) Evolution of Ganges-Brahmaputra western delta plain: clues from sedimentology and carbon isotopes. Quatern. Sci. Rev., 28, 2564–2581. SEGONZAC, G.D. (1970) The transformation of clay minerals during diagenesis and low-grade metamorphism: a review. Sedimentology, 15(3–4), 281–346. SINCAVAGE, R., GOODBRED, S. & PICKERING, J. (In press) Holocene Brahmaputra River path selection and variable sediment bypass as indicators of fluctuating hydrologic and climate conditions in Sylhet Basin, Bangladesh. Basin Res. SØMME, T.O., HELLAND-HANSEN, W., MARTINSEN, O.J. & THURMOND, J.B. (2009) Relationships between morphological and sedimentological parameters in source-to-sink systems: a basis for predicting semi-quantitative characteristics in subsurface systems. Basin Res., 21(4), 361–387. STANLEY, D.J. & HAIT, A.K. (2000) Holocene depositional patterns, neotectonics, and Sundarban mangroves in the western Ganges-Brahmaputra delta. J. Coastal Res., 16(1), 26–39. TENNANT, E. (2005) Late Quaternary climage and provenance signals of the Ganges-Brahmaputra Delta Plain, Bangladesh (Honors thesis). William and Mary, Williamsburg, VA, USA. THIEL, C., BUYLAERT, J.P., MURRAY, A., TERHORST, B., HOFER, I., TSUKAMOTO, S. & FRECHEN, M. (2011) Luminescence dating of the Stratzing loess profile (Austria)–Testing the potential of an elevated temperature post-IR IRSL protocol. Quatern. Int., 234(1), 23–31. UMITSU, M. (1987) Late Quaternary sedimentary environment and landform evolution in the Bengal lowland. Geogr. Rev. Japan, 60(2), 164–178. UMITSU, M. (1993) Late Quaternary sedimentary environments and landforms in the Ganges Delta. Sed. Geol., 83(3–4), 177–186. VELDE, B. (1992) Introduction to Clay Minerals: Chemistry, Origins, Uses and Envronmental Significance. 1st edn. Chapman & Hall, London. WEAVER, C.E. & POLLARD, L.D. (1973) The chemistry of clay minerals. Dev. Sedimentol., 15, 213. WILLIAMS, L.A. (2014) Late Quaternary stratigraphy and infilling of the Meghna River valley along the tectonically active eastern margin of the Ganges-Brahmaputra-Meghna delta (Masters thesis). Retrieved from Vanderbilt University ETD Repository. WILSON, C.A. & GOODBRED, S.L. (2015) Construction and maintenance of the Ganges-Brahmaputra-Meghna Delta: Linking © 2017 The Authors Basin Research © 2017 John Wiley & Sons Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists Pleistocene terraces of the Brahmaputra Process, Morphology, and Stratigraphy. Ann. Rev. Mar. Sci., 7, 67–88. WINTLE, A.G. (1997) Luminescence dating: laboratory procedures and protocols. Radiat. Meas., 27(5), 769–817. WITTMANN, H., von BLANCKENBURG, F., MAURICE, L., GUYOT, J.L., FILIZOLA, N. & KUBIK, P.W. (2011) Sediment production and delivery in the Amazon River basin quantified by in situ-produced cosmogenic nuclides and recent river loads. Geol. Soc. Am. Bull., 123(5/6), B30317–1. Manuscript received 11 April 2016; In revised form 10 November 2016; Manuscript accepted 6 February 2017. © 2017 The Authors Basin Research © 2017 John Wiley & Sons Ltd, European Association of Geoscientists & Engineers and International Association of Sedimentologists 567