depositional environment and osl chronology of the homeb silt deposits, kuiseb river, namibia

(2006) 478–491www.elsevier.com/locate/yqres

Quaternary Research 65

Depositional environment and OSL chronology of the Homeb silt deposits,Kuiseb River, Namibia

Pradeep Srivastava a,⁎, George A. Brook a, Eugene Marais b, P. Morthekai c, Ashok K. Singhvi c

a Department of Geography, University of Georgia, Athens, GA 30602, USAb National Museum of Namibia, P.O. Box 1203, Windhoek, Namibia

c Physical Research Laboratory, Navrangpura, Ahmedabad 380009, India

Received 21 April 2005

Abstract

Previous studies suggest that the Homeb silts of the Kuiseb valley, Namibia (i) accumulated in a dune-dammed lake, (ii) are end-point deposits,(iii) represent an aggrading river bed, and (iv) are slackwater deposits. Thus, they have been used alternatively as evidence of past drier conditionsor past wetter conditions. Lithostratigraphic analysis of two sediment sequences at Homeb indicates sedimentation by aggradation of the KuisebRiver triggered by a transition from an arid to humid climate. OSL ages for the sequences were obtained by the SAR protocol on aliquots of 9.6-mm and 4.0-mm diameter and on single grains. Four-millimeter aliquot minimum ages closely approximate the single-grain minimum ages andare younger than 9.6-mm aliquot minimum and central ages. Based on these results, the small-aliquot (4-mm) approach appears to provide agescomparable to those obtained by the more laborious and time-consuming single-grain method. Minimum ages indicate rapid deposition of theHomeb Silts in at least two episodes centered at ∼15 ka and ∼6 ka during climate transitions from arid to humid. Flash floods eroded the valleyfills during slightly more arid conditions.© 2006 University of Washington. All rights reserved.

Keywords: Namib Desert; Homeb Silt; Sedimentary environment; Luminescence dating

Introduction

The Homeb Silt Formation occurs in the canyon and valleysections of the Kuiseb River of Namibia, which forms, and tosome extent determines, the northern boundary of the NamibSand Sea (Ward, 1987). It is a regionally significant lateQuaternary sedimentary unit of controversial origin andtherefore of controversial paleoclimatic and paleohydrologicsignificance. It has been extensively studied for more than threedecades by both Namibian and international scientists becausethe type locality at Homeb is easily accessible from theGobabeb Research Station, now administered by the DesertResearch Foundation of Namibia. Research on the Homeb siltshas also influenced interpretation of relict fluvial sediments inthe valleys of many west-flowing rivers, such as the Khumib

⁎ Corresponding author. Wadia institute of Himalayan Geology, 33 GMSRoad, Dehradun-248001, India.

E-mail address: [email protected] (P. Srivastava).

0033-5894/$ - see front matter © 2006 University of Washington. All rights reservdoi:10.1016/j.yqres.2006.01.010

and Hoarusib draining to the Skeleton Coast (e.g., Eitel andZöller, 1996; Eitel et al., 2001; Srivastava et al., 2004, 2005).Because they are the most thoroughly researched of the relictfluvial sediments in Namibia, additional information on theconditions that led to the deposition of the Homeb silts, togetherwith a reliable chronology, would greatly facilitate interpreta-tion of other fluvial deposits in the region and help to establish areliable chronology for wet and dry climates in southwest Africaduring the Quaternary.

The Kuiseb is a westerly flowing ephemeral river in southernNamibia. It is 420 km long and has a catchment of 15,500 km2

(Jacobson et al., 1995). The headwaters of the Kuiseb are in thecentral Namibian Khomas Highlands reaching 2280 masl. Thisarea has a mean annual rainfall of ∼335 mm but only 5% of thecatchment receives more than 300 mm/yr and only 52% morethan 100 mm/yr (Jacobson et al., 1995). In its lower reaches nearthe coast, the river crosses the central Namib Desert whereannual rainfall is <20 mm. The Kuiseb drainage system can bedivided into four sections on the basis of catchment

ed.

mailto:[email protected]

http://dx.doi.org/10.1016/j.yqres.2006.01.010

479P. Srivastava et al. / Quaternary Research 65 (2006) 478–491

characteristics and valley morphology. There is (i) a headwatersection in the Khomas Highlands plateau where incision hasbeen minimal; (ii) a canyon section between the KhomasHighlands and Gobabeb superimposed into the calcrete-veneered Namib unconformity surface; (iii) a valley section,between Gobabeb and a few kilometers downstream ofSwartbank, where the river is confined by rocky banks withina clearly defined shallow (<10 m deep) valley; and (iv) a deltasection where the river braids into numerous undefined channelsthat spread over the coastal plain (Fig. 1; Marker, 1977).

Cosmogenic isotope dating indicates an episode of acceler-ated denudation in the central Namib Desert, with incision ofdeep canyons beginning around 2.8 Ma as a result ofincreasingly wet conditions in the region in response to Plio–Pleistocene global cooling (Van der Wateren and Dunai, 2001).Exposure ages of river-cut surfaces in the Kuiseb canyon showthat the phase of canyon cutting terminated or slowed downsignificantly between 1.3 and 0.4 Ma.

The type locality of the Homeb Silt Formation is at Homeb,and deposits in this general area have been studied intensively(e.g., Marker, 1977; Ollier, 1977; Marker and Müller, 1978;Rust and Wienke, 1980; Vogel, 1982; Ward, 1987; Smith et al.,

Figure 1. (A) Map showing the location of the Kuiseb catchment. Rainfall isohyetsDrainage basins are: (1) Khumib, (2) Hoarusib, (3) Hoanib, (4) Uniab, (5) Koigab,Tsauchab. (B) Map of the Kuiseb River basin showing the location of Homeb. (C) Prfrom a DEM based on U.S. Geological Survey GTOPO30 and topographic map ddivisions and the distribution of the Homeb silt deposits. The dotted line connecting thad a much steeper graded profile than today (A and B after Jacobson et al., 1995;

1993; Heine and Heine, 2002), particularly in regard tosediment characteristics, mode of deposition, and age. Thesilts have been interpreted as: (i) dune-dammed lake sedimentsindicating arid conditions (Goudie, 1972; Rust and Wienke,1980); (ii) river end-point deposits indicating arid conditions(Marker and Müller, 1978); (iii) floodplain deposits of anaggrading river indicating semiarid conditions (Ward, 1987;Smith et al., 1993 and references therein); and (iv) river floodslackwater deposits indicating wetter conditions and intenseprecipitation events in the headwaters (Heine and Heine, 2002).

In the present study, the lithofacies of the Homeb silts werereexamined and the sedimentological processes that resulted intheir deposition reassessed. Our analysis adds to reports byWard (1987) and Smith et al. (1993). We also provide newchronologic data including a comparison of the precision of thesingle-aliquot and single-grain OSL methods for dating fluvialsediments.

Study area and methodology

The Homeb Silt Formation is mostly preserved at thedownstream end of the canyon section of the Kuiseb as isolated

(mm) are shown in broken and continuous lines (after Jacobson et al., 1995).(6) Huab, (7) Ugab, (8) Omaruru, (9) Swakop, (10) Kuiseb, (11) Tsondab, (12)ofile along the thalweg of the Kuiseb River and the Namib unconformity surfaceata. (D) Partial longitudinal profile of the Kuiseb River showing geomorphiche upper surfaces of the silt remnants indicates that after silt deposition, the riverC after Van der Wateren and Dunai, 2001; D after Ward, 1987).

480 P. Srivastava et al. / Quaternary Research 65 (2006) 478–491

castle-shaped outcrops in the main channel and as terraced flat-topped outcrops in tributaries joining the river from the gravelplains (Smith et al., 1993). Remnants occur along the Kuisebover a distance of ∼62 km (Ward, 1987). We examined thesedimentary architecture and age of the silts at Homeb, the typelocality where most previous studies were conducted (Fig. 1).Here, the sediments are preserved as a series of terraces, where asingle exposure can be up to 25 m thick and up to 45 m abovethe present river bed.

After careful reconnaissance of the area, two thick sedimentsequences were studied in detail (Fig. 2). Section 1 is in aterraced deposit in the northern tributary channel at Homeb.Section 2, at a much lower elevation, is a castle-shaped remnantfarther east in a protected area of the main Kuiseb channel. Thelithofacies composition of each section was analyzed andsamples were collected in opaque pipes for luminescence dating.

Optically stimulated luminescence (OSL) dating on multiplegrains was carried out in the University of Georgia Lumines-cence Dating Laboratory under controlled red-light conditions.Five centimeters of sediment was removed from each end of thesample pipes for dose rate estimation. Luminescence measure-ments were made on the central section of the sedimentcylinder, which is least likely to have been exposed to sunlightduring sampling. Samples were washed with water, treated with10% HCl and 30% H2O2 to remove carbonates and organicmaterial, and then sieved to obtain the 120–150 μm particle sizefraction. Quartz and feldspar grains were separated by densityusing Na–Polytungstate (ρ = 2.58 g/cm3). The quartz fractionwas etched with 40% HF for 80 min, followed by 12N HCl for40 min to remove the alpha skin and any residual feldspar. Light

Figure 2. Map (A) and diagrammatic section (B) showing the Kuiseb and itstributaries in the vicinity of Homeb, and the locations of sediment sections1 and 2.

stimulation on quartz mineral extracts was carried out with aRISØ array of blue LEDs emitted at 470 nm. Detection opticscomprised Hoya 2 × U340 and Schott BG-39 filters coupled toan EMI 9635 QA photomultipier tube. Measurements weremade using a RISØ TL-DA-15 Reader. β-radiation was appliedusing a 25 mCi 90Sr/90Y built-in source. U and Th for annualdose estimation were measured on a thick source Daybreakalpha counting system. K was measured at the XRALlaboratory, Toronto, Canada by ICP-MS using the fusiontechnique for total K extraction. Gamma contribution fromcosmic rays was calculated following Prescott and Stephan(1982). OSL measurements on single grains were made at thePhysical Research Laboratory, Ahmedabad, India, using a RISØTL/OSL-15 Single Grain Reader with a NdYAG green laseremitting at 532 nm as the grain size was less than 150 μm.Although care was taken to insert a single grain into each slot inthe disc, it is likely that an occasional slot contained two grains.Equivalent dose estimation was based on the improved SingleAliquot Regeneration (SAR) protocol (Murray and Wintle,2003), with blue light illumination for 100 s at 280°C at the endof every cycle of measurement after a preheat of 260°C for 10 sand a cut heat of 240°C at a heating rate of 2°C/s. Green laserstimulated luminescence measurements were carried out at125°C for 1 s. The background signal was estimated using thelast 0.4 s of the shine-down curve and the intensity of the first0.2 s was used as the signal. Infrequently, grains exhibitedanomalously slow shine-down curves, indicating contaminationby feldspars. These grains were excluded from the analysis.

Paleodose measurements using multiple grains were madeon single aliquots of 9.6 mm and 4.0 mm diameter, using thesingle-aliquot regeneration (SAR) protocol of Murray andWintle (2000). A five-point measurement scheme was used withthree dose points to bracket the paleodose, a fourth zero doseand a fifth repeat-dose point. The repeat dose was measured tocorrect for sample sensitivity and to ensure that the procedurewas working correctly. Samples were preheated to 220°C for60 s followed by OSL readout at 125°C for 100 s. For allaliquots, the recycling ratio between the first and the fifth pointranged within 0.95–1.05. Data were analyzed using theANALYST program of Duller (1999). This approach producedthree age data sets that could be compared, particularly in regardto the accuracy of the multiple-grain (single-aliquot) approachcompared to the single-grain method.

The extent of bleaching of the sediments was assessed bytwo different methods outlined by Bourke et al. (2003). The firstutilizes the relationship between the natural luminescenceintensity and the paleodose of the individual aliquots.Incomplete bleaching results in higher luminescence and aproportionally higher paleodose with considerable variationfrom aliquot to aliquot (Colls et al., 2001). By contrast,paleodose measurements on aliquots from well-bleachedsediments are more consistent and do not correlate as wellwith natural luminescence. Thus higher R2 values (Pearson'scorrelation coefficient) between aliquot paleodose and naturalintensity may indicate only partial bleaching of the sample. Thesecond method of assessing the degree of bleaching assumesthat the range of aliquot paleodose values is a measure of

Table 1Results of bleaching studies

Sampleno

Following Colls et al. (2001) Following Clarke (1996)

R-square value Well bleached? Sn Well bleached?

HS-1 0.007 Yes 0.15 NoHS-3 0.01 Yes 0.07 YesHS-4 0.81 No 0.15 NoHS-6 −0.20 Yes 0.07 YesHS-7 0.44 No 0.18 NoHS-8 0.004 Yes 0.06 YesHS-9 0.010 Yes 0.08 YesHs-10 0.13 Yes 0.09 Yes

In the paleodose spread method of Colls et al. (2001), R2 is the correlationcoefficient between the paleodose and the natural intensity of individualaliquots. In the method of Clarke (1996), a coefficient of variation Sn > 0.1indicates poor bleaching. Three of the eight samples (HS-1, HS-4, and HS-7)show poor bleaching according to Clarke (1996) and two (HS-4 and HS-7)according to Colls et al. (2001).

Figure 3. Basal sediments at Section 1. (A) Well-sorted medium sand at thebottom of the sequence. (B) Parallel laminated silt. (C) Lensoidal fine sand. Thelensoidal nature of the unit is also shown in Figures 4 and 6. The scale in thephotograph is 80 cm long.


sample bleaching history (Clarke, 1996). The standarddeviation of the paleodose data set, divided by the meanpaleodose, produces a coefficient of variation (Sn). Sampleswith Sn < 0.1 are considered to be well bleached. In the samplesexamined in this study Sn varied from 0.06 to 0.18 (Table 1).

Sedimentary characteristics of the Homeb silts

Facies identification was based on the basic sedimentarystructure, degree of bioturbation, grain size, and geometry of thevarious sedimentary units. Lithofacies were identified andclassified into sedimentary sequences based on vertical andlateral associations. This detailed mapping enabled us todevelop a clear picture of the depositional environment.

Individual lithofacies

Five principal lithofacies were identified in the Homeb area:

Well-sorted medium sandA 0.5- to 2.0-m thick, reddish, moderately bioturbated,

medium sand lithofacies represents dune activity within asequence dominated by fluvial processes. In some locations, theunit exhibits parallel laminations, ripple cross-laminations, andsub-rounded to angular pebbles. There is an upward transfor-mation into gray fine sand (Fig. 3A). This unit indicates fluvialmodification of dune sand during floods and the establishmentof a river-channel environment.

Parallel-laminated siltThis unit consists of yellow, coarse to fine micaceous silt.

It is 0.5–3.5 m thick and has sharp basal and upper contacts(Fig. 3B). Individual laminae are 0.2–1 cm thick. The unit ismoderately bioturbated by roots and animal burrows (Smith etal., 1993). Individual laminae are coated with reddish ironoxides and desiccation cracks 0.5–5 cm deep are common.There are also lenses of gray rippled fine sand and reddishwell sorted sand a few cm thick within the unit, and gypsumcrystal aggregates occur at some locations (Smith et al.,

1993). This lithofacies is not associated with major sandbodies and also did not have a cyclic association with any ofthe other lithofacies.

The elements of this lithofacies indicate deposition in afloodplain or lacustrine environment (Miall, 1996). Parallellaminations and the occurrence of rippled fine sand lensesindicate deposition by sheet-like flows, where each laminapossibly represents an individual event. Iron coatings anddesiccation cracks in individual lamina indicate a break insedimentation. The sharp contacts between the upper and thelower units indicate that this unit is not genetically related to theoverlying and underlying units. Scattered horizons of bioturba-tion suggest short breaks in flood activity, as there is noevidence of pedogenic activity. Gypsum crystals may haveformed in pools left by exceptional floods. Crystallization mayhave been contemporaneous with clastic sedimentation nearby.

Lensoidal fine sandThis lithofacies consists of 0.2–1.5 m of micaceous, gray,

fine sand (Figs. 3C and 4A). Smaller units occur in thin lenseswhere the individual units are composed of suites ofsedimentary structures. Parallel laminations, climbing ripplesand small-scale ripple laminations, with no obvious cyclicity,are common sedimentary structures. The tops of the unitscommonly exhibit a high degree of bioturbation (Fig. 4C).These units are commonly overlain and underlain by parallellaminated silt or bioturbated clayey silt. Units thicker than0.5 cm are made up entirely of small-scale ripples and areoverlain by bioturbated clayey silt units (Figs. 4A, B).


Smaller units represent deposition in small runnels activatedon floodplains during flood events, and the larger, ripple-laminated units represent deposition in the major channels of anestablished floodplain. The fine grain size indicates that floodvelocities were likely <1 m/s (Miall, 1996).

Bioturbated clayey siltThis brownish-yellow unit is a highly bioturbated, clayey silt

0.2–1.5 m thick (Fig. 4B). Its association with sand bodies is themain diagnostic feature of this lithofacies. Bioturbation ismainly due to plant root and animal activity (Smith et al., 1993).In places there are crude laminations and lenses of rippled finesand 10–25 cm thick.

The characteristics of this lithofacies indicate deposition inthe vicinity of channels in a floodplain environment withenough moisture to support riparian vegetation and associatedanimal life that led to extensive bioturbation.

Angular gravelIn some locations there are 0.2–0.5 m thick, angular gravel

units with crude bedding structures and silt lenses (Fig. 6).Gravel size varies from 1 to 5 cm. The units are lensoidal, withsharp upper and lower contacts, and they occur sporadicallythrough the sediment sequence. They appear to be brecciasdeposited by sheet wash during breaks in regular flood activityand may record local desert rainfall events (Smith et al., 1993).

Lithofacies associations

Based on vertical and lateral lithofacies architecture, theHomeb silt deposits were classified into two main lithofaciesassociations.

Figure 4. Lithofacies composition of the Homeb silts. (A) Upper part of the sedimenoverlain by bioturbated clayey silt. (B) Bioturbated clayey silt capping sandy units. (C40, and 12 cm long, respectively. Inset in Figure 5 shows the location of panel A an

Channel facies associationThis forms a 0.5–3 m thick sequence composed of lensoidal

fine sand facies overlain by bioturbated clayey silts. The finesand units internally show ripple cross-laminations and parallellaminations, commonly with two channel events beingseparated by bioturbated horizons. This indicates depositionin smaller channels carrying fine sand with flow velocities<1 m/s. The reworked dune sand near the base of the sequenceindicates a period of aeolian sand deposition on the riverbed at atime of reduced river flow. This association forms the upper partof Section 1 (Figs. 5, 7).

Floodplain facies associationThis consists of parallel laminated silt, bioturbated clayey

silt, and thin lensoidal fine sand. The lensoidal fine sand unitsshow individually variable ripple cross-laminations, parallellaminations, and climbing ripples. Smith et al. (1993)determined that the flow direction in the vicinity of Section 1was into the tributary from the main river channel and thusreflected overbank flow during flood events. This associationrepresents floodplain aggradation with fine sand units, cappedby bioturbated clayey silt representing shallow channelsdeveloped on the floodplain. The parallel laminated silt wasdeposited by sheet flows on the floodplain. The silt may besupplied directly from overbank flow during floods or bereworked aeolian silt eroded from hillslopes during localprecipitation events. The iron staining and mud cracks betweenlaminations indicate periods of desiccation between sheet-flowevents. The degree of staining and size of the mud cracksreflect the intensity of drier spells. Gravel lenses in Section 2indicate that colluvial cones on local hillslopes extended intothe valley during breaks in fluvial activity. The lack of any

t sequence at Section 1 showing a thick unit of ripple cross-laminated fine sand) Bioturbation at the top of a sandy unit (Section 2). Scales in panels A–C are 80,d inset in Figure 6 shows the location of panel C.

Figure 5. Photograph of Section 1 and a sedimentological log with luminescence ages. Note the thin sand lenses. The arrow points to a prominent sand body near thetop of the sequence. Figure 4 shows more detail of this sand body.


pedogenesis in the sequence indicates rapid sedimentation.This association forms the basal part of Section 1 and a majorpart of Section 2.

Stratigraphic sections

Section 1 (23°37′46″S; 15°10′46″E)This is a 15-m-thick terraced sediment sequence located in

the tributary that joins the Kuiseb at Homeb (Fig. 5). The basalpart of the sequence consists of dune sand followed by a 5.7-m-thick floodplain facies association. The floodplain sedimentarypackage internally shows mud cracks of increasing size andmore pronounced iron oxide reddening between the layers ofparallel laminated silt from the bottom to the top. The topmostbed shows a concentration of penecontemporaneous gypsumcrystals and mud-crack polygons with diameters ∼1.5 m and∼5 cm deep. A 5.2-m-thick channel facies association,internally consisting of three sand body sequences, overliesthis. Each sequence consists of fluvially modified dune sand atthe base, overlain by ripple cross-laminated and parallel-laminated fine sand, and capped by bioturbated clayey silt.The upper stratigraphic package is the result of channel barformation within the channel and suggests channel aggradation.

Section 2 (23°37′39″S; 15°10′6″E)This 18-m-thick sequence is a castle-shaped remnant on the

northern bank of the main Kuiseb River channel at Homebupstream from Section 1 (Fig. 6). At the base is fluviallymodified dune sand that coarsens upward and becomes mixedwith gravels, indicating interplay between fluvial, aeolian, andcolluvial activity. This is followed by ∼15 m of the floodplain

facies association. Internally, the package consists of eight thinsand lenses capped with bioturbated clayey silt and five units ofangular gravel. The sand bodies were formed in shallowchannels that developed on an aggrading floodplain. The gravellenses indicate breaks or shifts in fluvial activity andprogradation of colluvial lobes from the valley walls duringlocal rainfall events.

Chronology

Based on morphostratigraphic evidence in and aroundHomeb, a mid to late Pleistocene age was postulated by earlyworkers (Marker and Müller, 1978; Ward, 1987). Radiocarbondates on calcareous deposits, molluscan shells and decayedwood fragments (Pta-1492, 1822, 1860–62, 2083, 2688)indicate sedimentation between 23,000 and 19,000 14C yr BP(Vogel, 1982). After calibration using data in Bard et al.(1990, 2004) and Hughen et al. (2004) (see below for details),Vogel's ages indicate deposition of the Homeb silts from27,000 to 22,000 cal yr BP. A possible problem with Vogel's(1982) ages is that all but one are for shell or secondarycarbonate, deposits that often give ages that are too oldbecause of contamination by old carbon. However, one age(Pta-2688) of 22,800 ± 800 14C yr BP (∼27,000 cal yr BP) ison organic matter, a plant stem 20 m below the top of oneHomeb sediment sequence, suggesting that there was a phaseof deposition at this time. Eitel and Zöller (1996) dated sandylayers at the base and near the top of a Homeb sequence,obtaining bracketing thermoluminescence (TL) ages of23.3 ± 3.2 ka and 19.3 ± 1.8 ka. However, opticallystimulated luminescence (OSL) dating is preferred over TL

Figure 6. Photograph of Section 2 and a sedimentological log with luminescence ages.


when dealing with fluvial sediments, as the trapped chargesassociated with OSL signals take significantly less time toreset (e.g., Aitken, 1998).

Using OSL, Bourke et al. (2003) dated two sections ofHomeb silt, one in the main channel and the other near themouth of a tributary upstream. They applied the single-aliquotregeneration (SAR) method by using quartz grains mountedon 10-mm stainless discs. The results indicate sedimentdeposition 9.8–6.3 ka and show that six of the eight sampleswere only partially bleached when deposited. Mean sample

Table 2Dosimetry data, depth, and computed dose rate of analyzed samples

Sample no. Sample Depth (m) U (ppm) Th (ppm) K (%

1 HS-1 13.2 3.0 ± 0.7 6.6 ± 2.3 1.612 HS-3 8.6 3.8 ± 1.1 8.9 ± 3.8 2.143 HS-4 6.0 3.4 ± 1.2 10.4 ± 3.4 2.784 HS-6 3.5 1.3 ± 0.5 7.3 ± 1.8 1.315 HS-7 1.2 3.3 ± 1.0 8.6 ± 3.4 2.466 HS-8 0.3 2.5 ± 0.7 8.8 ± 2.3 3.097 HS-9 17.8 2.1 ± 0.7 7.6 ± 2.3 1.678 HS-10 12.5 4.1 ± 1.3 14.6 ± 5.4 1.62

paleodose was used to estimate the ages of well-bleachedsamples and the minimum paleodose to date poorly bleachedsamples.

In this study, we dated eight samples by OSL (Figs. 5, 6;Tables 2, 3), six from Section 1 (HS-1, 3, 4, 6–8), and two fromSection 2 (HS-9, 10). As Bourke et al. found most of theirHomeb sediment samples to be only partially bleached, weadopted a rigorous dating plan in order to obtain the bestpossible chronology for our sections. Samples were dated inthree ways, first by the single-grain method and then by the

) Water content (%) Cosmic ray (μ Gy/ka) Dose rate (Gy/ka)

5 ± 2 150 ± 30 2.6 ± 0.35 ± 2 150 ± 30 3.4 ± 0.45 ± 2 150 ± 30 3.9 ± 0.55 ± 2 150 ± 30 2.1 ± 0.35 ± 2 150 ± 30 3.6 ± 0.45 ± 2 150 ± 30 3.9 ± 0.45 ± 2 150 ± 30 2.6 ± 0.35 ± 2 150 ± 30 3.5 ± 0.5

Table 3Comparison of central and minimum paleodoses and ages based on multiple-grain 9.6-mm and 4-mm aliquot and single-grain analyses

Sample IDandbleaching(W = well,P = poorly)

Ages(ka) ⁎

Multiple grain Single grain

9.6-mm aliquot 4-mm aliquot

Central P(Gy)

Min. P (Gy) CentralAge(ka)

Min. Age(ka)

Central P(Gy)


Min. Age(ka)

Central P(Gy)


Min. Age(ka)

HS-1 (P) 12.4 ± 1.5 54.5 ± 6.5 34.8 ± 1.2 16.3 ± 2.8 12.5 ± 1.2 44.4 ± 9.1 34.7 ± 1.3 15.9 ± 3.6 12.4 ± 1.2 64.2 ± 0.9(90)

32.2 ± 3.0(10)

24.7 ± 2.9 12.4 ± 1.8

HS-3 (W) 14.2 ± 1.7 63.6 ± 4.8 56.3 ± 2.7 18.6 ± 2.6 16.5 ± 2.3 57.3 ± 7.0 50.0 ± 2.0 15.8 ± 2.7 13.8 ± 1.7 68.2 ± 2.2(22)

49.7 ± 0.3(3)

20.1 ± 2.4 14.6 ± 1.7

HS-4 (P) 16.0 ± 2.8 87.1 ± 13.0 81.7 ± 9.3 21.8 ± 4.2 20.4 ± 3.3 – – – – 97.2 ± 3.0(37)

62.5 ± 6.2(5)

24.9 ± 3.3 16.0 ± 2.6

HS-6 (W) 16.3 ± 2.6 44.1 ± 3.4 37.9 ± 1.6 21.4 ± 2.9 18.5 ± 2.2 43.5 ± 8.2 35.6 ± 1.7 19.9 ± 4.1 16.3 ± 1.6 45.8 ± 2.6(7)

34.3 ± 3.3(1)

21.8 ± 3.4 16.3 ± 2.8

HS-7 (P) 17.6 ± 3.0 79.3 ± 14.5 62.8 ± 7.3 22.2 ± 4.9 17.4 ± 2.9 90.8 ± 12.9 64.5 ± 10.4 23.9 ± 4.2 16.9 ± 3.3 97.5 ± 2.0(37)

66.6 ± 6.6(4)

27.1 ± 3.1 18.5 ± 2.8

HS-8 (W) 16.3 ± 2.6 70.7 ± 4.6 62.3 ± 2.8 17.8 ± 2.4 15.7 ± 1.8 69.9 ± 6.3 62.6 ± 3.2 16.5 ± 2.1 14.8 ± 1.5 81.9 ± 3.9(6)

66.0 ± 11.8(1)

21.0 ± 2.4 16.9 ± 3.5

HS-9 (W) 4.8 ± 0.7 33.2 ± 2.8 25.0 ± 1.3 12.8 ± 1.9 9.8 ± 1.2 24.9 ± 8.7 11.9 ± 0.9 9.2 ± 3.3 4.5 ± 0.5 20.6 ± 0.5(38)

13.0 ± 1.8(4)

7.9 ± 0.9 5.0 ± 0.9

HS-10 (W) 6.8 ± 1.1 42.7 ± 4.0 33.9 ± 3.1 12.2 ± 2.2 9.9 ± 1.8 38.5 ± 8.5 25.9 ± 1.7 10.6 ± 2.8 7.1 ± 1.1 35.9 ± 0.6(33)

23.5 ± 2.4(4)

10.3 ± 1.5 6.7 ± 1.2

Ages are ka. Note the good correspondence between the multiple-grain 4-mm aliquot minimum ages and the single-grain minimum ages.⁎ Ages in ka are weighted means of 4-mm aliquot minimum and single-grain minimum ages.

485P.

Srivastavaet

al./Quaternary

Research

65(2006)

478–491


multiple-grains single-aliquot method using two aliquot sizes of9.6 mm and 4.0 mm.

We applied two tests to determine if our samples were poorlyor well bleached (Clarke, 1996; Colls et al., 2001). Theserevealed that samples HS-1, 4, and 7 were not properly bleachedat the time of deposition (Table 1).

Table 3 presents the single- and multiple-grain age data forthe eight samples analyzed, including central and minimumages. Given the likelihood of partial bleaching, we assume thatminimum ages in all analyses are a more accurate reflection ofthe true age of deposition, and that single-grain minimum agesare the most accurate (Bourke et al., 2003). In single-aliquotanalysis, the minimum age is the age of the aliquot withminimum paleodose; in single-grain analysis, it is the average ofthe youngest 10% of grains measured. In fact, our analysesshow that 4-mm aliquot minimum ages closely approximate thesingle-grain minimum ages and both are significantly youngerthan 9.6-mm aliquot minimum and central ages (Fig. 8).Furthermore, the 4-mm aliquot minimum ages of some samplesare younger than the single-grain minimum ages (e.g., HS-3, 7,8, and 9). The difference in stimulation wavelength,470 ± 30 nm in the case of single aliquot and 532 nm in thecase of single grains, could be one possible explanation for thisbecause the stimulation wavelength affects the proportion of themedium and slow components in the initial signal (Thomas etal., 2005). Based on our results, we believe that the small-aliquot (4-mm) approach provides ages comparable to thoseobtained by the more laborious and time-consuming single-grain method. Our final ages for the samples were calculated asweighted averages of the 4-mm aliquot minimum age and thesingle-grain minimum age.

Ages for Section 1 range from 12.4 ± 1.8 ka (HS-1, thelowermost sample) to 16.9 ± 3.5 ka (HS-8, the uppermostsample), not considering HS-4 and 7 that were partially

Figure 7. Lateral lithology of Section 1 showing lenses of fine sand commonly cappednot coupled with sandy units. Note the increasing thickness of the sand bodies towa

bleached according to criteria of both Clarke (1996) and Collset al. (2001). These overlapping ages suggest a fast rate ofsedimentation centered at 14.9 ± 2.1 ka (weighted mean of HS-1and HS-8).

Section 2, within the main channel of the Kuiseb, yieldedtwo ages, 4.9 ± 0.8 ka (HS-9) near the base, and 6.8 ± 1.1 ka(HS-10) 12.5 m from the top. Bourke et al. (2003) obtained acomparable age for samples taken 38 cm from the top of a 19.4-m-thick sequence in the Kuiseb (sample 2/1), suggesting thatthe sediments in the main channel were deposited rapidlyaround 6 ka (weighted mean of HS-9 and HS-10).

Depositional environment of the Homeb silts

The fine silts at Homeb are located in protected areas of themain channel and in tributaries of the lower Kuiseb River, andthey represent a graded river profile. Section 1 shows 1.5–0.5 mthick channel sequences towards the top (Figs. 4, 5 and 7),∼7 m above the bed of the tributary channel and ∼35 mabove the Kuiseb River bed. This indicates channel activityand aggradation within the tributary. It is reasonable tosuggest that this aggradation was in response to aggrada-tion of the main Kuiseb channel. However, the sequence inthe main channel (Section 2) contains no major sandbodies except at the base where fluvially modified dunesand forms the plinth, with small lenses of rippled sandcapped with bioturbated clayey silt. This indicates minorchannel activity on a floodplain environment. This modelof floodplain aggradation is also supported by earlierobservations (e.g., Ollier, 1977; Ward, 1987; Smith et al.,1993).

Heine and Heine (2002) suggest that the Homeb siltsrepresent a series of slack-water deposits and envisage floodsrising to more than 45 m above the level of the present channel

by bioturbated clayey silt. The units of parallel laminated silt are thicker and arerds the top of the sequence and the gypsum layer in the middle of the section.


between 23,000 and 19,000 cal yr BP. Slack-water deposits aretypically composed of fine-grained sand and silt couplets thatsettle out in areas of reduced flow velocity during large floods(Baker, 1987). The contact between these sandy and silty layersis gradual because of temporary ponding of the floodwaters(Kale et al., 2000).

In fact, the Homeb silt sequences are made up of thick unitsof parallel laminated silt where individual laminae representsheet-flow events. Each lamina is stained with an iron oxidecoating and shows mud cracks without being individuallycoupled with sandy units (Fig. 7). The sandy units are lensoid,commonly capped by bioturbated clayey units, and do not showany gradual contact with silty units. Thus, these representshallow channel deposits. The 1.5–0.5 m thick sand bodiestowards the top of Section 1 were deposited in relatively deeperchannels and under somewhat wetter conditions. This change insediment characteristics from the base to the top of Section 1suggests a transition from drier to wetter conditions duringdeposition.

We suggest that the Homeb silts formed due to channelaggradation and that the bulk of the sediment was depositedby sheet flows charged with silty sediments and by shallowchannel fills under a transitional climate from dry to wetterconditions. At times of extremely dry climate and sparsevegetation, sediment is readily available in the catchment andis transported downvalley as rainfall and river dischargeincrease. The transition to a wetter climate favors channelaggradation because this is when the stream has enoughhydrologic energy to transport sediment and there is novegetation to prevent its removal from hillslopes. Theavailability of sediment decreases as the climate becomeswetter and there is a more complete cover of vegetation. Thispromotes an increase in the water/sediment ratio and theincision of previously deposited sediments.

If the Homeb silts are slack-water deposits, this impliesprogressively higher floods and wetter conditions in the Kuisebbasin with increasing thickness of the deposits. However,evidence of desiccation in silty units in the basal package ofSection 1 (mud cracks, iron oxide staining and gypsumcrystallization) indicates relatively dry conditions. Silt depositswithin the main channel of the Kuiseb are unlikely to be slack-water deposits (Eitel et al., 2001, Smith et al., 1993; Marker andMüller, 1978).

As the Homeb silts are preserved in sheltered irregularities inthe valley walls of the Kuiseb (Ward, 1987), they could havebeen deposited in canyon expansion environments, particularlydownstream of Homeb. Canyon expansion environments aredistinct sites of deposition where the canyon widens to such adegree that the width of the canyon exceeds the average widthand creates an expanding jet of flow emerging from a reachimmediately upstream. This results in flow separation with asharp decrease in flow velocity, accumulating sedimentarysequences similar to slack-water deposits (Baker and Kochel,1988; Benito et al., 2003). Such deposits generally consist of10–50 cm thick units of medium to fine sand exhibiting a seriesof physical structures such as trough cross beds, parallellaminations, and massive sand bodies (Benito et al., 2003).

However, the Homeb silts are found not only in canyonexpansion zones but also within tributaries and upstream ofnarrow valley reaches. Furthermore, they consist primarily ofparallel laminated silts and lensoidal sand units, which rules outany possibility of their deposition only in canyon expansionenvironments.

Stratification of the Homeb silts follows the downstreamgradient (Fig. 1C) of the river profile, and silty units do notshow any varved or deltaic facies typical of lake sediments (seealso Heine and Heine, 2002). Therefore, it is unlikely that thesilts were deposited as a result of large dunes blocking the riverchannel and damming river flow, as suggested previously(Goudie, 1972; Rust and Wienke, 1980).

Marker andMüller (1978) have argued that the silt sequencesat Homeb are river end-point accumulations with depositionunder extremely low-energy conditions. Such a scenario isimprobable as the silts occur in the main Kuiseb valley andextend deep into its tributaries, and they rise to more than 45 mabove the riverbed.

In our view, the Homeb silts are found in tributaries and insheltered embayments of the main Kuiseb channel, because inthese locations the sediments that formally filled the valley wereprotected from later stream erosion. The massive floodspostulated by Heine and Heine (2002) to explain silts ∼45 mabove the present Kuiseb channel are not necessary ifbackflooding of tributaries was accompanied by aggradationof the main channel, as we suggest. The sediments at ClayCastles in the Hoarusib River of the northern Namib are similarto those at Homeb and also suggest an origin due to riverbedaggradation rather than an origin as slack-water, canyon-expansion blocked-riverbed (lake) or river end-point deposits(Srivastava et al., 2005).

River aggradation may result from several climatic andtectonic conditions. The graded river profile exhibited by theupper surfaces of residual silts in the Kuiseb indicates that at thetime of accumulation, the river could transport excess sedimentin a range of energy conditions. This implies that sedimentaccumulation started during a phase when the Kuiseb carried ahigher sediment load and terminated when an energy balancebetween erosion and transportation was reached.

On the other hand, tectonic uplift downstream from Homebwould have increased the accommodation space upstream in thevalley, thus promoting aggradation and subsequent incisionduring wetter conditions. Marker (1977) has shown that theaverage gradient of the Namib unconformity surface nearHomeb, into which the river is incised, and that of the mainchannel are similar. However, ∼50 km downstream fromHomeb, and about 80 km from the sea, the average gradient ofthis surface and the gradient of the Kuiseb channel increasesignificantly (Fig. 1C), an increase that Rust and Wienke (1980)suggest may indicate warping.

An alternative to warping is that the convex long profile of theKuiseb River, like that of many arid zone rivers, is due toincreasing desiccation seawards and loss of stream energy (e.g.,Goudie, 1972; Rust and Weineke, 1980). However, in thecanyon section of the Kuiseb, formed from 2.8 to 0.4 Ma (Vander Wateren and Dunai, 2001), the profile is decidedly concave

Figure 8. Graph of single and multiple grain ages. Note that with the exceptionof one sample, all ages are within the error limits of the ages produced by bothtechniques.


suggesting formation by a stream increasing in dischargedownvalley and probably reaching the sea (Fig. 1D). Onepossible reason for the Kuiseb canyon system ending ∼80 kminland is that since its formation there has been tectonic upliftalong the coast that has sharply increased the gradientdownstream of Homeb. Marker (1977) notes that all western

Figure 9. The Homeb silt record compared with other regional paleoclimate records fiTo correct for dead carbon, 1000 yr was subtracted from the age before calibration as sRust et al. (1984), Deacon and Lancaster (1988), Vogel (1989), Teller et al. (1990),Higher values suggest increased fluvial activity on land. (III) Summary climate recordare from Gingele (1996). (IV) The Homeb silt record (this paper). (V) Increasedgroundwater (Heaton et al., 1983). (VI) Lake development at the Tsodilo Hills, NW

river courses in Namibia are progressively younger seawards,since the coastal platform across which they flow constitutes aseries of emergent marine terraces. She interprets the generalconvexity of the lower Kuiseb as a function of emergence ratherthan of desiccation (Marker, 1977, p. 204). Significantly, theHomeb silts are located in the section of the Kuiseb channel withthe lowest thalweg gradient, where stream velocities should belower and therefore sediment preferentially deposited (Fig. 1D).

The locations and elevations of sediments exposed inSections 1 and 2 indicate that they were laid down during thefill stages of a series of cut-and-fill episodes. After a phase ofriver incision and sediment removal, there was a period ofaggradation at Homeb ∼15 ka when the Kuiseb River began tocarry a higher sediment load, probably during an arid-to-humidclimate transition as suggested by the change in sedimentcharacteristics from the base to the top of Section 1. Sometimebetween ∼15 and 6 ka, flash floods removed most of the oldersediment fill, leaving remnants only in sheltered embaymentsalong the valley (Fig. 8).

Around 6 ka, a second fill was rapidly emplaced in somelocations against bedrock, and in others against the erodedremnants of the earlier deposit, as at Homeb. The dominance offloodplain facies association sediments with only moderatechannel activity in Section 2 (∼6 ka), in contrast to majorchannel facies sediments in the upper part of Section 1

xed by calibrated radiocarbon ages. (I) Frequency of tufa and pan carbonate ages.uggested by Vogel (1989). Data from Lancaster (1979), Vogel and Visser (1981),Brook et al. (1999). (II) Percent sand in the GeoB 1023-4 Atlantic Ocean core.from the GeoB 1023-4 core indicating maximumwetness at 7–6 ka. (II) and (III)recharge in the western Kalahari indicated by the isotopic characteristics ofBotswana (Thomas et al., 2003).


(∼15 ka), suggests that the climate was somewhat wetter andriver discharge somewhat greater during the earlier phase ofsedimentation than during the latter. Subsequently, the ∼6-kadeposits were also largely removed by flash floods movingdown the Kuiseb valley. However, this more recent phase oferosion was not as complete as was the erosion of the older fill,leaving remnants in the main Kuiseb channel as well as insheltered embayments. High-energy flash floods in the valleymay be responsible for the erosion sequences with major sandbodies corresponding to those in Section 2. The cut and fillnature of the Homeb deposits argues for a climatic explanationfor their deposition and erosion. However, we cannot rule outthe possibility that sedimentation was at least in part a reactionto crustal warping near the coastal zone.

The Homeb silts and regional climate

OSL and TL chronologies for fluvial and aeolian deposits insouthern Africa rely on single-aliquot rather than single-grainanalysis and use mean ages rather than youngest ages. For thisreason, it is difficult to compare the Homeb chronologypresented here with these records. However, OSL data fromrelict terrace sediments in the Khumib valley of the northernNamib suggest a wetter climatic phase from >20.8 to 15.6 kafollowed by incision into the Khumib sediment sequence after8.1 ka (Srivastava et al., 2004). The Khumib record is based onmean aliquot ages that are very similar to the mean 9.6-mmaliquot ages we obtained on the Homeb deposits. Therefore, thiscorrespondence suggests that the climate changes that led todeposition and later erosion of the Homeb silts were widespread(Fig. 9).

Our minimum age record from Homeb can also becompared with climate records based on calibrated radiocar-bon ages for organic matter and secondary carbonate.Following Vogel (1989) we subtracted 1000 years from theoriginal radiocarbon age of secondary carbonates beforecalibration to account for the incorporation of old, deadcarbon. Ages were then calibrated to calendar years BP (cal yrBP) using OxCal version 3.9 (Ramsey, 1995, 2001). Agesbeyond the range of the dendrochronology-based OxCalprogram were corrected by interpolation from graphs showingradiocarbon and calendar year ages for Barbados corals (Bardet al., 1990; Figs. 1, 2) and planktonic foraminifera for theIberian Margin (Bard et al., 2004; Fig. 4) and Cariaco Basin(Hughen et al., 2004; Fig. 2). Together, these graphs providecalibration data for the last 50 kyr but because of some spreadof data our interpolated ages probably have an estimation errorof at least ±0.5 kyr.

Several radiocarbon-dated records indicate wetter conditionsfrom ∼18,000 to 12,000 cal yr BP and at ∼7000 cal yr BP (Fig.9). For example, the Nata Gamtes and Hudaob tufas in theKuiseb Valley date to 17,800 and 11,500 cal yr BP (Vogel,1989), the Blässkrantz tufa in the Naukluft Mountains to 17,000and 11,500 cal yr BP (Brook et al., 1999), the Namutoni tufa inEtosha to ∼9000 cal yr BP (Rust et al., 1984), and spring tufasat Gobabis to ∼19,000 and 13,000–11,500 cal yr BP (Deaconand Lancaster, 1988).

Pan sediments can also record past wetter conditions. AtOtjimaruru Pan, calcified organic deposits date to 14,000 cal yrBP (Lancaster, 1989), while at Urwi Pan spheroidal stromato-lites formed in permanent wave-agitated waters 18,300–17,500 cal yr BP (Lancaster, 1979). Lancaster (1989) suggeststhat there was a period of increased moisture in thesouthwestern Kalahari from 19,000 to 13,000 cal yr BP andthat the early Holocene was dry. Twenty-one ages on carbonatefrom eight playa sequences in the northern Namib Sand Seaindicate three wet intervals in the past, the most recent beingfrom 16,800 to 10,000 cal yr BP (Vogel and Visser, 1981; Telleret al., 1990). The Tsodilo Hills in northwest Botswana currentlylack surface water but a lake existed there from 19,000 to12,000 cal yr BP (Thomas et al., 2003).

A sharp increase in excess air in the Stampriet aquifer southof Windhoek, Namibia, around 7000 cal yr BP may record atransition from a dry to a wet climate (Stute and Talma, 1997).Pollen in spring deposits near Windhoek also suggests moistconditions from 8000 to 7000 cal yr BP (Scott et al., 1991).

Marine core evidence points to arid conditions in Namibiafrom 22,000 to 18,000 cal yr BP, but the period 18,000–12,000 cal yr BP was a time of increased moisture and riverrunoff due to a stronger monsoon. A second period of increasedriver flow occurred from 11,000 to 6000 cal yr BP, indicating amaximum humidity in the source area from 7000 to 6000 cal yrBP and corresponding to the Holocene climatic optimum.Present conditions were reached about 5000 cal yr BP (Gingele,1996). Studies of the isotopic composition of groundwater inthe western Kalahari indicate increased rainfall at 17,000–9000 cal yr BP (Heaton et al., 1983).

Conclusions

The Homeb silts are the result of channel aggradation inthe Kuiseb River valley. They were deposited by sheet flowscharged with silty sediments and by shallow channel fills.The cut-and-fill nature of the deposits suggests a climatictrigger for deposition during transitions from a drier to a morehumid climate. The sediments were later incised by low-frequency intense flash floods under slightly more aridconditions.

Luminescence dating of the Homeb sediments by the single-aliquot and single-grain techniques indicates that the small-aliquot (4-mm) method provides ages that are comparable tothose obtained by the more time-consuming and laborioussingle-grain method. OSL ages for two Homeb sedimentsections indicate rapid sedimentation centered at ∼15 and∼6 ka. Bleaching analysis suggests that five of the eightsamples analyzed were properly bleached prior to deposition.This is also demonstrated by the correspondence betweencentral and minimum ages of multiple- and single-grainanalyses (Fig 8, Table 3). The ages presented here for theyounger phase of sediment deposition are comparable to thoseof Bourke et al. (2003). However, the earlier phase of sedimentaccumulation at ∼15 ka is significantly younger that previousTL and 14C chronologies that indicate sediment deposition∼27,000–22,000 cal yr BP (Vogel, 1982) and ∼23–19 ka (Eitel


and Zöller, 1996). The TL ages may be too old because ofincomplete bleaching and use of central rather than minimumaliquot ages. However, as mentioned above, the 14C ages ofVogel (1982) include one age of 22,800 ± 800 14C yr BP(∼27,000 cal yr BP) for a plant stem 20 m below the top of aHomeb sediment sequence (Pta-2688). If this plant material hasnot been contaminated by older carbon, it may record an olderphase of sediment accumulation considerably earlier than 15 ka.The two phases of sedimentation at Homeb (∼15 ka and ∼6 ka)follow arid phases of climate at 21,000–17,500 cal yr BP and11,000–8900 cal yr BP recorded in the sediments of SouthAtlantic Ocean core 1023-5 off the Kunene River mouth (Shi etal., 2000). This supports our interpretation of the Homeb silts ashaving been deposited during transitions from an arid to a morehumid climate with subsequent erosion when more aridconditions returned.

Acknowledgments

This research was funded by NSF grant BCS-0002193 andwas conducted under permits issued by the National Monu-ments Council and the Ministry of Environment and Tourism ofNamibia. PS dedicates this work to the devout memory of hisfather Mr. Narendra Deo Srivastava (1943–2004). We thankNicholas Lancaster and Klaus Heine for helpful suggestions thatimproved the manuscript. PS thanks to the Director, WadiaInstitute of Himalayan Geology, Dehradun, India for providingnecessary facilities during the revision of the manuscript.

References

Aitken, M.J., 1998. An Introduction to Optical Dating. Academic Press,London. 359 pp.

Baker, V.R., 1987. Palaeoflood hydrology and extreme flood events. Journal ofHydrology 96, 79–99.

Baker, V.R., Kochel, R.C., 1988. Flood sedimentation in bedrock fluvialsystems. In: Baker, V.R., Kochel, R.C., Patton, P.C. (Eds.), FloodGeomorphology. John Wiley, New York, pp. 123–137.

Bard, E., Hamelin, B., Fairbanks, R.G., Zindler, A., 1990. Calibration of the 14Ctimescale over the past 30,000 yr using mass spectrometric U–Th ages fromBarbados corals. Nature 345, 405–410.

Bard, E., Rostek, F., Ménot-Combes, G., 2004. Radiocarbon calibration beyond20,000 14C yr BP by means of planktonic foraminifera of the Iberian Margin.Quaternary Research 61, 204–214.

Benito, G., Sanchez-Moya, Y., Sopena, A., 2003. Sedimentology of high stageflood deposits of the Tagus River, Central Spain. Sedimentary Geology 157,107–132.

Bourke, M.C., Child, A., Stokes, S., 2003. Optical age estimates for hyper-aridfluvial deposits at Homeb, Namibia. Quaternary Science Reviews 22,1099–1103.

Brook, G.A., Marais, E., Cowart, J.B., 1999. Evidence of wetter and drierconditions in Namibia from tufas and submerged speleothems. Cimbebasia15, 29–39.

Clarke, M.L., 1996. IRSL dating of sands: bleaching characteristics at depositioninferred from use of single aliquots. Radiation Measurements 26, 611–620.

Colls, A.E., Stokes, S., Blum, M.D., Straffin, E., 2001. Age limits on the lateQuaternary evolution of the upper Loire River. Quaternary Science Reviews20, 743–750.

Deacon, J., Lancaster, N., 1988. Late Quaternary Environments of SouthernAfrica. Oxford University Press. 220 pp.

Duller, G.A.T., 1999. Luminescence Analyst computer programme V2.18.

Department of Geography and Environmental Sciences. University ofWales,Abersystwyth.

Eitel, B., Zöller, L., 1996. Soils and sediments in the basin of Dieprivier–Uitskot(Khoriax District, Namibia) age, geomorphic and sedimentological inves-tigation, paleoclimatic interpretation. Palaeoecology of Africa 24, 159–172.

Eitel, B., Blümel, W.D., Hüser, K., Mauz, B., 2001. Dust and loessic alluvialdeposits in Northwestern Namibia (Damaraland, Kaokoveld) sedimentologyand palaeoclimatic evidence based on luminescence data. QuaternaryInternational 76/77, 57–65.

Gingele, F.X., 1996. Holocene climatic optimum in Southwest Africa—Evidence from the marine clay mineral record. Palaeogeography, Palaeo-climatology, Palaeoecology 122, 77–87.

Goudie, A., 1972. Climate, weathering, crust formation, dunes and fluvialfeatures of Central Namib Desert, near Gobabeb, Southwest Africa.Madoqua II 1 (54–64), 15–31.

Heaton, T.H.E., Talma, A.S., Vogel, J.C., 1983. Origin and history of nitrate inconfined groundwater in the western Kalahari. Journal of Hydrology 62,243–262.

Heine, K., Heine, J.T., 2002. A paleohydrologic reinterpretation of Homeb Silts,Kuiseb River, central Namib Desert (Namibia) and paleoclimatic implica-tions. Catena 48, 107–130.

Hughen, K., Lehman, S., Southon, J., Overpeck, J., Marchal, O., Herring, C.,Turnbull, J., 2004. 14C activity and global carbon cycle changes over the past50,000 years. Science 303, 202–207.

Jacobson, P.J., Jacobson, K.M., Seely, M.K., 1995. Ephemeral Rivers and theirCatchments: Sustaining People and development in Western Namibia.Desert Research Foundation of Namibia, Windhoek. 160 pp.

Kale, V.S., Singhvi, A.K., Mishra, P.K., Banerjee, D., 2000. Sedimentaryrecords and luminescence chronology of Late Holocene palaeofloods in theLuni River, Thar Desert, Northwest India. Catena 40, 337–358.

Lancaster, N., 1979. Evidence for a widespread late Pleistocene humid period inthe Kalahari. Nature 279, 145–146.

Lancaster, N., 1989. Late Quaternary palaeoenvironments in the southwesternKalahari. Palaeogeography, Palaeoclimatology, Palaeoecology 70, 367–376.

Marker, M.E., 1977. Aspects of the geomorphology of the Kuiseb River, SouthWest Africa. Madoqua 10, 199–206.

Marker, M.E., Müller, D., 1978. Relict vlei silts of the middle Kuiseb Rivervalley, South West Africa. Madoqua 11, 151–162.

Miall, A.D., 1996. The Geology of Fluvial Deposits. Springer, Berlin. 582 pp.Murray, A.S., Wintle, A.G., 2000. Luminescence dating of quartz using

improved single-aliquot regenerative-dose protocol. Radiation Measure-ments 32, 57–73.

Murray, A.S., Wintle, A.G., 2003. The single aliquot regenerative protocol:potential for improvements and reliability. Radiation Measurements 37,377–381.

Ollier, C.D., 1977. Outline geological and geomorphological history of theCentral Namib Desert. Madoqua 10 (3), 207–212.

Prescott, J.R., Stephan, L.G., 1982. Contribution of cosmic radiation toenvironmental dose. PACT 6, 17–25.

Ramsey, C.B., 1995. Radiocarbon calibration and analysis of stratigraphy: theOxCal program. Radiocarbon 37 (2), 425–430.

Ramsey, C.B., 2001. Development of the radiocarbon program OxCal.Radiocarbon 43 (2A), 355–363.

Rust, U., Wienke, F., 1980. A reinvestigation of some aspects of the evolution ofthe Kuiseb River valley upstream of Gobabeb, South West Africa. Madoqua12 (3), 163–173.

Rust, U., Schmidt, H.H., Dietz, K.R., 1984. Palaeoenvironments of the presentday arid South Western Africa 30000–5000 BP: results and problems.Palaeoecology of Africa 16, 109–148.

Scott, L., Cooremans, B., de Wet, J.S., Vogel, J.C., 1991. Holoceneenvironmental changes in Namibia inferred from pollen analysis ofswamp and lake deposits. The Holocene 1, 8–13.

Shi, N., Dupont, L.M., Beug, H.-J., Schneider, R., 2000. Correlation betweenvegetation in southwestern Africa and oceanic upwelling in the past 21,000years. Quaternary Research 54, 72–80.

Smith, R.M.H., Mason, T.R., Ward, J.D., 1993. Flash flood sediments andichnofacies of the Late Pleistocene Homeb Silts, Kuiseb River, Namibia.Sedimentary Geology 85, 579–599.


Srivastava, P., Brook, G.A., Marais, E., 2004. A Record of fluvial aggradation innorthern Namib Desert during the Late Quaternary. Zeitschrift fürGeomorphologie N.F., Suppl.-Vol. 133, 1–18.

Srivastava, P., Brook, G.A., Marais, E., 2005. Depositional environment andluminescence chronology of the Hoarusib River Clay Castles sediments,northern Namib Desert, Namibia. Catena 59, 187–204.

Stute, M., Talma, A.S., 1997. Glacial temperature and moisture transportregimes reconstructed from noble gases and *18O, Stampriet aquifer,Namibia. International Symposium on Isotope Techniques in the Study ofPast and Current Environmental Changes in the Hydrosphere and theAtmosphere, Vienna, Austria, pp. 307–318.

Teller, J.T., Rutter, N., Lancaster, N., 1990. Sedimentology and paleohydrologyof late Quaternary lake deposits in the northern Namib Sand Sea, Namibia.Quaternary Science Reviews 9, 343–364.

Thomas, D.S.G., Brook, G.A., Shaw, P., Bateman, M., Haberyan, K., Appleton,C., Nash, D., McLaren, S., Davies, F., 2003. Late Pleistocene wetting and

drying in the NW Kalahari: an integrated study from the Tsodilo Hills,Botswana. Quaternary International 104, 53–67.

Thomas, P.J., Jain, M., Juyal, N., Singhvi, A.K., 2005. Comparison of singlegrain and small aliquot OSL dose estimates in <3000 years old riversediments from south India. Radiation Measurement 39, 457–469.

Van der Wateren, F.M., Dunai, T.J., 2001. Late Neogene passive margindenudation history—Cosmogenic isotope measurements from the centralNamib Desert. Global and Planetary Change 30, 271–307.

Vogel, J.C., 1982. The age of Kuiseb River silt terrace at Homeb. Palaeoecologyof Africa 15, 201–209.

Vogel, J.C., 1989. Evidence for past climatic change in the Namib desert.Palaeogeography, Palaeoclimatology, Palaeoecology 70, 355–366.

Vogel, J.C., Visser, E., 1981. Pretoria radiocarbon dates II. Radiocarbon 23,43–80.

Ward, J.D., 1987. The Cenozoic succession of Kuiseb Valley, Central NamibDesert. Geological Survey of Namibia Memoir. 124 pp.

depositional environment and osl chronology of the homeb silt deposits, kuiseb river, namibia

Documents