Submerged Prehistoric Archaeology, Landscapes in prehistory, Palaeoenvironment, and Marine Geophysics
15 | Between the Lines – enhancing methodologies for the exploration of extensive, inundated palaeolandscapes
Simon Fitch, Vincent Gaffney, Benjamin Gearey and Eleanor Ramsey Abstract: In recent years there has been an increasing appreciation of the archaeological potential of European coastal shelves. This interest has not, however, simply been associated with the progress of traditional maritime history but rather the development of a wider marine archaeology aimed at exploring the settlement and cultural sequences of the immense, prehistoric landscapes that lie off some of our coasts. Inaccessible until quite recently, the development of a variety of remote sensing technologies has made exploration of these inhospitable landscapes a real possibility. The information that is being provided from such work is fundamentally changing our perception of the archaeology of the Palaeolithic and Mesolithic. This paper discusses the application of legacy seismic data to map these palaeolandscapes and contrasts the value of 2D and 3D data for these purposes. The results of new research in the North Sea and off the west coast of the United Kingdom are presented and used to test wide sampling strategies that have the potential to explore areas of the sea that have not been mapped using 3D technologies. The paper argues that a range of prospection strategies should be employed within these environments and that they should be optimised to support specific research goals. In the light of marine development and current economic uncertainty, the paper argues that heritage curators and researchers must take full advantage of data sets that have cost billions to capture and that provide unparalleled opportunities for research and management.
Introduction In recent years there has been a burgeoning appreciation of the archaeological potential of the European coastal shelves. This interest has not, however, simply been associated with the progress of traditional maritime history but rather the development of a wider marine archaeology aimed at exploring the settlement and cultural sequences of the immense, prehistoric landscapes that lie off some of our coasts (Bailey et al. 2010; Coles 1998; Peeters et al. 2009). Inundated following the last glacial period, it is increasingly understood that many aspects of early prehistoric settlement evidence across the north western European littoral are incomplete and, perhaps, not truly comprehensible without an adequate understanding of these enigmatic and largely unexplored regions. It may therefore be surprising that, prior to the last decade, the landscape potential of the marine archaeological record was rarely appreciated – even within the discipline (Oxley & O’Regan 2001; Roberts & Trow 2002; Fleming 2004). The opportunity for marine sediments to contribute significantly to our knowledge of more than c. 900,000 years of intermittent hominin occupation of the northwest Europe should, perhaps, have been self evident1. For much of this time the area that would become mainland Britain was not actually an island but a peninsula of the contiguous continental land mass (Stringer 2006).The dramatic topographic changes indicated by the evidence for inundation and periodic emergence were largely the result of a series of glaciations during which global temperatures declined, ice sheets expanded and sea levels fell, at times by as much as 125m (Wenban-Smith 2002). Reconstruction of the nature and pace of change is, however, complex and there are a number of competing models frequently cited within the literature. These include Jelgersma (1979), Lambeck (1995), Peltier et al. (2002) and Shennan et al. (2000, 2006). However, these studies tend to generate data that are relatively coarse and concerns have been expressed as to whether their output is adequate for the requirements of archaeological landscape reconstruction (Ward & Larcombe 2008). Archaeologically, there are equally difficult interpretational issues relating to these emergent plains. In most instances, the lack of evidence for settlement, and the unlikely success of exploration, has led to the marginalisation of the area within the literature or, if considered, the relegation of inundated regions such as that within the North Sea to the status of a land bridge supporting colonisation routes into the landscape that would eventually become Britain. There
1
As this paper was concluded finds from Happisburgh in Norfolk suggested that hominin occupation of the United Kingdom might be extended as par back as 900,000 (letter to Nature. Parfitt et al. (2010) Early Pleistocene human occupation at the edge of the boreal zone in northwest Europe. Nature 466, 229-33 (8 July 2010)
2
EAC OCCASIONAL PAPER NO. 5
was relatively little appreciation that such areas could support a vast, resource-rich countryside that was available for settlement and, indeed, may often have been more attractive for occupation than the terrestrial zones that now provide much of our information on early prehistory. For the period under consideration here, the Holocene, the largest of the inundated areas within the North Sea basin has been christened Doggerland (Coles 1998; Gaffney et al. 2007). This was a vast plain, bisected by large rivers that, after the end of the Devensian glaciation was briefly available for settlement and then gradually covered by the rising sea until about 7,000 years ago, when the current shorelines around the North Sea were established. Understanding the archaeology of this period is a major challenge even though available evidence hints that substantial deposits and evidence may be preserved within the marine environment (Bailey et al. 2010; Dix & Westley 2006; Fleming 2004). There are individual archaeological finds from as far north as the Viking Banks (Long et al. 1986), significant concentrations of archaeological and palaeoenvironmental finds across the southern North Sea (Reed 1913; Louwe Kooijmans 1970) and records of submerged sites including that at Bouldnor Cliff in the Solent (Momber 2000). Indeed, the relict landscapes identified in the North Sea, off the south coast of England and within the Irish Sea, are so large that it cannot be certain that the societies that inhabited these regions are directly comparable to those attested within terrestrial contexts (Gaffney et al. 2009). Marine landscapes offer new and intriguing opportunities for exploration and discovery not previously faced by terrestrial archaeologists. Moreover, development and dewatering on land threatens those terrestrial ecosystems that retain substantial palaeoenvironmental deposits essential to our understanding of the environments that supported early prehistoric societies. Consequently, the increasing evidence for extensive caches of palaeosediments within marine environments achieves huge importance. However, the areas under investigation are supra-national in scale and may be masked by tens of metres of water or sediment. This exceptional archaeological resource provides those archaeologists who wish to explore the landscape, and the heritage managers who seek to preserve them, with a unique set of legal, technical and methodological challenges2.
2
It is no wonder that one reviewer recently suggested that exploration of the inundated landscapes that exist across the globe may well prove to be one of the last great challenges for archaeology (Bailey 2010). A step change in our attitude towards these landscapes has been the consequence of technological development (COWRIE 2010; see Firth this volume). Marine archaeology, of course, has had recourse to a variety of data sources when exploring marine environments and these may include excavation, where conditions permit, seabed sampling, shallow coring, bathymetric survey, and a variety of remote sensing technologies including seismic reflection profiles. Some of these datasets may have been acquired for a variety of non-archaeological purposes and possess differing characteristics and utility for archaeological application. For example, surveys collecting seabed samples or involving shallow coring can provide detailed chronological, sedimentological and environmental data but frequently have a relatively poor spatial framework. High-resolution bathymetry can provide excellent images of the seabed topography but cannot represent submerged features that lack a bathymetric expression. For early prehistory, the requirement for regionally extensive data across the entire area of the continental shelf is such that, aside from precision and contiguity, issues of scale and resolution are also of considerable importance. Currently, it is only seismic reflection datasets are likely to provide maps for inundated later prehistoric landscape features at a regional level. However, marine seismic acquisition is also undertaken for a variety of purposes and involves varying data densities, coverage, depths of penetration and resolution. Consequently, there is often a choice to be made when using such data and it is entirely possible that individual surveys may not be appropriate for use by archaeologists with specific research agendas. The decision to use such information will therefore depend upon archaeological requirement and the fit of available data on the grounds of resolution or scale of survey. In many ways this position is not so different to that experienced by terrestrial archaeologists who often have valid reasons to choose spatially extensive, low-resolution sensors in preference to high-resolution sensors (Gaffney & Gater 2003). The latter technologies may often only operate at site level and have little relevance to research that is concerned with the investigation of geomorphology or behaviour at landscape level. Within a marine context, extensive datasets, often characterised by their low resolution, may not initially appear to support the requirements of detailed archaeological investigation. However, they can provide an invaluable topographic framework to guide detailed work or into which higher resolution survey, shallow boreholes, seabed samples and bathymetric data can be integrated (Gaffney et al. 2007, 2009). These extensive data may also be used within modelling programmes which may not be supported by less extensive datasets. We can explore some of these issues by considering the nature of these data sets and some examples of their recent use (see also Firth this volume for a discussion of survey techniques).
The wider legal issues associated with management of cultural resources are complex and will not be discussed here. For Britain these have recently been summarised in a forthcoming guidance note produced by COWRIE (Collaborative Offshore Wind Research into the Environment) dated 17/May/2010, and also extensively in a succession of Strategic Environmental Assessment (SEA) reports which can be consulted and downloaded at http:// www.offshore-sea.org.uk/site/scripts/sea_archive.php. Whilst much discussion in the past has tended to equate marine archaeology with the study of wrecks, recent guidance has increasingly emphasised marine environments as landscapes in their own right and the relevance of SEA directives and the Council of Europe European Landscape Convention to promote the protection, management and planning of European marine landscapes is increasingly evident (Roberts & Trow 2002).
15 Between the Lines
3
Figure 15.1: Typical marine seismic reflection acquisition. From Gaffney et al. 2009, figure 3.4.
Seismic reflection survey Seismic reflection surveying involves the transmission of acoustic energy into the subsurface and recording the energy reflected from acoustic impendence contrasts. The reflections produced as acoustic impendence contrasts are predominantly the product of changes in lithology. With appropriate processing this allows the production of pseudo-depth sections of the subsurface structure with the vertical axis being the two-way travel time to the reflector. Although the basics of this technique are common, the details vary according to a range of applications including the investigation of deep crustal structure, hydrocarbon exploration and near seabed sediment structure (e.g. Salomonsen & Jensen, 1994; Velegrakis et al., 1999; Praeg 2003; Bulat 2005). These diverse applications dictate different acquisition parameters that in turn determine the resolution and depth of penetration of the survey as well as the costs involved in acquiring the data. Consequently, the relative merits of a range of available seismic reflection data types needs to be assessed when considering the investigation of submerged, and partially buried features. Standard marine acquisition involves towing an energy source and a cable (streamer), containing pressure sensitive receivers, to record the reflections from the underlying strata (Figure 15.1). In single fold data, only one reflection is received from any point in the subsurface. However, many seismic profiles are multifold and reflections can then be summed in order to increase the signal-to-noise ratio of the seismic profile. Traditional seismic reflection data is generally referred to as 2D as it is acquired as a series of discrete vertical profiles using a single streamer towed behind the vessel (Figure 15.1). In contrast, 3D reflection seismic data involves the towing of multiple streamers which support the rapid collection of a series of closely spaced lines (Figure 15.2). This survey configuration provides significant advantages. Seismic response
Figure 15.2: Typical 3D marine seismic reflection acquisition. From Gaffney et al. 2009, figure 3.5.
4
EAC OCCASIONAL PAPER NO. 5
Figure 15. 3: A cube of conventional 3D data with a section removed to demonstrate the ability to slice the data in the horizontal and vertical dimensions and to reveal landscapes of differing age. From Gaffney et al. 2009, figure 5.14.
is correctly positioned in space and, in the case of data acquired for hydrocarbon exploration, is ‘binned’ within data volumes with a resolution of 12.5m × 12.5m × 4 milliseconds, or multiples thereof. Once treated in this manner a feature can be mapped from bin to bin, removing the potential errors involved in the interpretation of 2D data. Moreover, instead of relying on vertical profiles, the volume can be sliced in any direction. Of particular importance to the investigation of relatively shallow, and flat, landscape features is the ability to produce a horizontal slice (time slice) through the data as this can, in many cases, be interpreted as a map showing a range of sedimentary features (Figure 15.3). The interpretation of 3D seismic data has improved significantly in recent years due to the development of a range of new techniques originally designed to improve geological interpretation for
Figure 15. 4: A volume model of the data illustrated in Figure 3. A Holocene river channel (blue) overlies an older tunnel valley (gold and infilling sediments in purple). From Gaffney et al. 2009, figure 3.9.
hydrocarbon exploration and production. Once a stratigraphic marker of interest has been identified, it can be mapped across the 3D seismic volume to produce a horizon that may have a geomorphological or chrono-stratigraphic value and, in some cases, the output can approximate the original land surface itself. The value of such data for the interpretation and analysis of inundated landscapes and modelling past settlement or land use should be clear. Another advance in 3D seismic interpretation has been the development of opacity rendering techniques (Kidd 1999). Following conversion of conventional 3D seismic data into a voxel (3d pixel) volume, each voxel contains information from the original portion of the 3D seismic volume that it occupies together with an additional user-defined variable that controls
15 Between the Lines
Figure 15. 5: Study areas referred to within the text 1) North Sea Palaeolandscape Project 2) Humber REC 3-4) West Coast Palaeolandscape Project. ASTER DEM is a product of METI and NASA. ETOPO2v2 is the property of the National Geophysical Data Centre, NOAA, US Dept of Commerce.
5
its opacity. The opacity of individual voxels can then be varied as a function of their seismic amplitude (or any other attribute), allowing the user to examine only those voxels that fall within the particular amplitude (or attribute) range of interest. By using appropriate opacity filters it is possible to image depositional systems such as buried fluvial channels. This exploits seismic characteristics, which are in part lithologically dependent, and different from the surrounding materials, thus permitting the surrounding strata to be made transparent whilst preserving all but the smallest channels as opaque features (Fitch et al. 2005). In archaeological terms such processing also provides further insight into the stratigraphic relationship of features identified and, through their volume and sedimentary characteristics, the opportunity to assess whether such features have the potential for preservation of archaeological or environmental data (Figure 15.4). Generally, the ideal dataset for the investigation of submerged prehistoric landscapes within the region would be high-resolution (>100Hz) 3D seismic data with appropriate borehole control. Such a dataset would provide high (metre or less) vertical and lateral resolution and a laterally continuous coverage.
Unfortunately, such systems involve slower survey rates, higher costs and do not usually provide the extensive output required to explore landscapes at a supra-national scale. Commercial 3D seismic datasets, which possess a significantly coarser resolution, may appear to be less suitable for archaeological exploration but even these can provide maps containing important information from shallow deposits. Consequently, even with a bin spacing of 50m, the spatial coverage of such datasets and published outputs demonstrate that these data have the potential to provide an extensive reconnaissance tool for the investigation of submerged landscapes. The application of such technologies, within the Holocene at least, has been amply demonstrated by a number of projects carried out within British waters (Figure 15.5). Regional survey within British maritime waters Much recent research on British marine palaeolandscapes, beginning with the North Sea Palaeolandscapes Project (NSPP), has been carried out by the authors, and through the IBM Visual and Spatial Technology Centre (VISTA) and Birmingham ArchaeoEnvironmental (BAE) laboratory at the
6
EAC OCCASIONAL PAPER NO. 5
Figure 15.6: General plan of topographic features identified through the North Sea Palaeolandscapes Project (Gaffney et al. 2007, 2009). ASTER DEM is a product of METI and NASA. ETOPO2v2 is the property of the National Geophysical Data Centre, NOAA, US Dept of Commerce.
University of Birmingham. Much of the detail of this research has been published and need not be repeated here but the NSPP research team undertook mapping of the submerged European Mesolithic landscape known as Doggerland over an area of about 23,000km2 of the English sector of the North Sea (Gaffney et al. 2007, 2009). The opportunity for landscape investigation in such unprecedented detail was facilitated by access to 3D seismic data collected mainly for use by the petroleum industry. The primary dataset for the NSPP was provided by PGS Ltd and was represented by the merged 3D seismic dataset known as the ‘Southern North Sea Megamerge’ (PGS 2005). In archaeological terms, the data has a relatively coarse resolution (between 12.5m and 50m), although the intrinsic 3D nature of the data, and its landscape scale, facilitates the production of maps containing information from several metres of Holocene strata. Consequently, the data demonstrated that the Dogger Bank formed an emergent plain during the Holocene with complex meandering river systems and associated tributary channels and lakes dominating the region.
This project is ongoing. The National Atmospheric and Oceanographic Administration (NOAA) recently agreed to support further work focusing on the Dutch Sector, with the aim of producing a unified map of the landscape across the southern North Sea for the early Mesolithic period (Figure 15.6). There can be few who would argue that the results of the NSPP have been little less than revolutionary to our understanding of these landscapes, although developments in petroleum industry data collection suggest that the current situation can be further improved through the implementation of extensive, high definition 3D (HD3D) survey. These surveys potentially offer greatly improved vertical resolution and feature definition (Figure 15.7; Long 2003). This has been achieved by technology that supports acquisition of data through a denser 3D spatial sampling grid and improved frequency bandwidth recovery than was available to traditional 3D seismic reflection surveys. Although not widely accessible at present, archaeological research undertaken at the University
15 Between the Lines
Figure 15.7: A Shallow time slice from HD3D data in Qatar. The resolution of the complex meandering channel systems is outstanding, and signalto-noise quality is a step improvement over historical seismic data in such shallow waters. Data courtesy of Maersk Petroleum (Qatar) Ltd.
7
of Birmingham on HD3D data from the Arabian Gulf suggests that this data may be eminently suitable for the exploration of areas where noise or water depth are issues (Cuttler et al. 2010; Mueller et al. 2006). The acquisition of such data is becoming more frequent within the mineral sector and, as access to surveys is improved, future archaeological applications are likely to result in finer resolution, broad area, palaeolandscape investigations. Unfortunately, extensive HD3D datasets are only rarely available for archaeological study, whilst contiguous, traditional 3D surveys are not universal in their coverage or always accessible to archaeologists for a variety of commercial or logistical reasons. These are important points. We can be confident that past, habitable environments on the European coastal shelves are considerably more extensive than the available 3D data sets and in the absence of such information we remain essentially uninformed of the conditions or significance of these wider regions. Consequently, there is an imperative not simply to refine our methodologies relating to 3D seismic interpretation, but to explore and implement novel methodologies for areas where 3D data does not exist or is restricted in access. If this is not attempted we risk missing data critical to our understanding of earlier prehistoric settlement. Heritage managers, unable to manage a resource that is defined essentially by what we do not know, will be prevented from protecting what may be a one of the largest and best-preserved cultural resources in Europe, or indeed the world. Researchers at Birmingham have begun to explore this problem using 2D surveys, as these have a greater geographical spread than 3D surveys and therefore,
potentially, can be used to supplement contiguous surveys or provide data in their absence (Figure 15.8). With respect to the use of 2D data there are clearly two issues that deserve mention. Initially, it is generally true that 2D surveys have a greater vertical resolution than extensive 3D surveys. This suggests that these data can be used to clarify and improve the interpretation of 3D data where the data sets, and features of interest, are congruent. Unfortunately, when comparing the interpretation of extensive horizontal 3D data with 2D surveys, the acquisition pattern for 2D data often appears problematic. This frequently involves the collection of multiple profiles with a spacing that may be several orders of magnitude greater than the trace spacing (i.e. the horizontal sampling interval along the profile). This method of acquisition has significant disadvantages for archaeological prospection. Firstly, the reflected seismic energy is assumed to have originated from a point directly beneath the profile even though it could have originated from a point laterally offset from it. Secondly, the spacing between lines may be so wide that it can be difficult to map the position of a feature across the region of interest. Figure 15.9 demonstrates how wide line spacing can lead to several equally valid interpretations. There are therefore a number of issues with the use of 2D data as an exploratory landscape tool and these include: ● whether 2D data can be used to provide landscape interpretations in areas where 3D data is either unavailable or absent and ● whether 2D data can be used to refine interpretation based on traditional 3D survey.
8
EAC OCCASIONAL PAPER NO. 5
Figure 15.8: 2D lines (grey) and 3D survey (polygons) availability around the mainland UK. ASTER DEM is a product of METI and NASA. ETOPO2v2 is the property of the National Geophysical Data Centre, NOAA, US Dept of Commerce.
Figure 15.9: (a-d) Four possible interpretations of a channel morphology based on a coarse 2D seismic grid. From Gaffney et al. figure 3.6.
15 Between the Lines
Figure 15.10: Location of the WCPP pilot area (red outline), shown in relation to the coast of England and Wales. ASTER DEM is a product of METI and NASA. ETOPO2v2 is the property of the National Geophysical Data Centre, NOAA, US Dept of Commerce.
9
Using 2D data A specific output of the NSPP of particularly significance to this paper is the data audit carried out to identify the sources of remote sensed data potentially available for the study of inundated landscapes around the English coastline (Bunch et al. 2010). This study identified several areas around the United Kingdom where sufficient data, often in the form of older 2D survey, might support studies similar to the NSPP and which could contribute to our understanding of the development of the Mesolithic and, potentially, Palaeolithic periods in England and Wales, Ireland, Scotland and the Isle of Man. The west coast of Britain was identified as a key area that possessed a significant data resource and, through the work of Professor Martin Bell, had also been subject to a significant regional research programme on early Holocene archaeology (Bell 2008). In 2009 English Heritage commissioned a pilot project in the Irish Sea through the Marine Aggregates Levy Sustainability Fund (Figure 15.10). The aim of this work was to investigate the utility of available 2D survey data for extensive mapping and this was anticipated as a precursor to a larger project off the west coast of the United Kingdom. The situation of the Irish Sea is comparable to that of the North Sea in the existence of a significant coverage of traditional 2D and high-resolution 2D seismic lines, but differs with respect of the lesser availability of 3D data. It therefore provides an ideal test bed to develop a methodology for investigating coastal areas with variable data availability (Figure 15.10). The overall aim of the pilot study, known as the West Coast Palaeolandscapes Pilot (WCPP), was therefore to develop a methodology that might be utilised in areas where existing 3D seismic data coverage was limited or absent. Such a methodology had to be able to provide baseline data to facilitate future management of the submerged prehistoric resource, through the limited
10
EAC OCCASIONAL PAPER NO. 5
Berr/Dti Object of interest Line Name H751-11 H751-13 H751-15 H751-28N H751-30N H751-32N H751-17 H751-34 H751-36 Object of interest Line Name SW81-720(2) SW81-720(3) SW81-720(4) SW81-721 SW81-722 SW81-723 U038-87-10A U038-87-11 U038-87-12 U038-87-12A/1 U038-87-13 Available geomatics sensors and data Length 14.1 25.9 25.7 31.7 22.12 20.4 20.84 18.8 20.6 Available geomatics sensors and data Length 4.04 4.06 10.12 13.0 19.4 15.8 4.31 9.33 2.1 3.69 2.58
Table 15.1: 2D lines acquired for use within the West Coast Palaeolandscapes Pilot.
CS9NAME HY752D1002 HY752D1002 HY752D1002 HY752D1002 HY752D1002 HY752D1002 HY752D1002 HY752D1002 HY752D1002 UKOGL Operator SHELL SHELL SHELL SHELL SHELL SHELL ULTRAMAR ULTRAMAR ULTRAMAR ULTRAMAR ULTRAMAR BGS Survey BGS BGS BGS
contractor Centrica, and provided to the University of Birmingham for research purposes. The resulting digital survey has a bin spacing of 12.5m. Quality of the data was good and it responded well to serial time slicing. The data also proved to be of suitable for full processing and archaeological interpretation. Initial results suggest that the main limitation of this dataset for archaeological research is due to the relatively ephemeral characteristics of the Holocene deposits within the area. Information was confined to a relatively small number of slices, and thus a small vertical resolution. This was, however, an issue of the prevailing geology, rather than a feature of the dataset itself. The earliest landscape features identified by this project were the two broad areas of Late Upper Palaeolithic fluvio-glacial plains (Figures 15.11 & 15.12). These were present beneath palaeo-periglacial outwash and are significant as they were formed by drainage from nearby glaciers, possibly in the English Lake District. In landscape terms, they potentially acted as a barrier to human movement and, as a consequence, suggest a low potential for archaeological remains. However, the identification of a mammoth tusk during monitoring in the Humber region (Wessex Archaeology report to BMAPA 2006) suggests that these areas are not necessarily devoid of archaeological potential. The Early Mesolithic was represented within the study area by two large river systems (Figure 15.13), along with several other smaller river systems. These have been identified as braided or anastomosed rivers (Rosgen 1994). These types of rivers provide rich and varied environments with the potential to sustain a wide variety of animal and plant resources. As a consequence, these represent highly attractive areas for Mesolithic activity. Other features identified within the data include several areas of higher ground (Figure 15.13). These may well have acted as focal points in the wider landscape or provided opportunities to observe game. Such upstanding features are also of importance as they represent areas that would have formed islands during inundation and represent, the final possibly habitable areas within the region. The largest of these upstanding features in Figure 15.13 would also have been attractive to human populations due to the proximity of two large river systems, one to the south of the area of higher ground, and another cutting through it. 2D data sets acquired for use by the project derived from three sources (Table 15.1 & Figure 15.14). The first, obtained by the British Geological Survey between 1968 and 1972, comprised three shallow seismic surveys which ran across extensive sections of the study area. Data from these surveys were derived from a combination of common seismic sources (‘sparker’ and ‘pinger’). The seismic data were made available in the form of scanned paper rolls in TIFF format and the corresponding survey track plots available on DVD. There was some variability in quality within these surveys. The selected sparker datasets were adequate
Line Name Line 11 1972 Line 3 1972 1968
Length 51 33.5 34.5
mapping and identification of the Late Palaeolithic and Mesolithic landscapes. After consideration, the WCPP team adopted the following methodology. 1) A standard 3D seismic dataset within the pilot study area was utilised to identify archaeological features within the study area. 2) A small number of 2D seismic data over and around this area was obtained for comparative purposes. 3) The 2D datasets were then investigated to determine if they contained the features identified within the 3D dataset and the intersections recorded. 4) The spatial configuration of all existing 2D datasets within the region were then assessed against known features to ascertain their capacity to provide adequate landscape data across the pilot study area. The selected 3D dataset used within the project consisted of a single standard seismic survey covering an extensive area of the west. This survey, known as ‘Morecambe and Satellites’, is a 3D seismic reflection survey acquired using standard technology by the
15 Between the Lines
Figure 15.11: Timeslice image of the 3D dataset at 0.076s. Data courtesy of Centrica UK Ltd.
11
for full processing and archaeological interpretation based on the frequency, range and filter settings chosen during acquisition. The pinger datasets visualise a very shallow section of the seabed and the indistinct images were less reliable for archaeological interpretation. A further selection of traditional petroleum industry 2D seismic surveys was guided by the requirement for data that would intersect with several of the most significant features identified from the 3D analysis. These data were originally obtained by S&A Geophysical Ltd for Hydrocarbon Resources Ltd in 1975, and released to the British Department of Trade and Industry (DTI) in 1980. The original data were obtained from BERR/ Phoenix (representing the DTI’s data store), which holds a significant number of surveys from this area. The data is stored on original paper survey rolls which were subsequently scanned by Phoenix Data Solutions Ltd for conversion to standard SEG-Y files. The rolls were also scanned and provided in TIFF format. However, data compression techniques used on the images prevented these images from being displayed reliably. Although the data was of variable quality, even when the selected lines originated from the same survey, the digital format allowed application of a range of processing techniques to optimise the data for interpretation. Unfortunately, some lines remained unsuitable for analysis and it is acknowledged that
the scanning/conversion of the paper records into digital formats also potentially introduced an unknown element of error into the process of interpretation. The final 2D dataset consisted of two UKOGL shallow seismic surveys comprising 11 survey lines covering the onshore sections of the study area. These were obtained during 1981 and 1987 respectively. The 1981 survey was undertaken by Prakla Seismos for Shell UK Ltd and transcribed for the UKOGL Veritas data services UK Ltd. The survey data was acquired utilising a standard airgun as a source and recorded directly to tape as SEG-Y 32 bit floating-point data. The 1987 survey was undertaken by Horizon Ltd for Ultramar Exploration and processed at Horizon Exploration Ltd. Data from both surveys was provided to Birmingham for research purposes by UKOGL. Initial examination of the 2D seismic lines suggested that a combination of standard interpretation procedures coupled with associated GIS recording should be employed during analysis. Initially, the data was examined utilising standard seismic-stratigraphic procedures (Mitchum et al. 1977). Digital 2D data were imported into SMT Kingdom 8.2 (64bit) seismic analysis software and a seismic attribute analysis performed. Generation of this information, however, failed to
12
EAC OCCASIONAL PAPER NO. 5
Figure 15.12: The Early Holocene landscape within the WCPP study area. The light green zone reflects near shore areas which would have represented higher ground, whilst the dark green areas for a lower lying plain. ASTER DEM is a product of METI and NASA. ETOPO2v2 is the property of the National Geophysical Data Centre, NOAA, US Dept of Commerce.
identify any new features and there was only minor refinement of the identified features. Once completed, interpretation of anomalies was undertaken directly within the seismic analysis software and recorded as a series of ‘culture’ files. This information was exported directly into the project GIS for further processing. The scanned analogue (paper) data were converted into digital SEG-Y files using the Chesapeake ImagetoSEGY software. As the corresponding survey
track log data had been provided this was added to the image file during conversion. Once this had been achieved, processing followed the standard digital 2D interpretation process with individual incised features, and possible landscape features, being recorded. It is important to note that this method does not provide as precise a location for these features as might be achieved with the original survey data. Despite this, the error margin associated with the locations is in the range of metres and the accuracy achieved is sufficient
15 Between the Lines
13
Figure 15.13: Map of palaeolandscape features identified within the 3D seismic data set against Key: Blue = Probable Holocene fluvial channels and related features: Red = Geological features forming regional highs: Green = Late Upper Palaeolithic fluvioglaical floodplains. ASTER DEM is a product of METI and NASA. ETOPO2v2 is the property of the National Geophysical Data Centre, NOAA, US Dept of Commerce.
to permit future surveys to target features of interest with relative ease. Examination of the 2D datasets chosen for detailed study demonstrated their capacity to identify features of interest (Figure 15.15). Whilst a number of deeper features were not observed within the BGS datasets, the fluvioglacial plain and several of the target palaeochannels, which clustered in the deep-water
sections of the study area, were identified during analysis. The results, however, were undoubtedly constrained by the availability of data for analysis, which was restricted to only a small number of lines for exploratory purposes and also by data quality in some areas. The assessment suggested that results were not strongly dependent upon the age of the data. Analysis of the earliest survey line available, from 1968, provided some of the best results obtained. Those from 1972
14
EAC OCCASIONAL PAPER NO. 5
Figure 15.14: 2D lines used within the West Coast Palaeolandscapes Pilot Project study area. ASTER DEM is a product of METI and NASA. ETOPO2v2 is the property of the National Geophysical Data Centre, NOAA, US Dept of Commerce.
contained some noise and discontinuous reflectors. Furthermore, it was observed that a pinger dataset from 1968, although characterized by poor penetration, was still able to provide information on features of interest (Figure 15.16). Despite this, the legacy data did, initially, provide a worryingly high failure rate in terms of lines that might not be susceptible for analysis. Initially, this was assessed as high as 40% from the small sample of lines selected for analysis as part of the pilot project. However, these lines were themselves from a limited number of surveys and later work suggests that the overall failure rate is considerably lower when data is available from multiple surveys. Figure 15.17 illustrates the line failures in a larger sample of data within the Irish Sea and currently being analyzed as part West Coast Palaeolandscapes Main Project3. This includes data on 239 lines in total, 46 of which were corrupted, or partially corrupted, and unusable. This represented
3
a 19.25% failure rate overall. However, there is a lower failure rate in terms of line length which amounted to 14.75% in total or 797km from a total of 5,404km. Examination of the Seazone Solutions bathymetric data4 as a supporting data source suggested that the only major deep recognisable within the dataset, and which may have had palaeogeographic significance, was one associated with Morecambe Bay, and this was confirmed by the UKOGL 2D datasets which crossed this area. Consequently, whilst bathymetric data acted as a valuable background for the work, it only possessed a minor capacity to identify features of archaeological significance outside the shallow marine zone. The results indicated that it was possible to identify landscape features within the available legacy dataset. The primary features within the 3D data sets were identified and, with a degree of caution, it is likely that the overall trend of the main fluvial system in this area could be identified. What was also apparent was that the fluvioglacial plains were less well resolved within the 2D data, primarily because of their extensive nature. This may, in part, also be due to the issues of noise and
4
The West Coast Palaeolandscapes Main Project is funded by the Aggregates Levy Sustainability Fund through English Heritage and the Welsh Assembly. Partners include the IBM Visual and Spatial Technology Centre (University of Birmingham), Dyfed Archaeological Trust, and the Royal Commission on the Ancient and Historical Monuments of Wales
Seazone data licence no. 022010.003
15 Between the Lines
15
poorer spatial coverage across these areas. However, the results overall suggested that the most important issue relating to the use of 2D data sets for analysis is not generally their capacity to provide information so much as their overall density and availability. Consequently, it seemed reasonable to assess what the impact of varying 2D data availability might be in relation to the known results within the study area. To support this it was necessary to generate a map of
the intersections derived from a range of potentially available 2D line configurations within the area. This required the following steps. ● Digital GIS layers containing the locations of all 2D surveys within the area were imported, irrespective of whether these had been analysed or not, along with associated track log information.
Figure 15.15: Observed intersections between features observed within the 3D dataset and the 2D seismic datasets. Fluvial features are given in blue. The fluvioglacial plain is shown in green. ASTER DEM is a product of METI and NASA. ETOPO2v2 is the property of the National Geophysical Data Centre, NOAA, US Dept of Commerce.
16
EAC OCCASIONAL PAPER NO. 5
Figure 15.16: Palaeochannel feature located with the 1968 pinger data. Data courtesy of BGS.
● A point shapefile was generated of all of the 2D lines that intersected the features identified within the 3D seismic data. Points were created along each of the cropped lines at no less than 50m intervals, a resolution comparable to most 3D datasets. ● If the points overlay a feature identified from the 3D data the point was assigned an attribute relating to the underlying feature (Figure 15.18). This procedure generated an ‘ideal’ feature dataset of the maximum number of features that might be recorded using the existing 2D dataset. The results indicated, not surprisingly, that the numerous intersections of the 2D data with underlying features suggest, visually at least, that full access to all available 2D data would provide an adequate approximation of the underlying archaeological landscapes (Figure 15.18). This dataset was investigated further by decimating the available lines to explore the impact of varying availability of data. A numeric (double) attribute field – RandomS – was created for lines, and filled with random values between 0 and 1. Sub-samples of 10%, 20% and 50%, designed to reflect various data availabilities, were then generated by selecting by attribute at RandomS <= 0.1, 0.2, and 0.5. As during the previous procedure, points at no less than 50m were again generated for each of these sub-samples (Figures 15.19 & 15.20). Visual inspection of the data suggested that the definition of underlying features varied considerably in relation to line sample size. It was determined that at 10%, although features were clearly present, definition of size and alignment was poor. However, definition was considerably improved by a 20% data sample, and proved nearly as good as extensive coverage at 50%. It should be noted, however, that the study area has a very dense line coverage that is not necessarily representative of the line coverage for the wider area covered by the whole of the 2D line shapefile. In addition, the area within which the features were identified was particularly dense. It can be suggested that the definition of features is therefore related to the relative 2D seismic line density of a particular area. The line density (determined by [Sum line length/ area
of study area] 100), for the entire pilot study, area was 0.9%. The density of sub-samples within the study area is presented in the Table 15.2.
Sample % of lines within Study Area Whole Area Targeted over features 50% 20% 10% Line Density (%) 0.90 1.23 0.47 0.18 0.09
Table 15.2: Line density within sample area.
Several other factors are important in discussing these data. It is clear that the availability and configuration of 2D lines is not likely to be uniform around the United Kingdom coast and that some assessment of the impact of varying configurations of 2D lines would be invaluable. Consequently, it was decided to resample the pilot area 2D data set using line configurations from a series of sample areas selected around the English and Welsh coasts in order to ascertain what we might expect if we were to undertake a comparative survey in these areas using all available data. For this purpose, four additional areas around the western coast were selected for further analysis. Choice of sample areas was based solely on visual identification of concentrations of lines (Figure 15.21). The coverage of the lines within these areas ranged from relatively uniform to clustered. A polygon shapefile was created over the known features within the original study area (Area 1) and this used as a mask over selected locations. The lines within each of these areas were then exported and placed over the identified features (Figure 15.22, A to D), and a 50m point shapefile was created for each. The line density for the additional areas was calculated and is presented in the Table 15.3. With the exception of the Bristol Channel, all these sample configurations exceed line densities of 20% for the sample survey area and therefore approximate the minimum suggested line density. All the images
15 Between the Lines
17
Figure 15.17: Line failures (in red) recorded for data currently processed as part of the West Coast Palaeolandscapes Main Project. ASTER DEM is a product of METI and NASA. ETOPO2v2 is the property of the National Geophysical Data Centre, NOAA, US Dept of Commerce.
Area Number 1 (Part of original pilot study area) 2 (Welsh sector of the Irish Sea) 3 (Welsh sector – Cardigan Bay)) 4 (Scottish sector of the Irish Sea) 5 (Bristol Channel)
Table 15.3: Additional areas with line density.
Line Density (%) 1.16 0.95 0.21 0.23 0.18
suggest that an approximation of the landscape data generated by the 3D study might be acquired if the 2D line data was available within the configurations used. Qualitative variation was significant. Not surprisingly the Welsh sector of the Irish Sea, which has the highest line density, provides a visually coherent picture. The gridded configuration in the Scottish sector of the North Sea (Figure 15.21, area 4), although second lowest in terms of line density, provides a reasonable approximation of the extensive fluvioglacial plains in the south. Elsewhere, in Cardigan Bay and the Bristol
18
EAC OCCASIONAL PAPER NO. 5
Channel (Figure 15.21, areas 3 & 5), line densities and configurations are such that the derived image would have value but would not be comprehensive in output. Clearly, lower line densities would be less likely to provide images that were sufficient for extensive landscape interpretation, although they might still locate individual features, including palaeochannels, and these would still have considerable interpretative value for other archaeological activities including sediment sampling. In most cases, however, supporting
data sets might be required. For example, ALSF funded Regional Environmental Characterisation data (held by the BGS) is available in the Bristol Channel. This contains further digital 2D seismic profiles that could assist interpretation and line density in this area If these semi-quantitative outputs have value then the experience provided by the West Coast pilot study suggests that we can use the procedure outlined above to investigate the opportunities for detailed survey
Figure 15.18: Results of the intersection of all of the 2D lines that intersected the features identified within the 3D seismic data. ASTER DEM is a product of METI and NASA. ETOPO2v2 is the property of the National Geophysical Data Centre, NOAA, US Dept of Commerce.
15 Between the Lines
19
Figure 15.19: Intersection of features using the sub-samples at 10% of lines (top left), 25% of lines (top right) and 50% of lines (bottom left). ASTER GDEM is a product of METI and NASA’ ETOPO2v2 Global 2 Minute Database, National Geophysical Data Center, National Oceanic and Atmospheric Administration, US Dept of Commerce.
elsewhere in UK or European territorial waters. It seems reasonable to suggest that areas with line densities of 20%, but preferably 50%, or more, of the densities in the original study area, are required to provide data that might support extensive landscape interpretation. Appropriate areas can be located using the distribution of 2D line data held in the Birmingham GIS. Figure 15.8 illustrates the density of 2D line coverage around the United Kingdom in relation to primary 3D surveys. This data supported the creation of a grid of 10km × 10km squares generated around the UK coast which could be used to map areas where the line density was potentially suitable for sub-sampling and also the proportion of available lines that should be incorporated into any analysis.
The line density within each grid square was calculated as a percentage (line length to area), and added to the grid square polygon as an attribute (Figure 15.23). Images were created to show the percentage of lines in each grid square needed to be included in any analysis to ensure a maximum line density of 0.5% and 0.2% respectively (Figure 15.23, A & C). The impact of failures of data has already been stressed. Consequently, data was also mapped assuming a worst-case scenario where only 80% of the lines might prove amenable or available for analysis (Figures 15.23, B & D). This suggests that if a research project were solely reliant upon 2D data for mapping purposes then, in many circumstances, it might be necessary to acquire all available datasets to achieve a reasonable level of line coverage that would guarantee results comparable to those provided by the West Coast pilot project In considering Figures 15.8 and 15.23, it is not surprising that the distribution of both 2D and 3D data largely reflects the interest of energy companies and the intensity of energy exploitation in the southern and northern North Sea. This means that although extensive areas around the coast may always require significant acquisition of new data, specific, archaeologically sensitive areas are likely to be accessible for study because there is sufficient seismic coverage. This includes, for instance, the site of the Viking Bank lithic find and the northern shore of Europe during the maximum extent of exposed land c. 18k BP (Long et al. 1986; Fitch et al. 2007, Figure 9.2). Moreover, extensive data-rich areas off the Moray Firth provide real opportunity for
20
EAC OCCASIONAL PAPER NO. 5
Figure 15.20: Intersection of 2D line (length) with features identified from 3D data by sample value.
Figure 15.21: Locations of the selected sample areas of line concentrations around the western coast. ASTER DEM is a product of METI and NASA. ETOPO2v2 is the property of the National Geophysical Data Centre, NOAA, US Dept of Commerce.
15 Between the Lines
21
Figure 15.22: West Coast Palaeolandscapes pilot study area sampled using schemes derived from the Welsh sector of the Irish Sea (top left, Area 2), Cardigan Bay in the Welsh sector (top right, Area 3), the Scottish sector of the Irish Sea (bottom right, Area 4) and the Bristol Channel (bottom right, Area 5). ASTER DEM is a product of METI and NASA. ETOPO2v2 is the property of the National Geophysical Data Centre, NOAA, US Dept of Commerce.
exploring the landscapes associated with Scotland’s currently contentious later Palaeolithic history (Anon 2007). Despite this, it remains true that data availability is limited or absent over much of the territorial waters of the United Kingdom and that the majority of these areas may contain extensive palaeolandscapes and this is likely to be true of comparable areas elsewhere within Europe, including the Baltic and Adriatic. Specific, sensitive areas, including the sea area off the north east coast of England, which is associated
with important early Holocene settlement at Howick, remains to be surveyed to the level required for detail landscape survey (Waddington 2007; Waddington & Pederson 2007). The results of this limited retrospective sampling indicate that extensive projects based upon 2D data are indeed viable in areas where data is available and conducive to analysis. Where these conditions are met primary outlines of landscape features, required
22
EAC OCCASIONAL PAPER NO. 5
Figure 15.23: Data availability for marine prospection within UK territorial based upon line percentage requirements and line failure rates. Top left: The percentage of lines in each grid square needed to be included in any analysis to ensure a maximum line density of 0.2% assuming that all lines are useable. Top right: The percentage of lines in each grid square needed to be included in any analysis to ensure a maximum line density of 0.2% assuming that only 80% of the lines may be useable. Bottom right: The percentage of lines in each grid square needed to be included in any analysis to ensure a maximum line density of 0.5% assuming that all lines are useable. Bottom right: The percentage of lines in each grid square needed to be included in any analysis to ensure a maximum line density of 0.5% assuming that only 80% of the lines may be useable. ASTER DEM is a product of METI and NASA. ETOPO2v2 is the property of the National Geophysical Data Centre, NOAA, US Dept of Commerce.
by archaeologists and heritage managers, can be identified. However, it is equally clear that, given the variation in data availability, then a pragmatic approach may be required in some areas to acquire sufficient 2D data to achieve a positive outcome. In support, it is generally true that our territorial waters
have often been surveyed for a variety of reasons and that complimentary supporting data from a variety of public and private sources may be available within many of the areas identified for future research. It is also important to stress that, even if access to data becomes an issue in respect of the scale of future research, it is a
15 Between the Lines
23
Figure 15.24: HLC areas defined within the NSPP research area. ASTER DEM is a product of METI and NASA. ETOPO2v2 is the property of the National Geophysical Data Centre, NOAA, US Dept of Commerce.
fact that little or no archaeological baseline data exists for the submerged prehistoric resource across most of the offshore areas surrounding the United Kingdom and Europe more generally. There is, therefore, a real value in almost any information that can be acquired. Certainly, the results presented here suggest that there is no reason why mapping should not be attempted in the majority of marine areas within European territorial waters. Refining 3D interpretation using 2D data: Historic Landscape Characterisation Although lacking the full and comprehensive framework associated with 3D/HD3D survey, the fusion of 2D with 3D seismic data has the benefit of being able to combine legacy datasets to produce a higher definition interpretation, maximising the information value of known data assets and, potentially, reducing the need for re-survey. The option to use 2D data to refine relatively low resolution 3D data was highlighted
in the North Sea Palaeolandscapes study (Gaffney et al. 2007). There, supplementary detail of erosion surfaces from 2D lines across the Outer Silver Pit was used to suggest that not only had that feature been transformed from a relatively placid lake environment to a fast flowing estuary, but that these currents had also removed the basin’s potential for sediment sampling (Briggs et al. 2007). Then value of such an observation in terms of management or planning for coring should be clear. However, the potential for refinement of interpretation can also be demonstrated at a landscape level through the use of the North Sea data as part of a historic landscape characterisation (HLC) programme (Figure 15.24). HLC programmes are now an established management tool within British terrestrial archaeology and, increasingly, in marine foreshore or shallow marine contexts. There is now a considerable literature relating to HLC methodologies, although here it is worth stressing that the primary characteristic of an HLC programme is to support overall landscape
24
EAC OCCASIONAL PAPER NO. 5
Figure 15.25: The analytical and management process within the NSPP data flow. Gaffney & Thomson 2007, figure 9.4.
management and the curation of contiguous areas, rather than isolated points or polygons representing archaeological sites or areas of interest within a landscape (Aldred & Fairclough 2003; Clark et al. 2004; Fairclough 2001; Fairclough & Rippon 2002). Broad characterisation zones, on land, may incorporate many contemporary factors including landscape aesthetics as well as archaeological or historic data (Barratt et al. 2007). In curatorial terms, the output from such work assists managers in assessing the overall impact of change across the entire region rather than privileging artificial constructs such as sites or conservation zones. In the case of the North Sea, the relative inaccessibility of the landscape demanded modification of the application – if not the underlying philosophy or methodology of HLC. The results of the HLC analysis of the NSPP data contrast with terrestrial HLC analyses. Whilst the latter reflect a historical palimpsest of features overlain on a topographic backdrop, the North Sea HLC zones were obtained primarily from topographic and morphological data and the interpretation of these zones derives, in part, from their significance as part of a presumed hunter-gatherer economy. Examples of such a monolithic land use are rare in terrestrial contexts but they are not unknown (i.e. military landscapes in North America with have proven susceptible to HLC analysis – Barratt et al. 2007). Within the North Sea, the landscape was classified into fourteen broad areas based upon their depositional history and major landscape features (Fitch et al. 2007). The dividing lines for many of the landscape zones
observed coincided broadly with known watersheds between observed fluvial features. This data probably represents the best general zoning in terms of potential Holocene land use achievable using the available data, but also, to the extent that it may correlate with broad economic activity, the data may carry considerable potential to act as the basis for more detailed behavioural modelling. In managerial terms this data was further refined using the workflow outlined in Figure 15.25 and the detail of the outputs has already been published (Fitch et al. 2007, 110–8.) However, the HLC data from the NSPP has recently been used and refined by another survey project off the Humber estuary. This provides a novel example where legacy data has been used to guide new data capture and, in a cost-effective manner, where the new 2D data was able to refine the earlier 3D interpretation (Fitch et al. 2010; Benike et al. 2004; Fitch et al 2010; Novak & Bjorck 2002). The Humber Regional Environmental Characterisation (REC) project was funded by the Marine Environment Protection Fund under the wider Aggregate Levy Sustainability Fund as administered by DEFRA (http:// www.alsf-mepf.org.uk/). A consortium, including the British Geological Survey (BGS), Birmingham University’s Institute of Archaeology and Antiquity, Marine Ecological Surveys (MES) and Gardline Environmental, was created to collect regional data on seabed habitats, species and features of archaeological interest that exist at a regional scale (Figure 15.26). The
15 Between the Lines
25
Figure 15.26: The location of fluvial (blue dots) and Holocene (green dots) landscape feature identified within the Humber REC survey lines. ASTER DEM is a product of METI and NASA. ETOPO2v2 is the property of the National Geophysical Data Centre, NOAA, US Dept of Commerce.
scope of work for this project called for the acquisition of geophysical and acoustic data amounting to about 3,000km of line data within an area of 11,221km2. The data consists mainly of widely spaced lines covering the greater part of the survey area, and four detailed
survey areas over known or potential archaeological and biological sites of interest. The shallow seismic geophysical dataset provided for the Humber REC consisted of 2D Boomer lines collected by Gardline Surveys Ltd and provided to Birmingham University in
26
EAC OCCASIONAL PAPER NO. 5
Figure 15.27: Submerged prehistoric landscape characterisation resulting from the use of Humber REC Survey datasets. ASTER DEM is a product of METI and NASA. ETOPO2v2 is the property of the National Geophysical Data Centre, NOAA, US Dept of Commerce.
standard SEG-Y format. The data was inspected in detail at Birmingham where anomalies to determine the depth and extent of identified features and anomalies. For this paper, the important point of this work is that the survey area overlaps with the area of the NSPP and the results of the NSPP provided the basis for interpreting the marine landscape in advance of the Humber REC fieldwork. However, the high resolution survey over the landscape features recorded by the NSPP in turn represents an important validation of previous work and allows for the expansion of the submerged landscape record to the west of the original NSPP study area. With respect of the HLC outputs of the REC consortium it is hardly surprising that that the new study area can be characterised into broadly similar areas to that produced by the NSPP (Figure 15.27). However, notable modifications to the zones have been made in the near shore areas to reflect the additional, detailed information provided by the REC assessment (Figure 15.28 & Table 15.4). However, although the Humber REC used eight of the existing 14 categories defined by the NSPP, two additional categories or landscape zones were created. Within the areas already characterised by the NSPP, some 103 Holocene channels and 37
areas relating to Holocene land surfaces were located by the Humber REC project. These were utilised to corroborate the existing characterisation, and where necessary a reclassification of HLC zones. In the areas not covered by NSPP data some 120 areas containing archaeologically significant landscape features and 47 areas of Holocene landscape were identified and utilised to assist the classification of new areas. The proportional relationship between the distribution of archaeologically significant Holocene features across character zones is shown in Table 15.5. By utilising the area of survey covered by the geophysical lines it becomes possible to approximate the density of archaeologically significant landscape features within each of the character zones. The sum of the survey lines (not taking into consideration multiple passes) was approximately 1,300km for the offshore survey area, and 530km for the near shore survey area. In numeric terms, the Humber REC led to a reclassification of the NSPP data which amounted to about 1,150km2 of about 6,548km2 of the area characterised by the NSPP, or about 17.5% of the total area (Figure 15.28). Clearly, a significant gain in precision can be achieved by integrating 2 and 3D data in this manner.
15 Between the Lines
Table 15.4: Landscape characterisation as defined through the Humber REC.
27
Landscape characterisation as defined by the Humber REC Description Area of extensive Holocene landscape with numerous channel systems Area of reuse of Pleistocene features Area of smaller Holocene Channels Area with large lacustrine features Dominated by fluvial features Dominated by geology with fluvial systems Lacustrine Geology controlled landscape Landscape influenced by underlying glacial deposit Low or absent Holocene cover, archaeological potential is concentrated in localised incised systems Used in NSPP NEW YES YES YES YES YES YES YES YES NEW Area KM2 2,390 1,570 1,070 350 170 1,080 160 1,070 590 2,590
Table 15.5: Length of Survey Lines for Humber REC character areas.
Length of Survey Lines for Humber REC character areas Length of Survey Lines for Humber REC character areas Total Of LENGTH Inner Survey Lines Length of Features/ landscapes along lines 46.1 11 4.6 3 0.06 1 0.1 0.9 5.8 Total Of LENGTH Outer Survey Lines
Number of Features /landscapes Identified along lines
Description
Area of extensive Holocene landscape with numerous channel systems Area of reuse of Pleistocene features Area of smaller Holocene Channels Area with Large Lacustrine features Dominated by Fluvial Dominated with Geology with Fluvial systems Lacustrine Landscape Geology Controlled Landscape influenced by underlying glacial deposit Low or absent Holocene cover, archaeological potential is concentrated in localised incised systems
2392 1567 1074 356 167 1077 158 1074 591
143 270 148 53 25 118 76 203 120
210
172 34 13 9
23
1 9
4
5 5 22
258
136
291
35
8.5
Projected % of character area containing Holocene features/landscape (km2) 13% (312km2) 4% (64km2) 3% (33km2) 5% (20.15km2) 1% (2km2) 0.8% (9km2) 0.1% (0.2km2) 0.4% (5km2) 5% (29km2) 2% (5km2)
Area KM2
28
EAC OCCASIONAL PAPER NO. 5
Figure 15.28: Refinement of HLC zones based on 2D data. Existing (unaltered) NSPP character areas are shown in green, whilst re-characterised NSPP areas are shown in pink. The newly characterised are is shown in red. ASTER DEM is a product of METI and NASA. ETOPO2v2 is the property of the National Geophysical Data Centre, NOAA, US Dept of Commerce.
A final aspect of the use of 2D data provided by the Humber REC relates to its detailed use to guide ‘ground observation’ of features. A specific example of such an activity relates to the sediment cores recovered from a transect across palaeochannel features some 50km offshore from Titchwell in the study area (survey number: 18_100N). This was carried out using a 5m vibrocore rig operated from the Gardline vessel Sea Profiler. The cores were recovered and transported to the University of Birmingham for stratigraphic recording and sub-sampling (Figure 15.30). The suite of sediments from the palaeochannel fill consists of basal clays overlain by humified peats sealed by organic silts, which are in turn overlain by marine sands of the current seabed. These deposits infill a channel incised into the basal boulder clay and reach a maximum thickness of over 3m. Analysis of pollen, plant macrofossil, beetles and ostracod/forams is currently in progress, but initial results demonstrate excellent preservation of these sub-fossil remains. Radiocarbon and Optically Stimulated Luminescence dating provides a chronological framework for the palaeoenvironmental records and demonstrate that the basal peats were accumulating during the 7th–8th millennium BC. The pollen, macrofossil and beetle
records are beginning to shed valuable light on the landscape of Doggerland during this period, whilst the ostracod/forams are providing information regarding the timing and nature of the marine transgression that resulted in the eventual inundation of the area. This work is significant for two reasons: firstly, the acquisition of stratigraphic information allows further ‘fine tuning’ of geophysical data in terms of the character of the submerged deposits represented by this information. For example, in the case of the palaeochannel discussed above the stratigraphic boundary between the basal boulder clay which forms the predominant drift deposit in the REC study area and the overlying silts and clays is apparent as a distinct positive reflector. The correlation of geophysical and associated stratigraphic records in this manner should permit more robust interpretation of seismic data and gives us greater confidence in the characterisation of the features identified using such survey. Given the logistical difficulties and cost implications of offshore work, as well as likely future development pressures on the marine zone, the ability to reliably identify locations of high palaeoenvironmental potential is highly significant.
15 Between the Lines
29
Figure 15.29: Seismic profile showing palaeochannels and core locations within the Humber REC. Image Copyright: The University of Birmingham.
The capacity to accurately identify in situ deposits of palaeoenvironmental potential cannot be overemphasised. However, an outstanding question is how future work might locate and investigate any archaeological remains preserved in or sealed beneath deposits of this kind. Currently, the majority of palaeoenvironmental sequences that provide information regarding changes in relative sea level have been sampled from coastal and near shore environments (e.g. Shennan & Horton 2002). Few palaeoenvironmental sequences are available from distances far off shore (up to 50km) or in relatively deep water (25m+). Understanding the character of the environment of Doggerland itself as well as the pattern and process of the early Holocene inundation requires datasets from across as wide a geographical area of the landscape as possible. This may become more important as preliminary results from the REC analysis suggest that currently established rates of early Holocene relative sea level change may not apply especially well to it, perhaps raising questions regarding the utility of regional sea level curves (e.g. Kiden et al. 2002). This is critical in terms of understanding human activity and for assessing cultural responses to the flooding and ultimate loss of Doggerland. Conclusions It should be clear that the benefits of a strategy that can exploit a greater range of legacy data than has previously been used are considerable. At a purely practical level, archaeologists and heritage managers could never hope to replace or replicate this resource if it had not been provided for other purposes. Indeed, the investment required to acquire data for the area of the North Sea Palaeolandscapes project alone has been estimated as equivalent to a century of heritage funding for a national service equivalent to English Heritage (Powlesland 2010). We are unlikely, therefore, to have any alternative to legacy data sources to explore the majority of areas associated with prehistoric habitation and marine inundation. Although there is an emerging consensus that the coastal shelf may be key to understanding the process of prehistoric colonisation
Figure 15.30: Collecting sediment samples using a vibrocorer during the Humber REC. Copyright Dr Ben Gearey.
of the region, and to contextualise and interpret the settlement pattern within the terrestrial record of all the surrounding countries, our current knowledge of the majority of the area remains intensely speculative. Consequently, any opportunity to explore those marine areas previously available for human habitation and associated extensive 2 or 3D data should be accepted with some alacrity (Figures 16.8 & 15.23). In such circumstances, remote sensing, in one form or another, is likely to remain the only practical route towards the investigation of the majority of the inundated landscapes surrounding our coasts. However, it should be emphasised that there is no single data source or methodology that will satisfy the requirements of all archaeologists or heritage managers. The methodologies and technologies chosen for research and management will always depend upon the nature of the archaeological
30
EAC OCCASIONAL PAPER NO. 5
questions being posed. Some projects, including those which seek detailed sediment sampling, proxy or even direct evidence for settlement or land use, may well require high resolution survey and demand the acquisition of new data in areas which have not previously attracted survey. In other circumstances, the availability of the extraordinarily large, pre-existing data sets that have been acquired around our coasts for other purposes have the capacity to inform and guide the development of European research agendas in their own right. In the case of research involving supranational behavioural or settlement modelling, the relatively coarse 2D and 3D data sets that are available may well be adequate for such purposes. Together with improved data on sea level rise and geomorphological change, these data have the potential to provide dramatic, new insights into landscapes which may be key to our regional models. It should also be noted that as national governments become more conscious of their responsibility to preserve the immense heritage landscapes that exist within national waters, there will be a greater requirement for data that is fit for management purposes and, one suspects, can be defended legally. Europe’s oceans have always been contested areas and their continued development as strategic economic zones (driven by fishing, energy, aggregates, telecommunication and shipping) will demand ever more accurate mapping. Detailed information on the distribution of heritage resources will increasingly be required by planners from countries with marine possessions. However, it remains true that much current research in marine contexts is made possible by the generosity of commercial institutions that often provide data at no cost and governmental agencies which fund processing or data acquisition. As the European economy falters during 2010, such support will become even more important. Unfortunately, current financial stringencies may force some data suppliers to take a harder position on charging researchers for data. In the future, costs may become a limiting factor for research within UK waters at least. At the moment that archaeologists had begun to overcome the issues of exploring the vast, inundated prehistoric landscapes preserved within the region, it would be tragic if access to data prevented European heritage managers realising their cultural potential. Consequently, the development of cost-effective methodologies is of considerable importance. We must promote methodologies that integrate existing, extensive and relatively coarse data with spatially restricted but higher resolution data and use these to formulate future directed survey and prospection. We hope that the results presented within this paper provide one opportunity to support a strategic shift in marine heritage management and prospection. Acknowledgements We would like to thank the following companies, institutions and individuals for their support for the projects whose results have contributed to this publication. English Heritage, the Marine Aggregates Levy Sustainability Fund, the University of Birmingham,
PGS, Centrica UK Ltd, DEFRA, British geological survey, MES, Gardline, Maersk Oil (Qatar), Chesapeake Technology Inc., Seismic Micro-Technology Ltd (SMT). Pheonix, Seazone. IBM, VSG and Mechdyne the Qatar Museums Authority, Prof. Dave Tappin (BGS), Dr. Bryony Pearce (MES), Vince Grove (Gardline), Dr Andrew Bellamy (BMAPA, UMD), Huw Edwards (PGS), Dr Kath Buxton (English Heritage), Dr Magnus Alexander (English Heritage), Dr Nic Flemming (University of Southampton), Mr Paul Hatton (Information Services, University of Birmingham), Ms Valerie Scadeng (VSG), Dr Jim Roche (IBM), Mr Richard Cashmore (Mechdyne), Esbern Hoch (Maersk Oil Qatar), Faisal Abdulla Al-Naimi (QMA). His Excellency Sheikh Hasan Bin Mohammed AlThani (QMA). ETOPO2v2 Global 2 Minute Database, National Geophysical Data Center, National Oceanic and Atmospheric Administration, U.S. Dept. of Commerce, http://www.ngdc.noaa.gov/mgg/global/etopo2.html The original data of ASTER GDEM is the property of METI and NASA http://www.gdem.orsdac.or.jp/ GDEM/E/index.html References Aldred, O. & Fairclough, G. 2003: Historic Landscape Characterisation: taking stock of the method. English Heritage and Somerset County Council, London. Anon 2007. Orkney finds confirm early Scottish colonisation. British Archaeology 97. http://www. britarch.ac.uk/ba/ba97/news.shtml. Bailey G. 2010: Review of Europe’s Lost World: the Rediscovery of Doggerland, by Vince Gaffney, Simon Fitch & David Smith, 2009. York: Council for British Archaeology. (CBA Research Report 160). Cambridge Archaeological Journal 20, 145–6. Bailey, G., Westley, K., Davies, W., Gaffney, V., Gibbard, P., Firth, A., Flemming, N., Schreve, D., Hosfield, R., White, M. & Bates, M. 2010: Maritime and Marine Historic Environment Research Framework: Draft Palaeolithic Resource Assessment. http://www. southampton.ac.uk/archaeology/research/ projects/maritime_research_framework.html. Barratt, G., Dingwall, L., Gaffney, V., Fitch, S., Huckerby, C. & Maguire, T. 2007: Heritage management at Fort Hood, Texas: Experiments in Historic Landscape Characterisation. Archaeopress, Oxford. Bennike, O., Jensen, J. B., Lemke, W., Kuijpers, A. & Lomholt, S. 2004: Late- and postglacial history of the Great Belt, Denmark. Boreas 33, 18–33. Briggs, K., Thomson, K. & Gaffney, V. 2007: A Geomorphological Investigation of Submerged Depositional Features within the Outer Silver Pit, Southern North Sea. In Gaffney, V., Thomson, K. & Fitch, S. (eds): Mapping Doggerland. Mapping Doggerland: The Mesolithic Landscapes of the Southern North Sea. Archaeopress. Oxford. 43–60. Bulat, J. 2005: Some considerations on the interpretation of seabed images based on commercial 3D seismic in the Faroe-Shetland Channel. Basin Research 17, 21–42.
15 Between the Lines
31
Bunch, M., Gaffney, V. & Thomson, K. 2007: Coordinating marine data sources. In Gaffney, V., Thomson, K. & Fitch, S. (eds): Mapping Doggerland. Mapping Doggerland: The Mesolithic Landscapes of the Southern North Sea. Archaeopress. Oxford. 11–21. Clark, J.G.D. 1936: The Mesolithic Settlement of Europe. Cambridge University Press, Cambridge. Clark, J., Darlington, J. & Fairclough, G. 2004: Using Historic landscape characterisation: English Heritage’s review of HLC Applications 2002 – 03. English Heritage and Lancashire County Council. Coles, B.J. (1998): Doggerland: a speculative survey. Proceedings of the Prehistoric Society 64: 45–81. COWRIE, 2010: Offshore Geotechnical Investigations and Historic Environment Analysis: Guidance for the Renewable Energy Sector. Draft document (July 2010). Cuttler, R., Fitch, S. & Al-Naimi, F.A. (2010): Considering the ‘Terra Incognita’ and the implications for the Cultural Resource Management of the Arabian Gulf Palaeolandscape. In Potts, D., Al Naboodah, H. & Hellyer, P. (eds): Archaeology of the United Arab Emirates: Proceedings of the Second Conference on the Archaeology of the UAE. Trident press, London Dix, J., Quinn, R., & Westley, K. 2004: A Re-assessment of the Archaeological Potential of Continental Shelves. Located at: http://www.arch.soton.ac.uk/Research/ Aggregates/shelve-report.htm. Fairclough, G.J. 2001: Boundless Horizons – Historic Landscape Characterisation. English Heritage Conservation Bulletin 40(March 2001), 23–6. Fairclough, G.J. & Rippon, S.J. (eds) 2002: Europe’s cultural landscape: archaeologists and the management of change. EAC Occasional Paper no 2, Europae Archaeologiae Consilium and English Heritage, Brussels and London. Fitch, S., Thomson, K., & Gaffney, V.L. 2005: Late Pleistocene and Holocene depositional systems and the palaeogeography of the Dogger Bank, North Sea. Quaternary Research 64, 185–96. Fitch S., Gaffney, V. & Thomson K. 2007: Heritage management and the North Sea Palaeolandscapes. In Gaffney, V., Thomson K. & Fitch S. (eds): Mapping Doggerland: The Mesolithic Landscapes of the Southern North Sea. Archaeopress, Oxford. 105–8. Fitch, S., Gaffney, V., & Ramsay, E. 2010: West Coast Palaeolandscape Pilot Project. Unpublished manuscript. Aggregates levy Sustainability Fund/ English Heritage. Fitch et al. 2010: http://www.barch.bham.ac.uk/ projects/pdfs/Humber%20REC.pdf. Flemming, N.C. 2004: Submarine Prehistoric Archaeology of the North Sea. CBA Research Report 141. Gaffney, C. & Gater, J. 2003: Revealing the Buried Past: Geophysics for Archaeologists. Stroud, United Kingdom: Tempus.
Gaffney, V., Fitch, S. & Smith, D. 2009: Europe’s Lost World: the rediscovery of Doggerland. CBA Research Report. No. 160. Gaffney, V., Thomson, K. & Fitch, S. (eds) 2007: Mapping Doggerland: The Mesolithic Landscapes of the Southern North Sea. Archaeopress. Oxford. Jelgersma, S. 1979: Sea-level changes in the North Sea basin. Acta Universitatis Upsaliensis, Symposia Universitatis Upsaliensis Annum Quingentesimum Celebrantis, 2, 233–48. Lambeck, K. 1995: Late Devensian and Holocene shorelines of the British Isles and North Sea from models of glacio-hydro-isostatic rebound. Journal of the Geological Society, London, 152, 437–48. Louwe Kooijmans, L.P. 1970: Mesolithic bone and antler implements from the North Sea and from the Netherlands, Ber. Rijksdienst oudheidk. bodemonderz. 20/21, 27–73. Kidd, G.D. 1999: Fundamentals of 3D seismic volume visualization. The Leading Edge 18, 702–9. Kiden, P., Denys, L. & Johnston, P. 2002: Late Quaternary sea-level change and isostatic and tectonic land movements along the Belgian-Dutch North Sea coast: geological data and model results. Journal of Quaternary Science 17, 535–46. Long, A. 2003: Removing seismic noise and increasing resolution in the East Java Sea: High Density 3D acquisition and processing. PetroMin 29(8), 26–9. Long, D, Wickham-Jones, C.R. & Ruckley, N.A. 1986: A flint artefact from the northern North Sea. In Roe, D. (ed.): Studies in the Upper Palaeolithic of Britain and North Western Europe, BAR International Series, 296, 55–62. Mueller, C., Luebke, H., Woelz, S., Jokisch, T., Wendt, G, & Rabbel, W. 2006: Marine 3-D seismic investigation of a late Ertebølle settlement in Wismar Bay (SEAMAP-3D case study). Geophysical Research Abstracts, 8, 08447, SRef-ID: 1607-7962/gra/ EGU06-A-08447. Novak, B. & Bjorck, S. 2002: Late Pleistocene–early Holocene fluvial facies and depositional processes in the Fehmarn Belt, between Germany and Denmark, revealed by high-resolution seismic and lithofacies analysis. Sedimentology 49, 451–65. Oxley, I. & O’Regan, D. 2001: The Marine Archaeological Resource. Institute of Field Archaeologists Paper No. 4. PGS, 2005: Visualise the Big Picture with PGS MegaSurveys. http://www.pgs.com/ upload/31163/ data.pdf. Peeters, H., Murphy, P.L. & Flemming, N. 2009: North Sea Prehistory Research and Management Framework 2009. Amersfoort: NSPRMF. Peltier, W.R., Shennan, I., Drummond, R., & Horton, B. 2002: On the postglacial isostatic adjustment of the British Isles and the shallow viscoelastic structure of the Earth. Geophysical Journal International 148(3), 443–75.
32
EAC OCCASIONAL PAPER NO. 5
Powlesland, D. 2010: Review of Gaffney V., Fitch S. & Smith D. 2009: Europe’s Lost World: the rediscovery of Doggerland. CBA Research Report. No. 160. Praeg, D. 2003: Seismic imaging of mid-Pleistocene tunnel-valleys in the North Sea Basin – high resolution from low frequencies. Journal of Applied Geophysics 53, 273–98. Reid, C. 1913: Submerged Forests. Oxford. Roberts, P. & Trow, S. 2002: Taking to the Water:English Heritage’s Initial Policy for The Management of Maritime Archaeology in England. Unpublished policy statement, English Heritage, (http://www. englishheritage. org.uk/server/show/nav.8381). Salomonsen, I. & Jensen, K.A. 1994: Quaternary erosional surfaces in the Danish North Sea. Boreas 23, 244–53. Shennan, I., Bradley, S., Milne, G.A., Brooks, A., Basset, S.E. & Hamilton, S. 2006: Relative sea-level changes, glacial isostatic modelling and ice-sheet reconstructions from the British Isles since the Last Glacial Maximum. Journal of Quaternary Science 21(6), 585–99. Shennan, I & Horton, B. 2002: Holocene land- and sealevel changes in Great Britain. Journal of Quaternary Science 17, 511–26. Shennan, I., Lambeck, K., Flather, R., Horton, B., McArthur, J., Innes, J., Lloyd, J., Rutherford, R., & Wingfield, R. 2000: Modelling western North Sea palaeogeographies and tidal changes during
the Holocene. In Shennan, I. & Andrews, J. (eds): Holocene Land-Ocean Interaction and Environmental Change around the North Sea. London: Geological Society, 299–319. Stringer, C. 2006: Homo Brittanicus: The Incredible Story of Human Life in Britain, Penguin Books, London. Velegrakis, A.F., Dix, J.K. & Collins, M.B. 1999: Late Quaternary evolution of the upper reaches of the Solent River, Southern England, based upon marine geophysical evidence. Journal of the Geological Society, London 156, 73–87. Ward, I. & Larcombe, P. 2008: Determining the preservation rating of submerged archaeology in the postglacial southern North Sea: a first-order geomorphological approach. Environmental Archaeology 13(1), 59–83. Waddington, C. (ed.) 2007: Mesolithic Settlement in the North Sea Basin: A Case Study from Howick, NorthEast England. Oxford, Oxbow. Waddington, C. & Pedersen, K. (eds) 2007: Mesolithic Studies in the North Sea Basin and Beyond: Proceedings of a Conference Held at Newcastle in 2003. Oxford, Oxbow. Wenban-Smith, F. 2002: Palaeolithic and Mesolithic Archaeology on the Sea-bed: Marine Aggregate Dredging and the Historic Environment. Wessex Archaeology, http://www.wessexarch.co.uk/files/ projects/BMAPA-Protocol/mads2.pdf.