A Sub-Centennial-Scale Optically Stimulated Luminescence Chronostratigraphy and Late Holocene Flood History from a Temperate River Confluence Ben Pears1, Antony G

https://doi.org/10.1130/G47079.1

Manuscript received 7 October 2019 Revised manuscript received 22 January 2020 Manuscript accepted 31 March 2020

A sub-centennial-scale optically stimulated luminescence chronostratigraphy and late Holocene flood history from a temperate river confluence Ben Pears1, Antony G. Brown1,2, Phillip S. Toms3, Jamie Wood3, David Sanderson4 and Richard Jones5 1Palaeoenvironmental Laboratory, Department of Geography and Environmental Science, University of Southampton, Highfield Campus, University Road, SO17 1BJ Southampton, UK 2Natural Sciences, Tromsø University Museum, Arctic University of Norway, 9013 Tromsø, Norway 3Luminescence Dating Laboratory, School of Natural and Social Sciences, University of Gloucestershire, Swindon Road, GL50 4HZ Cheltenham, UK 4Scottish Universities Environmental Research Centre, Rankine Avenue, Scottish Enterprise Technology Park, G75 0QF East Kilbride, UK 5Centre for English Local History, University of Leicester, Salisbury Road, LE1 7RH Leicester, UK ABSTRACT In upland reaches, 14C probability density River confluences can be metastable and contain valuable geological records of catchment functions can yield depositional histories from response to decadal- to millennial-scale environmental change. However, in alluvial reaches, flood channel and flood deposits (Macklin et al., stratigraphies are particularly hard to date using 14C. In this paper, we use a novel combination 2010), but many 14C-derived chronostratigra- of optically stimulated luminescence and multiproxy sedimentological analyses to provide a flood phies terminate before 1000 yr B.P. as a result record for the confluence of the Rivers Severn and Teme (United Kingdom) over the past two of sediment reworking and organic transloca- millennia, which we compare with independent European climate records. The results show that tion, leading to problems with the calibration by ca. 2000 yr B.P., the Severn-Teme confluence had stabilized and overbank alluviation had of dates. Additionally, in clastic-dominated allu- commenced. Initially, this occurred from moderately high flood magnitudes between ca. 2000 vial sequences, suitable organic material is only and 1800 yr B.P. (50 BCE–150 CE), but was followed from 1800 to 1600 yr B.P. (150–350 CE) occasionally encountered, and, where it is, its by fine alluvial deposition and decreased flood intensity. From 1600 to 1400 yr B.P. (350–550 reliability for dating is also questionable. These CE), the accumulation rate increased, with evidence of large flood events associated with the factors have made it difficult to relate the sedi- climatic deterioration of the Dark Age Cold Period. Following a period of reduced flood activity mentary record to documentary or instrumen- after ca. 1400 yr B.P. (ca. 550 CE), larger flood events and increase in accumulation rate once tal flood histories, which in Europe rarely go again became more prevalent from ca. 850 yr B.P. (ca. 1100 CE), coincident with the start of back further than 100–250 yr (Macdonald and the Medieval Climate Anomaly, a period associated with warmer, wetter conditions and in- Sangster, 2017; Longfield et al., 2018). In this creased land-use intensity. This state persisted until ca. 450 yr B.P. (ca. 1500 CE), after which paper, we show how the confluence of the Rivers increased flood magnitudes can be associated with climatic variations during the Little Ice Age. Severn and Teme (United Kingdom) has evolved We demonstrate that from the combination of high-resolution dating techniques and multiple over the past two millennia. We reconstruct analytical parameters, distinctive phases of relative flood magnitude versus flood duration can overbank sedimentation and a detailed flood be determined to a detailed chronological precision beyond that possible from 14C dating. This history using extensive optically stimulated permits the identification of the regional factors behind floodplain sedimentation, which we luminescence (OSL) dating, alongside detailed correlate with the intensification of land-use and climatic drivers over the last two millennia. stratigraphic and sedimentological analyses.

INTRODUCTION et al., 2007; Brown et al., 2013; Dixon et al., CONFLUENCE METASTABILITY River floodplains contain archives of chang- 2018). Additionally, oscillatory channel behav- The Severn-Teme confluence (centered ing climatic and catchment conditions through ior can also occur, stimulated by large floods on 52.166947°N, 2.2303104°W) has >5 m of their extensive sedimentary records. In partic- (Brown et al., 2013). However, some conflu- overbank sediments deposited unconformably ular, confluence zones act as nodal points for ences can be surprisingly stable, becoming fixed on channel gravels. Both the Severn (4325 km2) flooding, because floodplains are generally at or pinned when confined by tall banks created by and Teme (1648 km2) catchments drain the their widest and integrate flood peaks from sev- high rates of levee and/or overbank deposition Cambrian Mountains, receiving mixed-intensity eral catchments. Confluence migration can hap- (Dixon et al., 2018). It is these large volumes of precipitation from west-to-east cyclonic activ- pen through channel-belt movements as well as overbank sediment stored in confluence zones ity and depression systems. The 3.5 km2 con- from meander-migration and/or through mean- that act as archives of flooding and past catch- fluence zone is bounded by Pleistocene gravel der cutoffs (Best and Lane, 2004; Camporeale ment conditions. terrace deposits with alluvium deriving from the

CITATION: Pears, B., et al., 2020, A sub-centennial-scale optically stimulated luminescence chronostratigraphy and late Holocene flood history from a temperate river confluence: Geology, v. 48, p. 819–825, https://doi.org/10.1130/G47079.1

Geological Society of America | GEOLOGY | Volume 48 | Number 8 | www.gsapubs.org 819

Downloaded from http://pubs.geoscienceworld.org/gsa/geology/article-pdf/48/8/819/5095985/819.pdf by guest on 25 September 2021 3.0°W 2.5°W 2.0°W aries, which continue to follow the present river Catchment location Shrewsbury A - Severn catchment course, as well as the location of medieval mills, N Buildwas N leats (artificial watercourses or aqueducts sup- plying water to watermills), and weirs (low head 52.5°N dams) all indicate channel stability from as early Worcester as 1188–951 yr B.P. (762–999 CE) (Fig. 1). The UK extensive depths of sandy silts blanketing the River Broadwas paleochannels can be traced across the conflu- Severn A ence and further afield, including the River Avon catchment Severn-Teme B confluence Tewkesbury catchment to the east. This unit was originally 52.0°N designated the “buff red silty clay” (Shotton 1978) and shown to postdate ca. 3600 yr B.P. B - Severn-Teme confluence and Powick environs Gloucester 2.25°W 2.235°W 2.22°W (ca. 1650 BCE) in all locations. B 52.175°N Section S1 km The stratigraphy of the sampled Powick location 0 25 50 75 section (52.169259°N, 2.2376946°W) (Fig. 2)

S1 WSW Sample section ENE S N consists of a lower unit of 0.6 m of yellowish- S2 20 location on River Teme gray medium to fine sandy silt (unit A), over- ) 18 Powick Hams D b O 16 lain by 2.3 m of dark-yellow to reddish-brown A RF e d m d d d 52.166°N ( f 14 fine silt (unit B), covered by 1.1 m of reddish- S1 n

i t brown medium sandy silt (unit C), capped by S2 a 12

and e Powick l

E 0.8 m of dark reddish-brown coarse sandy silt S3 10 Powick Hams KEY: and a 0.4 m soil horizon (unit D). Faint, but Cores and transects 8 Paleochannels 0 0.2 0.4 0.6 0.8 11.2 1.4 1.6 1.8 2 clear, sand-silt laminations were identified and Prehistoric/Roman S2 Distance (km) Settlement (a-c) NW SE W E 52.157°N 25 River River recorded throughout. These could be traced the Medieval manor Teme Severn Church ) entire length of the section (150 m) and corre- Medieval bridge D 20 S4 SEVERN O Powick Hams Strip fields A b spond to a sequence previously recorded ∼500 m

Parish boundary ( 15 e S4

Water mill n

o upstream (Brown, 1985). This confirms that the

Weir Dated t

a 10 Fish pond/fishery Callow End 0 0.25 0.5 0.75 1 N v sequence dated and analyzed here is representa-

e 1651 Bridge of boats l p

paleochannel E Modern bridge S3 km 5 tive of the confluence as a whole, and that the C - Confluence floodplain transects (S1, S2, S3, S4) 0 0.4 0.8 1.21.6 2 2.4 2.6 rate of overbank deposition was high enough to Distance (km) S4 W S3 E SW NE W E Floodplain transects key: outpace rare bioturbation by earthworms. 25 Carey’s Callow End River 25 Carey’s River Topsoil

) brook paleochannel Severn ) brook Severn Made ground

D 20 D 20 Dark red/brown alluvium

O O Red/brown alluvium

A RF A SAMPLING AND ANALYSIS

m d Yellow alluvium d m

(

15 15 e d d dd d d d Grey alluvium n d n Eight OSL dates and in situ dosimetry were

o Silty clay paleochannel fill

t t Organic paleochannel fill

a 10 a 10

v 4 v Sand and gravel terrace taken at 0.4–0.8 m intervals up-sequence (see

¹ C - B e

l l Sidmouth mudstone

E E 1 5 ¹4C - A 5 d Boundary ditch the Supplemental Material , Data Set S3). Sedi- e Flood embankment 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0 0.4 0.8 1.2 1.6 2 2.4 f Fishpond ment u-channels (method to obtain longitudinal Distance (km) Distance (km) p Paleochannel RF Ridge and furrow cultivation ¹4C - A: 4826-4407 yr B.P. (2876-2457 BCE), 14C - B: 1188-951 yr B.P. (762-999 CE) (Brown, 1987) b Worcester bypass in-situ sediment samples from cylindrical cores OSL sample positions and sections) ranging from 0.3 to 0.4 m in length Figure 1. (A) River Severn catchment (UK), major settlements, and other sample locations (see Fig. 4). were extracted for detailed sedimentary analysis Base map is 50-m-resolution lidar digital terrain model (DTM) (UK Ordnance Survey, 2016). (B) River (loss on ignition [LOI], magnetic susceptibility Severn–River Teme confluence on 1-m-resolution lidar DTM base map SO85SW, SO85SE (UK Environ- ment Agency, 2016), showing core locations (British Geological Survey, 2017), river section location (red dot), and paleochannels (dated, pink; undated, orange). Historic land use is plotted from 1 m lidar, 1Supplemental Material. Supplemental Sections Coventry map (ca.1648 CE, Herefordshire and Worcestershire Record Office, BA940, folio 970.7.73), S1–S16; Figures S1 (sediment accumulation rate mod- UK Ordnance Survey (UK Ordnance Survey, 1886), and Worcestershire Historic Environment Record eled by OxCal and Bacon) and S2 (relative moisture (http://www.worcestershire.gov.uk/archaeology). (C) Reconstructed and modeled floodplain cross values between OSL and LOI analytical methods); sections (S1–S4) with surface topography extracted from 1 m lidar DTM and shown as elevation Table S1 (OSL procedure from the Rivers Severn- in meters above Ordnance Datum (mAOD) against distance (km) across floodplain. Sedimento- Teme confluence at Powick, UK); and Data Sets S1 logical data were extracted from core data (Brown, 1987; British Geological Survey, 2017) and, (raw data for the modeled calendric dates, sediment historic features were marked from a county historic environment record (Worcestershire Historic accumulation rate, and sedimentological analyses), Environment Record). S2 (raw and log normalized data for ITRAX XRF analysis and key elements Zr, Rb, Fe, Mn, and heavy metals illustrated in Fig. 2), S3 (individual raw data basal Triassic mudstone, siltstone, and sandstone sion to the north between 4826 and 4407 yr B.P. sets for each 5 cm pOSL run alongside a background lithologies (Fig. 1). Deep clastic alluvial deposits (2876–2457 BCE). The remodeling of borehole sediment sample and a summary sheet of all data and of the Severn-Teme confluence were first identi- data across the confluence shows an undated replicates), S4 (raw data, log normalized data, and sta- fied more than 30 years ago (Brown, 1985) but paleochannel below the overbank units, which tistical analysis used in the agglomerative hierarchical were undatable due to a lack of suitable organic most likely demonstrates an intermediate posi- cluster analysis illustrated in Fig, 2), S5 (calculated 14 log data of sedimentary analyses by 50 yr period and material for C dating. However, a chronostratig- tion prior to the creation of the large gooseneck- the statistical analysis used in the principal component raphy of channel change in the reach was estab- type meander characteristic of the present Teme analysis illustrated in Fig. 3), and S6 (20 yr grouping lished by dating paleochannel sediments within channel. Alongside the physical evidence, his- for the sediment deposition models for the Severn- the confluence zone (Brown, 1987). The dating torical data also confirm channel stability. The Teme confluence at Powick, Broadwas, and Buildwas and climatic datasets illustrated in Fig. 4). Please visit of a large meander of the Teme, which must once earliest detailed cartographic evidence shows https://doi.org/10.1130/GEOL.26213S.12250157 to have been close to the junction with the Severn, that the Teme had adopted its current course by access the supplemental material, and contact edit- shows that it was abandoned by progressive avul- 1648 CE, and the permanency of parish bound- [email protected] with any questions.

820 www.gsapubs.org | Volume 48 | Number 8 | GEOLOGY | Geological Society of America

Downloaded from http://pubs.geoscienceworld.org/gsa/geology/article-pdf/48/8/819/5095985/819.pdf by guest on 25 September 2021 [MS], and particle size), alongside multi-elemen- derive from eroded bank sediment and mobi- these demonstrate intensive phases of deposi- tal analysis using an ITRAX X-ray fluorescence lized during high-energy flow events. Periods tion in the Severn-Teme confluence, particularly (XRF) scanner (Supplemental Material Sections of constant flood rate with equally sized events 400–350 yr B.P. (1550–1600 CE), 250–200 yr S5–S9; Data Sets S1 and S2). Additional bulk would produce a uniform pOSL curve. B.P. (1700–1750 CE), and ca. 100 yr B.P. (1850 samples were taken every 5 cm, from 10 cm into CE), and appear conterminous with the period the cleaned river section to shield them from DISCUSSION of maximum recorded historical flood events light, and subjected to luminescence profiling From the age-depth model and sedimentary (Marsh et al., 2016; Macdonald and Sangster, using a portable reader (portable optically stimu- properties, it is possible to propose a flood his- 2017) and the abandonment of the open-field lated luminescence, pOSL) (Section S10; Data tory for the midpoint of the Severn catchment system. The data support the identification of Set S3). Age-depth Bayesian modeling of the over the past 2000 yr (Fig. 2). From 2000 to periods of shorter-duration but higher-mag- dated alluvial sequence was initially conducted 1800 yr B.P. (50 BCE–150 CE), medium-silt nitude flood-dominated regimes caused by using the OSL settings provided by OxCal soft- alluvium with elevated proportions of sand and increased snowmelt and storminess (Rumsby ware version 4.3 with the IntCal13 calibration pOSL are present and possibly reflect increased and Macklin, 1996; Macklin et al., 2012). curve (Bronk Ramsey 2008, 2009) and then soil erosion from the Late Iron Age and Romano- Statistically there are clear relationships using Bacon software version 2.2 (Blaauw and British cultivation, similar to other major United between the key analytical results (Fig. 3). Christen 2011) to provide a cross-reference of Kingdom lowland floodplains (Robinson and There is a positive correlation between grain chronological quality as well as improved calen- Lambrick, 1984; Brown, 2009). Between 1800 size, sand content, and Zr:Rb associated with a dric date output (2σ [95.4%] and 1σ [68.2%] and 1600 yr B.P. (150–350 CE), there is a reduc- higher coarse component within the alluvium. precision) (Fig. 2; Sections S3 and S4; Table S1) tion in sand and increase in carbonate and fine A positive trend is also present for pIRSL and and sediment accumulation rate (∼1.8–3.2 mm/ particulates, indicating a reduction in fluvial pOSL, although there is less-clear correlation yr). Statistical analysis of data was conducted to depositional conditions. After 1600 yr B.P. between the overall particle size and pOSL, clarify depositional processes (Sections S11 and and continuing until 1400 yr B.P. (550 CE), possibly as a result of variations in sediment S12; Data Sets S4 and S5), and sediment index fine alluvial sedimentation continued, but an provenance. Clear negative correlations are pres- models created to compare against climatic data increase in accumulation rate and clear peaks ent between (1) the aforementioned classes, and (Section S14; Data Set S6). in MS demonstrate large, lower-energy flood (2) carbonate and, to a lesser extent, magnetic events at the start of the Dark Age Cold Period susceptibility. SEDIMENTARY PARAMETERS in response to climatic downturn (Lamb 1972). The stratigraphic variation at the Severn- The sequence is entirely clastic except for The development of extensive floodplain mead- Teme confluence can now be compared to the top 0.4 m associated with contemporary owlands across many river systems at this time that of other recently established deep alluvial soil formation and increased organic matter may have, in part, been as a result of suitable sequences along the Rivers Teme and Severn (Fig. 2). High-temperature LOI shows several “well-watered” conditions (Williamson 2013, (Pears et al., 2020), providing further evidence peaks within periods of finer sedimentation 204). There followed from 1400 to 1100 yr B.P. for flood regimes in other parts of their catch- reflecting carbonate, but also small increases (550–850 CE) a fall in accumulation rate and ments, as well as to European-scale climatic in mixed layer, clays that are present in the basal lower-energy alluviation, with increases in car- models to assess potential drivers (Fig. 4). All Mercia Mudstone geology. MS peaks are most bonate and pOSL after 1250 yr B.P. (700 CE). three fluvial sequences demonstrate predomi- likely due to increases in Fe minerals associated From 1100 to 600 yr B.P. (850–1350 CE), the nantly fine-grained deposition to 1450 yr B.P. with ironstones, heavy minerals, and coatings sedimentological evidence suggests higher flu- (350 CE) during the Roman Warm Period asso- of quartz grains typical of the aeolian-derived vial activity with an increase in high-magnitude ciated with warmer, drier conditions across the Triassic sands in the Teme catchment. Particle events. The accumulation rate increases markedly United Kingdom, Ireland, and Europe (Charman size fractions are taken to reflect flow veloci- alongside more numerous peaks in MS, higher et al., 2006; Swindles et al., 2013; Wilson et al., ties delivered to the floodplain and would reflect sand content, and increases in pOSL and elemen- 2013; Büntgen et al., 2011; Fig. 4A). In contrast, flood velocity if the location of the channel were tal indicators. This clear change corresponds with from 1500 to 1000 yr B.P. (400–1100 CE), the constant relative to the site. The power func- the change to warmer, wetter conditions typical sedimentary models from the upstream profiles

tion combines the coarse sediment fraction (D90, of the Medieval Climate Anomaly alongside at Broadwas and Buildwas demonstrate much- coarsest grain size percentile), a proxy for flow increased land-use intensity on the floodplain. coarser sediment deposition than at the Severn- velocity, with organic matter percentage (an At this time, arable cultivation covered 30%–50% Teme confluence, suggesting that the variable inverse function of sedimentation rate). Mea- of the total land area in those English counties climatic conditions of the Dark Age Cold Period surements of mineralogy were undertaken using within the Severn-Teme catchment (Broadberry and Medieval Climate Anomaly had more local- the ITRAX XRF Scanner (Croudace et al., 2006) et al., 2015; Rippon et al., 2015) and beyond (Hey ized effects upon the fluvial activity in the upper to calculate log Zr:Rb ratios. This provides a 2004), corresponding to peak medieval popula- reaches of these river systems (Fig. 4B). The measure of the weathered nature of the sediment tion levels in midland England of >10 persons/ onset of significantly coarser sedimentation in and the relative contribution of detrital grains km2 (Goldewijk et al., 2010). the Severn-Teme confluence from 900 yr B.P. versus clay minerals. From 600 to 400 yr B.P. (1350–1550 CE), (1000 CE), which accelerated after 400 yr B.P. In order to refine the depositional history river conditions appear to have calmed, and a (1550 CE), is mirrored in the other sequences and determine flood events, pOSL analysis sharp decrease in accumulation rate alongside albeit with longer phases of finer sediment depo- was conducted (Sanderson and Murphy, 2010). increase in carbonate and reduction in other sition, and can be associated with climatic vari- The technique has proven success in fluvial proxies suggest fewer high-energy depositional ability during the Little Ice Age (Wanner et al., contexts (Muñoz-Salinas et al., 2010) (Section events. After 400 yr B.P. (1550 CE) and con- 2008; Phipps et al., 2013), leading to prolonged S10), and demonstrates that sudden changes and tinuing through to the present, there is another phases of wetter conditions across the United particularly peaks in accumulated pulsed infra- rise in accumulation rate alongside a significant Kingdom and Ireland (Charman et al., 2006; red stimulated luminescence (pIRSL) dose that increase in grain size to coarse silt, consider- Swindles et al., 2013; Wilson et al. 2013) and diverge from general trends identify sources ably higher sand content, and major increases cooler European climates (Esper et al., 2014; of sediment that are relatively unbleached and in pOSL and elemental indicators. Together Fig. 4C), as well as continued intensification of

Geological Society of America | GEOLOGY | Volume 48 | Number 8 | www.gsapubs.org 821

Downloaded from http://pubs.geoscienceworld.org/gsa/geology/article-pdf/48/8/819/5095985/819.pdf by guest on 25 September 2021 20 01 10 1 2 3

Clas s Clas s Clas s (Dissimilarity)

AHC Pruning line = 16 dendrogram

853) (1 02 886)

2 1 8 7 6 5 4 3 9 18 17 16 15 14 13 12 11 10

f, g Zone 1852 , (1

er 2004 ) of e 007 st PO W- PO W- PO W- PO W- PO W- PO W- PO W- 4 PO W- PO W-

PO W- PO W- PO W- PO W- PO W- PO W- PO W- PO W- PO W-

Ga Higher ,

ster & 2 3. 1

1886 a, c tcps) gi 995) 975) ( 1895) and (1 Re b (1 a 2. 9 1795 a,b, c

ll 1947-48

2000 1852 a,b, d

As, Sr tt 23 ri

Metals ces : al

58 83 76 ha (Log average average (Log n

.7 RC rr io

ma ut nu 1236 a Heavy Heavy 12 1770 c 1672 14 1526 15 Middle River Severn fe re

Cd, Sb, Ba, Pb Lower Cu, Zn, 2. 52 Recorded floods/yr (CE) 01 e - NE Re a - Da b - Ma c - So d - An

f - Marsh and Dale (2002) g - Marsh and Hannaford (2007)

Higher

) 853)

(x10¯¹) 007 (2

886) Log Zr:Fe Log rd (1 1852,

fo 11

er of Lower 4 st

nna

5- me

ster 1886 c, e Coarser ¯¹) Ha 88 7) Te gi 1795 b

(1 ( 1895)

and (x10 Re 1852 a,b,c,d ( 1995) ll and t es: 23 ri al

h ot ha 13

nc Log Zr:Rb Log 1236 a 1483 1526 a River rs ma 1615 1947 rri 1770 b ut 2007 f

nnu Finer

fe re ) -1 4 Re 01 a - Da b - So e - Symons f - Ma d - A c - Ma Recorded floods/yr (CE) Historic flood records (CBHE; Black and La w, 23

, 2007 )

per cycle (x10 cycle per otal counts counts otal

T Net pIRSL Net Severn 01

Net pIRS L Replicate s Agency

5³

) Milward (2005)

D /

(

power index power Sediment Sediment

)

D (mm)

( fraction

eme confluenc e verage coarse verage A

) C - Prehistoric and Romano-British settlement (Horizontal scale 2m, vertical 1m )

) ) )

% 20 (

µm

5 m-2 mm) m-2 -1 mm )

20cm Orthophotography (UK Environment 10

µ (63

51 Sand fractions Sand

vFS (63-125 µm FS (125-250 µm MS (250-500 ) CS (500 µm me

Powick

(% Te

m-2 mm) m-2 B - Powick medieval bridge with modern behind

A - 2007 flood in the Severn-T 00

otal sand otal (63 µ (63 i T 02 g j e h e (B ) e e 01

e Fine

)

φ 67 ( f e (B )

Silt Medium

Fine particulates Fine (10th century) d

Coarse

) 45 ¹ i - Damari (1995 )

h - Cook (1996 ) j - Marriott and Gastor (1886 ) g - Edwards and Cook (2000) f - Atkin (1998, 2004)

3 e (B )

mm/yr

( Abbey in ca.1040s f rate 2.5 Abbey by Edgar in 972 e

Poincguuic 2

accumulation 2. 2 Sediment Projected 4.3 eme confluence environs (CE) eme ca. late 13th cent.

1.5

) ¹ bridges of boats 1651 estminster

Bacon v. OxCal v.

‾ ‾ 200 mg³kg

x10 ( rceste r,

100

20 yr running mean susceptibility Magnetic Magnetic ces: n fe re 20

e - Page and Willis-Bund (1924) a - Milward (2005) b - Rogers (2014) c - Vaughan and Wainwright (2012) d - Mawer and Stenton (1927) 950 °C) 950 Powick first recorded as Re Bransford-Henwick railway bridge destroyed 1886 2 mills at Powick, 45 ploughs and meadow 1086 Powick given to Pershore Battle of Wo Major flood of 2007 (A) Powick given to W Leigh and Laughern Brooks bridges destroyed 1852 Major ironworkings, slitting mill, and forge ca.1725 Powick given to Great Malvern Priory c.12th cent. Battle of Powick bridge 1642 Prehistoric and Romano-British settlement on terrace gravels (see C and Fig. 1) a,b,c Cut mill on Laughern Brook demolished ca.1760s Third mill built on the Powick water corn mill in decay 1626 The Black Death 1348-50 Repairs to Powick medieval bridge c.1447 Powick bridge in poor condition 1598, 1604, 1633 Earliest record of Powick leat c.1475

New Powick Bridge built 1837

.5 (% LOI at LOI (% Carbonate Historic events in Severn-T

550 550

(% LOI at LOI (% Organics 26 15

154 CE ) (%)

content 154-1485 CE) Moisture Moisture Presen t 05 Age (100 BCE-43CE)

Subdivisions Stuart, and Hanove r

(1485-1750 CE ) (CE) Age 50 0 Anglo-Saxon (660-899 CE ) 25 0 20 0 600 500 900 700 550 450 400 350 100 650 750 300 150 80 0 950 850

1800 1750 1700 1650 1550 1450 1350 1250 1950 2000 1900 1850 1600 1500 1400 1300 1150 1100 1200 1050 1000 Anglo-Saxon (899-1066 CE) Anglo-Saxon (560-660 CE )

udo r,

Norman (1066-1

) P. 3rd century (200-300 CE ) T ictorian (CE1837-1900 CE ) 0 Plantagenet (1 V 20th century (1900-1999 CE )

Migration period (410-560 CE ) 20 0 40 0 50 0 70 0 90 0 10 0 30 0 60 0 80 0 Late Iron Late

(yr B. (yr 1600 1700 1900 1200 1300 1400 1800 1000 1100 1500 Early Age Middle Roman military phase (43-100 CE ) Late Romano-British (300-410 CE ) Romano-British expansion (100-200 CE) Ind. and agric. revolution (1750-1840 CE ) 2001 0.04k a 0.03k a 0.04k a 0.06k a 0.06k a 0.09k a 0.10k a 0.09k a dates 1501 1 1.52ka ± OSL Modern Age (

0.15 0.10 (2 σ ) 95.4% 0.05 (1 σ ) 68.2%

1) (red) 0 Density Accumulation rate (yr/cm) Accumulation shape: 1.5 Accumulation mean: 5

OxCal. v (yellow) Bacon v. (201 (cm) 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 50 0

10 20 30 40 50 60 70 80 90 Depth 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 Coarser grained horizon Finer grained horizo n Anomaly ca. 900-1300 CE A B ca. 536 CE even t Climatic phase D C – D Stratigraphic units

rm Period ca. 250 BCE-400 CE

A Horizons lf Minimum ca. 1289-1350 CE Age Cold Period ca. 400-765 CE Oort Minimum ca. 1040-1080 CE Wo

Dalton Minimum ca. 1790-1820 CE Spörer Minimum ca. 1460-1550 CE image Maunder Minimum ca. 1645-1715 CE Dark

Roman Wa Medieval Climate Stratigraphic Stratigraphic

822 www.gsapubs.org | Volume 48 | Number 8 | GEOLOGY | Geological Society of America

Downloaded from http://pubs.geoscienceworld.org/gsa/geology/article-pdf/48/8/819/5095985/819.pdf by guest on 25 September 2021 Figure 2. Stratigraphic image with interpreted horizons and stratigraphic units; Age-depth model using OxCal. v.4.3.2 (yellow) (Bronk Ramsey, 2017), and Bacon v.2.2 (red) (Blaauw and Christen, 2011) of each optically stimulated luminescence (OSL) date with entire date range (black horizontal line), and calculated statistical ranges at 1σ (68.2%) and 2σ (95.4%) variation (black brackets). Purple and grey shading demonstrates statistical range across sequence at 1σ and 2σ. The density-accumulation rate graph of the Bacon model demonstrates an overall uniform rate of deposition, with a mean accumulation of 5 yr/cm. However, more detailed variation in sediment accumulation rate is also presented. Both the OxCal and Bacon methods show a similar overall pattern but more discrete variation is illustrated in the Bacon model. Also shown are loss on ignition (LOI) for moisture (at 105 °C for 12 h), organics (at 550 °C for 2 h) and, carbonate content (at 950 °C for 4 h); magnetic susceptibility; fine and coarse sediment textural characteristics, defined by particle size analysis (vFS—very fine sand; FS—fine sand; MS—

medium sand; CS—coarse sand; D90 —coarsest sediment fraction); sediment power index, calculated as coarse fraction divided by the cubed 3 proportion of fine fraction (D90 /D10 ); sediment texture also defined by portable optically stimulated luminescence (pOSL) as shown by net pulsed infra-red stimulated luminescence (pIRSL); ITRAX XRF log Zr:Rb (finer to coarser); ;og Zr:Fe (lower to higher flood event indicator), and ;og average total counts per second (tcps) heavy metals (lower to higher) (Cu, Zn, As, Sr, Cd, Sb, Ba, Pb). The Agglomerative Hierarchical Analysis (AHC) demonstrates the statistical classification of sediment types in the sequence and their dissimilarity from each other. At Powick, UK, three clear classes have been determined (Class 3—ca. 50 BCE to 1000 CE; Class 2—ca. 1000 to 1375 CE; Class 1—ca. 1375 to present, and specifically after 1550 CE). This reflects increasing up-sequence variation in texture and grain-size from increasingly higher energy flood deposition. Vertical dashed line represents the calculated pruning line from which the cluster variation has been computed. Alongside the main classes, refined statistical zones (POW-1 to POW-18) are marked, reflecting intra-class cluster variation across the sequence. Bottom row of the figure shows mainly qualitative data sets including climatic phase (colder—light blue; warmer—yellow) with the name and period of duration. UK historical periods and subdivisions (Industrial and agricultural revolution [Ind. and agric. Revolution]); historic events within the Severn-Teme confluence; orthophotography of extent of the 2007 flood, recently matched or exceeded by 2020 flood; images of Powick medieval and modern bridges, and an example of prehistoric and Romano-British settlement evidence on the floodplain edge (location shown in Fig. 1). Final section shows the number of recorded floods per year from the Teme and Severn, with exceptional events illustrated by dates. Both models show a clear increase in larger floods after ca. 1500CE which corresponds with the period of greatest fluvial activity in the sediment model. See the Supplemental Material (see footnote 1) for methods, data, and references.

Biplot (axes F1 and F2 : 91.00 %) Log pIRSL (red light) KEY: Log magnetic 3 Log pOSL (blue light) Major chronological period susceptibility Average 50 year data point Late Modern Early medieval Iron Age Post-medieval Modern period (n = 64) period (n = 176) (n = 16) Medieval 2 Early medieval Romano-British Late Iron Age

Romano-British 1 period (n = 92) Medieval period (n = 104)

0 -6 -4 -2 04268 Finer Coarser sediment sediment

Log Zr:Rb - Log sand content F2 (18.76%) Log carbonate -1 2

is (LOI at 950 ºC) Ax 6 100 5

ue 80 -2 (%) ve ti al 4 ty

60 i l 3 la Log sand content 40 bi a mu

2 i Post-medieval genv r

Ei 20 a 1 Cu period (n = 69) Log grain size v

0 0 1243 567 -3 Components Axis 1 - Log grain size F1 (72.24%)

Figure 3. Principal component analysis (PCA) using XLSTAT software (version 2019.2.3, https://www.xlstat.com/en/) conducted across seven key proxy texture variables (red lines): log pulsed infra-red stimulated luminescence (pIRSL, red light); log pulsed optically stimulated luminescence (pOSL, blue light); log Zr:Rb; log sand; ;og grain size; log magnetic susceptibility; and log carbonate derived by loss on ignition (LOI) at 950 °C. Stratigraphic observations (n = 521) were clustered across 50 yr time spans from 20 BCE to present and placed into the biplot based upon the Eigenvalues of the two major components lLog grain size versus log % sand), and the results classed into six Late Holocene time periods (black ellipses). The biplot percentage (91%) represents the proportion of the total data used in the model and the percentages of each axis (log grain size 72.24% and log sand content 18.76%) show the proportion across the two components.

Geological Society of America | GEOLOGY | Volume 48 | Number 8 | www.gsapubs.org 823

Downloaded from http://pubs.geoscienceworld.org/gsa/geology/article-pdf/48/8/819/5095985/819.pdf by guest on 25 September 2021 YEAR (CE) The pOSL analysis was conducted using equipment Sedimentary 0 250 500 750 1000 1250 1500 1750 2000 owned by the University of Tromsø, Norway. We also sequences RWP DACP MCA O W S M D G thank the three anonymous reviewers for comments (A–C) 536CE 2 CS on the original manuscript. 1 No Data Powick, A 0 Severn/Teme -1 REFERENCES CITED CS 2 -2 FS Best, J.L., and Lane, S.N., 2004, Confluence, channel, Broadwas, 1 and river junction, in Goudie, A.S., ed., Encyclo- 0 B River Teme pedia of Geomorphology: New York, Routledge, -1 CS 2 -2 p. 180–183. FS 1 Buildwas, Blaauw, M., and Christen, J.A., 2011, Flexible pal- C 0 River Severn aeoclimate age-depth models using an autore- -1 -2 gressive gamma process: Bayesian Analysis, v. 6, FS p. 457–474, https://doi.org/10.1214/11-BA618. Black, A.R., and Law, F.M., 2004, Development and a.. b.b c.c Climatic Data HWT utilization of a national web-based chronology of (D–I) 1 hydrological events: Hydrological Sciences Jour- Charman nal, v. 49, p. 237–246, https://doi.org/10.1623/ D 0 et al. (2006) hysj.49.2.237.34835 .

HWT -1 British Geological Survey, 2017, Geoindex onshore 2 LWT borehole scans for the Severn-Teme confluence, 1 Swindles SO85SW, SO85SE: Keyworth, Nottingham, UK, 0 E et al. (2013) -1 British Geological Survey, http://mapapps2.bgs. -2 HP ac.uk/geoindex/home.html. 2 LWT Broadberry, S., Campbell, B.M.S., Klein, A., Overton, 1 Wilson F No Data M., and van Leeuwen, B., 2015, British Economic 0 et al. (2013) -1 Growth, 1270–1870: Cambridge, UK, Cambridge -2 University Press, 461 p., https://doi.org/10.1017/ HP 2 LP 1 CBO9781107707603. Büntgen 0 G Bronk Ramsey, C., 2008, Deposition models for et al. (2011) -1 chronological records: Quaternary Science -2 HST LP 2 Reviews, v. 27, p. 42–60, https://doi.org/10.1016/ 1 Esper j.quascirev.2007.01.019. No Data H 0 et al. (2014) Bronk Ramsey, C., 2009, Bayesian analysis of radio- -1 carbon dates: Radiocarbon, v. 51, p. 337–360, ISA 3 -2 LST 2 https://doi.org/10.1017/S0033822200033865. Phipps 1 Brown, A.G., 1985, Traditional and multivariate tech- 0 No Data I et al. (2013) -1 niques in the interpretation of floodplain sediment -2 -3 grain size variations: Earth Surface Processes DSA 2000 1750 1500 1250 1000 750 500 250 0 and Landforms, v. 10, p. 281–291, https://doi Legend: YEAR (B.P.) .org/10.1002/esp.3290100310. Climatic periods: Coarser sediment Brown, A.G., 1987, Holocene floodplain sedimen- CS and FS - Coarser and finer sedimentation RWP - Roman Warm Period; 536CE - 536CE event; Finer sediment tation and channel response of the lower River Wetter HWT and LWT - Higher and lower water table DACP - Dark Age Cold Period; O - Oort Minimum; Dryer HP and LP - Higher and lower precipitation MCA - Medieval Climate Anomaly; W - Wolf Minimum; Severn, U.K.: Zeitscrift für Geomorphologie, Cooler HST and LST - Higher and lower summer temperatures S - Spörer Minimum; M - Maunder Minimum; v. 31, p. 293–310. Warmer ISA and DSA - Increased and decreased solar activity D - Dalton Minimum; G - Gleissberg Minimum Higher solar activity Brown, A.G., 2009, Colluvial and alluvial response All raw data have been mean normalized Lower solar activity to land use change in Midland England: An inte- with a 20-year smoothed average grated geoarchaeological approach: Geomorphol- Figure 4. Sediment deposition models from the River Severn–River Teme confluence (United ogy, v. 108, p. 92–106, https://doi.org/10.1016/ Kingdom) at Powick (A) with comparative models from sites upstream at Broadwas (Teme) (B) j.geomorph.2007.12.021. and Buildwas (Severn) (C), classified into three phases: (a) 20–400 CE; (b) 400–1100 CE; (c) Brown, A., Toms, P., Carey, C., and Rhodes, E., 2013, 1100 CE–present. Also shown for comparison are upland peatland water table depth and proxy Geomorphology of the Anthropocene: Time- wetter-drier conditions in United Kingdom (D) and Ireland (E); precipitation in southern and transgressive discontinuities of human-induced central England (F) and central and southern Europe (G); summer temperatures in northern alluviation: Anthropocene, v. 1, p. 3–13, https:// Europe (H); and solar activity variation (I). doi.org/10.1016/j.ancene.2013.06.002. Büntgen, U., et al., 2011, 2500 years of European climate variability and human susceptibility: Sci- ence, v. 331, p. 578–582, https://doi.org/10.1126/ land use and sediment erosion across the conflu- tinct variation in depositionary conditions in science.1197175. ence landscape. the late Holocene, with phases of low- and Camporeale, C., Perona, P., Porporato, A., and Ridolfi, high-magnitude events associated with chang- L., 2007, Hierarchy of models for meandering rivers and related morphodynamic processes: CONCLUSIONS ing land-use intensity in the surrounding local Reviews of Geophysics, v. 45, RG1001, https:// A combination of detailed topographic landscape in the medieval, post-medieval, and doi.org/10.1029/2005RG000185. and sedimentological mapping, OSL dating, modern periods, and exacerbated by variable Charman, D.J., Blundell, A., Chiverrell, R.C., Hen- and stratigraphic modeling show that the river climatic drivers, especially during the Dark don, D., and Langdon, P.G., 2006, Compilation channels forming the Severn-Teme confluence Age Cold Period, Medieval Climate Anomaly, of non-annually resolved Holocene proxy climate records: Stacked Holocene peatland palaeo-water zone have been almost completely fixed for and Little Ice Age. table reconstructions from northern Britain: the past 2000 yr. This has produced the con- Quaternary Science Reviews, v. 25, p. 336–350, tinuous sedimentological record of levee and/ ACKNOWLEDGMENTS https://doi.org/10.1016/ j.quascirev.2005.05.005 . or overbank flooding. High-resolution OSL This work was undertaken as part of the Lever- Croudace, I.W., Rindby, A., and Rothwell, R.G., hulme Trust–funded “Flood and Flow” project 2006, ITRAX: Description and evaluation of a dating of the sedimentary sequence has pro- (RPG-2016-004). We thank the landowners for access, new multi-function X-ray core scanner, in Roth- duced an alluvial flood record for the last two the British Geological Survey for borehole records, well, R.G., ed., New Techniques in Sediment millennia. The sedimentary record shows dis- and the UK Environment Agency for lidar data. Core Analysis: Geological Society of London

824 www.gsapubs.org | Volume 48 | Number 8 | GEOLOGY | Geological Society of America

Downloaded from http://pubs.geoscienceworld.org/gsa/geology/article-pdf/48/8/819/5095985/819.pdf by guest on 25 September 2021 Special Publication 267, p. 51–63, https://doi Marsh, T., Kirby, C., Muchan, K., Barker, L., Hen- Shotton, F.W., 1978, Archaeological inferences from .org/10.1144/GSL.SP.2006.267.01.04. derson, E., and Hannaford, J., 2016, The winter the study of alluvium in the lower Severn-Avon Dixon, S.J., Sambrook Smith, G.H., Best, J.L., floods of 2015/2016 in the UK—A review: Wall- valleys, in Limbrey, S., and Evans, J.G., eds, The Nicholas, A.P., Bull, J.M., Vardy, M.E., Sarker, ingford, UK, Centre for Ecology & Hydrology, Effect of Man on the Landscape: The Lowland M.H., and Goodbred, S., 2018, The planform 37 p. Zone: Council for British Archaeology Research mobility of river channel confluences: Insights Muñoz-Salinas, E., Bishop, P., Sanderson, D.C.W., Report 21, p. 27–32. from remotely sensed imagery: Earth-Science and Zamorano, J.-J., 2010, Interpreting lumines- Swindles, G.T., et al., 2013, Centennial-scale climate Reviews, v. 176, p. 1–18, https://doi.org/10.1016/ cence data from a portable OSL reader: Three change in Ireland during the Holocene: Earth- j.earscirev.2017.09.009. case studies in fluvial settings: Earth Surface Pro- Science Reviews, v. 126, p. 300–320, https://doi Esper, J., Düthorn, E., Krusic, P.J., Timonen, M., and cesses and Landforms, v. 36, p. 651–660, https:// .org/10.1016/j.earscirev.2013.08.012. Büntgen, U., 2014, Northern European summer doi.org/10.1002/esp.2084. UK Environment Agency, 2016, Raw lidar data, 1 m temperature variations over the Common Era Pears, B., Brown, A.G., Carroll, J., Toms, P., Wood, J., DTM (2016) for SO85SW, SO85SE: Department from integrated tree-ring density records: Journal and Jones, R., 2020, Early medieval place-names for Environment, Food and Rural Affairs, HMSO, of Quaternary Science, v. 29, p. 487–494, https:// and riverine flood histories: A new approach and https://environment.data.gov.uk/DefraDataDown doi.org/10.1002/jqs.2726. new chronostratigraphic records for three English load/?Mode=survey. Goldewijk, K. K. Beusen, A., and Janssen, P., 2010, rivers: European Journal of Archaeology, p. 1–25, UK Ordnance Survey, 1886,UK Ordnance Survey Long-term dynamic modeling of global popula- https://doi.org/10.1017/eaa.2019.72. County Series 1st Edition, 1886, [TIFF geospatial tion and built-up area in a spatially explicit way: Phipps, S.J., McGregor, H.V., Gergis, J., Gallant, data], scale 1:2500, Tiles: worc-so8149-1,worc- HYDE 3.1: The Holocene, v. 20, p. 565–573, A.J.E., Neukom, R., Stevenson, S., Ackerley, so8150-1,worc-so8151-1,worc-so8152-1,worc- https://doi.org/10.1177/0959683609356587. D., Brown, J.R., Fischer, M.J., and van Ommen, so8249-1,worc-so8250-1,worc-so8251-1,worc- Hey, G., 2004, Yarnton: Saxon and Medieval Settle- T.D., 2013, Paleoclimate data-model compari- so8252-1,worc-so8349-1,worc-so8350-1,worc- ment and Landscape—Results of Excavations son and the role of climate forcings over the so8351-1,worc-so8352-1,worc-so8449-1,worc- 1990–96: Oxford Archaeology Thames Valley past 1500 years: Journal of Climate, v. 26, so8450-1,worc-so8451-1,worc-so8452-1. Landscapes Monograph 20, 465 p. p. 6915–6936, https://doi.org/10.1175/JCLI- Updated 30 November 2010 (downloaded from Lamb, H.H., 1972, Climate: Present, Past and Future: D-12-00108.1. EDINA Historic Digimap Service, http://digi- London, Routledge, 613 p. Rippon, S., Smart, C., and Pears, B., 2015, The Fields map.edina.ac.uk, on 2017-07-20 17:40:25.966). Longfield, S.A., Faulkner, D., Kjeldsen, T.R., Macklin, of Britannia: Oxford, UK, Oxford University UK Ordnance Survey, 2016, Raw lidar data, 50 m M.G., Jones, A.F., Foulds, S.A., Brewer, P.A., Press, 416 p. DTM for grid squares SO85SW, SO85SE, and Griffiths, H.M., 2018, Incorporating sedi- Robinson, M.A., and Lambrick, G.H., 1984, Holocene 1:50,000, https://digimap.edina.ac.uk/roam/ mentological data in UK flood frequency estima- alluviation and hydrology in the upper Thames download/os (extracted and downloaded from tion: Journal of Flood Risk Management, v. 12, basin: Nature, v. 308, p. 809–814, https://doi Edina digimap [2017-07-20]). e12449, https://doi.org/10.1111/jfr3.12449. .org/10.1038/308809a0. Wanner, H., et al., 2008, Mid- to Late Holocene Macdonald, N., and Sangster, H., 2017, High-magnitude Rumsby, B.T., and Macklin, M.G., 1996, River climate change: An overview: Quaternary Sci- flooding across Britain since AD 1750: Hydrology response to the last neoglacial (the ‘Little Ice ence Reviews, v. 27, p. 1791–1828, https://doi and Earth System Sciences, v. 21, p. 1631–1650, Age’) in northern, western and central Europe, .org/10.1016/j.quascirev.2008.06.013. https://doi.org/10.5194/hess-21-1631-2017 . in Branson, J., et al., eds., Global Continen- Williamson, T., 2013, Environment, Society and Macklin, M.G., Jones, A.F., and Lewin, J., 2010, tal Changes: The Context of Palaeohydrology: Landscape in Early Medieval England: Time River response to rapid Holocene environmen- Geological Society of London Special Publica- and Topography: Woodbridge, UK, The Boydell tal change: Evidence and explanation in Brit- tion 115, p. 217–233, https://doi.org/10.1144/ Press, Anglo-Saxon Studies, v. 19, 280 p. ish catchments: Quaternary Science Reviews, GSL.SP.1996.115.01.17. Wilson, R., Miles, D.., Loader, N.J., Melvin, T., Cun- v. 29, p. 1555–1576, https://doi.org/10.1016/ Sanderson, D.C.W., and Murphy, S., 2010, Using ningham, L., Cooper, R., and Briffa, K., 2013, j.quascirev.2009.06.010. simple portable OSL measurements and labo- A millennial long March–July precipitation Macklin, M.G., Lewin, J., and Woodward, J.C., 2012, ratory characterisation to help understand com- reconstruction for southern-central England: The fluvial record of climate change: Philo- plex and heterogeneous sediment sequences for Climate Dynamics, v. 40, p. 997–1017, https:// sophical Transactions of the Royal Society A, luminescence dating: Quaternary Geochronol- doi.org/10.1007/s00382-012-1318-z. v. 370, p. 2143–2172, https://doi.org/10.1098/ ogy, v. 5, p. 299–305, https://doi.org/10.1016/ rsta.2011.0608. j.quageo.2009.02.001. Printed in USA

Geological Society of America | GEOLOGY | Volume 48 | Number 8 | www.gsapubs.org 825

Downloaded from http://pubs.geoscienceworld.org/gsa/geology/article-pdf/48/8/819/5095985/819.pdf by guest on 25 September 2021