The bottom sediments of the turbidity maximum area of the Scheldt estuary were mapped in 1999 using echo sounding, sidescan sonar and grain-size analyses of bottom sediments. Four sediment types, sand, muddy sand, sandy mud and mud were recognised. Mud, with very little sand, occurs mainly in the access channels to the sluices giving access to the harbour docks of Antwerp. The sediments of the main channel have a sandier texture. One might conclude that the total mud stock in the middle estuary has increased, both between 1964–1986 and 1986–1999, but on the contrary the mud supply from the river, the mud stock in the river channel and the mud supply to the lower estuary have all decreased. The increase in the mud stock in the area as a whole was completely at the expense of mud deposition in the access channels to the sluice gates giving access to the harbour of Antwerp.
The mud stock in the river channel decreased over the years because of a decreasing mud supply from the river. The mud stock in the river channel shows variations that are directly related to fluctuations in the river load. When the suspended matter decreased during a certain year the supply of silt and clay particles decreased correspondingly and the resuspension-deposition mechanism caused a relative increase of the sand fraction in the bottom sediments in that year.
Sustained management of an estuary needs all estuarine processes to be comprehensively monitored. These processes are influenced by the existence of mud, a fine material composed of particles mostly below 63 μ m, a substantial part of which is of organic origin, and a variable amount of carbonates partly of local origin (Wartel and Faas, 1986). Mud deposits concentrate in areas where suspended matter concentrations are highest. Along the estuary of the Scheldt the highest suspended matter concentrations occur in the middle estuary (turbidity maximum area; Chen, 2003) and, through mechanisms of deposition and resuspension, high concentrations contribute to the persistence of a turbidity maximum and thus indirectly influence the residence time of mud in the estuary. In this context knowledge of the mud volume in the bottom (mud stock) and its regional distribution become very important, especially when a sediment budget has to be established.
It was the aim of this study to investigate the changes in the mud stock between 1964 and 1999 in the turbidity maximum area of the Scheldt estuary (Figure 1). Furthermore the study used successive recordings to compare the changes in the mud stock to: river discharge; suspended matter supply from the river basin; and morphological changes resulting from the construction of deep access channels to the sluice gates that give access to the docks of the Antwerp harbour.
Maps of bottom sediments, called bottom maps, in the Scheldt estuary were constructed at different time intervals. Bottom maps of the middle estuary were constructed by Bastin (1973), Bastin (1987) and Wartel et al. (2000). Bottom maps of the lower estuary were constructed by de Looff (1978a, b). These bottom maps could be used to estimate the quantity of mud in the bottom that is available for resuspension. To gain a clear insight in the evolution of the mud stock it was necessary to compare the change in mud stock to river discharge and to the supply of fluvial and marine mud. Annual averages of river discharge and suspended matter supply from the river basin are available from the Administration Waterways and Maritime Affairs, Section Maritime Entrance, Antwerp (Taverniers, 2000) as well as the mud content as organic matter content of the dredged material. Both have been reported on a yearly basis since 1989.
The macrotidal Scheldt estuary is more than 160 km long extending from its mouth at the North Sea (west of Flushing) to Ghent where the tide is stopped by a weir. The average tidal range increases from 3.8 m near Flushing to a maximum of 5.2 m near Schelle. Near Ghent an average tidal range of 1.9 m occurs. A more detailed description of the estuary is given in Chen et al. (2005). The estuary can roughly be subdivided in three parts: lower, middle and upper. The lower estuary extends from the mouth to 50 km upstream of Flushing covering the Westerschelde. The middle part extends from 50 to 100 km upstream of Flushing and encompasses the area were the highest suspended matter concentrations prevail (Chen et al., 2005). The upper part extends from 100 km upstream of Flushing to Ghent and has in general lower suspended matter concentrations except for the most landward part where an increase in suspended matter concentration has been observed (Chen et al., 2005).
The area covered by the bottom map, recorded in August 1999, extends from Zandvliet at 56 km upstream of Flushing up to the upper limit of the road of Antwerp at approximately 80 km upstream of Flushing and covers exactly the same area as previously published bottom maps (Bastin et al., 1988). It is located at the seaward part of the turbidity maximum area and thus resuspension-deposition processes coupled to the reworking of bottom sediments are expected to be important.
In order to construct a map of the bottom sediments it was necessary to classify the bottom sediments. However, no strictly separated sediment classes exist in nature and thus a number of criteria were laid down. The bottom sediments of the study area were classified into four categories applying a multivariate extension of the entropy concept (Wartel et al., 1998) to 150 samples of bottom sediments. The nature of the sediment was determined based on six parameters including four size fractions: gravel (> 2 mm), sand (2 mm to 63 μ m), silt (63 μ m to 2 μ m) and clay (< 2 μ m), the organic matter content and the calcium-carbonate content. A hundred sampling sites were selected using the 1986 bottom map of Bastin (Bastin, 1973) as a reference such that all four categories of that bottom map were equally sampled. Eleven supplementary sampling sites were selected to function as the calibration of the Quester Tangent Corporation (QTC) software and of the “Lithoprobe” (Wartel et al., 2000). Pure mud was calibrated using samples from the access channel to the Kallo sluice gate. At every sampling site minimum 3 and maximum 6 samples were taken and analysed for calibration.
The nature of the bottom sediments was recorded using the characteristics of the reflected signal emitted by an echosounder (ATLAS DESO 22) at a frequency of 33 kHz. After digitizing the analogue echo, captured between the transducer (sender/receiver) and the amplifier, it was analysed with QTC-VIEW software and classified based on 3 energy-descriptive parameters. The calibration of the QTC-classification was done on eleven predefined sites selected on the basis of the analyses of 50 bottom samples.
After an initial bottom map was constructed it was compared to the results of simultaneously recorded sidescan sonar observations covering the whole area. This was done using a C-MAX CM800 sidescan sonar at a frequency of 325 kHz and a slant range of 75 m.
A compilation of the QTC-map with the sidescan-sonar map gave the final bottom map.
The “Lithoprobe” was used for the in situ characterisation of bottom sediments (Wartel et al., 2000). The probe was inserted in the bottom at a speed of 0.2 m s−1 by its own weight. Maximum penetration was 5 m and depended on the weight used and on the compaction of the bottom. The sensors of the probe could discriminate between sediment layers differing in grain-size composition and exceeding a thickness of 5 cm. The movement of the lithoprobe was followed by a pressure sensor, an altimeter and a clinometer to allow accurate determination of the vertical depth of penetration. The data were stored using a high-speed data logger.
Results and discussion
Sediment properties and distribution
The median grain-size of the sediment decreased from Flushing (d50 > 300 μ m) towards Antwerp (d50 < 50 μ m) and was finest in the estuarine turbidity maximum area (Chen et al., 2005). In the study area four categories were recognised. These categories ranged from almost pure sand (86 ± 6% > 63 μ m) with a low organic matter content (2.2 ± 2%) and a bulk density of 1.8, to almost pure mud (82 ± 2% < 63 μ m) with a high organic matter content (11 ± 1%) and a very low bulk density (1.2). Pure mud sediments were observed mainly on the upper tidal flats (mud flats) and in harbours with a free opening to the river (e.g. access channels to sluice gates). Also in the river channel were muddy sand (40 ± 19% < 63 μ m) and sandy mud (70 ± 4% < 63 μ m) with bulk densities ranging from 1.4 to 1.6. These sediments differ from the pure mud that was observed in the access channels to sluice gates by the amount of sand (less than 5% in the access channels and up to 30% in the river channel), by a lower content in organic matter and by a higher bulk density.
As a rule it could be postulated that the Scheldt estuary bottom sediments show a good correlation between their clay content and the amount of organic matter (Figure 2).
The spatial distribution of the sediment types had a patchy appearance as a result of the heterogeneous bottom morphology. Nevertheless, a major trend in the sediment distribution could be recognised: sand and muddy sand occurred on the shoals between ebb-and flood channels, the shallower inner parts of the river bends consisted mainly of sandy mud, and pure mud occurred almost exclusively in the access channels to sluice gates and in harbours with an open connection to the river. The quasi absence of sand in the access channels suggested deposition from a uniform suspension (Sas, 1989).
The quantification of the mud stock required not only the spatial distribution but also the known thickness of the “recent” sediment layer. In this study “recent” sediment was understood to mean sediment that was available for erosion and resuspension. Bottom maps showed the spatial distribution of sediments deal with the surface of the sediment but gave no information on the thickness of the uppermost resuspendable sediment layer. The sediment composition and its bulk density changes with depth and hence also the thickness of the erodible layer varies with the sediment properties. Therefore, the “recent sediment” layer was defined using the lithoprobe data, density measurements on sediment cores, and grain-size analyses. The lithoprobe penetrated deeper in muddy sand and sandy mud (with an average of 0.6 m) than in sand (with an average of 0.25 m). In some areas, located in outer bends of the main channel and below the 10 m isobath, there was no penetration of the lithoprobe. The “recent sediment”, if any, in these areas was very thin. The amount of mud in the access channels to the sluice gates, showing an extremely low bulk density, was calculated using acoustical data giving the thickness of the mud layer and, in combination with topographic records, giving the spatial extension of the layer.
Evolution of bottom mud from 1964 to 1999
Mud is a very complex material mainly composed of mineral silt and clay particles with a small amount of fine sand, organic substances, carbonates, iron-and aluminum-hydroxide complexes and microorganisms. Therefore it is difficult to determine the proportion of mud in sediments. For this study the mud proportion in every predefined sediment type was calculated based on the average size-spectrum of the calibration samples of mud. A mud-coefficient was then determined by calculating the proportion of this spectrum in the size spectrum of every sediment-type and then this mud-coefficient was applied to the four pre-defined sediment types of the bottom maps.
The total amount of mud between the Belgian-Dutch border and Antwerp, calculated using the 1964-bottom map, amounted to 3,282 × 103 t dry matter (see Table 1). Applying the same method to the 1986-bottom map gives a total of 5,082 × 103 t. During that period the mud content, that differentiates both maps, had increased by 55% corresponding to a yearly increase of 82 × 103 t. Roughly 42 × 103t (55%) of this mud were deposited in the access channels of the large sluice gates (Zandvliet, Kallo). The remainder 45% (40 × 103 t) were deposited in the river channel and in the interdital areas. The supply of suspended river load during the same period (1964 and 1989) was estimated between 375 × 103 t (Verlaan et al., 1997) and 420 × 103 t per year (van Maldegem et al., 1993). Comparing the only increase in mud-volume in the river channel to the average supply of fluvial load (400 × 103 t) indicated that only 10% of the supplied load was captured in the bottom sediments of the river channel. A small part of the fluvial load was deposited at the tidal marshes. Assuming an average sedimentation of 500 g m− 2 per neap-spring cycle (Temmerman et al., 2003) on a tidal marsh surface of 24 km2 gave an annual deposition of 12 × 103 t on the tidal marshes of the middle estuary. The remainder of the fluvial import, approximately 306 × 103 t, was transported partly to the lower estuary and partly passed through the sluice gates into the harbour docks. No data were available for the transport towards the harbour docks. The mud transport to the lower estuary has previously been evaluated at 300 × 103 t per year by van Maldegem et al. (1993) which was very close to the estimate obtained from this study.
The same analytical approach was applied to the 1999 bottom-map giving a total amount of 5,594 × 103 t dry matter. However, between 1992 and 1999 1,187 × 103 t have been dredged from the access channel to the Kallo-sluice gate and the dredged soil was removed from the system and dumped outside of the river. This removed quantity had to be added to the amount of mud found from the bottom 1999 bottom map which gave a total of 6,781 × 103 t of mud resulting in an increase of 1,699 × 103 t or 33% compared with the 1989 bottom-map. The corresponding average yearly increment was 131 × 103 t. The major part of this mud, 111 × 103 t or 85%, was deposited in the access channels to the major sluice gates and only 20 × 103 t or 15% was on the river bottom of the middle estuary or transported to the lower estuary. These data clearly showed the effect of the access channels on the mud budget.
Although during the period 1986-1999 the mud content increased by a factor 1.3 compared to the period 1964–1986, the average annual supply of fluvial suspended load had decreased from 400 × 103 t to approximately 230 × 103 t (Taverniers, 2000) and only 20 × 103 t or 9% of which was captured in the bottom sediments assuming that river supply was the only source of suspended matter in the middle estuary. An equal proportion was obtained for the period 1964–1986 and thus from the comparison of the three bottom maps it could tentatively be postulated that approximately one tenth of the suspended load supplied from the river was deposited in the bottom sediments between Antwerp and Zandvliet. The transport to the lower estuary could be estimated from these data. Subtracting the yearly mud deposit in the middle estuary (131 × 103 t) plus the deposition on the tidal marshes (12 × 103 t) from the river supply (230 × 103 t) resulted in a surplus of 87 × 103 t, the major part of which was transported into the lower estuary.
This value is much lower than the previously mentioned 300 × 103 t (van Maldegem et al., 1993) or than the 131 × 103 t given by Baeyens et al. (1998) or the 136 × 103 t mentioned by van Eck (1991). It follows that the sediment transfer from the middle to the lower estuary must have drastically decreased compared to the period prior to 1986. An argument in favour of this decrease is furnished by the estimation of the fluvio-marine equilibrium based on stable isotopes. It shows that from 1993 to 1998 the marine component had increased by 10% near the Belgian-Dutch border and by 20% upstream of Antwerp (Chen et al., 2005). This increase in marine component was relative and resulted from a decrease in fluvial supply as well as from an increase in marine supply. Indeed, the amount of suspended matter that entered the lower estuary (Westerschelde) from the North Sea could be estimated at 130 × 103 t (van Maldegem et al., 1993; Vereeke, 1994), however, the majority of it settled out in the Westerschelde (van Alphen, 1990) and only a minor part will reach the middle estuary and thus can hardly explain an increasing marine influence. Nevertheless, deepening of the Westerschelde (lower estuary) since 1970 must have enhanced the marine influence further landward.
Summarizing one can conclude that the mud stock in the middle estuary had increased but that on the contrary the mud supply from the river, the mud stock in the river channel and the mud supply to the lower estuary had all decreased. It followed that the increase in the mud stock in the middle estuary was completely at the expense of mud deposition in the access channel to the Kallo sluice gate and to a lesser degree in the access channels to the other sluice gates. The gyratory water flow in these access channels (Sas, 1989; Fettweis and Sas, 1999) was such that during each half tide the water is renewed twice. The current velocity in the access channel sharply decreased from the river channel to the sluice gate allowing fast settlement of flocculated suspended matter (Chen, 2003).
The amount of mud that was deposited in the estuary is proportional to the surface of the depositional area and to the prevailing hydraulic conditions. The construction of access channels to the sluices not only constituted a substantial increase in the surface of the depositional area but also, due to the reduced current velocity inside these access channels, constituted an ideal environment for the deposition of mud. Mud deposited in the access channels was subtracted from the river mud supply and, for a constant supply of river mud, caused a decreasing export to the lower estuary. It may be questioned to what degree this deposition also caused a decreasing sedimentation on the salt marshes but at present no information is available to support this.
Silt and clay fractions in sandy mud and muddy sand sediments
Supplementary information to assess the mud stock was obtained from the analyses of dredged sediments. Most dredging operations occur on shoals separating ebb-and flood channels. Since 1989 these areas have been monitored and the silt-clay fraction of the mineral part of the sediment analysed. The mineral part represented on the average 80% by weight of the total sediment. The silt-clay fraction ranged between 1 and 4% near the mouth of the estuary and stayed well below 2% in the lower estuary. In the middle estuary the silt-clay fraction increased above 25% showing a well pronounced maximum near Lillo (Figure 3).
The silt-clay fraction not only varied with distance from the mouth but also with time. This was illustrated with an example from a dredging site near Lillo where the highest concentrations of silt and clay were observed (Figure 4). It can be seen that between 1989 and 2002 the silt-clay fraction reached a maximum in 1991 and in 2001 and a minimum in 1989 and 1996. This fluctuation appears to be directly related to the fluctuation of the river load (Figure 5). During periods of high discharge of fluvial mud the silt-clay content appeared to be high and during periods of low fluvial mud discharge the silt-clay content appeared to be low. The correlation could be expressed as Ps + c = 0.06Qs (Ps + c = percent silt and clay, Qs = fluvial suspended matter load) with a correlation coefficient of R = 0.8. From this correlation it could be postulated that during periods of low fluvial supply, and thus less supply of mud, reworking of bottom sediments by resuspension resulted in a relative higher sand content in the bottom sediment. The amount of silt and clay particles in the bottom sediment is a function of the supply of particles from the suspended load which in turn is composed of particles brought in from outside and from resuspension. The amount of silt and clay particles that were exported from the system is a function of the amount of particles in the suspended load, and which increased with the amount of resuspended particles that were not immediately redeposited. If the concentration of particles brought in from outside becomes low then the fraction that is resuspended and exported will eventually become more important, leading to an impoverishment of silt and clay particles in the bottom sediment.
In respect to this it is important to notice that the residual currents in this area (Lillo) are in equilibrium giving rise to a residence time of several months (Wollast, 1988; Chen et al., 2005) and thus the resuspension process will become more important reinforcing the process of removal of silt and clay particles. Thus, when the suspended matter supply decreases the supply of fine silt and clay particles decreases correspondingly and the resuspension-deposition mechanism will cause a relative increase of the sand fraction in the bottom sediment.
Previous studies (Salden and van Maldegem, 1998; Baeyens et al., 1998) have described the mud budget in the Scheldt estuary giving special attention to the exchange of mud between the turbidity maximum area, chiefly occurring in the middle estuary, and the lower, marine dominated, estuary (Westerschelde). The basic principles for these studies were the estimated supply of suspended load from the river basin, suspended matter concentrations measured at several places in the estuary, monitoring data of dredging and dumping operations and exchange of suspended matter between the river and the docks through the sluice gates. The exchange with the bottom was never measured but was assumed to be the balance of the suspended load data.
This study attempted to verify these budgets starting from the calculated mud quantities in the bottom sediments, the mud stock, with the use of bottom maps built in 1964, 1986 and 1999 respectively and from a classification of the bottom sediments with the use of measured sediment properties (grain-size, carbonates, and organic matter).
The different bottom maps showed that mud deposits were most important in the middle estuary. Analyses of the successive maps from the middle estuary suggested an increase of the total mud stock in the middle estuary between 1964 and 1999. In the river channel, on the other hand, the mud stock decreased and this decrease was not steady but changed to the same extend as the supply of fluvial mud from the river. Besides a decreasing mud stock, the output to the lower estuary had decreased over the same period. Assuming none or only minor changes in intertidal sedimentation it seems that the majority of the increase of the total mud stock in the middle estuary must have been, in essence, at the expense of mud deposition in the access channels to the sluice gates of the harbour of Antwerp and in particular in the access channel to the Kallo sluice gate.
The ratio of the mud accumulation in the river channel to the fluvial supply seemed quite constant and could be estimated at 1:10. Furthermore it appeared that the silt-clay content of the bottom sediment in the river channel was correlated with the river mud supply. During years with a low average mud supply the silt-clay content was low and, inversely, during years with a high mud supply the silt-clay content was high. Resuspension-deposition mechanisms were presumably responsible for this. Resuspended material did not necessarily return to the river bottom but may have been removed from the river by settling in the access channels where a gyratory water movement and a drop in water velocity cause fast sedimentation. The final result was that, notwithstanding a decreased fluvial supply, the total accumulation of mud in the study had almost doubled. The construction of access channels to the sluices substantially increased the surface of the depositional area in the estuary and in addition the sluices were areas of low energy constituting ideal environments for the deposition of fine mud. Obviously the access channels acted as mud-traps at the cost of mud deposition in the main channel as well as at the cost of the output to the lower estuary, thus influencing in a non-negligible way the natural sediment budget. This process was not detected in previous sediment budgets which underestimated the mud accumulation in the middle estuary. Consequently, the export of suspended load to the lower estuary (Westerschelde) was lower than in the studies of Salden and van Maldegem (1998) and Baeyens et al. (1998).
The authors are very much indebted to ir. E. Taverniers (Administration Waterways and Maritime affairs, Section Maritime Entrance, Antwerp, Belgium) for his valuable help during this study. They also thank Dr. Reg Parker (Blackdown Consultants, UK) for his assistance during this research and for putting lithoprobe data at their disposal. The assiduous cooperation of F. Francken (Royal Belgian Institute of Natural Sciences) and of the captain and the crew of the RV Veremans during the field work is very much appreciated.