To increase our understanding of the relative control of antecedent topography, sediment supply and sea-level change on grain size and organic content in the Humber Estuary, the composition of its Holocene fill over space and time has been evaluated. A model based-upon bed-by-bed description of over 3500 boreholes was sub-horizontally ‘sliced’ to analyse the composition of sediments in different parts of the estuary across the entire fill, as well as solely within estuary marginal sequences. Results demonstrate that different physiographic parts of the estuary are characterized by distinctive vertical sediment profiles that reflect the relative control of antecedent topography as well as sea-level upon them. The results raise questions about the processes controlling sand and gravel abundance in marginal sequences, where the estuary is physiographically most constrained. This semi-quantitative evaluation, the first of its kind on an estuarine fill, provides a technique for interpreting the relative importance of controls on an estuarine system, but highlights the need for improved modelling of estuarine channel form and dynamics over centurial to millennial time-scales.
Scientists and policy-makers attempt to predict how coasts will respond to enhanced rates of sea-level rise over periods of up to hundreds of years, using models developed on present day processes as well as records of change over historical and Holocene (last 10 000 years) time-scales. Within estuaries, such records may be determined by sedimentary analysis, though interpretation is complex as the sediment record reflects not only sea-level history, but also the influence of antecedent topography within the estuary basin as well as the history of sediment supply. Disentangling these is pertinent to management of the Humber Estuary on England's North Sea coast, which has a significant sediment supply and occurs in a region of considerable morphological variation.
(a) Humber setting
The estuary (figure 1) has a catchment of about 24 000 km2 (about 20% of the area of England). From its 6.6 km-wide mouth at Spurn Point, its open waters extend 62 km landwards to Trent Falls, the confluence of the Rivers Ouse and Trent, though the tidal limit on these rivers is 120 and 145 km from the mouth. The mean tidal range at mouth is ca 5.8 m (springs; when the volume of the estuary ranges between 2.5 and 1.1 km3) and about 2.8 m (neaps) (ABP-NUDCM 1998).
The estuarine region is sub-divided into several distinctive areas that are informally named (figure 1). These fall into four parts (i–iv) which are physiographically distinctive today, though the differences between them will have been more marked in the Holocene (figure 2).
The Isle of Axholme and Dutch River areas at the landward end of the system occur within the basin, formerly occupied by the Late Pleistocene glacial Lake Humber (Gaunt 1994).
The Reads Island, Humber Bridge, Hull Roads, Foulholme and Sunk Island areas occur within the Humber Valley System, confined by higher ground, most notably the Chalk escarpments of the Yorkshire and Lincolnshire Wolds in the Humber Bridge area.
Spurn Bight, where the valley opens-out at the seaward end, represents the estuary mouth.
The Lincolnshire Marsh, which is not part of the estuary, has an open aspect, facing the North Sea.
The changing distribution of sedimentary environments (facies) during Holocene sea-level change has been addressed by many programmes (Long et al. 1998; Van de Noort & Ellis 1998), including the UK Natural Environment Research Council (NERC)-funded Land–Ocean Interaction Study (LOIS) (NERC 1994; Wilkinson et al. 1997; Shennan & Andrews 2000 and references therein). The latter study examined how environments have changed over time (Metcalfe et al. 2000) and identified distinctive diachronous estuarine sediment facies overlying freshwater post-glacial sediments. Rees et al. (2000) describe the facies as suites on the basis of their distinctive chemistry, and include the brackish Butterwick and Newland Suites, the saltmarsh-dominated Garthorpe Suite and the tidal mud flat, sand flat and channel systems associated with the Saltend and associated Suites (figure 3). Detailed analysis of the fabrics, geochemistry, heavy mineralogy and clay mineralogy of these was undertaken within several boreholes (Ridgway et al. 1998). However, on the basis that the suites represent a lithostratigraphic combination of facies, they were also identified within many records of existing rotary or percussion boreholes drilled in reclaimed parts of the estuary and archived at the National Geosciences Data Centre (NGRC). The amount and quality of information the boreholes supply varies enormously; many have only very basic lithological descriptions, but others have been logged in detail and provide comprehensive textural information. A comprehensive database (LOITH; Rees et al. 2000) of the Holocene sediments in over 3500 boreholes (the sites of which are shown in figure 1) was constructed, based upon detailed coding of all recorded characteristics of the Holocene sediments at a bed-by-bed level. The database allowed modelling of the three-dimensional distribution of the sediment suites within the estuary and demonstration of a progressive increase in the amount of marine sediment stored through the Holocene. Notably, modelling shows that much of the later Holocene sequence is typified by sediments deposited by high-energy tidal facies such as channel deposits, sandflats and sparsely vegetated mudflats (Rees et al. 2000). Today, saltmarsh only accounts for about 6.3 km2 of peripheral parts of the ca 100 km2 inter-tidal area, but in the Mid-Holocene, it probably extended over large parts of the estuary. The LOIS investigations also showed that the estuary is likely to have had a constant sediment supply throughout the Holocene (Ridgway et al. 1998; Rees et al. 2000); the erosion and tidal transport of muds, from coasts north of the estuary into the Humber, is likely to have prevailed for much of the last 8000 years (Shennan et al. 2003). Today, the estuary contains about 0.8 million tonnes in suspension in the summer, and about 1.6 in the winter (ABP-NUDCM 1998).
The manner in which the estuary is likely to respond to increased rates of change in sea-level, climate and human intervention is of concern because of the high value of coastal assets, particularly over 900 km2 of land and 300 000 people living below the 5 m contour (largely within the cities and towns of Hull, Immingham and Goole; figure 1). Extensive reclaimed areas are defended by sea-walls that are already expensive to maintain but with rising sea-levels (annually ca 1.1 mm historically, though projected to increase over the next 100 years; Woodworth et al. 1999), compounded by the likelihood of increased storminess and storm surges (Flather & Smith 1998), they will require considerable future maintenance and development (Rees et al. 1998b). It is within this context that the UK Environment Agency commissioned the Humber Estuary Geomorphological Studies phase 2 (Geo2) consortium to evaluate likely patterns of change over tens to hundreds of years in the estuary (Barham et al. 1999) to test, for instance, the potential for landward migration of the entire estuary system (comprising a range of facies and hydrodynamic regimes) with relative sea-level rise (‘rollover’; Allen 1990; Townend & Pethick 2002).
As part of the studies, the consortium decided to determine the extent to which long-term (Holocene) sediment compositions are influenced by antecedent topography, external sediment supply and sea-level change. The consortium initially decided that the entire estuary sequence cored in boreholes drilled in reclaimed parts of the estuary (hereafter referred to as the entire fill) should be evaluated, however, the boreholes drilled for LOIS show that the character of the sequences is substantially influenced by erosion associated with the active channel and subsequent deposition of the channel-related Saltend and associated Suites (figure 3). It was considered that it would also be useful to separately evaluate grain-size distributions in parts of the estuary removed from substantial channel influence. Consequently, sediments within the estuary marginal sequences associated with the Garthorpe Suite of Rees et al. (2000) were analysed independently. This saltmarsh-dominated suite was selected because it was deposited furthest from the active channel and is least likely to be affected by erosion. It also contains most indicators of peri-marine influence including reedswamps and fens (giving rise to peats) that are likely to give the best indication of subtle-changes in sedimentation in response to sea-level rise.
(a) Isolation of data relating to the sediment bodies
To enable calculation of the composition of the entire fill and marginal sequences, sub-sets of the LOITH database were isolated using grids of the sediment suite boundaries generated within LOIS (Rees et al. 2000). To interrogate only the estuarine sediments, the underlying freshwater sequence was removed from the database through the deletion of all lithological records below the point where each borehole intersects the grid representing the top of the sequence (figures 2 and 3). To interrogate only marginal sequences, all lithologies below and above the Garthorpe Suite were removed from the database in a similar fashion.
(b) Slicing of the sediment bodies
The scarcity of dated sediments within the Humber region precludes a quantitative (sample analysis based) assessment of sediment depositional patterns over time and space. Instead, more substantial, though lower-quality, sediment composition data contained within the lithological descriptions of boreholes in sub-sets of the LOITH database were used. By ‘sub-horizontally’ slicing, the fill of the estuary (represented by the sub-sets of the database) using a plane based on the coastal physiographic environmental continuum it may be expected that in parts of the estuary away from the active channel (for instance within the estuarine floodplain) similarly aged sediments from different areas should occur within the same slice. To do this, a joint topographic (Rees et al. 2000) and bathymetric (Institute of Geological Sciences 1968) surface model (Rees et al. 2000) was moved incrementally through the Holocene fill by the amount of sea-level rise calculated to have occurred during 200-year periods based on the smooth Holocene sea-level curve generated by Shennan et al. (2000) for the estuary as part of the LOIS (figure 4). Where each slice intersected the surface representing the top of the freshwater sequence (see above and figure 2) is shown in figure 4. The ‘slicing’ was performed within each of the boreholes in the sub-sets of the database using a series of queries.
(c) Calculation of relative grain size and organic content
To enable calculation of the relative grain size and organic content of sediments within each slice, it was necessary to use semi-quantitative methods. As fewer than 20 Humber boreholes have had both particle size and organic analysis undertaken on them, the descriptions provided by the many individuals who logged the borehole (including many non-specialists, and technicians working before the advent of semi-standardized nomenclature) have been used to determine the relative proportions of sediment components on a bed-by-bed level. The identification by most individuals logging cores of components, such as clay or silt (within mud), was recognized as probably unreliable, but it is likely that were able to identify muds (less than 0.063 mm) on the basis of their cohesive character and to distinguish gravels (above 2.0 mm) from sands. On this basis, it was considered reasonable to be able to estimate the relative proportions of peat, mud, sand and gravel (expressed as a percentage) contained in each slice for each area of the estuary. However, the quantification of mixed components within sediments, based on nomenclature, is not simple. The description of mixed sediments has been the subject of considerable attention (SLTT 2004 and references therein), but while standards have been established for future description (e.g. British Standards Institute 2000), little assessment has been made of ‘historic’ descriptions made before standards were set. Variation is significant, particularly in the analysis described here, where most boreholes used in the modelling were logged before the 1990s. Consequently, to derive information about the relative proportions of components in a sediment for this study, a group of 50 UK geologists, engineers and technicians were asked how they would name a mixed sediment example provided to them. Of the group, 56% describe lithologies by first describing the least common constituent, and successively describing constituents in ascending order until they finally describe the main constituent (e.g. a gravelly, muddy, sandy peat=gravel<mud<sand<peat). However, 44% of the group describe the dominant lithology last, but the qualifiers at the beginning of a name in descending order of abundance (e.g. a gravelly, sandy, muddy, peat=mud<sand<gravel<peat). To evaluate the error caused by the incorrect quantification of a lithology described in a descending format in an ascending fashion, the descriptions made by two individuals of mixed sediments in a Humber borehole in both formats were compared against measured components. Borehole HMB18 [SE 8288 0744] was selected as it is the Humber borehole for which most particle-size analysis (PSA) data (Malvern Diffraction Particle Sizer) on muds, sands and gravels (N=47) and loss on ignition data has been collected. Drilled at West Butterwick near the Trent (figure 1), it penetrated the Saltend, Garthorpe and Butterwick Suites (figure 3) before entering freshwater deposits (terminal depth 13.0 m). The difference between the content of samples as described and as determined by PSA was calculated as a percentage. Differences were averaged and standard deviations calculated (table 1).
The analysis shows that the mean average error in the estimate of peat, sand and gravel is less than 16%, except for muds which may reach 23%. These figures suggest that across the entire Humber dataset the average error in the estimate sediment components is likely to be less than 25% and that of peats, sands and gravels less than 20%. The standard deviations are large, but given the number of boreholes in each estuarine area, are likely to be of little significance.
The relative proportions of sediment components in the modelling were based on the assumption that descriptions were in an ascending format. To maintain simplicity and consistency, the thickness of each sediment component in a slice was multiplied by rwhere n is rank i of the ith component (table 2). Peats were taken to represent organic content (for a comprehensive review of Humber organic sediments, see Andrews et al. 2000); other biotic structures, such as rootlets or shells, were not considered within the lithology composition. These data were compiled within spreadsheets and vertical profiles, displaying the variations in the proportion of each constituent plotted out for the periods 8500 years BP to present (entire fill) and 7000–2000 years BP (marginal sequence) for each of the estuarine areas. Data from outside these ranges were too sparse to be useful for analysis (though have been considered in the analysis of the combination of all estuarine areas; see below).
Results of the analysis are presented in figures 5–7, and show variations in sediments preserved below reclaimed parts of all areas of the Humber, both for the entire fill and marginal sequences. Vertical profiles by area are shown in figures 5 and 6, and the variation along the length of the estuary in figure 7a,b. The vertical profiles, for the entire fill (figure 5), demonstrate considerable variation regionally, but, when averaged out across all areas (figure 7c), show an apparently simple, almost linear, decrease in sand upwards through the profiles. The main longitudinal patterns that may be recognized are the landwards increase of peat and the substantial sand content in the Isle of Axholme, Humber Bridge and Spurn Bight areas. The variations in vertical profiles across different areas in the marginal sequence alone (figure 6) are more variable than in entire fill, largely because of the peat-rich slices. Two peat-rich ‘events’ may be recognized, one in the lower part of the column (ca 5000–5900 years BP) and a smaller one in the upper part (ca 3400–3800 years BP), associated with the ‘Middle Peat’ (Rees et al. 2000). That the ‘events’ associated with each do not appear to occur at the same ‘time’ in all profiles reflects the likelihood that the peats developed at slightly different times locally, but also the fact that each slice is likely to be diachronous. (Seeing the variation in topography or bathymetry of the modern estuary, sub-horizontal slicing will clearly provide a simplistic model of changing sediment storage through time). Consideration must also be made of the different degrees of consolidation across the estuary fill; contemporaneously deposited sediments occur at different elevations because of differing degrees of autocompaction of underlying sediments, usually because of lithological differences (Allen 1999). While it is acknowledged that use of a modern topographic–bathymetric surface as a slicing mechanism probably generates artefacts when slicing sediments deposited in depositional regimes and geographies notably different to those of today, it is unlikely to have been particularly significant given the number of boreholes in the database, and the fact that they were drilled away from the main channel. Another distinguishing feature of most marginal sequence profiles is the higher proportion of sand and gravel in them before 4800 BP (see §5). The main feature of the longitudinal variation in the marginal sequence is the lower proportion of sand and gravel in the Isle of Axholme, Humber Bridge and Spurn Bight areas.
Analysis of profiles in figures 5 and 6 suggests that, despite considerable variability, they can be easily be related to the physiographic parts of the estuarine region. The Isle of Axholme and Dutch River areas are, when compared with others, typified by an abundance of peat and sand. The latter is likely to have been caused by reworking of sand from lacustrine, fluvial and aeolian Pleistocene deposits associated with Lake Humber and (to a lesser degree) sand-runoff from slopes. The predominance of sediments sourced from the areas of the Late Pleistocene Lake Humber in non-marine Holocene sediment supply was noted on geochemical grounds by Rees et al. (2000). The sustained supply of local sands through the Holocene is illustrated by the sand content in the marginal sequences being maintained. The distribution of sand in the area is not interpreted to be chiefly the result of channel-related processes because of the similar trends in the entire fill and marginal sequence profiles, which instead suggest local sediment supply and because Holocene channel events (Macklin 1999) cannot be identified. The scarcer sand in the Humber Valley System profiles (Reads Island, Humber Bridge, Hull Roads, Foulholme and Sunk Island areas) probably results less from the abundance of local sand and gravel entering the system, than the lower storage capacity and channel dynamics. The valley system exhibits across the entire fill, a notable upwards-fining, except in the uppermost metre in many cases—probably as a result of aeolian winnowing since reclamation. Given that the facies traversed from the edge of the modern estuary to the centre of the present channel (from marginal freshwater peats and brackish muds, through marine mudflats [including saltmarshes], and sandflats to channel sands; BGS 1990) progressively coarsen and that the same succession of facies is broadly seen from the base to the top of the ‘typical’ Humber Holocene sequence (figure 2), it may be expected that vertical profiles would also coarsen-upwards. That this is not the case in the entire fill vertical profiles (figures 5 and 7c) is because these contain many channel bodies that fine-upwards from sand-covered gravel lags, several of which were drilled within the LOIS programme (Rees et al. 2000). The base of these (which are commonly less than 2000 years old) generally occurs between about 8 and 10 m below ordnance datum (Rees et al. 1998a). This explains the abundance of sand and gravel towards the base of most of the vertical profiles in figure 5, and particularly in the Humber Bridge area where tidal energies may have been greatest where the valley system is most laterally constrained, but not the depletion of sand in the associated marginal sequence (see §5). The abundance of sand and gravel within the latter sequence before 4800 years BP also requires explanation (see §5). Compared with the entire fill and marginal sequence profiles of the Humber Valley System, those of Spurn Bight differ in that the entire fill exhibits notable coarsening-upwards over the last 4000 years and the marginal sequence suggests that this trend started earlier. This is interpreted here to result from the landwards migration of the Humber tidal delta system (the modern equivalent of which is represented by coastal sand complexes to the southeast of Grimsby; figures 5 and 6). It is the presence of the delta that is likely to cause the high sand content of the area compared with others in the system (figure 5). The upwards-coarsening trend in the upper part of the Lincolnshire Marsh profiles may also be attributable to the landward migration of the delta system. Otherwise, the marsh profiles are notable in containing little peat and a marked contrast between the sand content of the entire fill and marginal sequence vertical profiles.
(a) Relative impacts of sea-level, topography and sediment supply
The clear distinction between the profiles (both entire fill and marginal sequence) in different physiographic parts of the system suggests that antecedent topography has significantly impacted upon the character of sediments deposited in the estuary. This is likely to have affected channel form and fill. Whilst topographic controls may be pre-dominant with regard to controls on composition of the estuary fill, the impact of sea-level change is also seen. Vertical changes in the peat content of profile reflect variations (falls, or still-stands) in relative sea-level, particularly in the marginal sequence. In this respect it is unlikely that the Shennan et al. (2000) curve is entirely realistic. There is also reasonable evidence for landward ‘roll-over’ of the estuary system in relation to increasing relative sea-level. This is seen in the migration of the tidal delta system in entire fill, as well as the marginal sequence vertical profiles. In contrast to the good evidence for topographic and sea-level controls on the estuary, there is little evidence that the transport of water-borne sediments from outside the system has controlled the volume or timing of sand or gravel accumulation within the system. Whilst most sediment in the estuary comes from external, marine sources (Shennan et al. 2003), the apparent lack of synchronicity between coarse-grained accumulations either in entire fill or marginal sequence vertical profiles suggests this was not event-driven.
Despite limited uncertainties relating to nomenclature and diachronous slicing, when applied consistently, the methodology gives a picture of the relative abundance of peat, mud, sand or gravel (to within 25%) vertically through the estuarine sequence in any area, or between different areas of the estuary. The size of the dataset gives confidence that the variations in the profiles are real.
The analysis highlights some sediment distributions that invite further analysis with regard to their environmental controls, particularly regarding sand and gravel in marginal sequences of the Humber Valley System. While sediment distributions may be seen to have been dominantly controlled by the antecedent topography, the mechanisms and processes governing channel form, migration, erosion and sediment transport in the main estuary channel and sediment interchange with adjacent marginal parts of the estuary remain uncertain. This is highlighted by two questions posed by the profiles:
The first relates to why, in areas where channel deposits are sand-rich, are marginal sequences depleted in sand? In the Humber Bridge area where the valley is most constrained, the channel is sand and gravel rich, while the associated marginal sequence is notably depleted in these components (compare figure 7a with b). While this may relate to the smaller development of saltmarsh in this area because of physical constriction, leading to a lower channel density, this is uncertain. Saltmarsh may also be unlikely to form in the vicinity of the channel because of strong inter-tidal flows; it will occur in upper, storage, areas on the flanks of the reach that are supplied by weaker transverse flows on and off the storage area and hence only supplying finer sediment.
The second question addresses why is the marginal sequence of the valley system sand and gravel-rich before 4800 years BP? Most muds and sands are transported onto saltmarsh surfaces during high-water, particularly during spring tides and flood events; the gravels in the marginal sequences are accounted for by the presence of tidal channels within the saltmarshes (Allen 2000 and references therein). The gravels, dominated by till-derived pebbles, are typically less than 0.2 m thick, and overlain by saltmarsh muds. The decrease in their abundance after 4800 years BP could be accounted for by less: bedload sediment supply, high-energy events (such as storm surges), channel migration or stream transport of terrestrial material through marsh areas (with a decreasing base-level gradient).
Although bedload supply to the estuary from major river sources (Coulthard et al. 2000; Shennan et al. 2003) has varied, it is unlikely that it explains the trend shown here, particularly as most significant sediment pulses occurred since 4800 years BP (Macklin 1999). Furthermore, it would be expected that event signatures would be noted in all other parts of the system (they do not appear to be present in the Spurn Bight area), and be broadly synchronous; this is not the case. The same reasoning would also suggest that the enhanced sand and gravel content was not caused by high-energy events. It seems more plausible that channel migration caused the feature. However, as accommodation space (the volume in which sediment can be deposited, as determined by antecedent topography of the basin, sea-level change and sediment supply; Rees et al. 1998b, 2000) decreased rapidly only after 4800 years BP, it may be expected that associated channel migration would be more common in estuary marginal sequences after this time.
Given the energy needed for transport of sands and gravels into saltmarsh areas from the main channel, it may be more likely that the sands and gravels were transported into saltmarsh areas by streams adjacent to the estuary, in which case the abundance of sands and gravels in saltmarsh channels before 4800 years BP probably simply reflects greater transport potential of such streams during periods of lower base-level associated with sea-levels at least 4.5 m lower than today.
These questions highlight our poor understanding of estuarine channel morphology and processes over centurial to millennial time-scales. This may result from the small number of studies that have addressed estuarine dynamics over Holocene time-scales and the lack of ancient examples through their erosion during sea-level lowstands. The analysis described here, however, illustrates the need to understand better how sands and gravels are transported from the main estuary channel and adjacent catchments into channels draining saltmarshes. It also highlight the need to improve our understanding of channel dynamics where constrained by topography or accommodation space—probably through modelling. While the evaluation detailed here deals with a natural estuary, before extensive modification, such questions are not academic given the constrictions put onto the modern estuary by flood defences.
The described methodology provides a good three-dimensional picture of the relative abundance of peat, mud, sand or gravel (to within 25% by abundance) through the estuarine fill. The size of the dataset gives confidence that the variations seen in the profiles are real.
While the estuary fill volume has been determined by cumulative sediment accretion in line with sea-level, variations in grain size and organic content have almost totally been controlled by antecedent topography and sea-level. Topographical control is demonstrated by the internal similarity of profiles within different physiographically distinct parts of the estuary. Sea-level control may be seen in the position of the estuary mouth delta system, which appears to have migrated landwards, suggesting estuarine rollover, and in the distribution of peats.
The controls on the sand and gravel composition of estuary marginal sequences recognized in the analysis remain uncertain. Further modelling is required, particularly with regard to transport of sands and gravels from the main estuary channel and adjacent catchments into saltmarsh channels and into channel dynamics, where constrained by topography or accommodation space.
This work was commissioned by the Environment Agency, though would not have been possible without the substantial resources provided through the LOIS Humber Holocene investigations by NERC. Thanks are given to Patrick Bell, Rhonda Newsham (BGS) and Annette Parkes (Hull University) who carried out much of the work on the LOITH databases, the gridding and the PSA, respectively. Constructive comments on the manuscript from Ian Townend (ABP), Dave Prandle (POL), Peter Balson (BGS) and two anonymous referees were greatly appreciated. This paper is published with permission from the Director of the British Geological Survey (NERC).
One contribution of 20 to a Theme Issue ‘Sea level science’.
- © 2006 BGS/NERC