Holocene floodplain deposition and scale effects in a typical European upland catchment: A case study from the Amblève catchment,Ardennes (Belgium)

Abstract

This study quantifies Holocene alluvial sediment deposition in the Belgian Amblève catchment (1080 km²), situated in the Ardennes uplands. An extended coring data set is used for a quantitative description and a quantification of the alluvial depositions. The floodplains fall into three main types: the upper and lower floodplains and the steep reaches. Total Holocene alluvial sediment deposition amounts to 32 Tg, or 0.029 Tg/km². Dating of sediments using iron slag as tracer shows that on average 42% of contemporary storage was accumulated during the last 600 years. Radiocarbon dating of fluvial deposits at eight sites in the catchment confirms that the majority of the sediment is relatively recent. The increased sedimentation rates for this period are related to anthropogenic land use, possibly enhanced by climatic variations. A fluvial sediment budget was constructed for this 600 yr time period and shows that export from the catchment has about the same importance as storage in the floodplains, while lateral reworking of existing floodplain deposits only affects half the quantity of sediments. Overall, floodplain sediment storage in the Amblève catchment, comparable with other upland catchments, is of lesser importance compared with catchments dominated by loess. This can mainly be explained by lower sediment supply resulting from less intense anthropogenic land use, despite the higher sediment transport efficiency through the fluvial system. Floodplain sediment storage is scale-dependent, with initially a sharp increase in catchment area-specific sediment deposition, followed by a steady decrease with increasing catchment area.

Keywords

Ardennes Belgium fluvial land use change sediment budget

Introduction

In many catchments of the temperate zone of Europe, soil erosion and sediment redistribution is the major landscape-forming process during the Holocene. Most studies emphasise the important role of anthropogenic land use in inducing these processes, although climatic variability may be superimposed as a driving factor, and major erosion and deposition phases result often from the combination of intense anthropogenic land use with periods with a higher rainfall erodibility (e.g. Dotterweich, 2008; Notebaert and Verstraeten, 2010).

Such deposition phases, traditionally studied in single-site studies, offer site-specific insight into sedimentation rates. Different studies (Notebaert et al., 2011b; Trimble, 1999) have shown how rates may differ between sites and depositional environments, and that a direct connection between erosion rates and site-specific sedimentation rates proves controversial. Therefore a catchment-wide approach, which takes into account spatial variability, is more suitable to gain insight into the environmental drivers of erosion and deposition. One approach is the calculation of catchment-averaged sedimentation rates, combined with a more qualitative interpretation of catchment-wide sedimentation events, based on a statistical analysis of radiocarbon dates (e.g. Hoffmann et al., 2009; Macklin and Lewin, 2003). This approach remains controversial, however (e.g. Chiverrell et al., 2011), and depends on the availability of a large number of dates. Another approach is the construction of sediment budgets. Such sediment budgets quantify the different sources and sinks of sediment in a catchment over a given time period. As a quantitative method, sediment budgets allow objective comparison of sediment deposition between different spatial and temporal units. Moreover, when these are constructed for different time periods more insight into temporal and spatial sedimentation patterns can be gained compared with a simple compilation of site-specific sedimentation rates (e.g. Trimble, 2009). A main drawback of sediment budgets is that they require a large data set, and data are often hard to collect, especially when the studied time period extends beyond historical times. Over the last decades, an increasing number of sediment budgets have become available for small to middle sized catchments in West- and Central Europe (e.g. de Moor and Verstraeten, 2008; Förster and Wunderlich, 2009; Fuchs et al., 2011; Macaire et al., 2002).

Soil erosion will result in colluvial deposition, alluvial deposition, and sediment export towards the marine system. As such, rivers play an important role in deposition and transport of sediments, and as the link between hillslope processes and the marine system. Because rivers typically have a larger catchment area compared with colluvial dry valleys, they also reflect soil erosion dynamics for a larger catchment. In addition, Holocene river deposits are relatively easy to distinguish in the field, owing to their nature and obvious position in often inherited valleys. In regions where colluvial deposits are shallow, stony, and difficult to identify, the quantification of alluvial sediment storage is often the only part of the sediment budget which can be done. The quantification of fluvial deposits in a catchment allows an insight into Holocene erosion processes, although the connectivity between the erosion locations and the floodplains, including intermediate colluvial sediment deposition, is not taken into account. Recently, Holocene floodplain deposition has been quantified for different catchments in temperate Europe (e.g. Fuchs et al., 2011; Hoffmann et al., 2007; de Moor and Verstraeten, 2008; Notebaert et al., 2009; Seidel and Mackel, 2007; Stolz, 2011) resulting in a large variation in the catchment area-specific amount of Holocene floodplain deposits. Floodplain deposition is smaller in upland catchments (e.g. Stolz, 2011) than in the loess regions (e.g. Notebaert et al., 2009). Nevertheless, a seemingly good relationship can be found between catchment area and floodplain storage within the Rhine catchment (Hoffmann et al., 2007).

Differences between uplands and loess regions can be at least partially explained by environmental differences, as they have often a different (less intense) land use history, and also their physical catchment settings differ. Where most studies link sediment deposition to anthropogenic and/or climatic pressure, the role of the physical environment in transporting sediment and subsequent deposition on a longer timescale has been largely ignored. With the growing interest in quantification of long-term sediment and nutrient transport over large spatial scales (Hoffmann et al., 2007; Van Oost et al., 2012), there is an increasing need in large-scaled quantification, which should also include the physical environment as a predicting factor. Based on this, scaling relationships between catchment area and sediment storage (such as in Hoffmann et al., 2007) may have to be differentiated for such physical properties.

The main objective of this paper is to quantify the Holocene floodplain deposition in the Amblève catchment located in the Belgian Ardennes upland region, and to understand the relationship with the physical environment. Total Holocene floodplain deposition is quantified and compared with other catchments in temperate Europe. In addition, we aim to quantify and understand scale and along-river long-profile effects in (catchment area specific) sediment storage. Finally, floodplain deposition is dated and the results are related to known past environmental variations in land cover and climate.

Regional settings

This study focuses on the part of the Amblève catchment (French name; German name: Amel) located in the Ardennes, upstream Remouchamps (c. 1080 km²; Figure 1). The river has a total valley length of c. 70 km, and has three major tributaries: the Lienne, Salm and Warche. Soils are developed in local thin loessic deposits (<1 m), Pleistocene solifluction deposits and weathering material of the bedrock material. Shallow stony loam soils are dominant, which can be classified (Food and Agriculture Organization, 2006) as Cambisols, while locally also Histosols occur. Fluvisols occur in colluvial and fluvial valleys. The average annual rainfall spatially varies between 900 and 1400 mm. The average January temperature varies between −1.5°C and 1.5°C and the average July temperature varies between 13°C and 16°C. Just upstream at the confluence with the Salm, average discharge is 8.3 m³/s, with a peak discharge of 158.9 m³/s (median discharge: 4.9 m³/s; measuring period 1992–2012; catchment 491.6 km²; Aqualim, 2013). Just downstream of the study area, average discharge is 19.3 m³/s, with a peak discharge of 329 m³/s (median discharge: 12.68 m³/s; measuring period 1974–2011; catchment 1068 km²; Voies Hydrauliques, 2013).

Figure 1.

Overview maps of the study area. (a) Catchment topography, coring locations and major streams. (b) Main villages. (c) Situation of the catchment within Belgium. (d) Location of the different floodplain types. (e) Major geological zones (based on Rixhon and Demoulin, 2010). Coring locations of Figure 2 are indicated as REM2 (Remouchamps 2), BG (Bevercé Gorges) and BUL4 (Bullingen 4). Radiocarbon samples also come from Nidrum (NID) and Stavelot (STA) (colour figure available online).

Quaternary tectonics are responsible for the uplift of the Ardennes Massif, with a maximal uplift rate in the northeastern part of the catchment (e.g. Demoulin and Hallot, 2009), although this has not influenced fluvial processes on a Holocene timescale. The lowest terraces date from the pre-Holocene, as is evidenced by their tephra content (Juvigne, 1979).

Geology of the Amblève catchment is diverse. The eastern part as well as a small zone in the west are geologically located in the Ardennes anticlinorium, which has a heterogeneous lithology of lower and middle Devonian conglomerate, arkose, sandstone and slate. The central part is located in the Stavelot Massif with mainly quartzite and slate from Cambro-Ordovician age. Permian conglomerates occupy the Malmedy Graben in which the Warche tributary flows. A detailed description of the geology is given by Rixhon and Demoulin (2010). The Ardennes massif was strongly eroded during the Cretaceous and Cenozoic periods, resulting in peneplaine formation and (often strongly tropically weathered) Cenozoic marine deposits locally fill depressions.

The catchment has an undulating plateau topography, with altitudes of c. 350–694 m a.s.l. This plateau is deeply incised (often > 100 m altitude difference) by river valleys, with several sets of Quaternary river terraces recognizable along the steep valley edges. The upper reaches of most rivers have less deeply incised river valleys (< 50 m), and are separated from the lower reaches by a knickpoint in the river long profile. These knickpoints are located at the level where the Mid-Pleistocene terraces cross the river length profile, and as such represent the division between the upper inherited Mid-Pleistocene valleys (Younger Main Terrace (YMT) or older terraces) and the lower-Quaternary incised valleys (Demoulin et al., 2012; Rixhon and Demoulin, 2010). For the Warche two knickpoints are present: a well expressed one related to a Late-Pleistocene capture, and one further upstream related to the incision phase (see Demoulin et al., 2012, and references therein).

Contemporary land use on the plateaus is dominated by large areas of grassland mixed with some cropland fields surrounding the villages in a typical pattern of exploitation patches. Forest stands, mainly plantations, dominate the area between these exploitation patches, as well as the steep valley slopes. The highest parts of the catchment are covered with extensive moorland (Hautes Fagnes) and adjoining forests. Floodplains are mainly used for grassland and locally for cropland, while the Warche floodplain in the Malmedy graben is almost entirely built-up.

Data on historical land use are limited and point to a rather extensive land use compared with the nearby Belgian lowland regions. Pollen data from the Hautes Fagnes plateau and the Lienne subcatchment show first traces of agriculture for the end of the Atlantic Period and the Subatlantic Period, respectively, which is ascribed to Neolithic farming (Damblon, 1969; Gullentops et al., 1966). During the Iron Age and especially the Roman Period, cereal pollen became more common, but still in low concentrations (Damblon, 1969; Gullentops et al., 1966). The abbeys of Malmedy and Stavelot were erected in ad 651, and functioned as local centers of anthropogenic land occupation from the Medieval Period. Pollen data show a more prominent human influence on land cover from ad 1200 (Damblon, 1969, 1978). Iron industries settled in the valleys from the 15th century, in the study area mainly in the Lienne subcatchment (Houbrechts and Petit, 2004). These industries are associated with widespread deforestation, and the density of the iron industries suggests that a major part of the forest was regularly deforested to provide wood (Houbrechts and Weber, 2007). Overall arable land use remained rather limited, as is demonstrated by the historical de Ferraris map (ad 1774). Deciduous forests were gradually replaced by coniferous forest in the 19th century (Petit and Lambin, 2002). The combined grassland-cropland area increased during the 19th and 20th centuries, mainly before 1957, while the moorland area decreased (Petit and Lambin, 2002). A transition from cropland to grassland occurred during the 20th century (Mols, 2004), explaining the dominance of grassland in the contemporary agricultural landscape.

Methodology

River cross-sections were constructed at 55 locations, using 655 hand corings, a channel cutbank profile description for each site, and c. 20 profile pits. All corings and profile pits reach the total thickness of Holocene deposits. For each coring a detailed in-field description was made with a vertical resolution of c. 5 cm, describing sediment texture, color, quantity, nature and size of gravel, presence of plant material, peat or other inclusions, and soil horizons. For each cross-section the fluvial architecture was established (for detailed descriptions see Notebaert et al., 2011a).

Based on the detailed coring data the average Holocene floodplain sediment thickness was established for each cross-section. This paper focuses on identifying the fine sediment introduced into the fluvial system during the Holocene, and does not take into account the gravel bedload sediments. These gravels are Pleistocene deposits which are reworked by the Holocene rivers as bedload, and can easily be distinguished (Notebaert et al., 2011a). Although part of the gravel may result from Holocene soil erosion processes, this fraction cannot be differentiated in the field from the reworked Pleistocene deposits, and as such is not considered in the sediment quantifications. From these sediment thickness data, masses of deposited sediment per unit floodplain surface area (MDFSA; mass per m² of the floodplain; Mg/m²) were calculated taking into account the bulk density of the sediments and corrected for estimated gravel content. The mineral sediment mass per unit floodplain surface area for each coring (MDFSA; Mg/m²) was calculated by ( Notebaert et al., 2010; Verstraeten and Poesen, 2001):

M D F S A = \sum_{i = 1}^{j} (d_{i} (1 - C F_{i}) \frac{1}{\frac{1}{D B D_{MS}} + \frac{O M_{i}}{(1 - O M_{i}) D B D_{OM}}})

(1)

where j is the number of floodplain units in the core, d_i the thickness of floodplain unit i (m), CF_i the volume percentage of the coarse fraction in unit i, OM_i the percentage organic matter in unit i, DBD_OM the dry bulk density of the organic matter (Mg/m³) and DBD_MS the dry bulk density of the mineral sediment mass (Mg/m³). The different floodplain units are based on their fluvial architecture and are discussed in detail in Notebaert et al. (2011a). A value of 1.42 was taken for DBD_MS, consequent with previous studies (Notebaert et al., 2010). DBD_OM values vary between 0.224 t/m³ and 0.56 t/m³ (e.g. Van Asselen, 2011), depending on compaction. In this study we consider a value of 0.35 t/m³, consequent with previous studies (Rommens et al., 2006; e.g. Notebaert et al., 2010). Sediments were categorized into two categories: clastic floodplain deposits and organic floodplain deposits. The latter occur only in a few corings in the upper valleys. Values for OM were determined on 14 samples by using the loss-on-ignition method, whereas the values for dry bulk density were taken from Rommens et al. (2006) for floodplain deposits a value of 4% is used, while 8% is used for the organic deposits. For each cross-section an average MDFSA (Mg/m²) was calculated as the average of the available corings, weighted by the coring interdistance in cases of unequal coring spacing.

Floodplains were manually delineated (see Notebaertet al., 2010) using 1:40,000 scaled soil maps, 1:10,000 scaled topographical maps, detailed field observations and where available LIDAR data (1 m resolution; see Notebaert et al., 2010). Floodplains were classified in six major types based on floodplain and valley morphology, (theoretical) stream power (see below for formulae). This was further subdivided into 13 subtypes according to position in the catchment, to differentiate between the different major subcatchments. The classification is catchment-specific and is only used to extrapolate data to units which represent a homogenous morphology. For each floodplain type the average MDFSA (Mg/m²) was calculated based on the available average values for each cross-section, and applying the cross-section width as weight factor to obtain an average value (using width × mass).

Further GIS-based calculations are performed in ArcGIS. These include the calculation of catchment-specific alluvial sediment storage along the river long profile. For equally spaced data points along the stream line, the upstream catchments were calculated in the ArcHydro package using a 20 m pixel resolution DTM. Delineated catchments were manually controlled and corrected where necessary. Manually digitized floodplains were combined with MDFSA values, to provide a map of floodplain deposition. Next, for each streamline point the total upstream alluvial sediment deposition was calculated. When the results are plotted along a river long profile, the total sediment deposition is mainly a function of catchment area as the main scaling factors. Therefore sediment deposition is expressed as sediment deposition per unit upstream (catchment) surface area(SDUSA, Mg/km²):

S D U S A = \frac{S D}{A}

(2)

where SD is the total sediment deposition in the catchment (Mg), and A the catchment area (km²).

Calculation of additional floodplain attributes is based on field measurements and a GIS anaylis. Floodplain and channel width were measured in field. Floodplain slopes are calculated based on 1:10,000 scaled topographic maps with a equidistance of 5 m. Specific stream power (ω; W/m²) is calculated by:

ω = \frac{g ρ Q_{b} S}{W}

(3)

where ρ is the mass density of water (kg/m³), g is gravitational acceleration (9.81 m/s²), Q_b is the discharge at bankfull stage; S is the floodplain slope; W is the floodplain width. Normally, for S and W, respectively, the channel slope and the channel width are taken. We introduce these values as a new parameter, as we are interested in floodplain processes, related to flooding events, rather than in-channel processes. Using floodplain width as value W is a simplification, as the width of the flooding zone will depend on floodplain topography and flooding height, but nevertheless this parameter provides insight in the possible inundated width for the floodplain and associated floodplain deposition events. Values for Q_b are based on a regional scaling relationship, based on gravel-bed rivers with an impermeable substrate (Petit and Pauquet, 1997; Petit et al., 2005):

Q_{b} = 0.128 A^{0.981} (R^{2} = 0.96; n = 38)

(4)

where A is the upstream catchment area (km²).

Temporal control on sediment deposition was achieved using radiocarbon dating and dating through tracers (Notebaert et al., 2011a). Samples for radiocarbon dating were sieved, dried, and manually searched for datable material, mainly charcoal or plant remains. In addition, iron slag was used as a tracer to date floodplain sedimentation (Houbrechts and Petit, 2003, 2004; Notebaert et al., 2011a). Using such a tracer has the main advantage that insight is gathered into the (lateral) spatial variability in floodplain deposition since introduction of the tracer, and as such into the importance of lateral reworking of existing deposits and the net aggradation of the floodplain (Notebaert et al., 2011a).

Results

Floodplain typology and architecture

Floodplains can easily be grouped into four major groups (Table 1, Figure 1d): the floodplains of the Malmedy graben, the upper floodplains, the steep reaches which are associated with knickpoints, and the lower floodplains (Demoulin et al., 2012; Notebaert et al., 2011a). These are further divided into floodplain types based on floodplain slope and catchment size to differentiate between the smaller upstream floodplains, while we also differentiate between subcatchments.

Table 1.

Properties of the main floodplain types in the Amblève catchments. These types are further divided according to subcatchment.

Reach	Floodplain slope angle (%)	Sinuosity	Width–depth ratio	River bed	River banks
Upper valleys (smallest)	0.8–3.5	1.0–1.5	<10	Gravel	Silt & loam
Upper valleys	0.4–1.0	1.2–1.6	<10	Gravel	Silt & loam
Steep stretches	>1	c. 1.0	> 10	Gravel	Sand
Malmedy graben	0.4–0.6	1.0–1.2	> 10	Gravel	Silt & loam
Lower valleys	0.2–1.0	1.0–1.2	> 10	Gravel	Silt & loam

For the upper floodplains, floodplain slopes decrease downstream from c. 3.5% to 0.4%. Valleys are slightly incised (<50 m, often <20 m) in the surrounding plateau (Cenozoic erosion surface), and although variations exist, valley slopes are often gentle (<10%). These upper floodplains are located in a remnant of broad late-Pliocene valleys, slightly incised in the Cenozoic erosion surface (Demoulin et al., 2012). River channels meander slightly in the upper headwaters, but with decreasing floodplain slopes they develop a strongly meandering pattern with compound meander bends. River channels have a low width–depth ratio (<10), and have a typical pool-rifle structure (Petit et al., 2008). The upper valleys are further differentiated for the extrapolation of the sediment budget, based on catchment area and floodplain slope. Floodplains are defined in a homogenous way for the entire catchment, based on the presence of Holocene deposits and the presence of a permanent stream, and any apparent difference in drainage density network (Figure 1) is a result of upstream propagation of the tectonic erosion wave, and subsequent stripping of the landscape.

In the steep reaches (gorges) floodplains are most often absent or very small (< 20 m). These steep reaches are located at downstream regressive knickpoints, most of which are linked to the upstream propagation of tectonic-induced erosion (Demoulinet al., 2012), although also a tectonic knickpoint occurs. Channels have a large width–depth ratio (>10), with the local occurrence of steps, especially on the smaller tributaries.

The lower floodplains are deeply incised in the surrounding plateau (>100 m, most often > 200 m) with floodplain (and valley) width strongly variable and controlled by local lithology and valley form. The narrow floodplain and deeply incised valleys are the result of regressive knickpoint erosion induced by late-Quaternary increase in tectonic uplift. The subsequent base-level changes are responsible for the upstream propagation of an incision wave which cuts into the broader Mid-Pleistocene valleys (Demoulin et al., 2012). The resulting valley slopes are often steep (>10%). The valleys have an inherited meandering pattern with large meanders, but within the floodplains the rivers have a straight course. River channels are shallow (width–depth ratio >10) with a riffle-pool typology and many exposed bars.

Floodplains of the Malmedy graben are very broad, located downstream of a steep reach, up to c. 700 m, with a slightly meandering river channel. River channels are comparable with those of the lower floodplains where they are not rectified. The floodplains of the Malmedy graben are almost entirely covered with World War II debris, and a major part is also built-up today, which limits the information on the natural floodplain environment.

Although there are morphological differences between these different floodplain types, they have a comparable fluvial architecture, except for the steep reaches (Figure 2). Floodplain architecture is already described in detail in Notebaert et al. (2011a), and we consequently follow their unit numbering. Floodplains (Figure 2, cross-section Remouchamps 2 and Bullingen 4) consist of a basal gravel layer (unit 1), which is a Pleistocene deposit that is being reworked in the contemporary river beds. On top of this are fine deposits, which consist generally out of two main layers. The lower part (unit 2) shows a textural fining up from sandy deposits to silty clay loam. Near its base this unit often contains small twigs or other material which is deposited on channel bars. We interpreted this unit as a (point) bar deposit. On top of it is a layer of silty clay loam to loam (unit 3), which is difficult to distinguish from unit 2 because of the gradual transition, but which we interpreted as an overbank deposit. Units 2 and 3 form the bulk of the Holocene deposits, and are between 0.3 and 2.5 m thick, varying within and between transects. Owing to the gradual transition their individual thickness is difficult to establish. Sometimes a gravelly or sandy layer (unit 4, not represented on Figure 2), related to a major flood, is situated within these units 2 and 3. Organic deposits (unit 6) occur very locally in infilled older cut-off channels, or as distal floodbasin infillings in the upper valleys. A gravelly lower terrace, with a height of 0.2 to 0.8 m and probably of a pre-Holocene age (Notebaert et al., 2011a), can be recognized at several places in the lower floodplains, and is sometimes covered by a thin layer of fine Holocene deposits (unit 7, not represented on Figure 2, see Notebaert et al., 2011a). When floodplains are present in the steep reaches (Figure 2, Bévercé Gorge cross-section), they consist of the typical basal gravel layer (unit 1) covered with poorly sorted sandy deposits (unit 5). At many locations also Holocene colluvial foodslope deposits (unit 8) are present in the floodplains.

Figure 2.

Typical floodplain cross-sections for the main floodplain types. Cross-section Bullingen 4 represents a typical upstream floodplain, while Remouchamps 2 is located in the lower floodplains. The cross-section Bévercé Gorge is located in a steep floodplain section. Vertical exaggeration 10× (Bullingen 4 and Bevercé Gorges) and 30× (Remouchamps 2) (colour figure available online).

The Holocene fluvial architecture, and especially the presence of a basal (point) bar layer over the entire width of the floodplains, indicates the importance of lateral reworking of the flood-plain (Notebaert et al., 2011a). As such, the early-Holocene fluvial record was eroded and is no longer present (see also below). The limited thickness of the floodplain deposits shows that vertical aggradation was of minor importance for these floodplains during the Holocene.

Quantification of floodplain deposits

Average MDFSA was calculated for each of the floodplain types (Table 2) and spatially represented (Figure 3). Average floodplain deposition varies between 0.42 and 1.17 Mg/m² for the different floodplain types. Deposition is lowest for the steep reaches and the smallest upper floodplains (Figure 3), which have the thinnest Holocene deposits and narrowest floodplains. Thickness of sediment is largest for the lower floodplains (Table 2, Figures 3 and 4). Major depositional zones occur in the large floodplains of the Malmedy graben, and in the large upper floodplains, especially of the Amblève.

Table 2.

Results for the different floodplain types: sediment deposition (Mg/m²), total floodplain area in the catchment (m²) and total sediment deposition (×10⁶ Mg).

Zone	Sediment deposition (Mg/m²)	Area (m²)	Total deposition (× 10⁶ Mg)
Lower valleys
Lienne	1.12	1,468,051	1.6
Other	1.17	8,388,221	9.7
Malmedy Graben	0.81	3,238,571	2.6
Steep sections	0.51	1,778,304	0.9
Upper valleys (smallest)
Warche	0.80	151,898	0.1
Amblève	0.41	2,358,043	1.0
Other	0.51	7,829,161	4.0
Upper valleys (intermediate)
Warche	1.05	3,135,238	3.3
Amblève	0.94	2,265,830	2.1
Other	0.99	2,071,335	2.0
Main upper valleys
Warche	1.01	1,627,410	1.9
Amblève	1.17	1,180,727	1.2
Other	1.08	1,386,016	1.5
Total	–	36,878,805 (=36.9 km²)	32.0

Figure 3.

Floodplain sediment deposition in the Amblève catchment. Dots indicate site-specific average floodplain deposition for the different coring sites. The shading of the floodplain areas represent the average floodplain deposition for each floodplain type.

Figure 4.

River long profiles for the Amblève, Lienne and Warche. Background colors represent the floodplain types. The x-axis gives the upstream distance from Remouchamps (the place where the catchment exits the Ardennes) along the valley axis. (a) Data for the different coring locations: local sediment storage (Mg/m² floodplain area), upstream catchment area (km²),local sediment storage as a function of floodplain width × local sediment storage (W × M, Mg/m), specific stream power (W/m²). (b) Upstream properties for systematic created points along the stream axis: total (sub)catchment alluvial sediment storage (Tg) and sediment deposition per unit upstream surface area (SDUSA, Mg/km²) (colour figure available online).

From these values the total amount of deposited sediment for the entire catchment was calculated. Total floodplain deposition in the Amblève catchment (1080 km²) amounts to 32 × 10⁶ Mg. For a set of generated subcatchments (see methods) each time the total upstream catchments sediment deposition was calculated, which allows the assessment of spatial scaling relations to estimate floodplain deposition for catchments ranging from 3.5 km² to 1000 km² (Figures 4, 5). These calculations are based on the average values for each floodplain type, and may as such lose some detail on the smallest spatial scales. Results (Figures 4, 5) show low SDUSA values for the smallest catchments, as these catchments have only a limited alluvial area, and a (unknown) larger fraction exists of dry colluvial valleys. With increasing catchment area the SDUSA increases to a certain value from where it remains constant or slightly decreases, with some variability. A confluence effect is present where tributaries with different SDUSA values flow together, such as at the confluence of the Warche and the Amblève (see Amblève profile). SDUSA for the entire catchment amounts 0.030 × 10⁶ Mg/km².

Figure 5.

SDUSA (Tg/km²) in function of the catchment upstream area (m²) (a) for the different generated subcatchments of the Amblève catchment; (b) for the Amblève catchment and reported values for other European catchments (based on de Moor and Verstraeten, 2008; Hoffmann et al., 2007; Notebaert et al., 2010; Stolz et al., 2012). Note logarithmic scale on x-axis (colour figure available online).

Age of floodplain deposits

As a consequence of the important lateral floodplain erosion, vertical aggradation profiles for the entire Holocene are absent, and core-specific age–depth models represent sedimentation history since the last time the channel was positioned at that specific location, rather than sedimentation history over the entire Holocene. Notebaert et al. (2011a) have indicated the importance of understanding the fluvial architecture in such a setting, and have shown how dating using tracers can contribute in calculating net sedimentation rate for the floodplains. In addition, also specific-point dates such as those yielded with radiocarbon dating can be used when carefully considering the fluvial architecture.

Within the Amblève catchment, iron slag, originating from medieval iron industries (Houbrechts et al., 2011), has been used for dating floodplain deposition in the Lienne subcatchment (Houbrechts and Petit, 2004, 2006; Notebaert et al., 2011a). Although medieval iron-processing industries may have been present near the Amblève-Warche confluence (Houbrechts and Petit, 2004), an exploratory study showed too small amounts of slag present in the sediments to make a reliable reconstruction. Results for the Lienne are reported by Notebaert et al. (2011a) based on five sites in the downstream valleys, one for the Chavanne tributary, and one for a small upstream valley. Metal industries are for the different locations initiated between 1421 and 1537 (Houbrechts and Weber, 2007), and between 17 and 69% of contemporary sediment was net accumulated after this moment, with an average of 42 ± 17% (Notebaert et al., 2011a). The considered time period consists of only 5% of the total Holocene time period. The same results indicate that in addition to this net sediment accumulation, between 14 and 54% of the floodplain was laterally reworked during this same time period (average 28 ± 14%, n=6).

Chronological information for the upstream valleys is largely absent because of the absence of datable organic material in the overbank deposits. Iron slag was only found at one upstream site (Monty), indicating c. 40% of the Holocene deposits were deposited since c. ad 1600. Only at the Bullingen 4 site, organic deposits could be dated, and the results show that the upper 30–50% of the floodplain deposits date from the last 600–700 years, which is in line with the findings of the Lienne subcatchment. At one other site (Nidrum), no conclusions could be drawn, as samples from different depths yield the same age (Table 3).

Table 3.

Results of radiocarbon dating of floodplain sediments at eight sites in the Amblève catchment. Dates were calibrated using oxcal 4.1 (Bronk Ramsey, 2009) and the Intcal 09 curve (Reimer et al., 2009).

Sample	Lab code	Material	Age	Calibrated age	Location	Position
NID 1 51^*	UTC14833	Plant mat.	−1461 ± 31	Modern	Nidrum	Top floodplain Deposits, 5 cm depth
NID 1 53^*	UTC14834	Plant mat.	3666 ± 44	2196–1923 bc	Nidrum	Floodplain, 50 cm depth
NID 1 55^*	UTC14835	Plant mat.	3745 ± 41	2287–2032 bc	Nidrum	Base floodplain deposits, 75 cm depth
STA 3 41^*	UTC14836	Wood	157 ± 34	1665–modern	Stavelot	Base of Holocene deposits, 90 cm depth
STA 3 142^*	UTC14837	Wood	152 ± 31	1666–modern	Stavelot	Base of Holocene deposits, 180 cm depth
BUL 4 96^*	UTC14845	Plant mat.	139 ± 35	ad 1669–1945	Bullingen	See Figure 2; floodplain, 45 cm depth
BUL 4 43^*	UTC14846	Plant mat.	959 ± 35	ad 1017–1160	Bullingen	See fig. 2; floodplain, 120-130 cm depth
BUL 4 52^*	UTC14847	Plant mat.	646 ± 37	ad 1280–1397	Bullingen	See fig. 2; floodplain 55-70 cm depth
BUL 4 71^*	UTC14848	Plant mat.	724 ± 37	ad 1220–1385	Bullingen	See fig. 2; floodplain, 50–55 cm depth
Lienne 1°	Beta-246256	Charcoal	490 ± 40	ad 1324–1465	Trou de Bra	Floodplain, 59–69 cm depth; no slags
Lienne 2°	Beta-248832	Charcoal	1020 ± 40	ad 1324–1465	Chession	Floodplain, 60–70 cm depth; no slags
Lienne 3°	Beta-246257	Wood	1020 ± 40	ad 897–1153	Targnon	Abandoned channel; 98–108 cm depth
Lienne 4°	Beta-246258	Charcoal	660 ± 40	ad 1274–1397	Targnon	Abandoned channel; 97 cm depth
Lienne 5°	Beta-246259	Charcoal	660 ± 40	ad 1274–1397	Targnon	Abandoned channel; 100–110 cm depth

Notes: °dates from the Lienne catchment come from Notebaert et al. (2011a); * new dates for this study.

Discussion

Age and relationship with land use and climate

Notebaert et al. (2011a) show that the last c. 600 years was a major depositional period for the lower valleys of the Lienne, with c. 42% of the total Holocene net sediment accumulation for c. 5% of the Holocene period. Dating results from the upper Warche floodplains show comparable results.

Based on simple synchronicity, this major sedimentation phase can be related to the major anthropogenic land use during this period. These results point towards a major sedimentation phase during the last c. 600 years, driven by land use, although a (partial) role of climate cannot be excluded. Palynological data show no major changes in natural vegetation at the end of the Holocene (e.g. Damblon, 1978). Different studies have shown that soil erosion is limited under a dense forest cover (e.g. Notebaert et al., 2011c), as is the natural vegetation in the Ardennes, even with an increase in rainfall (erosivity). As a consequence, important increases in floodplain deposition should be attributed to forest clearings. The last 600 years coincide with major forest clearings for agriculture and for wood production related to the early (14th–18th centuries) industrialization of this region. Also other studies in nearby comparable upland regions find a relationship between increased late-Holocene fluvial sediment deposition and anthropigenic land use. In the Belgian–Dutch Geul catchment, sedimentation rates are reported to increase from the Roman Period on, and increased rates are being related to agricultural activities and mining activities (de Moor and Verstraeten, 2008; de Moor et al., 2008; Stam, 1999). Even if for both catchments a detailed temporal framework is missing, a comparison between the Geul and Amblève catchments shows major differences in sedimentation history, with an important sedimentation phase during the Roman Period and a major sedimentation phase related to very large-scale mining in the 19th century for the Geul. These differences can be explained by the major differences in historical land use, related to the thicker loess cover in the Geul catchment and its proximity to major Roman and Medieval cities (Liège and Aachen). In the West German uplands (Westerwald and Taunus regions), it is reported that catchments with an early and more intensive occupation history also have older and more extensive floodplain deposits (Stolz, 2011; Stolz and Grunert, 2008; Stolz et al., 2012). For the catchments with a less intense land-use history, results are comparable with this study. In the Lieser (Stolz et al., 2012) and Nister (Stolz, 2011) catchments, respectively c. 54 and 55% of the Holocene floodplain deposits date from the the late Medieval to Modern period. These percentages are comparable with the Lienne and Amblève, where c. 42% and 30–50% of the floodplain deposits date from more or less the same period. It is also reported for northern Bavaria that floodplain deposition increases mainly during the medieval time period, related to anthropogenic land use (Fuchs et al., 2011). The reported decrease in sedimentation rates somewhere after the medieval time period is not observed in this study because of the lower temporal resolution.

In order to get more solid insight into the relationship between fluvial sediment deposition and land use or climatic variations, a sediment budget approach with a modeling study can be combined (e.g. Notebaert et al., 2011c). Such studies could allow better insight into the intermediate pathways and sinks of sediment, including buffer effects. In addition, more detailed historical data of both land use and sediment deposition would allow a better understanding of the relationship between land use and soil erosion.

Spatial patterns and scale effects of sediment deposition

The results of this study provide an ideal tool for assessing scale effects on sediment deposition. Results show how the SDUSA varies with increasing catchment area (Figures 4, 5). For the smallest catchments the upstream floodplain area (UFA) will be limited, and in fact an important part of Holocene deposits will be stored in the colluvial valleys. With increasing catchment area also the UFA will increase, and as such the SDUSA. Where for the smaller valleys the percentage increase in UFA is still important for each increase in catchment area, this becomes less and less important as the UFA gains a large value, explaining the downstream flattening of the SDUSA function. These upper floodplains are not incised in the Mid-Pleistocene terraces, which results in wide floodplains, relatively low floodplain slope gradients and a direct connection to the agricultural areas. As a result of the large accommodation space (wide valleys), the low specific stream power and possibly also the direct connection to the sediment sources, these are mainly sedimentary sinks.

Further downstream the SDUSA decreases again, starting at the point where floodplains narrow because of the incision of the Mid-Pleistocene terraces (e.g. Demoulin et al., 2012). Here, floodplain slopes increase and floodplains narrow, while discharge increases as a function of increasing catchment area. The resulting increased specific stream power is responsible for an increased within-floodplain sediment transport and decrease in local sediment deposition. The floodplain is also less well connected to the agricultural area, which is situated on the flatter parts of the Quaternary river terraces and the plateau, and not the on the steep valley slopes. However, past deforestation related to metal industries, may also have occurred on the lower steep slopes, which are well connected to the floodplains. This decrease in SDUSA is for the Amblève at the steep stretch, and for the Warche already for the most downstream parts of the upper valleys, where the first floodplain narrowing occurs, upstream the major narrowing at the steep reach. This downstream decrease is interrupted for the Warche river by an important increase, which is due to the sediment sink in the Malmedy graben, where floodplains are wide, resulting in a low specific stream power.

SDUSA values are also influenced by the SDUSA values of tributaries, especially when the catchment surface area of these tributaries is relatively large. The higher SDUSA value of the Warche is reflected on the Amblève long profile as an increase in SDUSA at the confluence. The general pattern of upstream floodplain area and floodplain sediment thickness results in differences between the main tributaries. The SDUSA of the Amblève upstream its confluence with the Warche is 0.033 10⁶ Mg/km², while at this location the SDUSA of the Warche is 0.048 10⁶ Mg/km². This is explained by differences in catchment morphology, with a larger fraction of the Warche catchment existing of floodplains (5.1% compared to 4.3%), partially caused by the wide floodplains in the Malmedy graben.

The SDUSA–upstream area relationship (Figure 5) confirms the scale pattern in SDUSA. This relationship indicates how the catchment area specific fluvial sediment storage varies over different (catchment area) scales. It provides insight into the importance of fluvial sediment storage at different spatial scales (Figure 5a) and between different catchments and/or environments (Figure 5b). A large scatter exist in SDUSA values for small catchments, expressing the importance of local variations at this scale. These local variations are averaged out for increasing catchment sizes. For the main stem of the Amblève river, SDUSA values are decreasing downstream as a function of the relatively low deposition in the steep reaches and this main stem, and the lower SDUSA values for most tributaries which join the lower floodplains.

Comparison with other catchments

Sediment quantifications for catchments with a comparable geological and topographical setting in the Western German Uplands (Stolz et al., 2012) show overall values which fall in the same order of magnitude as for the Amblève. When comparing these SDUSA values with values of rivers from the Belgium loess belt, the later are c. 10 × higher (Figure 5b): it amounts 0.4 × 10⁶ Mg/km² for the Dijle catchment (758 km²; Notebaert et al., 2010), while slightly lower values are found for some of its tributaries (0.23–0.40 10⁶ Mg/km²; Notebaert et al., 2010). In the Geul and Gulp catchments, at the edge of the loess region, SDUSA values are 0.10–0.11 × 10⁶ Mg/km² (Notebaert et al., 2010). Also Hoffmann et al. (2007) report comparable values for subcatchments of the Rhine (0.21–0.36 × 10⁶ Mg/km², with an outlier of 0.96 × 10⁶ Mg/km²; calculation based on reported values). As such, it is clear that sediment deposition in the steeper upland catchments, such as the Amblève or those reported by Stolz et al. (2012), is lower than for those other catchments. This can be explained by the balance model for floodplain aggradation and incision (Lane, 1955), which explains floodplain aggradation through river properties (discharge, quantity and size of transported sediment) and topography (river channel slope).

Floodplains of the Amblève have, compared with the rivers of the (Belgian) loess belt, (1) steeper floodplain slopes, (2) higher discharge for major flood events, (3) lower sediment load. Following the balance model, these three parameters should lead to decreased aggradations or increased incision. Sediment supply (5–27 Mg/km² per yr, mostly around 20 Mg/km² per yr for comparable rivers in the same region; Petit, 1995) is lower than for the loess-region rivers in Belgium (Geul, 38–81 Mg/km² per yr; Leenaers, 1989; and Dijle, 90–210 Mg/km² per yr; Steegen, 2001; Verstraeten et al., 2006). This can be attributed to the less intensive land use in the Amblève catchment and/or a difference in connectivity between hillslopes and the fluvial system. Although hillslope sediment deposition has not been quantified for the Amblève, soil maps show a smaller extent of colluvial valleys compared with the Dijle. From this we assume that also hillslope sediment deposition will be lower for the Amblève catchment. The markedly lower sediment load of the Amblève suggests that lower SDUSA values are rather caused by lower sediment delivery to the fluvial system. Floodplain slopes are higher for the Amblève (lower floodplain 0.2–0.9%) than for the Dijle (lower floodplain 0.06–0.08%). Bankfull discharge for the Amblève is about 85 m³/s (catchment area of 760 km²), while it is c. 24 m³/s for the Dijle (catchment area of 760 km²; e.g. Notebaert et al., 2011c). This will result in a higher stream power, and combined with the smaller Amblève floodplains in a much higher specific stream power, and as a result the floodplain sediment transport can be expected to be much higher for the Amblève.

In addition to these factors, also continuing floodplain aggradation may result in an increase of river bank height when there is no river bed aggradation, resulting in an increased river channel cross-section, an increase in bankfull discharge and a decrease in flooding events.

Quantification of sediment fluxes for the agricultural period

Based on the available datings, a quantification for the last 600 years can be made, which corresponds more or less to the period of increased anthropogenic land use. During this time period c. 42% of the sediment currently present in the floodplain was deposited, which is 13.4 × 10⁶ Mg (Table 4). Sediment-yield values of the Ardennes (see above) can be extrapolated to the same time period, as these reported values represent catchments with land use variations comparable with those which can be expected for the Amblève during this time period. The results provide only a very rough estimate, as sediment load may well have varied within this period, and indicate that between 3 and 17 × 10³ Mg/km² (catchment area) would have been exported. Information on the extent and thickness of colluvial deposits is missing for the Amblève, and as such a colluvial sediment budget and the quantity of hillslope erosion cannot be estimated.

Table 4.

Simplified fluvial sediment budget for the last 600 years for the Amblève catchment. For export out of the catchment a broad range is indicated together with a best estimate (BE). Details on the methods used are explained in the text.

Sediment sink	Total amount (Mg)	Area specific (Mg/km²)	Area specific rate (Mg/km² per yr)
Colluvium	?	?	?
Floodplain deposition	13.4 × 10⁶	12.4 × 10³	20.7
Export	3–17 × 10⁶ (BE: 13 × 10⁶)	3–16 × 10³ (BE: 12 × 10³)	5–27 (BE: 20)
Reworking by channel	7.1 × 10⁶	6.5 × 10³	10.8
Sediment delivery to the fluvial system	17–31 × 10⁶ (BE: 26 × 10⁶)	15–29 × 10³ (BE: 25 × 10³)	25–48 (BE: 41)

For the last 600 yr also the importance of lateral erosion (and deposition) by the river channel can be established. Three assumptions have to be made in these calculations: (1) a constant net floodplain aggradation rate for the last 600 years; (2) all the sediments which are reworked, are only reworked once during this period (i.e. the meander belt passed only once at each location), which means that the estimate is a minimal estimate; and (3) the rates which are calculated for the Lienne catchment are representative for the entire catchment. Based on these assumptions, 28% (see above) of the sediments present before ad 1400 were reworked during the considered period, while 14% (28% × 0.5) of the net aggradation within this period is also reworked within this same period. As a result, 22% of present floodplain deposits, or 7.1 × 10⁶ Mg, were reworked during this 600 year period. This estimate might be rather a maximum estimate than a minimum estimate because these percentages are attained for the lower floodplains, while the percentage of reworked floodplain deposits is probably lower for the upper floodplains. These upper floodplains are often wider, with smaller channels, while the river banks are relative high (low width–depth ratio) and consolidated (silty and clayey deposits), while also stream power is lower, which results probably in lower meandering speeds and a lower fraction of the floodplain being reworked. It remains unclear whether these high lateral erosion rates for this type of river has an influence on Holocene floodplain aggradation.

Snijders et al. (2006) report a meandering speed of 12 m per century for the lower Amblève (study area also partially downstream our study area), based on historical maps since ad 1850. The same study also used metal slags for the lower Amblève, resulting in average meandering speeds of 0.7–3.3 m per century since ad 1390 (Snijders et al., 2006). Their study area overlaps our study area for a floodplain length of 13.8 km, and their values can be extrapolated using the river channel length and floodplain area for this overlapping stretch (excluding the reaches that lack floodplains). Results indicate that 87% of the floodplain would be reworked when using the map-based rates, while 5–23% would be reworked when using the slag-based rates. The map-based estimates do not, however, take into account that the same part of the floodplain can be reworked multiple times during the considered time period, which lowers the proportion of the floodplain that was effectively reworked. The rates reported by Snijders et al. (2006) confirm that calculated rates for the Lienne are realistic.

The resulting sediment budget for the last 600 years (Table 4) indicates that for this time period vertical floodplain aggradation or net floodplain deposition was more important than lateral reworking of floodplain deposits by the river. There is a large uncertainty on the sediment load, but when considering a value of 20 Mg/km² per yr as a best estimate, sediment export from the catchment is comparable to sediment storage. Based on this rough estimate, about 50% of the sediment that is delivered to the fluvial system is exported from the catchment. This indicates that the fluvial system in these upland catchments is an important transport system for sediments. Data for the Dutch Geul catchment and the Belgian Dijle catchment, both in a nearby loess regions, show that for the last 1000 years, between 35 and 36% of the total sediment added to the river system is exported (meaning 64–65% is deposited as alluvium (de Moor and Verstraeten, 2008; Notebaert et al., 2011b). For Coon Creek (USA), Trimble (1999) has shown how (absolute) sediment export remains more or less constant for three time intervals between 1853 and 1993, but with varying sediment delivery to the fluvial system the fraction of exported sediment varies between 9 and 32%.

Based on the calculations for sediment export and floodplain storage, the sediment delivery to the fluvial system is 26 × 10⁶ Mg for this 600 year period, which equals 41 Mg/km² per yr (best estimate; range: 26–48 Mg/km² per yr). When the same calculations are made for the last 1000 years for the Dijle catchment, 434 Mg/km² per yr was delivered to the fluvial system (based on Notebaert et al., 2011b). As a consequence, the relative low fluvial sediment storage in the Amblève catchment can be stated to be a consequence of (1) mainly a lower sediment delivery to the fluvial system owing to less intensive anthropogenic land use, and (2) also a larger fraction of the delivered sediment being exported out of the catchment, because of higher stream power.

Sedimentary cascade model

Based on the results of this study a conceptual sediment cascade model for the Amblève catchment can be constructed (Figure 6). For the upper valleys, sediment thickness, floodplain width and SDUSA are increasing downstream, with decreasing floodplain slope angles. Floodplains are directly connected to the sediment-producing slopes which are under cropland or grassland, and to large colluvial valleys, while also several tributaries join the valleys. Although the slopes and colluvial valleys are directly connected to the fluvial system, within-slope connectivity is rather low with the occurrence of many (small) colluvial steps.

Figure 6.

Conceptual model describing the sediment cascade in the fluvial system of the Amblève catchment: (a) represents different properties of the fluvial system with a dimensionless axis: altitude (black), floodplain sediment thickness and range (blue line and shaded area), instream sediment connectivity (red). (b) Plan view of the floodplain: 1: floodplain and river; 2: tributary; 3: colluvial sediment input; 4: low connectivity between colluvial valleys and the fluvial system. The different floodplain types are indicated at the top. (c) Cross-section of the upper valleys (A–B) and a cross-section of the lower valleys (C–D), with simplified and generalized topography and land use and hillslope connectivity. The upper valleys are less incised and the floodplains are directly connected to the agricultural area on the valley slopes. The lower valleys are deeply incised in the mid-Quaternary river terraces, with steep valley slopes. The floodplains are not directly connected to the agricultural area. Note that forests are both (plantation) coniferous forests and (natural and plantation) deciduous forests. Both cross-sections have different vertical and horizontal scales (colour figure available online).

Steep valley stretches occur downstream the major knickpoint, where floodplains are mostly absent. Owing to the increased stream power, all the sediments are transported downstream. Additional sediment may be delivered by tributary streams. Locally produced sediment may be directly delivered to the fluvial system from the slopes, especially in the past during periods when these steep slopes were deforested.

Sediment is again deposited in the lower valleys. Here, there is a direct connection between the floodplains and the steeper valley edge slopes, but these are (at present) almost entirely forested, producing limited amounts of sediment, and large colluvial valleys are missing. During past major deforestation phases, sediment production on these slopes may have been much higher with direct sediment input into the fluvial system. The larger agricultural areas on the plateau are however not directly connected to the main lower valleys, and sediment is transported through tributaries towards the main streams.

Conclusions

This study has quantified the alluvial sediment storage in the Belgian upland Amblève catchment (1080 km²) to be 32 Tg (= 32 × 10¹² g). About 42% of Holocene fluvial deposits were net accumulated during the last 600 years, and this high recent sediment rate can be explained by increased anthropogenic land use during this period. When calculating sediment deposition per unit (upstream catchment) surface area (=SDUSA), downstream variations are caused by (1) scale effects for the smallest streams related to a downstream-increasing importance of the floodplain area; (2) scale effects related to the lower sediment storage for the lower (deeply incised) valleys compared with the upper valleys; (3) confluence effects. SDUSA values fall within the same range, but at the low side, as other (German) upland catchments. SDUSA value for loess-region catchments are much larger than those for these upland regions.

Based on the findings, a fluvial sediment budget for the last 600 years was constructed. Even as major assumptions have to be made, the budget allows to conclude that floodplain deposition and export of sediment out of the catchment were about equally large and lateral reworking of sediments by the river channel was about half as important as each of these two other processes. About 26 × 10⁶ Mg (best estimate; range: 16–30 × 10⁶ Mg) of sediment was delivered to the fluvial system, which equals 41 Mg/km² per yr (best estimate; range: 26–48 Mg/km² per yr). This value is about 5–10 times lower than for the Belgian loess region catchments, and most of the catchments in the Rhine catchment.

Finally, we can conclude that for the Amblève catchment, low SDUSA values are caused by low sediment delivery from the slopes to the fluvial system. Also a larger fraction of the sediments delivered to the catchment from the slopes is exported out of the catchment. Although catchment settings and catchment topography are responsible for higher potential sediment transport by the fluvial system, we argue that the less intensive land use proves to be the major reason for low SDUSA values.

Footnotes

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

References

Aqualim (2013) Réseau de surveillance hydrométrique: Aqualim. Available at: http://aqualim.environnement.wallonie.be.

Bronk Ramsey

(2009) Bayesian analysis of radiocarbon dates. Radiocarbon 51: 337–360.

Chiverrell

Thorndycraft

Hoffmann

(2011) Cumulative probability functions and their role in evaluating the chronology of geomorphological events during the Holocene. Journal of Quaternary Science 26: 76–85.

Damblon

(1969) Etude palynologique comparée de deux tourbières du plateau des Hautes Fagnes de Belgique: La Fagne Wallonne et la Fagne de Clefay. Bulletin du Jardin botanique national de Belgique/Bulletin van de National Plantentuin van België 39: 17–45.

Damblon

(1978) Etudes paléo-écologiques de tourbières en haute Ardenne. Laboratoire de Palnyologie et de Phytosociologie.

de Moor

JJW

Verstraeten

(2008) Alluvial and colluvial sediment storage in the Geul River catchment (The Netherlands) — Combining field and modelling data to construct a late Holocene sediment budget. Geomorphology 95: 487–503.

de Moor

JJW

Kasse

van Balen

. (2008) Human and climate impact on catchment development during the Holocene – Geul River, the Netherlands. Geomorphology 98: 316–339.

Demoulin

Hallot

(2009) Shape and amount of the Quaternary uplift of the western Rhenish shield and the Ardennes (western Europe). Tectonophysics 474: 696–708.

Demoulin

Beckers

Rixhon

. (2012) Valley downcutting in the Ardennes (W Europe): Interplay between tectonically triggered regressive erosion and climatic cyclicity. Netherlands Journal of Geosciences-Geologie En Mijnbouw 91: 79–90.

10.

Dotterweich

(2008) The history of soil erosion and fluvial deposits in small catchments of central Europe: Deciphering the long-term interaction between humans and the environment – A review. Geomorphology 101: 192–208.

11.

Food and Agriculture Organization (2006) World Reference Base for Soil Resources 2006: A Framework for International Classification, Correlation and Communication. Rome: World Soil Resource Reports.

12.

Förster

Wunderlich

(2009) Holocene sediment budgets for upland catchments: The problem of soilscape model and data availability. Catena 77: 143–149.

13.

Fuchs

Will

Kunert

. (2011) The temporal and spatial quantification of Holocene sediment dynamics in a meso-scale catchment in northern Bavaria, Germany. The Holocene 21: 1093–1104.

14.

Gullentops

Mullenders

Schaillee

. (1966) Observations géologiques et palynologiques dans la vallée de la Lienne. Acta Geographica Lovaniensia 4: 192–204.

15.

Hoffmann

Erkens

Cohen

. (2007) Holocene floodplain sediment storage and hillslope erosion within the Rhine catchment. The Holocene 17: 105–118.

16.

Hoffmann

Erkens

Gerlach

. (2009) Trends and controls of Holocene floodplain sedimentation in the Rhine catchment. Catena 77: 96–106.

17.

Houbrechts

Petit

(2003) Utilisation des scories métallurgiques en dynamique fluviale: Détermination de la compétence effective des rivières et estimation des vitesses de progression de leur charge de fond. [Utilisation of métallurgie slags for the study of fluvial dynamics: Determination of the effective competency of rivers and estimation of the transit time of sedimentary waves.] Géomorphologie: relief, processus, environnement 9: 3–12.

18.

Houbrechts

Petit

(2004) Etude de la dynamique fluviale des rivières ardennaises grâce aux scories métallurgiques. De la Meuse à l’Ardenne 36: 57–67.

19.

Houbrechts

Petit

(2006) Utilisation des microscories métallurgiques comme traceur de la sédimentation dans les plaines alluviales des rivières ardennaises. L’érosion entre société, climat et paléoenvironnement–Table ronde en l’honneur du Professeur René Neboit-Guilhot, Presses Universitaires Blaise Pascal, Collection Nature et Société, pp. 91–96.

20.

Houbrechts

Weber

(2007) La sidérurgie proto-industrielle dans le bassin de la Lienne. De la Meuse à l’Ardenne 39: 35–63.

21.

Houbrechts

Levecq

Vanderheyden

. (2011) Long-term bedload mobility in gravel-bed rivers using iron slag as a tracer. Geomorphology 126: 233–244.

22.

Juvigne

(1979) L’encaissement des rivières ardennaises depuis le début de la dernière glaciation. Zeitschrift Fur Geomorphologie 23: 291–300.

23.

Lane

(1955) Design of stable channels. Transactions of the American Society of Civil Engineers 120: 1234–1260.

24.

Leenaers

(1989) The Dispersal of Metal Mining Wastes in the Catchments of the River Geul (Belgium – The Netherlands). Netherlands Geographical Studies 102, 200 pp.

25.

Macaire

J-J

Bellemlih

Di-Giovanni

. (2002) Sediment yield and storage variations in the Négron River catchment (south western Parisian Basin, France) during the Holocene period. Earth Surface Processes and Landforms 27: 991–1009.

26.

Macklin

Lewin

(2003) River sediments, great floods and centennial-scale Holocene climate change. Journal of Quaternary Science 18: 101–105.

27.

Mols

(2004) Dynamique fluviale en réponse aux changements d’affectation du sol des bassins versants de l’Euregio Meuse-Rhin. Mémoire de diplôme d’études approfondies, Université de Liège, 55 pp.

28.

Notebaert

Verstraeten

(2010) Sensitivity of West and Central European river systems to environmental changes during the Holocene A review. Earth-Science Reviews 103: 163–182.

29.

Notebaert

Houbrechts

Verstraeten

. (2011a) Fluvial architecture of Belgian river systems in contrasting environments: Implications for reconstructing the sedimentation history. Netherlands Journal of Geosciences-Geologie En Mijnbouw 90: 31–50.

30.

Notebaert

Verstraeten

Govers

. (2010) Quantification of alluvial sediment storage in contrasting environments: Methodology and error estimation. Catena 82: 169–182.

31.

Notebaert

Verstraeten

Rommens

. (2009) Establishing a Holocene sediment budget for the river Dijle. Catena 77: 150–163.

32.

Notebaert

Verstraeten

Vandenberghe

. (2011b) Changing hillslope and fluvial Holocene sediment dynamics in a Belgian loess catchment. Journal of Quaternary Science 26: 44–58.

33.

Notebaert

Verstraeten

Ward

. (2011c) Modeling the sensitivity of sediment and water runoff dynamics to Holocene climate and land use changes at the catchment scale. Geomorphology 126: 18–31.

34.

Petit

Lambin

(2002) Long-term land-cover changes in the Belgian Ardennes (1775–1929): Model-based reconstruction vs. historical maps. Global Change Biology 8: 616–630.

35.

Petit

(1995) Régime hydrologique et dynamique fluviale des rivières ardennaises. In: Demoulin

(ed.) L’Ardenne: Essai de Géographie Physique. Livre en hommage au Professeur A. Pissart. Université de Liège, pp. 194–223.

36.

Petit

Pauquet

(1997) Bankfull discharge recurrence interval in gravel-bed rivers. Earth Surface Processes and Landforms 22: 685–693.

37.

Petit

Gob

Houbrechts

. (2005) Critical specific stream power in gravel-bed rivers. Geomorphology 69: 92–101.

38.

Petit

Hallot

Houbrechts

. (2008) La typologie et les caractéristiques hydromorphologiques des cours d’eau wallons. Bilan d’une décennie d’ingénierie écologique. Service Public de Wallonie, pp. 7–16.

39.

Reimer

Baillie

MGL

Bard

. (2009) IntCal09 and Marine09 radiocarbon age calibration curves, 0–50,000 years cal BP. Radiocarbon 51: 1111–1150.

40.

Rixhon

Demoulin

(2010) Fluvial terraces of the Amblève: A marker of the Quaternary river incision in the NE Ardennes massif (Western Europe). Zeitschrift Fur Geomorphologie 54: 161–180.

41.

Rommens

Verstraeten

Bogman

. (2006) Holocene alluvial sediment storage in a small river catchment in the loess area of central Belgium. Geomorphology 77: 187–201.

42.

Seidel

Mackel

(2007) Holocene sediment budgets in two river catchments in the Southern Upper Rhine Valley, Germany. Geomorphology 92: 198–207.

43.

Snijders

Hallot

Houbrechts

. (2006) Approche méthodologique pour la délimitation des espaces de liberté des rivières du massif ardennais. Bulletin de la Société géographique de Liège 48: 7–17.

44.

Stam

(1999) The dating of fluvial deposits with heavy metals, ²¹⁰Pb and ¹³⁷Cs in the geul catchment (The Netherlands). Physics and Chemistry of the Earth, Part B: Hydrology, Oceans and Atmosphere 24: 155–160.

45.

Steegen

(2001) Sediment deposition in and export from small agricultural catchments. Unpublished PhD Geography Thesis, KU Leuven, Faculty of Sciences, 220 pp.

46.

Stolz

(2011) Budgeting soil erosion from floodplain sediments of the central Rhenish Slate Mountains (Westerwald), Germany. The Holocene 21: 499–510.

47.

Stolz

Grunert

(2008) Floodplain sediments of some streams in the Taunus and Westerwald Mts., western Germany, as evidence of historical land use. Zeitschrift für Geomorphologie 52: 349–373.

48.

Stolz

Grunert

Fulling

(2012) The formation of alluvial fans and young floodplain deposits in the Lieser catchment, Eifel Mountains, western German Uplands: A study of soil erosion budgeting. The Holocene 22: 267–280.

49.

Trimble

(1999) Decreased rates of alluvial sediment storage in the Coon Creek basin, Wisconsin, 1975–93. Science 285: 1244–1246.

50.

Trimble

(2009) Fluvial processes, morphology and sediment budgets in the Coon Creek Basin, WI, USA, 1975–1993. Geomorphology 108: 8–23.

51.

Van Asselen

(2011) The contribution of peat compaction to total basin subsidence: Implications for the provision of accommodation space in organic-rich deltas. Basin Research 23: 239–255.

52.

Van Oost

Verstraeten

Doetterl

. (2012) Legacy of human-induced C erosion and burial on soil–atmosphere C exchange. Proceedings of the National Academy of Science USA 109: 19,492–19,497.

53.

Verstraeten

Poesen

(2001) Variability of dry sediment bulk density between and within retention ponds and its impact on the calculation of sediment yields. Earth Surface Processes and Landforms 26: 375–394.

54.

Verstraeten

Poesen

Goossens

. (2006) Belgium. In: Boardman

Poesen

(eds) Soil Erosion in Europe. John Wiley & Sons, pp. 385–411.

55.

Voies Hydrauliques (2013) Voies hydrauliques. Station Martinrive. Available at: http://voies-hydrauliques.wallonie.be.