Late quaternary evolution of the Meuse fluvial system and its sediment composition : a reconstruction based on bulk sample geochemistry and forward modelling

All fluvial systems ultimately drain into alluvial basins, where the weathering products of their upstream drainage areas accumulate over a time-span varying from 10 <sup>0</SUP>to 10 <sup>6</SUP>years. Most silted-up alluvial basins are low-gradient deltas that are densely populated, because their high fertility maintains a high agricultural potential. Global warming due to increased concentrations of greenhouse gases in the Earth's atmosphere will very likely affect the quantity and quality of both water discharges and fine-grained sediment fluxes to these alluvial basins. The long-term interplay between tectonics, climate and sea level determines the frequency, the timing, the allocation and magnitude of erosional and depositional events in fluvial systems. Erosional or depositional events are the direct consequence of changes in the quantity of discharge and sediment fluxes and thus will influence the sediment composition too. Therefore, a thorough fundamental study of the response of fluvial systems to climatic change is indispensable to understand the natural erosional and depositional dynamics and to assess possible future changes in the bulk geochemical composition of sedimentary sequences. In this respect, the expression of past climate changes in the composition of fluvial sedimentary records may serve to predict future fluvial response to climate change: the past as a key to the future.This thesis presents the research results for a case study of the River Meuse that combines geomorphological, bulk geochemical and forward modelling techniques. The study focuses on the impact of climate change on the natural composition of clastic river sediments, <em>in casu</em> variations in the bulk geochemical composition of fine-grained residual channel infillings on a temporal scale of 10 <sup>3</SUP>-10 <sup>5</SUP>years. The main topic was split up into two main parts, each dealing with individual queries. <strong>Part I</strong> addresses the first goal of the research, namely to determine which fluvial sediments are likely to register a palaeoclimatic signal, where these sediments are found and how the climatic signal is expressed in these sediments. Earlier work directed the focus towards fine-grained sediments, because the clay and fine-silt size fractions are the most likely candidates to register evidence of changing weathering and transport pathways. Chapter 2, 3 and 4 describe the results of geomorphological fieldwork and subsequent laboratory analyses by zooming in on fine-grained deposits in the Meuse lower reach, which date to the 15-0 ka BP time-frame.<strong>Chapter 2</strong> describes the general geological setting of the study area that forms part of the Meuse lower reach in North Limburg (southern Netherlands). The response of the River Meuse to Late Glacial climate change was reconstructed to provide insight into the origin and age of the residual channels and their infillings. Ice-cores drilled within the framework of the Greenland Ice Core Project (GRIP) indicate frequent, profound and abrupt climatic change over the last 250,000 years Before Present (250-ka BP) in the North Atlantic region and Northwest Europe. The Late Glacial part of the GRIP ice-core demonstrated that mean annual temperature rose dramatically at the beginning of the Late Glacial (<img src="/wda/abstracts/ongeveer_enkel.gif"/>14.5-ka BP; 5-°C) and in the Early Holocene (<img src="/wda/abstracts/ongeveer_enkel.gif"/>10.2-ka BP; 7-°C). Quantitative climate reconstructions based on palynological and coleopteran data for the upstream Ardennes (where the major part of the River Meuse discharge and sediment load originates) and in the Netherlands corroborated these temperature rises. The Ardennes data also indicated a fourfold increase in mean annual precipitation. The rain-fed River Meuse is very sensitive to temperature and precipitation changes and accordingly responded by lowering its Late-glacial floodplain: glacially aggraded deposits were incised and several previously active channels were abandoned and turned into low-energy depositional environments. Subsequent flooding events left their traces in the form of a discontinuous clayey and silty sedimentary record within the residual channels. Periods of sediment by-pass led to gyttja and peat accumulation. This offered excellent opportunities to date the residual channels and the intercalated clastic sediments, because the organic material has been formed amply within the dateable <sup>14</SUP>C-range.Extensive conventional <sup>14</SUP>C-dating of strongly organic gyttja and peat intervals in the residual channels permitted the reconstruction of two Late-glacial phases of rapid vertical channel downcutting, preceding lateral river valley degradation. The first phase dated between<img src="/wda/abstracts/ongeveer_enkel.gif"/>13.3 and 12.5-ka BP (Early Bølling) and the second one between<img src="/wda/abstracts/ongeveer_enkel.gif"/>10.2 and 9.8-ka BP (Early Preboreal). The onset of channel downcutting lagged the climatic amelioration in the Ardennes some 500-1300 years, depending on initial landscape conditions. This suggests a major influence of interstadial and interglacial vegetation growth on landscape fixation and stabilisation and soil development, resulting in a decrease of the sediment supply relative to simultaneously increasing river discharges and hence channel downcutting. Following the downcutting phases, meandering river patterns suggest low-energy river dynamics during the Late-glacial Interstadial (Late Bølling, Allerød) and the Late Preboreal. Simultaneously, increasing landscape stability contributed to a gradual fining of the sediments. Renewed climatic deterioration around 11.3-ka <sup>14</SUP>C BP (Late Allerød) and during the severe Younger Dryas cooling event (10.8-10.2 ka BP) caused a short return to high-energetic braided river conditions. All in all, the results of Chapter 2 show that the Late-glacial geological setting of the Meuse lower reach in North Limburg provided the ideal environment for deposition and preservation of fine-grained clastic sediments.<strong>Chapter 3</strong> zooms in on a set of<img src="/wda/abstracts/ongeveer_enkel.gif"/>640 samples taken from non-polluted, largely unconsolidated, fine-grained residual channel infillings deposited in the Meuse lower reach. The <2000-µm fractions have been analysed for their granulometry (determined with a Laser Grainsizer), bulk geochemistry (using X-Ray Fluorescence Spectroscopy, XRFS) and part of the samples for their clay geochemistry and clay mineralogy (X-Ray Diffraction, XRD). The natural compositional variation in these fine-grained sediments was quantified using multivariate statistical analyses. This enabled distinguishing the palaeoclimatic signal from variation that is not directly related to climate, like post-depositional overprinting effects. Hydrodynamic sorting of minerals in different size fractions explained over 70% of the variation in sediment composition. Phyllosilicates in the clay and fine-silt size fractions hosted the major part of the major constituents Al ₂ O ₃ , TiO ₂ , K ₂ O, MgO and the trace elements Ba, Co, Ce, Cr, Ga, La, Nd, Ni, Pb, Rb, V and Zn. Early-diagenetic formation of siderite and vivianite in the strongly organic and clayey anoxic gyttja environment appeared to cause relative natural accumulations of Fe ₂ O ₃ , MnO, P ₂ O ₅ , Co, Ni and notably Zn above the phyllosilicate background values. High lime contents caused elevated contents of CaO and Sr, while Na ₂ O and Zr, Nb, Y and Nd were more or less strongly related to the occurrence of albitic feldspars and heavy minerals respectively in the coarse silt fraction. Only 13 samples out of 636 showed strong anomalies or accumulations of the trace elements Pb, Zn, Ni and Co, which in high concentrations can pose a potential threat to the environment. This confirmed the assumption that most samples did not suffer from post-depositional anthropogenic pollution. However, bio-accumulation and very early atmospheric pollution due to small-scale ore-mining and smelting in the Roman era might explain the elevated Pb-contents found in Subatlantic clays.<strong>Chapter 4</strong> deals with the same set of samples <strong></strong> and confirms theoretical considerations that the composition of fine-grained clastic sediments does not remain constant over a period of 10 <sup>3</SUP>-10 <sup>4</SUP>years. The <sup>14</SUP>C-dating of frequent-occurring organic intervals and additional palynological information enabled time labelling of samples taken from intercalated clastic layers. Bivariate scatterplots showed that the Pleniglacial, Late Glacial and Holocene sample groups differ considerably in their clay contents and in their contents of several main and trace constituents. Firstly, Holocene samples were found to have significantly higher clay contents, suggesting higher clay mineral supply. Secondly, Holocene samples contained more Al ₂ O ₃ and less K ₂ O, MgO and TiO ₂ relative to Pleniglacial and Late-glacial specimens within a comparable granulometrical range.The typically clay-related trace elements Ba, Cr, Rb and V showed similar chronological differentiation as for the main constituents, namely lower Holocene values relative to Al ₂ O ₃ .  However, these trace elements have higher ratios in Holocene samples relative to K ₂ O and MgO, owing to relative depletion of the latter constituents. Detailed clay mineralogical analysis of separated clay fractions and clay mineral weathering literature strongly suggested that this systematic shift in sediment composition could be ascribed to both an absolute and relative increase of the smectite and vermiculite contents and interstratifications of these minerals with illite in Late-glacial Interstadial and Holocene sediments. Because overprinting effects owing to post-depositional soil formation and anthropogenic effects could be excluded, the changes in detrital clay mineralogy have been interpreted as a systematic sedimentary palaeoclimatic signal.Climatic amelioration and increasing landscape stability during prolonged interstadials and interglacials increased the weathering intensity (rate) of phyllosilicates and lengthened weathering duration. Widespread soil formation on Palaeozoic metapelitic rocks in the Ardennes low mountain range as well as on loess deposits has most likely caused the clays were progressively depleted of the main constituents K ₂ O, MgO and TiO ₂ relative to Al ₂ O ₃ and Ba, Cr, Ga, Rb and V (Chapter 4). The resulting higher supply of the typically pedogenetic high-Al, low-K and low-Mg smectites and vermiculites ultimately constituted a palaeoclimatic signal in the clay and fine silt size fractions of Meuse sediments. Several early diagenetic post-depositional processes favour the formation of authigenic minerals in the gyttja-redox environment (siderite and vivianite). They exclude the use of the following major and minor constituents for reconstructing long-term detrital compositional changes: Fe ₂ O ₃ , MnO and P ₂ O ₅ . In the same manner, variations in heavy mineral and lime content exclude the use of Nb, Y, Zr and the CaO-Sr pair respectively.<strong>Part II</strong> encompasses the second subgoal of the research, namely to answer the questions why the sediments one is interested in have been laid down there in the first place, how they relate to long-term and large-scale fluvial dynamics and how sediment compositional changes relate to internal fluvial dynamics. Sediment composition is namely directly related to the sediment flux, which itself depends on the evolution of the fluvial system. Therefore, the long-term river dynamics had to be quantified at the spatial scale of the whole drainage basin to account for other factors than climate and to get a grip on long-term compositional changes. A forward modelling study in Chapter 5 serves to understand the 15-0 ka BP lower reach results of Chapter 2-4 within its fluvio-systematic context. This context concerns the influence of external forcing on the long-term evolution of the Meuse longitudinal profile on the spatial scale of the whole drainage basin. Chapter 6 attempts to give a finishing touch by combining sediment flux calculations with bulk geochemical data to provide insight in the long-term evolution of fluvial sediment composition.<strong>Chapter 5</strong> contains the interpretations concerning long-term fluvial dynamics resulting from sediment flux calculations in a well-calibrated semi 3-D forward modelling study. The longitudinal profile development of the River Meuse was simulated in response to changes in the external forcing variables tectonics, climate and sea level for the time-span 250-0 ka BP. The modelling results showed that a scenario of climate-controlled discharge and hillslope sediment supply (interstadial or interglacial increasing discharges and simultaneously decreasing hillslope sediment supply) is able to reproduce a phase of river valley degradation at the start of interglacial periods. The incisional phase is identical to the observed phase of Late-glacial incision in the Meuse lower reach, which set the favourable conditions for preservation of fine-grained sediments in the resulting residual channels (Ch. 2). This suggests that the followed forward modelling approach adequately simulates the long-term evolution of the Meuse fluvial system.The Profile Evolution Map visualises the long-term evolution of sediment fluxes and demonstrates the timing and allocation of erosional and depositional phases along the longitudinal profile. The downstream positions of sections along the longitudinal profile strongly determine how they respond to time-equivalent changes in the external forcing variables. The fact that the upstream and downstream sections appeared to mutually influence each other clearly indicated complex-response in the fluvial system. Tectonic uplift and climate change could be demonstrated to dominate fluvial response in the North French and Ardennes upper and middle reaches. Here, the model predicts erosional phases to be most pronounced during prolonged interstadials and interglacials, leading to climate-controlled river valley incision and degradation of the longitudinal profile. On the other hand, continuing tectonic subsidence within the Roer Valley Graben and the southern North Sea Basin and eustatic sea-level changes dominated the response in the lowermost reaches. Here, continuous deposition takes place, interrupted by an incisional phase at the beginning of interglacials. The subsequent rising Eemian and Holocene sea levels caused increased sediment aggradation, leading to gradient backfilling. This process generated a depositional wedge that protruded progressively upstream and shifted the terrace intersection land-inward over some 100-150 km.In <strong>Chapter 6</strong> , the bulk geochemical data of various upstream Ardennes tributary catchments have been coupled to the sediment supply arising from climate-controlled hillslope processes and from internal valley-erosion. The integration of bulk geochemical data with calculated sediment fluxes originating from the forward modelling study allowed simulating the effects of changes in the external forcing variables on sediment composition in the fluvial system. The Geochemical Evolution Map visualises the long-term evolution of sediment composition along the longitudinal profile. We modelled a scenario of changing weathering duration, namely alternating glacial weathering-limited (short weathering duration) and interglacial transport-limited sediment supply (long weathering duration. This scenario performed well in simulating the timing and direction of changes in the long-term sediment composition. Furthermore, the simulated changes in sediment composition were of the same order of magnitude as the measured changes recorded in the fine-grained sediments of the Meuse lower reach. However, a discrepancy with absolute values indicated that the direct effects of increased weathering intensity could not be excluded, especially in prolonged interstadial or interglacial periods with transport-limited sediment supply. Furthermore, circumstantial evidence indicated that loess influxes might play an important role as well, but these have yet to be quantified.