Archaeological_Applications_of_JIBS_Data_Progress_Report_09.pdf

Final report prepared for the Heritage Council under the Irish National Strategic 

Archaeological Research (INSTAR) Programme 2009 

[INSTAR Project 16702] 

Archaeological applications of the 

Joint Irish Bathymetric Survey [JIBS] data 

- Phase 2 - 

Principal Investigators 

Rory Quinn and Wes Forsythe 

Associate Investigators 

Ruth Plets, Kieran Westley, Trevor Bell, Fergal McGrath, Rhonda Robinson and Sara Benetti 

Project Researchers 

Ruth Plets, Annika Clements and Chris McGonigle

INSTAR Project 16702: Archaeological applications of the Joint Irish Bathymetric Survey [JIBS] data 

CONTENTS 

Section 

Page 

1 INVESTIGATORS 3 

2 INTRODUCTION 

[2.1] Background and motivation 

[2.2] Aims and objectives 

[2.3] Research questions 

3 THEORETICAL FRAMEWORK 

[3.1] Shipwreck archaeology 

[3.2] Submerged landscape archaeology 

4 DATA SETS 

[4.1] JIBS data 

[4.2] Shipwrecks 

[4.3] Submerged landscapes 

5 METHODOLOGY 

[5.1] Production of 1m resolution DEM 

[5.2] FM Geocoder processing of backscatter 

[5.3] Automated segmentation of backscatter 

[5.4] Seismic integration into SMT Kingdom Suite 

[5.5] GIS - spatial integration of data sets 

6 RESULTS 

[6.1] Refining palaeo-geographies – The Skerries 

[6.2] Target sites for direct sampling 

[6.3] Refining site formation models and site IDs 

[6.4] Dissemination of results 

7 IMPACT OF THE PROJECT 

[7.1] Impact on teaching 

[7.2] Impact on research 

[7.3] Impact on practise 

5 

5 

10 

11 

14 

17 

25 

26 

27 

29 

32 

40 

43 

49 

53 

60 

61 

101 

104 

104 

104 

8 SCHEDULE OF EXPENDITURE 106 

9 REFERENCES 107 

APPENDICES 

[ 2 ]


1. INVESTIGATORS 

Principal Investigators: Rory Quinn 1 and Wes Forsythe 1 

Associate Investigators: Fergal McGrath 2 , Sara Benetti 5 , Rhonda Robinson 3 , Ruth Plets 1 , 

Kieran Westley 1 and Trevor Bell 4 

Project Researchers: Ruth Plets, Annika Clements and Chris McGonigle 

1 Centre for Maritime Archaeology, University of Ulster (UU), Northern Ireland 

2 Marine Institute (MI), Galway, Ireland 

3 Northern Ireland Environment Agency (NIEA), DoE, Northern Ireland 

4 

Department of Geography, Memorial University of Newfoundland (MUN), Canada 

5 School of Environmental Sciences, University of Ulster (UU), Northern Ireland 

Rory Quinn is a Senior Lecturer in marine geoarchaeology at UU with 15 years 

experience in the acquisition, processing and interpretation of high-resolution marine 

sonar data for archaeological investigations. He has participated in shipwreck and 

submerged landscape projects in Ireland, UK, East Africa and Canada, and is currently a 

Coracle Research Fellow in submerged landscapes with Memorial University of 

Newfoundland, on the co-ordinating committee of the Submerged Landscapes 

Archaeological Network and on the editorial board of the Journal of the North Atlantic. 

Wes Forsythe is a Lecturer in maritime archaeology at UU with 10 years experience of 

coastal, intertidal and underwater work in Ireland and abroad. He has worked for state 

heritage agencies in the Republic and Northern Ireland and has had direct involvement 

with both the issues facing management in the marine environment and research needs 

in this field. His research work has varied from studies of submerged Late Mesolithic 

forests to macro- and micro-scale investigations of Ireland's shipwrecks. He has directed 

four major maritime archaeology research programmes and contributed to many others 

from inception to publication. 

Ruth Plets, a Research Associate in marine geoarchaeology at UU, obtained her first 

degree in marine geology at Ghent University (Belgium) before studying for an MSc in 

Oceanography and PhD at Southampton University. As part of her PhD, she looked at 

how high-resolution marine geophysical techniques can be used to image, characterize 

and visualize archaeological material, including both shipwrecks and submerged 

landscapes. Her research interests lie in sedimentology, high-resolution geophysics and 

marine geoarchaeology. 

Annika Clements, a Research Associate in Marine GIS at UU, obtained a BSc in Biological 

Sciences at the University of Oxford and an MSc in Oceanography at the University of 

Southampton. Annika specialises in Marine GIS applications, with research interests in 

seabed mapping, marine habitat characterisation and mapping (subtidal benthos) using 

underwater acoustics, underwater video techniques and fisheries research. 

Kieran Westley, a Research Associate in maritime archaeology at UU, studied for his 

undergraduate degree in Archaeology and Anthropology at Cambridge University, before 

going to Southampton University to do an MA in Maritime Archaeology followed by a PhD. 

His PhD focussed on coastal environmental changes and how they may have affected the 

migration and dispersal patterns of prehistoric humans. His broader research interests lie in 

the archaeology of continental shelves, in particular the role of coastlines in prehistory and 

the archaeological potential of submerged landscapes. Common to both is the extensive use 

of palaeo-environmental data alongside the archaeological evidence in order to examine the 

effect of changing climates, sea-levels and landscapes on prehistoric societies, and also to 

locate and study archaeological sites that have been inundated by sea-level rise. 

[ 3 ]


Chris McGonigle obtained a BSc in Marine Science at UU. He specialises in Marine GIS 

applications, with research interests in seabed mapping and backscatter segmentation, with 

particular reference to habitat characterisation. 

Fergal McGrath is Team Leader of the Advanced Mapping Services at the MI. He is 

Project Manager for the MI contribution to the Joint Irish Bathymetric Survey (JIBS), an 

INTERREG III funded cross-border program led by the Maritime and Coastguard Agency 

and is currently involved in project and data management of the national marine 

mapping programme (INFOMAR). Fergal has 10 years experience in marine geophysical, 

geological, geotechnical and hydrographic data acquisition, processing and interpretation, 

including site and debris clearance surveys, pipeline/cable route surveys, seismic 

prospecting and geohazard investigation surveys in the North Sea, West Africa, Gulf Of 

Mexico, Caspian Sea and South Atlantic. 

Sara Benetti is a Lecturer in Environmental Change at UU, researching abrupt and longterm 

climatic change and its effects on sedimentological processes in the context of 

glacial/interglacial cycles. Sara has previous experience with the British Antarctic Survey, 

the National Oceanography Centre (UK), Woods Hole (USA) and the Bedford Institute of 

Oceanography (Canada). 

Rhonda Robinson has worked for the Northern Ireland Environment Agency for 12 

years. Rhonda is a Senior Inspector with management responsibility for the Maritime 

Archaeology branch of the Service. In addition, she deals with all issues relating to 

archaeological management within the broader North Coast area. She has extensive 

experience in GIS analysis and of archival management, including digital resources. 

Trevor Bell is Professor of Physical Geography at MUN who studies landscape and seabed 

history from a variety of perspectives to address a range of research questions from 

theoretical to applied. His approach is strongly interdisciplinary and collaborative, involving 

analysis and expertise from a range of disciplines in the earth, life and social sciences. His 

current project titles illustrate this approach and provide a general picture of my research 

pursuits: Dynamics of the Newfoundland Ice Cap; Mapping postglacial sea levels and 

submerged landscapes; Prehistoric settlement, subsistence and environment in 

Newfoundland; Multibeam bathymetric mapping for seabed morphology and habitat 

classification; Climate sensitivity of tundra and boreal ecosystems in Labrador highlands, 

and Climate-change impacts in Arctic coastal communities. 

[ 4 ]


2. INTRODUCTION 

2.1 Background and motivation 

To address the need for high-resolution bathymetric data off the north coast of Ireland, the 

Joint Irish Bathymetric Survey (JIBS) was instigated as a partnership between the Maritime 

and Coastguard Agency (MCA) and the Marine Institute (MI), under INTERREG IIIA 

( 2,133,508). The JIBS project commenced in April 2007 and was completed in September 

2008, providing 100% multi-beam bathymetry coverage (Figure 2.1) within the 3nm 

coastal strip from the Fanad Peninsula (Co. Donegal) to Fair Head (Co. Antrim). 

Under INSTAR 2008, Phase 1 of this project (Archaeological applications of the Joint Irish 

Bathymetric Survey data) was funded to assess the JIBS data for archaeological 

applications. The objectives of this feasibility study were four-fold: [1] to develop a GIS for 

the study area comprising extant and new (JIBS) data, [2] to assess the viability of 

interpreting and mapping palaeo-shoreline features from the JIBS data in the context of 

past sea-level change [3] to record the locations and seabed conditions of wreck sites in the 

study area, and [4] to provide Government agencies with baseline data and 

recommendations for managing submerged archaeological sites. The final project report for 

Phase 1 of the project was submitted to the Heritage Council in December 2008. 

Figure 2.1 Coverage of the JIBS survey from Fanad Peninsula in the west to Fair Head in 

the east. 

Phase 1, the feasibility study, proved beyond any doubt that the archaeological 

applications of the JIBS data are many. While the quality and resolution of the JIBS data 

proved even better than expected, interpretation of these data uncovered many 

previously unrecorded submerged sites of archaeological potential. For example, 

research undertaken as part of Phase 1 successfully catalogued and identified palaeolandscape 

features, including a series of geomorphic signatures relating to submerged 

palaeo-landscapes (Figures 2.2 and 2.3). This phase also facilitated preliminary coarse 

[ 5 ]


palaeo-geographic reconstructions for the entire study area by integrating the JIBS data 

with information from sea-level studies and glacio-isostatic modeling (Figures 2.4 to 2.7). 

Figure 2.2 Relict beach ridge off Runkerry submerged in 8m water depth. 

Figure 2.3 Palaeo-seacliffs and possible submerged wave-cut terrace off Bengore Head. 

[ 6 ]


Figure 2.4 Overview of the study area coloured to represent the sea-level lowstand of - 

30m identified by Kelley et al (2006). Offshore bathymetry is provided by the JIBS 

multibeam data, terrestrial topography is provided by the SRTM digital elevation model. 

At this stage, Loughs Swilly and Foyle would have been freshwater or lagoonal systems, 

providing sheltered areas for resource exploitation. 

[ 7 ]


Figure 2.5 Palaeo-geographic reconstruction of Lough Swilly assuming a sea-level 

lowstand of -30m (Kelley et al, 2006). At this depth, the sea lough is transformed into an 

inland river valley or lake basin. The black areas in the littoral zone correspond to areas 

of no data. 

[8]


Figure 2.6 Palaeo-geographic reconstruction of the Skerries assuming a sea-level 

lowstand of -30m (Kelley et al, 2006). At this depth, the modern reefs and rocks form 

the shoreline of an extended coastal plain. 

[9]


Figure 2.7 Palaeo-geographic reconstruction of Rathlin Island assuming a sea-level 

lowstand of -30m. Note there is relatively little change to the island’s eastern and 

northern shores where bathymetric gradient is steep. However, the lower-lying southern 

portion of the island is extended by hundreds of metres to several kilometres. 

The provision of the JIBS data also presents a unique opportunity to address issues 

relating to the nature and extent of the wreck resource off the north coast, allowing for a 

unique study into shipwreck site formation processes at previously unattainable 

resolutions over a large geographic area. For example, work undertaken in Phase 1 

queried current entries in the existing Maritime Sites and Monuments Record (MSMR) 

against the multibeam data and screened the data for unrecorded wreck sites to provide 

more accurate positional information (Figure 2.8). A database comprising in excess of 

200 anomalies of archaeological potential was constructed (Figure 2.9), with the majority 

of these anomalies corresponding to previously unrecorded wreck sites. In addition, 

preliminary observations were made on general wreck distribution and of the broader 

environmental context (sediment type, hydrodynamic conditions) of the anomalies, the 

first step in producing more detailed site formation models (Figure 2.10). 

2.2 Phase 2 - Aims and objectives 

Phase 2 of the project was awarded funding in May 2009. Objectives for Phase 2 are sixfold: 

[10]


[1] To better constrain and/or verify palaeo-landscape interpretations from Phase 1 with 

extant seismic data. 

[2] To produce high-resolution palaeo-geographic reconstructions of areas identified as 

having high archaeological potential. These reconstructions will be done at 1,000-year 

time steps, from 14,000 years ago to present. 

[3] To identify targets in these landscapes for future marine coring programmes (e.g. 

Marine Institute Shiptime Programme). 

[4] To re-assess anomalies identified in Phase 1 so as to confine the research of Phase 2 

to sites positively identified as shipwrecks. 

[5] To produce Digital Elevation Models (DEMs) for each wreck site using the highestresolution 

(maximum 10cm) sonar data. 

[6] The final objective involves the dissemination of results, including distribution of the 

Heritage Council report and dissemination of the compiled databases to relevant 

Government agencies. 

2.3 Central research questions 

[1] What evidence is there for past sea-level change and remnants of palaeo-landsurfaces 

(e.g. lagoons, palaeo-channels) off the north coast? What are the implications 

for the Mesolithic colonization? 

[2] What is the potential submerged human settlement off the north coast of Ireland, 

particularly within the previously identified 30m lowstand? 

[3] What is the evidence for previously unknown archaeological sites, in particular 

shipwrecks? Do these sites pre-date the earliest known wrecks on this shore (1588)? 

[4] What is the extent of evidence for shipwrecking and how does this evidence correlate 

with the respective national shipwreck databases? 

[5] What is the archaeological evidence from known hazards for shipping (e.g. 

submerged rocks and sandbanks), in particular those associated with potential votive 

deposits? 

Figure 2.8 Overview GIS map of study area showing the relevant Admiralty Charts, 

colour-coded JIBS multibeam data, MSMR entries, shipwrecks known to sports divers and 

anomalies detected on JIBS data during Phase 1. 

[11]

Figure 2.9 Classification scheme of anomalies as potential shipwrecks used during 

Phase 1 of the project. (A) Probable wreck; (B) Possible wrecks (may be geological 

feature); (C) Possible wreck (may be data processing artefact). The easternmost wreck 

in Figure A is S.S. Lugano (1917). All other sites are unidentified.


Figure 2.10 Some results from Phase 1: (A) Scour features and 700m-long depositional 

feature associated with possible wreck indicates NE-SW direction of bottom currents. (B) 

Bedforms north of the possible shipwreck indicate an average W-E direction of bottom 

current. This wreck is located in a bowl shaped scour pit that opens into a deeper 

channel to the north. 

[ 13 ]


3. THEORETICAL FRAMEWORK 

3.1 Shipwreck 

Wreck sites are generally defined as sunken ships and aircraft, and any material 

associated with such vessels. For this report in particular, wreck sites will be 

synonymous with shipwrecks and their associated materials, unless stated otherwise. 

Shipwrecks are time capsules that not only provide a detailed image of the lives and 

work conditions of the people onboard the ship, but they also offer an insight into the 

broader social, economic and political history of a nation. Furthermore, artefacts found in 

association with shipwrecks, which are not necessarily unique to maritime sites (e.g. 

pottery, crockery, clothing, food, etc.), can complement the terrestrial record due to the 

difference in preservation conditions on or in the seabed compared to typical terrestrial 

environments. This is particularly true if artefacts have been preserved under anaerobic 

marine conditions (e.g. buried in estuarine silts and muds). In Ireland, however, a 

misconception has existed for a long time that preservation conditions around the 

dynamic Irish coast would be unfavourable for the survival of any underwater remains 

and that, if any did survive, they would be of little importance (Breen, 1996). As an 

example, the discovery of three Spanish Armada wrecks off the coast were originally 

seen as an exception to this rule and it was felt that these vessels had little to do with 

Irish history but were there by chance (Breen, 1996). The northern coast in particular is 

exposed to high wave states and storm activity, resulting in the dominance of coarsegrained 

seabed substrates, making shipwreck preservation generally unfavourable. 

However, deep recesses in the northern Irish coastline together with high bathymetric 

relief and the presence of estuaries, means micro-environments with higher 

archaeological preservation potential do exist (Wheeler, 2002). 

Continuous advances in technology, especially the development of SCUBA gear in the 

1960s and constant improvements of high-resolution marine geophysical methods in the 

1990s, have meant that increased numbers of maritime archaeological sites with high 

preservation potential have been discovered, located and imaged off the Irish coast. This 

should come as no surprise. People have lived on and around the Irish coast for over 

9,000 years, exploiting its resources and using the sea as a means of transport and 

communication (Breen and Forsythe, 2004). Furthermore, Ireland’s strategic position as 

an island on the edge of western Europe was one of the determining factors in the 

development of the nation as its location proved advantageous in military and defensive 

terms, later resulting in Ireland becoming a focal point for English naval activity (Breen, 

1996; Breen and Forsythe, 2001). Consequently, the sea has always played an 

important role in the country’s economic and political landscape and relicts of this are 

bound to be found on the seabed. 

Boats and ships have always played an important role in the lives of coastal people 

whether they were used as a means of transport and trade or a means of making a 

living. For Ireland in particular, early Mesolithic colonists must have travelled across a 

stretch of open water as early as 10,000 years ago using some sort of vessel to reach 

the Irish coast. Hence, a boat building tradition appears to have been part of Irish life 

since its earliest occupation. The nature of the vessels varied in different parts of the 

country and changed considerably through time. While dugout boats, bark– and skincovered 

craft were the dominant vessel types for the prehistoric period, the first 

evidence of other boat forms (i.e. plank built), in both foreign and native vessels, 

appeared in Ireland during the Iron Age. At the start of the early medieval period, before 

the arrival of the Vikings, large clinker-built vessels were in use in the English Kingdoms 

which the Irish must have been aware of through trading and raiding. However, the 

coracle (a small wickerwork boat covered with an animal hide) remains the most 

mentioned vessel in the first Irish documentary sources. The arrival of the Vikings had a 

great impact on Irish communities and the emerging dominance of the Nordic boat 

building tradition is evident on several iconographic depictions of ships of that time. 

[ 14 ]


Apart from documentary and iconographic evidence, numerous boat and ship timbers, 

graffito and boat models have been recovered from former Irish Viking towns. Despite 

the advances in technology brought to Ireland by the Vikings, the logboat was still used 

and its representation in the archaeological record indicates that it is the most enduring 

boat type used on the island. During the high medieval period, trade became ever more 

important and, despite the rarity of timber finds of vessels from that period in Ireland, 

the main Northern European vessel types (cog, hulc and keel) must have been a regular 

sight. Outside the major ports though, local traditional boats were still in use. In the late 

medieval period, profound changes in European shipbuilding meant that the carvel 

technique was now used for bigger ships (carracks, galleons and galleys), while the 

clinker technique was restricted to smaller vessels. During the early modern period, 

Ireland’s economy was completely dominated by England. The great innovations of the 

period in ship design were stimulated by large-scale mercantile business (the so-called 

‘Companies’) carrying goods from around the world to Europe. These vessels were 

characterised by high sterns, three masts, a combination of square and lateen sails and 

complex rigging. The vessels became larger in size and design, to over 600 tons by the 

mid-eighteenth century and over 1000 tons by the end of the century. A number of ships 

belonging to the various companies were wrecked around the Irish coasts. Nonetheless, 

currachs and logboats remained one of the most enduring vernacular boats in Ireland. A 

period of relative peace and prosperity in the 18 th century reflected in the increase of 

coastal trade and shipping. The early years of this coastal trade was dominated by wellestablished 

wooden sailing vessels such as schooners, brigs, brigantines and snows. 

Hundreds of wrecks from this period survive around the coast. Wooden sailing vessels 

continued to be used into the early twentieth century but were doomed once the 

technology of steam and iron began to develop. Commercial iron shipbuilding began in 

the first half of the nineteenth century. First, individual iron components were combined 

with wooden structural elements to form composite ships (e.g. many of the great clipper 

ships). Advances in the use of iron in shipbuilding continued and when steel eventually 

began to replace iron from the 1870s onwards, the old sailing ships went into decline. A 

more detailed history of the boats and ships of Ireland and its place in a wider European 

context can be found in Breen and Forsythe (2004). 

As the result of the work of individuals studying Ireland’s underwater cultural heritage, 

governmental bodies have started to recognize the archaeological potential of the 

marine environment and have established structured survey programs in recent years 

(e.g. MAP, Maritime Archaeology Project). This has resulted in the development of 

inventories for Ireland (both the Republic and Northern Ireland) based on historical 

cartographic, documentary and photographic material combined with contemporary 

marine survey data (Breen, 1996; Breen and Forsythe, 2001; Breen et al., 2007). In 

total, the records of over 10,000 shipwreck incidents off Ireland have been compiled 

(Forsythe et al., 2000), which led to the production of marine sites and monuments 

records (MSMRs) for each Irish County. 

Despite the long history of boat building in Ireland, the wreck record is heavily biased 

chronologically towards the post-medieval period (postdating the 16 th century). For 

Northern Ireland specifically 3% predate the 18 th century, 11% date from 1751 to 1800, 

66% date to the 19 th century and 20% date to before 1945 (Breen et al., 2007; Figure 

3.1) The high percentage for post-medieval period is caused by the increase in recording 

of ship losses by the authorities coupled with the upsurge in shipping and coastal trade 

after the 17 th century. The general bias towards the post-medieval period must be 

sought in the nature of wreck preservation and the history of seabed mapping and 

research. Firstly, older wrecks have been subject to physical, chemical and biological 

degradation for a longer time period and might therefore not have survived. Secondly, if 

they have survived, then they are most likely buried within sediment and have not yet 

been discovered using traditional surveying techniques. 

[ 15 ]


Figure 3.1 Extent of shipwreck loss in Northern Ireland shown in 50 year periods from 

the 15 th century to 1945 (Breen et al., 2007). 

Over 95% of vessels contained within the Irish shipwreck archive are those propelled by 

wind power and it was this source of power which often resulted in wrecking along the 

coast (Forsythe et al., 2000). The general distribution of Irish wrecks therefore reveals 

that most ships are concentrated around ports and notorious natural hazards. The 

spatial distribution of preserved wrecks in Northern Ireland specifically, in comparison 

with the total Irish coast, shows that 8% of wrecks are located on the Co. Down coast, 

3% off the Co. Antrim coast and less than 1% off Co. Derry (Breen et al., 2007). This 

distribution reflects the differences in shipping activity (largest in Co. Down and Antrim), 

presence of natural hazards (with extensive shoals, banks and rocks off the coast of Co. 

Down) and preservation potential (lowest in the exposed areas of the north coast) of the 

different areas. 

The compiled MSMRs are not only an aid for maritime archaeologists studying the 

marine archaeological resource, either on a local or regional scale. More recently the 

archive is being used as an integral part of Environmental Impact Assessments (EIAs) 

(Breen, 1996). Commercial developments such as marine construction or seabed cable 

laying, now have to take archaeology into account, and rely heavily on these MSMRs. 

However, there are some problems with the MSMRs as they stand. Firstly, the positional 

accuracy of the sites within the MSMR is often inaccurate. They frequently depend on 

post-18 th century documentary evidence and word-of-mouth reports, DECCA or latitudelongitude 

readings from sources varying from fishermen, sports divers and the 

Hydrographic Office (Quinn et al., 2002b). It is therefore high time that more exact and 

systematic surveys are conducted which can improve or confirm the positional data of 

existing entries in the MSMR and can quantify the nature of currently unknown 

submerged sites. Secondly, the entries in the MSMR do not offer any information on the 

wider context or the environment in which the site is located. However, such information 

is vital in understanding site formation processes and to assess the preservation 

potential and future of the site. Fully submerged shipwreck sites act as open systems, 

with the exchange of material (sediment, water, organic and inorganic objects) and 

energy (wave, tidal, storm) across system boundaries (Quinn, 2006). Formation 

processes at wreck sites are driven by some combination of chemical, biological and 

physical processes, with physical processes dominant in initial phases (Quinn, 2006). 

Depositional and erosional patterns that form in response to hydrodynamic forcing are 

[ 16 ]


often difficult to quantify at sites due to the spatial and temporal scales at which these 

processes occur. These are particularly difficult to understand if the information is 

founded on dive based survey data, which tend to concentrate on wreck and individual 

artefacts rather than the wider environment. To date, therefore, the CMA strategy for 

mapping submerged wreck sites has concentrated on the integration of single-beam 

bathymetric, side-scan, magnetometer and sub-bottom data to identify sites of 

archaeological potential (Bull et al., 1998; Quinn et al., 2000; Quinn et al., 2002a; 

Quinn, 2006; Quinn et al., 2007). These data have been interpreted in the context of a 

series of control experiments, designed to mimic the geophysical signatures of 

submerged archaeological material (Quinn et al., 1998a; Quinn et al., 1998b; Quinn et 

al., 2005). This experimental approach was developed to inform interpretation of 

geophysical data acquired over potential archaeological sites. However, the limited 

availability (and relatively low-resolution) of data off the north coast has currently 

hindered this approach. 

The provision of the JIBS data presents a unique opportunity to address these issues. 

Firstly, current entries in the MSMR will be queried against the multibeam data and the 

data screened for unrecorded wreck sites. For each of the sites a contact sheet will be 

produced detailing the exact position (WGS84), describing the appearance on the 

multibeam data together with a description of the wider context (sediment type, 

hydrodynamic conditions) and a preliminary interpretation. Secondly, it will allow us to 

study wreck site formation processes at a resolution and scale previously unobtainable, 

permitting us to model the hydrodynamics of each site in greater detail and on a 

regional scale. However, we are fully aware that, in order to validate both the 

interpretation of the wreck site and its surrounding environmental parameters, groundtruthing 

may still be necessary in the future. 

3.2 The submerged prehistoric landscape 

Submerged prehistoric landscapes are tracts of lands that were subaerially exposed in 

the past but have since been inundated as a result of sea-level rise. These nowsubmerged 

areas were potentially important landscapes for prehistoric humans as they 

offered a range of marine and terrestrial resources as well as access to transportation 

and migration routes along shorelines and estuaries and rivers (Flemming, 1998). The 

existence of submerged prehistoric landscapes with associated archaeological evidence 

in a particular area is contingent on three variables. Firstly, sea-level must have been 

lower at some point in the past. Secondly, prehistoric humans must have been present 

and occupied the exposed land. Thirdly, sedimentary processes accompanying 

inundation must have preserved rather than eroded the palaeo-landscape. Existing 

archaeological and palaeo-environmental records demonstrate that all three conditions 

can be met for Ireland. 

3.2.1 Archaeological and environmental background 

Ireland has a relatively short prehistory in comparison to other European countries, 

beginning after the end of the last Ice Age with the period generally referred to as the 

Mesolithic and terminating with end of the Iron Age at c. 500 AD. The earliest known 

archaeological site in Ireland is located at Mount Sandel (Northern Ireland) on the 

eastern bank of the River Bann, roughly 9km from the modern coast, and dates to 

between 9000 to 8500 radiocarbon years before present ( 14 C BP) (Smith, 1992). In 

calendar years this equates to a date of around 10,000 to 9500 BP. Evidence from this, 

and other sites, indicates that these earliest inhabitants were seasonally mobile huntergatherers 

exploiting a range of species including terrestrial mammals, birds, fish and 

shellfish (both fresh- and saltwater) and wild plants (Van Wijngaarden-Bakker, 1990). In 

particular, the majority of sites were situated on riverbanks, lakeshores and coasts, 

testifying to the importance of marine and freshwater resources. 

[ 17 ]


The absence of an earlier archaeological record has been attributed to the fact that 

Ireland was almost totally covered by an ice sheet several hundred metres thick at the 

height of the last Ice Age – referred to as the Last Glacial Maximum (LGM) - which 

occurred between 24,000 and 21,000 BP. Retreat of the ice began shortly after 21,000 

BP to the point at which the whole of Ireland was ice-free by 15,000 BP (Brooks et al., 

2007; Sejrup et al., 2005). With ice retreat came of a suite of environmental changes, 

most notably in sea-level and climate. 

When the great ice sheets of the Last Glacial formed, they drew water from the world’s 

oceans, causing a glacio-eustatic sea-level fall of ~120m (Peltier and Fairbanks, 2006). 

However, their great weight caused the areas on which they formed to subside. Thus, 

areas under the ice experienced higher sea-levels on account of this isostatic depression. 

As the ice retreated, the sea initially flooded into these isostatically depressed 

landscapes. Over time though, the release of the great weight of the ice caused them to 

isostatically uplift, often faster than the glacio-eustatic rise driven by the influx of 

meltwater into the oceans, creating a pattern of sea-level fall. Areas where ice was 

thickest experienced the greatest isostatic depression, and hence rebound. In some 

instances, this may have been sufficient to result in a pattern of constant sea-level fall 

right up till the present. Conversely, where the ice was thinner, rebound slowed and was 

eventually overtaken by the glacio-eustatic rise. The result of this was the exposure, and 

then submergence of habitable tracts of land (see overview in Lambeck and Chappell, 

2001). It is this pattern that characterizes much of Ireland, though on a local and 

regional scale, variations in the position and weight of the ice sheet and the timing of 

deglaciation meant that the history and magnitude of sea-level change differed across 

the country (see below and Brooks and Edwards, 2006). 

The retreat of the ice was induced by a period of global climate warming which ended 

the cold dry LGM and gradually transformed the arid polar deserts and steppe-tundra of 

the mid latitudes to more temperate parkland and forest. This was not to say that the 

warming trend was steady and unbroken. On the contrary, a number of cold stadials 

(e.g. the Younger Dryas – 12,900 to 11,500 BP) characterized by glacial conditions and 

ice advances can be seen in palaeo-environmental records (Rochon et al., 1998; Sejrup 

et al., 2005). However, by the time the first humans arrived in Ireland, the worst of 

these fluctuations were over, and they entered an increasingly temperate landscape of 

birch forests that gave way to elm and oak by 9,000 BP (Smith, 1992). Sea-levels 

though, still continued to fluctuate creating continual variations in coastal configuration 

and geomorphology. 

The end of the Mesolithic and start of the succeeding period – the Neolithic – has 

typically been defined, in Ireland as in the rest of Europe, by the transition from a 

hunter-gatherer to a settled agricultural way of life. Thus, evidence of seasonal mobility, 

temporary structures and ephemeral sites are replaced in the archaeological record by 

pottery, field systems, domesticated sheep and cattle, cereal crops and larger, more 

permanent structures such as megalithic tombs. That said, the boundary between the 

two periods is somewhat fuzzy owing to a lack of secure well-dated archaeological 

evidence for the intervening transitional interval. At present, what can be said for certain 

is that some typically Mesolithic tool assemblages can be dated to as late as 5500 ( 14 C) 

BP while the building of more substantial houses, megalithic tombs and adoption of a 

farming economy only took place at, or just after 5000 ( 14 C) BP (Woodman, 2000). 

Moreover, the nature of the transition between the two is still debated, with arguments 

ranging from indigenous adaptation of outside ideas, an influx of agricultural immigrants 

or a mix of the two (Woodman, 2000; Rowley-Conwy, 2004). Climatically, the rather 

stable and temperate climate begun in the Mesolithic continued, with the main difference 

being increasing forest clearance to create land suitable for agriculture. Sea-levels had 

still not entirely stabilized by this time; although the glacio-eustatic meltwater-driven 

rise had largely ceased, isostatic rebound was still taking place. 

By 4500 ( 14 C) BP new developments can be seen in the archaeological record, most 

notably the appearance of distinctive ‘Beaker’-style pottery and the replacement of stone 

[ 18 ]


tools by bronze implements. The agricultural way of life continued throughout this period 

– the Bronze Age – which also saw increases in social complexity, such as the building of 

defensive structures such as hillforts, increasing settlement sizes and increased trade 

and exchange networks with Britain and the Continent (Breen and Forsythe, 2004). 

Social complexity continued to increase through the end of the Bronze Age around 2,600 

( 14 C) BP and into the final phase of the prehistoric period; the Iron Age. As well as the 

introduction of iron, the hierarchal societies of the previous period developed even 

further with the construction of defensive earthworks marking out centres of power. 

Ireland remained connected to the British Isles and mainland Europe via sea trade 

routes and may have been subject to Roman military incursions though these were 

ultimately unsuccessful as it was never conquered as Britain was. Climate and sea-level 

trends were broadly the same as in the Neolithic. 

An important aspect to consider throughout Irish prehistory is the importance of coasts 

and waterways. The first Mesolithic colonists came from Britain, either from SW Scotland 

or via the Isle of Man into Ulster, or from north Wales into Leinster. This is currently 

believed to have been a water crossing as a land bridge probably did not exist despite 

lower sea-levels (Breen and Forsythe, 2004). Nonetheless, these would have shifted 

coastlines further out to sea, narrowing the distance between the two landmasses. 

Landbridges may however, have played a role in the migration of fauna into Ireland 

immediately after the post-LGM deglaciation and during warmer interstadials preceding 

the glacial maximum (Woodman et al., 1997). Once people were established in Ireland, 

the coast remained significant. Many Mesolithic sites were coastal and sited to exploit a 

range of resources such as inshore fish, shell middens and flint sources. Even in the 

Neolithic, despite the increasing dependence of agricultural products, marine and coastal 

resources may also have played an important part in subsistence patterns. In addition, 

specialized lithic workshop sites are located close to sources of flakeable stone in coastal 

areas (Bamford and Woodman, 2004) while evidence of axe trade between Britain and 

Ireland as well as the remains of boats show that coasts and water transport remained 

important throughout the Neolithic (Breen and Forsythe, 2004). This implies that the 

seabed off Ireland potentially contains a wealth of prehistoric material ranging from 

Mesolithic flint tools to Bronze Age boats. The submergence of this past archaeological 

landscape is dramatically illustrated by archaeological and palaeo-environmental 

evidence preserved offshore. For example, a Neolithic megalithic tomb is situated in the 

intertidal zone in Cork Harbour (Woodman, 1990) while surveys in the Shannon Estuary 

have revealed 165 sites exposed on low tide mudflats including Neolithic and Bronze Age 

material (O'Sullivan, 2001). 

3.2.2 Sea-level change and archaeology in the study area 

The broad Irish trends and patterns described above are equally applicable to the study 

area. The SMR database contains evidence of archaeological sites from all prehistoric 

periods, many of which are sited to take advantage of marine, coastal, or freshwater 

resources. Moreover, the Mount Sandel Mesolithic site is located in the study area, 

indicating that the timeframe of study goes back to the very earliest period of Irish 

prehistory. 

With respect to relative sea-level (RSL change) the study area shows the complex 

pattern associated with formerly glaciated areas subject to the combined forces of 

glacio-eustasy and isostatic rebound. The broad expected pattern of change is therefore 

an RSL highstand when the ice retreated, then an RSL fall as the crust uplifted faster 

than the glacio-eustatic rise, creating a sea-level lowstand. As uplift slowed, it was 

overtaken by the glacio-eustatic inflow resulting in RSL rising. This pattern could have 

continued till the present, or more likely given continued (albeit slow) crustal uplift, RSL 

fall could again have occurred towards the middle/end of the Holocene, when the glacioeustatic 

inflow ceased. 

Areas across the study region therefore underwent varying degrees of crustal rebound 

depending on the local thickness of ice and timing of deglaciation and consequently 

[ 19 ]


different patterns of RSL change. A database of radiocarbon dated RSL indicators and 

limits has recently been compiled (Brooks and Edwards, 2006). Four clusters of sea-level 

data cover the study area; North Donegal, Lough Swilly, Derry and North Antrim. The 

database also shows that with the exception of the westernmost portion of the study 

area (North Donegal), RSL index points are not available. Instead, knowledge of palaeosea-levels 

has been derived from limiting dates (Figure 3.2). Overall, these suggest an 

RSL rise from at least c. 10 ka and in some areas provide evidence of a Holocene 

highstand of 2-3m (see Figure 3.2; McDowell et al., 2005) or even as high as 5m above 

present (Kelley et al., 2006). 

To this can be added evidence that is less well age-constrained. Kelley et al. (2006) have 

documented an RSL lowstand in the form of submerged notches and wave truncated 

features in 32m water depth off Co. Antrim. This feature is not dated directly, but a set 

of submerged beach deposits in 30m water depth from Belfast Lough (some 90km to the 

southeast) are dated to 13.4 cal ka BP. Raised shorelines and marine deposits landward 

of the present coast also attest to formerly higher sea-levels created prior to the 

lowstand as the land was isostatically depressed (McCabe et al., 2007). 

In the absence of detailed RSL data, one alternative is to turn to computer models of the 

Earth’s isostatic response to ice loading. These estimate the magnitude and timing of the 

crustal uplift/subsidence using a model of the Earth’s rheology and isostatic response 

coupled with an ice sheet growth/decay history and a glacio-eustatic sea-level record, 

both of which are based on field data. RSL observations from the study area are used to 

calibrate the models, effectively adjusting them so that they fit the RSL data, but remain 

within the constraints set out by the Earth’s rheology, ice sheet history and glacioeustatic 

record. This allows RSL curves to be generated for areas that lack RSL data, or 

to fill gaps in extant curves within widely spaced data points. 

These Glacio-Isostatic Adjustment (GIA) models are well developed (see for example 

Lambeck, 1995; Peltier, 1994; Shennan et al., 2006) and have been specifically applied 

to Ireland (Brooks et al., 2008; Lambeck and Purcell, 2001). Nonetheless, there are 

claims that the models are inaccurate and fail to reproduce the sea-level changes seen in 

the Irish geological record. For example, McCabe et al. (2007) interpret a variety of 

geological evidence (including raised beaches and marine muds) as suggestive of rapid 

‘sawtooth’ fluctuations in RSL for NE Ireland between 21-17,000 BP created by rapid 

deglaciation and subsequent ice re-advances. In response, modellers have stated that 

the relationships of these data to past sea-level cannot be quantified and have also been 

assembled from across an area that likely experienced considerable local RSL variability, 

and therefore should not be combined into a single continuous record (Edwards et al., 

2008). Moreover, Kelley et al’s (2006) lowstand value of -30m is not replicated by 

modelled curves for Belfast or Antrim, which suggest values of -5 to -15m. The 

explanation for this is not yet clear, but either stems from an underestimate by the 

model or the fact that the material (shells) dated for the sea-level data are not in situ 

(Brooks et al., 2008). 

What should be clear from this is that at present neither the models, nor the geological 

evidence, provide direct quantifiable evidence of past sea-level. Rather they should be 

seen as providing estimates and constraints on the pattern of past RSL change. In any 

case, from an archaeological perspective, most of the controversies can be avoided as 

the period of focus is the Holocene not the Late Pleistocene (i.e. post- 10,000 BP). By 

this time, there is closer agreement between the RSL data and the models: the pattern 

of change at this point is one of an RSL rise from 10,000 BP to a highstand by 8 to 6,000 

BP followed by a fall to the present (Brooks et al., 2008; Kelley et al., 2006). GIA 

modeling suggests that the amount of RSL rise increased as one moves west across the 

study area (see isobase maps in Figure 3.3). For example RSL was between -4 to -8m 

on the North Antrim coast at 10,000 BP, but between -18 to -20m in Donegal at the 

same time. Conversely, the magnitude of the Holocene highstand decreased moving 

west across the study area. It reached Antrim by 8-7,000 BP and rose up to 2-5m by 

5,000BP, but did not penetrate as far as West Donegal. 

[ 20 ]


The estimates that will be used across within the study area can be summarized as 

follows: 

[i] Antrim 

GIA modeling indicates a lowstand of c. -12m on the North Antrim coast at c. 11 ka BP, 

and an RSL position of c. -6m by 10 ka BP (Figure 3.2). Conversely, RSL data can 

interpreted to show a lowstand of -30m by 13.5 ka BP rising to -12 to -14m by 10 ka 

(Kelley et al., 2006; McCabe et al., 2007). On this basis we will consider a maximum 

lowstand of -30m, but recognizing that by the time of the earliest known Mesolithic, RSL 

may have been higher; between -6 to -14m. 

[ii] Derry 

GIA modeling indicates a lowstand of c. -15m on the Derry coast at c. 11 ka BP and an 

RSL position of c. -8 to -10m by 10 ka BP (Figure 3.2). Once again though, these 

estimates are countered by apparent shoreline notches at -30m and 13.5 ka BP, putting 

RSL closer to -14m by 10 ka BP (Kelley et al., 2006; McCabe et al., 2007). Therefore, a 

maximum lowstand of -30m will still be considered, with the recognition that the 

lowstand may have been closer to -15 to -8m. 

[iii] Donegal 

GIA modeling indicates a lowstand of c. -20 to -25m on the Donegal coast at c. 11 ka BP 

and an RSL position of c. -15 to -10m by 10 ka BP (Figure 3.2). Unlike Northern Ireland, 

there are no lowstand estimates from geological data. If the -30m value is indeed 

correct, then GIA modeling suggests that the lowstand off Donegal should have been 

even deeper. In the absence of corroborating evidence, we will again use the -30m value 

as a maximum limit, but recognizing that the actual value could have fallen on either 

side of it. 

[ 21 ]


Figure 3.2 RSL data and GIA modeled curves for the study area. Data from Brooks and 

Edwards (2006); Brooks et al. (2008), additional lowstand point from Kelley et al. 

(2006). 

[ 22 ]


Figure 3.3 GIA-modelled RSL isobases for the Holocene. Negative values show where 

RSL was below present, positive values show where RSL was above present. The zero 

contour shows where RSL was at the same level as at present. Data from Brooks (2007). 

This implies that from an archaeological perspective, the seabed off the Antrim coast will 

contain landscapes dating to the early Mesolithic, but relatively little later material. By 

contrast, submerged prehistoric landscapes dating from the Mesolithic to the Neolithic 

could be found off Donegal. Derry, situated between the two, most likely had exposed 

landscapes during the Mesolithic and possibly the earliest Neolithic. 

This data, combined with topographic and bathymetric Digital Elevation Models (DEMs) 

allows the creation of palaeo-geographic reconstructions of the past landscape (Figure 

3.4). However, the available bathymetric DEMs are rather coarse; the most widely 

available model is the ETOPO-2 dataset which has a resolution of 2 minutes of 

latitude/longitude. Even conventional bathymetric charts are less than adequate due to 

the fact that they are constructed using widely spaced single-beam echosounder and 

lead line data, with considerable interpolation between individual data points. Any 

reconstructions which make use of these datasets can be used for illustrative purposes 

on large (i.e. hundreds of kilometers) spatial scales, but are insufficiently accurate for 

interpretative purposes on local to regional scales (i.e. tens of kilometers or less). 

[ 23 ]


Figure 3.4 Examples of low-resolution palaeo-geographic reconstructions created using 

GIA modeled RSL change and the ETOPO-2 DEM. Timesteps are 18,000 BP (left) and 

16,000 BP (right) (from Brooks, 2007). 

The high-resolution multibeam data acquired by the JIBS project will therefore allow the 

creation of more accurate palaeo-geographic reconstructions which will serve two broad 

purposes. Firstly, they will allow the identification of features on the seabed that relate 

to the past coastline and subaerial landscape, in particular those which were 

preferentially used or occupied by prehistoric humans (Bell et al., 2008; Bell et al., 

2006). Examples of such features include beaches, lagoons, shoreline notches, barriers, 

estuaries, deltas, lakes and river channels. This will facilitate future efforts to prospect 

for submerged archaeological sites by pinpointing areas of the highest archaeological 

potential. Secondly, the reconstructions will aid in the interpretation of the existing 

archaeological record by placing it in a more accurate palaeo-geographical context. For 

example, how might past palaeo-geography have influenced migration and trade routes 

and how did prehistoric social and settlement patterns on the coast respond to the 

palaeo-geographic changes wrought by long-term processes of marine transgression and 

regression? 

[ 24 ]


4. DATASETS AVAILABLE FOR THIS STUDY 

4.1 JIBS data 

The objective of the JIBS Project was ‘to promote joint action to survey the 

seabed in such a way as to satisfy the needs of many organisations’. The 

Maritime and Coastguard Agency (MCA)-appointed contractors in partnership with 

the Marine Institute of Ireland surveyed and acquired, for the first time, 

comprehensive multi-beam bathymetry data over prioritised areas within the 3nm 

coastal strip between Fanad Head and Torr Head. The survey was conducted to 

IHO Order 1 standard, ensuring 100% coverage of the seabed (a summary of the 

requirements are outline in Table 4.1). Bathymetry and backscatter data were 

provided over the world-wide web as ‘Generic Sensor Format (GSF)’. 

Table 4.1 Summary of minimum standards for hydrographic surveys IHO S-44 

5 th edition Order 1 (International Hydrographic Bureau, 2008) 

ORDER 1 

Examples of typical areas Harbours, harbor approach channels, recommended 

tracks and some coastal areas with depths up to 

100 m 

Horizontal Accuracy (95% 5 m + 5% of depth 

Confidence Level) 

Depth Accuracy for Reduced a = 0.5 m 

Depths (95% Confidence b.= 0.013 

Level) (1) 

100% Bottom Search Required in selected areas (2) 

System Detection Capability Cubic features > 2 m in depths up to 40 m; 

10% of depth beyond 40 m (3) 

Maximum Line spacing (4) Not defined, as 

full sea floor search is required 

Notes: 

(1) Recognising that there are both constant and depth dependent uncertainties that affect the 

uncertainty of the depths, the formula below is to be used to compute, at the 95% confidence level, the 

maximum allowable TVU. The parameters “a” and “b” for each Order, as given in the Table, together 

with the depth “d” have to be introduced into the formula in order to calculate the maximum allowable 

TVU for a specific depth: 

± [a2 +(b*d)2] 

Where: 

a represents that portion of the uncertainty that does not vary with depth 

b is a coefficient which represents that portion of the uncertainty that varies 

with depth d is the depth 

b x d represents that portion of the uncertainty that varies with depth 

(2) For safety of navigation purposes, the use of an accurately specified mechanical sweep to guarantee 

a minimum safe clearance depth throughout an area may be considered sufficient for Special Order and 

Order 1a surveys. 

(3) A cubic feature means a regular cube each side of which has the same length. It should be noted 

that the IHO Special Order and Order 1a feature detection requirements of 1 metre and 2 metre cubes 

respectively, are minimum requirements. In certain circumstances it may be deemed necessary by the 

hydrographic offices / organizations to detect smaller features to minimise the risk of undetected 

hazards to surface navigation. For Order 1a the relaxing of feature detection criteria at 40 metres 

reflects the maximum expected draught of vessels. 

(4) The line spacing can be expanded if procedures for ensuring an adequate sounding density are used. 

"Maximum Line Spacing" is to be interpreted as the: 

- Spacing of sounding lines for single beam echo sounders, or the 

[ 25 ]


- Distance between the useable outer limits of swaths for swath systems. 

The cleaned JIBS data was supplied to the CMA in a processed Fledermaus format 

(*.scene/*.sd) in September 2008. Two versions of the data off Co.’s Derry and 

Antrim were made available by the MCA: gridded to a resolution of 6m and a 

resolution of 4m. Similarly, two datasets were supplied to the CMA by the Marine 

Institute for Co. Donegal: gridded to a resolution of 5m and a resolution of 2m. 

The backscatter data has not yet been made available. 

For Phase 2 we were given the following: cleaned xyz data for the whole JIBS 

area, raw .all files for data acquired with Kongsberg multibeam systems and 

xtf/gsf files for the data acquired with the Reson multibeam system. 

The data was explored in Fledermaus and exported to a GIS for further analysis 

(Figure 4.1). 

Figure 4.1 Map showing the extent of the JIBS bathymetric data set. 

4.2 Shipwrecks 

In addition to JIBS data, the following data sets were available during Phase 1 

(updated versions of these were integrated in Phase 2): 

• Maritime Sites and Monuments Records (MSMRs) for Counties Donegal, 

Derry and Antrim. These were provided by the Department of 

Environment, Heritage and Local Government (DEHLG) for Co. Donegal 

(30 entries in total) and by the Northern Ireland Environment Agency 

(NIEA) for the Co.’s Derry and Antrim (886 entries in total). See Figure 

4.2 for the spatial distribution of these sites. 

• Raster and vector versions of all Admiralty Charts for the north coast. 

[ 26 ]


• Dive Records for the Counties Donegal, Derry and Antrim. These are 

publically available on http://www.irishwrecksonline.net/. 

• Sub-bottom data. The University of Ulster has been collecting Chirp 

seismic data since 1997 off the coast of Northern Ireland. This data can 

linked to the JIBS data using IVS Fledermaus, SMTKingdom Suite and 

ESRI GIS. This will allow us to link anomalies detected on the bathymetry 

to sub-surface features, adding an extra dimension to the data. 

Data were projected from latitude-longitude co-ordinates (WGS84) or Irish 

National Grid into Universal Transverse Mercator (UTM) Zone 29. 

4.3 Submerged prehistoric landscape 

In addition to JIBS data, the following data sets were available during Phase 1 

(updated versions of these were integrated in Phase 2): 

• (Terrestrial) Sites and Monuments Records (SMRs). The record with data 

on 3,129 archaeological sites and historic monuments for Co. Donegal and 

16,020 sites for Co.’s Derry and Antrim, ranging from prehistoric tombs to 

post-medieval settlements were provided to us by the Department of 


(available on http://www.archaeology.ie/smrmapviewer/mapviewer.aspx) 

and by (NIEA) for the Counties in Northern Ireland. See Figure 4.3 for the 

spatial distribution of these sites. 

• Shuttle Radar Topography Mission (SRTM) data. This data is freely 

downloadable through the USGS “The National Map Seamless Server” 

(http://seamless.usgs.gov/ or ftp://e0srp01u.ecs.nasa.gov/srtm/). SRTM 

is an international effort to obtain digital elevation models on a global 

scale using radar interferometry. Data used for this study consists of 

SRTM3 data with a sample spacing of 3-arc seconds. This data was used 

to link the terrestrial topography to the acquired marine bathymetry. 

• Glacial Isostatic Adjustment (GIA) Models for sea-level change between 

20,000 – 1,000 BP. The model used is the ‘best-fit’ model described in 

(Brooks et al., 2007). It was developed following an iterative process that 

compared model outputs against Irish RSL data to achieve the closest fit 

possible. The two major components of this model are its ice model and 

Earth model. For the former it uses a version of the BIM-1 British and Irish 

ice sheet model (developed by Shennan et al., 2006) modified to have 

thicker LGM ice over parts of Ireland and the Irish Sea and an earlier and 

more rapid pattern of deglaciation. The latter is a three layer model with a 

lithospheric thickness of 71 km, an upper mantle viscosity of 4 X 10 20 Pa s 

and a lower mantle viscosity of 4 X 10 22 Pas. Modelled isobases of RSL 

change were provided by courtesy of Robin Edwards and Tony Brooks. 

[ 27 ]


Figure 4.2Overview map showing the documented shipwrecks off the north coast 

of Ireland. 

Figure 4.3 Overview map of study area showing all terrestrial archaeological 

sites. 

[ 28 ]


5. METHODOLOGY 

The methodology descried below concentrates on the new methods developed 

during Phase 2 of the project. For details on methods relating to work done in 

Phase 1, please refer to the 2008 report. 

5.1 Production of 1m resolution DEM 

The main purpose of multibeam sonar systems is to accurately determine the 

depth of the seafloor using an acoustic process called ‘beam forming’. By using 

several beams at different angles, a large area of seabed can be covered in a 

relatively short survey time. From the angle and travel time of the returned 

signal, the position of each echo can be calculated and a bathymetric map (also 

called a Digital Elevation Model, DEM) can be created. 

• workflow 

Ungridded, cleaned .xyz data were delivered to the University of Ulster in January 

2009 by the UK Hydrographic Office (UKHO) and the Marine Institute (MI). In 

total 5333 .xyz bathymetric files (152GB) from the Northern Irish side and 890 

.xyz bathymetric files (142GB) from the Republic of Ireland side of the study area 

were used to create digital elevation models (DEMs) (Figures 5.1 and 5.2 showing 

lines of MI and MCA/UKHO). 

All individual .xyz lines for both study areas were imported into IVS FM DMagic 

software for gridding. However, because both datasets are so large, the area had 

to be divided into smaller zones (called ‘quadrants’ hereafter) in order to perform 

the gridding process (Figures 5.1 and 5.2). Subsequently, each quadrant was 

gridded using a weighted moving average gridding type (weight diameter 3) to a 

1m bin size and output to the UTM zone 29N coordinate system. FM DMagic 

automatically creates a DEM from the resulting grid which can be imported into 

IVS Fledermaus for enhanced visualization and interpretation. In addition, the 

gridded data were exported as an ASCII grid for use in ArcGIS. 

The workflow is summarized as follows: 

> Load all lines into FM DMagic. 

> Select a quandrant in FM DMagic (if too many lines are selected, the program 

tends to crash during processing). 

> Grid the selected lines (1m bin, weighted moving average 3, UTM29N) and 

choose the default shading parameters in a first instance – this will create an *.sd 

file. 

> Export the created grid as an ArcGIS ascii for further analysis in ArcMAP. 

> Import the created *.sd file into IVS Fledermaus for visualisation and 

interpretation – note that in this step, the shading parameters may have to be 

changed to get the best results. 

[ 29 ]


Figure 5.1 Overview map showing the survey lines acquired by the MCA and 

UKHO in Northern Irish waters and the associated quadrants used to create 

DEMs. 

Figure 5.2 Overview map showing the survey lines acquired by the Marine 

Institute in the waters off Donegal (Republic of Ireland) and the associated 

quadrants used to create DEMs. 

• Issues encountered with the data 

Relatively few issues were encountered during the processing of the bathymetric 

data. However, it is worth noting that all IVS programmes (FM Dmagic, Geocoder, 

Fledermaus) struggle with large datasets and, therefore, it was necessary to 

divide the study area into smaller zones. 

At the moment, all .xyz data received from the MI have been processed to very 

high quality. This is not necessarily true for the data in Northern Ireland. In 

several areas, evident processing features have been detected in the bathymetric 

[ 30 ]


data (Figure 5.3). At the time of the data delivery, the UKHO was still actively 

working on bathymetric corrections. At present, it is unclear if this process has 

been finalised and whether the data can be made publically available. 

Nevertheless, the data quality has improved significantly compared to the data 

we were given for Phase 1 (Figure 5.4) of the project and, as a consequence of 

the relatively short time given to complete Phase 2, we decided that we could not 

wait any longer for the delivery of the final data and chose to work with what was 

available at the time. 

Figure 5.3 Bathymetry map in the area of the Bann Mouth showing tidal 

correction issues resulting in processing features. 

[ 31 ]


Figure 5.4 Improvement in the data around the southern tip of Rathlin Island in 

terms of the reduction of processing errors and improved tidal correction 

5.2 FM Geocoder processing of backscatter data 

Until recently, multibeam systems were predominantly used to measure the 

depth to the seabed in order to create a detailed bathymetric map. However, as 

the acoustic beams propagate through the water, they eventually interact with 

the seafloor or objects lying upon it. Most of this energy is reflected away from 

the sonar system, a small portion of the energy is lost in the subsurface and 

[ 32 ]


another small portion, known as backscatter, reflects back to the sonar system. 

The amplitude of this returned signal is measured by the transducers, together 

with the travel time. The amount of backscatter is determined by three factors: 

the local morphology of the surface ensonified, the small-scale (sub-metre) 

roughness of the surface and the material properties of the seafloor. The data are 

presented as a black and white image showing the strength of the returning 

energy and stitched together to display a continuous image of the seabed, called 

a mosaic. 

• Workflow 

In order to plot the multibeam’s associated backscatter data, the raw multibeam 

data were requested and received from the UKHO and MI. In Northern Irish 

waters, the original data were acquired using three different systems from three 

different vessels, while the MI deployed two different multibeam systems (Table 

5.1). In the same way as for the bathymetric data, the lines for each of the 

systems were divided into smaller areas (called ‘quadrants’) to make the data 

more manageable during processing and mosaicing in IVS FMGeocoder. 

Table 5.1 Overview of different vessels, multibeam systems and acquisition 

settings 

Area Vessel Multibeam 

system 

Northern 

Ireland 

Republic 

of 

Ireland 

Sounding 

mode 

Centre 

frequency 

Pulse 

length 

File 

format 

No. 

lines 

(size 

data) 

Jetstream EM3002 1 (shallow) 293kHz 149μsec .all 4148 

(187GB) 

Meridian Reson 7125 - 200/400kHz 10- 

300μsec 

.xtf/.gsf 950 

(76.5GB) 

Victor EM710 0 (v. 71-97- 206μsec .all 

Hensen 

shallow) 83kHz 

1 (shallow) 71-83- 

77kHz 

500μsec .all 412 

(54.8GB) 

2 

71-77- 2000μsec .all 

(medium) 74kHz 

Celtic EM3002 1 (shallow) 293kHz 149 μsec .all 809 

Voyager 

(147GB) 

Celtic EM1002 1 (shallow) 95kHz 200μsec .all 971 

Voyager 

(40GB) 

FMGeocoder is a software program designed to visualize and analyze backscatter 

data from multi-beam and sidescan sonar. In processing the source files into 

mosaics, it is designed to perform as many corrections as possible to maximize 

the information content within the backscatter signals. Once the mosaic has been 

created, a number of statistics can be calculated (Fledermaus reference manual, 

2009; Figure 5.5). The final mosaic and associated statistics can subsequently be 

exported as *.sd files for further visualization in IVS Fledermaus or as ArcGIS 

ASCII files. For this project, a *.sd file and an ArcGIS ascii grid were created for 

each individual quadrant mosaic before being merged into a larger mosaic in 

ArcGIS. The created *.sd scalar files can also be draped on top of the bathymetric 

*.sd file in IVS Fledermaus (see contact sheets). 

[ 33 ]


Figure 5.5. Example of Backscatter mosaic and a 

series of statistical maps created in Geocoder. Note 

that at present Geocoder does not display a colour bar 

with the imagery; to get the actual statistical values, 

the grids must be exported as ascii files and imported 

into ArcGIS – this bug will be reported to IVS. 

[ 34 ]



> Based on the .xyz data loaded in FM DMagic, the raw backscatter data for each 

system were divided into smaller zones and copied into individual folders. 

>FM Geocoder was run for each individual area for each system. The Kongsberg 

*.all files were loaded as ‘Beam Time Series data’ while the Reson *.xtf files were 

loaded as ‘Sidescan + Bathymetry data’. 

> Parameters chosen for the mosaicing of the *.all data: 1m bin size for the 

backscatter mosaic, 2m bin size for the statistics calculation, lock the histogram 

dB range to -70 to 10dB, NO Tx/Rx Power Gain correction and Apply FLAT AVG 

correction. 

> Parameters chosen for the mosaicing of the *.xtf data: 1m bin size for the 

backscatter mosaic, no statistics calculation, lock the histogram range to 20 to 

130 (mean amplitude), NO Tx/Rx Power Gain correction and Apply FLAT AVG 

correction. 

> The resulting mosaic for each quadrant was exported as a *.sd file for 

visualisation and interpretation in IVS Fledermaus. This file represents a scalar 

which can be draped over the bathymetric data. 

> The resulting mosaic for each quadrant was exported as an ArcGIS ascii file for 

analysis in ArcGIS. 

Figure 5.6 Overview map showing the survey lines acquired with the EM3002 

onboard Jetstream in Northern Irish waters and the associated quadrants used to 

create backscatter mosaics. 

[ 35 ]


Figure 5.7 Overview map showing the survey lines acquired with the Reson7125 

onboard Meridian in Northern Irish waters and the associated quadrants used to 

create backscatter mosaics. 


onboard Victor Hensen in Northern Irish waters and the associated quadrants 

used to create backscatter mosaics. 

[ 36 ]



onboard Celtic Voyager in the waters off Donegal (Republic of Ireland) and the 

associated quadrants used to create backscatter mosaics. 


onboard Celtic Voyager in the waters off Donegal (Republic of Ireland) and the 

associated quadrants used to create backscatter mosaics. 


Issues encountered can be divided into two types: problems related to the 

processing software and problems caused by the acquisition settings. 

[ 37 ]


FM Geocoder 7.0 was officially released in July 2009. As with any new software, 

several bugs and limitations were encountered, some of which have been 

addressed by IVS in subsequent releases, while others are still unresolved. The 

first major issue is the inability to read the Reson7125 .xtf and .gsf file as a pair 

(although the manual claims it is possible to do this). As a consequence, only the 

.xtf data could be imported and was treated by the software as ‘sidescan solo’ 

data. So, although the data are still displayed and mosaiced, no bathymetric 

corrections can be made nor can any statistics be calculated. The second issue is 

the inability to compensate data acquired with different systems for the creation 

of a single mosaic (Figure 5.11), and, therefore, a separate mosaic needs to be 

created for each system. This causes problems when the resulting mosaic (scalar 

.sd file) needs to be draped over the bathymetric data, as it is not possible to 

drape more than one scalar over each bathymetric DEM. Yet, each bathymetric 

quadrant generally has corresponding backscatter data acquired with the different 

systems and thus, in theory, should have multiple scalars associated with it. This 

issue still remains unresolved. 

Figure 5.11 Example of the inability of FM Geocoder to compensate the 

backscatter of two separate multibeam systems and create a seamless mosaic. 

Note the higher backscatter values in the EM710 – these are intrinsic to the 

system and should not be confused with a change in seabed substrate. 

Two of the datasets appeared to show backscatter offsets in the mosaics that are 

the results of changes in the acquisition parameters, which FM Geocoder cannot 

compensate for. The EM710 can be operated at three different modes (very 

shallow, shallow and medium) with each of these modes emitting different 

frequencies (with decreasing frequency values for increasing depths). The 

frequency content of the emitted beam plays an important part in determining 

the amount of energy that will be returned after bouncing off the seafloor, with 

[ 38 ]


higher frequencies losing their energy faster than lower frequencies due to 

attenuation in the water column. Hence, differences in frequencies in the pulse 

will result in differences in the absorption of the sound in the water column, as 

well as differences in the interaction with the seafloor, all affecting the final 

backscatter value. In the same way that the backscatter data from the different 

systems could not be mosaiced together in FM Geocoder, it appears that the 

different modes of the EM710 should be mosaiced separately as well. Currently, 

FM Geocoder blends the two modes into a single mosaic, but the interpreter 

should be aware that apparent changes in backscatter are caused by changes in 

the acquisition mode and not the substrate of the seabed (Figure 5.12). Ideally, 

the backscatter data should always be collected using a single, consistent setting 

over the study area. The MI definitely has adopted this practice, but it seems that 

the hydrographers in charge of collecting the data on behalf of the MCA, were 

mainly concerned about acquiring high quality bathymetric data and did not 

consider the impact changes in frequency would have on the backscatter data. 

The University of Ulster does currently not have any software to deal with this 

issue or divide the raw data depending on the mode setting. As such, caution 

must be taken when using the backscatter data to derive substrate maps without 

groundtruthing. 

Figure 5.12 Issues encountered when mosaicing EM710 backscatter data in FM 

Geocoder: changes in frequency mode result in blending of the grey levels – 

changes in grey level are the result of frequency changes, not substrate changes. 

Similar issues were encountered with the Reson7125 data. Changes in 

backscatter response occur in adjacent lines and cannot be attributed to changes 

in seafloor substrate (Figure 5.13). Unfortunately, we have not been able to 

extract the acquisition parameter settings for the .xtf files and, hence, have not 

been able to identify which parameter is causing the change. A final issue related 

to the Reson7125 concerns layback (positioning). This becomes visible on the 

final mosaic when tracing features that should be continuous, but now show a jigsaw 

pattern (Figure 5.13). Again, the University of Ulster currently does not own 

[ 39 ]


the appropriate multibeam processing software to rectify this problem. We hope 

to address this issue in the future. 

Figure 5.13 Issues encountered when mosaicing Reson 7125 backscatter data in 

FM Geocoder: inability to account for acquisition setting changes and layback 

positioning errors. 

5.3 Automated segmentation of backscatter data 

5.3.1 Context 

Segmentation is the process of reducing complex information (in this case a digital 

sonograph) into a simplified, arbitarary representation. This is done primarily as an 

interpretive aid and is typically used to identify objects or boundaries between (or within) 

aggregations of similar units (in this case pixels). Classification is distinct from segmentation, 

as this involves assigning value(s) to the defined segments, on the basis of their abiotic 

characteristics. This method is commonly used for habitat mapping, but is used here to 

attempt to distinguish archaeological sites from natural sites. 

The segmentation effort was focussed around Rathlin Island owing to the 

significance of this site in terms of shipwreck site distribution and submerged 

landscape potential and the time constraints involved in processing data in this 

manner. Full details of the processing effort are included in Appendix A. 

[ 40 ]


The segmentation was approached from two different perspectives. In the first 

instance, the raw data (sensor outputs) were classified using a commercial 

platform for semi-automated classification based on an unsupervised technique 

(QTC-Multiview). In the second, an independent unsupervised classification was 

performed on the geometric and radiometrically compensated backscatter 

imagery using industry standard image processing software (ERDAS Imagine). An 

unsupervised method of classification was chosen in both approaches owing to 

the paucity of ground truth data over the site at large. 

In both instances, the classification process was complicated by variations in 

operational frequency and angular sector; resulting in both internal variations for 

systems such as the EM710; and across the survey area as a whole which was 

covered using several different systems (EM1002; EM3002D; EM710; Reson 

7125). There were additional complications owing to variations in pulse length 

and resultant operational frequency within sonar systems (particularly the EM710 

which has 3 pre-defined mode settings). In response to these issues, the 

segmentations were conducted in four separate classifications; two of which were 

variable by system, and three of which were variable by mode (pulse length). For 

the purposes of comparison, where possible the data were conditioned and 

processed in as close a manner as possible. 

5.3.2 QTC-Multiview 

The initial classification was achieved using Quester Tangent Corporation’s 

Multiview software package. This software has the capacity to perform an 

unsupervised classification of multibeam backscatter data based on the beamtime 

series backscatter imagery, and relative success using this approach has 

been reported in the literature (e.g. Robidoux et al., 2009; McGonigle et al., 

2009; McGonigle et al., in press; Preston, 2009). 

The data were subjected to the QTC-Multiview processing flow. This method for 

processing multibeam backscatter imagery has been well described in the 

literature (eg. QTC, 2005; Robidoux et al., 2009; McGonigle et al., 2009; 

McGonigle et al., in press; Preston, 2009). The data were processed in cognisance 

of the existing sub-divisions of the JIBS area (described above). 

All survey lines which intersected areas A1_5, A1_6, A2_5, A2_6 and A3_6 were 

loaded and cleaned using a threshold of 1000. The volume of data for each 

respective system within each area is presented in Table 5.2. Quality control for 

the data was performed in the absence of the CARIS reject flags, which would 

have allowed for less data redundancy at the cleaning stage. 

Table 5.2: Distribution of survey lines (by acquisition system) relative to 

processing areas. 

AREA EM3002D (lines) EM710 (lines) 

A2_5 323 38 

A1_5 213 67 

A1_6 19 31 

A2_6 175 108 

A3_6 53 24 

Total 783 268 

Unresolved Errors 9 0 

The software performs the segmentation based on the backscatter image data, 

which it does at one of a variety of scales. In this instance, the rectangle 

[ 41 ]


dimensions selected for the data were based on the footprint on the ground of a 

representative proportion of the survey lines. The average footprint (on the 

ground) of each rectangle was variable with depth. A 10-line sub-sample of 

survey lines with rectangle dimensions of 257x17 pixels showed an average 

footprint of 13.7m across track against 8.9 m along track. Vessel speed was 

variable, averaging at 2.5 – 3 ms -1 . Ping rate was similarly variable, ranging from 

5.9 - 6.9 Hz across the sub-sample of EM3002D data. 

The next stage in the processing flow is the generation of vectors (FFVs). At the 

chosen settings, vectors were generated which describe the variation in the 

backscatter imagery contained within each rectangular sample. These are then 

reduced by Principal Components Analysis (PCA), in the manner described (QTC, 

2005), to the three values most responsible for the variance in the dataset. The 

data are then subjected to a simulated k-means algorithm, which attempts to 

amalgamate the vectors into statistically logical groups (Preston, 2009). Finally, 

the data are described by confidence (the probability that a record belongs to the 

class to which it has been assigned) and probability (the measure of the 

closeness to the cluster centre in the 3-D vector space) (QTC, 2005). 

Problems encountered 

Retrospectively, the choice of rectangle dimensions for the EM710 dataset may 

have been less appropriate than initially expected. However, the need to impose 

an amount of continuity to the data was the reason this choice was made. This 

point is further substantiated by the statement in the software manual, which 

advises that “It is strongly recommended that the rectangle size be constant 

throughout all data sets in a survey” (QTC, 2005). The Reson7125 data 

(Meridian) were omitted from the classification process, owing to data format 

incompatibility with the software. The multibeam data were supplied in Generic 

Sensor Format (.gsf) and eXtended Triton Format (.xtf) and QTC-Multiview does 

not support either of these two data types. These data formats represent 

processed data types which have been exported from CARIS, or a similar 

hydrographic processing package (e.g. Hypack). The raw sensor data from the 

Reson7125 is logged as either .s7k or .pds format. It is possible to convert.s7k to 

.pds retrospectively. 

5.3.3 QTC-Clams 

The categorical interpolation was performed in proprietary software QTC-Clams 

The interpolation parameters were kept constant throughout each dataset. The 

final output grids were preserved at a 20 m cell size, as this was approximate to 

the average spatial footprint of the rectangular patches. This expedited the 

interpolation process significantly, without impacting the quality of the 

classification process. The interpolation parameters were tested iteratively, before 

accepting a 100m search radius based on the 35 points closest to the centre of 

each grid node. 

Problems encountered 

The main issues with QTC-Clams are those which have been commonly reported 

in the literature (e.g. McGonigle et al., 2009; McGonigle et al., in press). The 

persistence of range dependent artefacts, dependence related to the orientation 

of survey lines and disparate data density in either across-track or along track 

dimensions all present problems in these series of classifications. However, they 

are a limitation of this approach for seafloor characterisation and have not 

occurred as a result of procedural inaccuracies/inconsistencies through the course 

of this research. 

[ 42 ]


5.3.4 Erdas Imagine 9.2 

The second classification was more user-directed, and was based on an 

unsupervised classification of the corrected backscatter imagery in ERDAS 

Imagine v.9.2. This effort was focussed on Church Bay area, Rathlin Island. 

Image statistics were recalculated and bin-sizes redefined prior to commencing 

the classification process. The isodata clustering approach was then used to 

define 5 discrete classes based on 5 iterations of the image. The image was 

filtered using a low-pass 7x7 window prior to being subjected to the classification 

procedure. 

5.3.5 ArcGIS transfer 

In addition to the material presented for ArcGIS above, a series of stages of 

processing were undertaken in relation to the backscatter segmentation. The 

generation of classified vector data (QTC-Multiview output) and the classified 

interpolated surface (QTC-Clams) was achieved independently of ArcGIS, 

however, these data had to be conditioned in order to facilitate their import into 

the GIS environment. This was achieved using Geospatial Designs GridConvert 

(Evans, 2005) freeware to transform the data from Golden Software’s .grd format 

into generic ASCII grids. The classified vector data from QTC-Multiview also were 

used to produce inverse distance weighted interpolations of the Q1, Q2, Q3, 

Confidence and Probability values generated by processing. The provenance of 

these values is discussed above. Metadata was written in accordance with ISO 

19115, and the filing and naming conventions were standardised. The data 

generated in the Erdas Imagine environment were interchangeable with ArcGIS 

by using .img image format. However, the unsupervised classification performed 

in Imagine was subjected to manual cleaning in ArcGIS to remove nadir striping, 

and to suppress obvious acquisition artefacts. This was achieved using ESRI’s 

Spatial Analyst extension by converting the surface to polygonal vectors, and 

manually editing the class values to force accordance with the surrounding areas. 

5.4 Seismic integration into SMT Kingdom Suite 

Seismic systems called sub-bottom profilers are used to image sediment layers 

and bedrock beneath the seabed. Over the past ten years, the University of Ulster 

has acquired seismic data off the north coast of Ireland using a Chirp sub-bottom 

profiler. Such Chirp systems can be classified as high-resolution seismic sources 

and are ideal for shallower coastal waters. Depending on the frequency used and 

the type of sediment, penetration can vary from 3m in very coarse sands to over 

100m in fine-grained sediments, with a vertical resolution between 10–40cm. 

• Workflow 

In the first instance, all Chirp data acquired by the University of Ulster were 

converted from 8mm Exabyte tape to DVD. As the University currently has no 

appropriate tape drive connected to a Unix/Linux system, this process was 

outsourced to Adaptatech Ltd., a London-based company specialised in 

duplication and replication of digital data. In total 16 tapes were converted to 

DVD, containing 11.63GB of seismic data. 

On inspection of the header information, it became apparent that no navigational 

data were available in the header. For some surveys, there was a separate digital 

record with (D)GPS coordinates, as recorded on the survey vessel, while for other 

surveys only the printed record was available. Where a digital record was found, 

[ 43 ]


the position of each shotpoint was linked to a position using the GPS time in the 

header and GPS time from the navigation file. In the case of the paper trace, a 

new digital record was created with the shotpoints’ GPS time and location. There 

was no record of the offset between the ship’s (D)GPS antenna and the position 

of the Chirp system – it probably ranged from a few metres to up to several tens 

of metres. However, as the majority of the surveys were acquired using GPS 

positioning, with a horizontal accuracy of at best 30m or at worst in the order of 

50-100m, layback correction would be smaller than the actual GPS-induced error. 

In the next step, all Chirp data and related navigational data were imported into 

SMT Kingdom, a windows-based seismic interpretation and visualization software 

(some post processing is allowed). Navigational data were initially imported in 

WGS84, but were converted to UTM29N on request. All visualization and 

interpretation were made using UTM29N. A Geotiff of the multibeam bathymetric 

and backscatter data were imported, linking seabed features to sub-surface 

stratigraphy (Figure 5.14). 

Figure 5.14 Overview of the Chirp sub-bottom lines overlaid on a geotiff of the 

multibeam bathymetry as displayed in the base map in SMT Kingdom (note: the 

lack of colour bar associated with the bathymetry reflects the fact that this is a 

geotiff and not an actual grid with associated depth values). 

Subsequently, post-processing was performed, which aids interpretation of the 

seismic sections. On inspection of the data’s power spectrum, it became apparent 

that, although the spectrum should look like a bell shaped curve between 2- 

10kHz, two peaks were present in the data: one between 1-7kHz and one 

between 7.5-14kHz. From analysis of the data, it was decided that the higher 

frequency peak was probably caused by noise. Therefore, a bandpass filter was 

applied running from 2.5-3-7-7.25 kHz to select the lower frequency curve 

(Figure 5.15). Subsequently, automatic gain control was added to the filtered 

data. A filter length of 5, 10, 15 and 20 ms was tested – 10ms gave the best 

result, but all other filter lengths were kept as well. Finally, an envelope was 

applied to the processed data. This operator is commonly applied to Chirp data 

and enhances interpretability (Figure 5.16). 

[ 44 ]


Figure 5.15 Amplitude spectrum for the raw data (blue) and the bandpass 

filtered data using a 2.5-3-7-7.25kHz trapezoid filter (red). 

There was no record of tidal predictions or tide gauge data for any of the survey 

days, making it impossible to adjust the data accurately to Chart Datum. It was 

decided to link obvious features seen on the seismic lines, which in all likelihood 

have not changed significantly over the years (e.g. bedrock outcrops), to 

corresponding features and their depths as recorded from the multibeam data. 

Seismic cross-lines were used to further constrain the tidal height adjustment. 


> Create a navigation file with shotpoint, x, y, name of line 

> Load the navigation files into SMT Kingdom using the following workflow: 

surveys > import world coordinate > 2D by file > choose UTM29N 

> Load the Chirp data in SMT Kingdom using the following workflow: right click 

on the imported survey line in the work tree > import > segy. Select the correct 

line and the correct byte numbers and sample interval. When the ‘assign 2D 

shotpoints to traces’ pops up, define the shotpoint and trace numbers by copying 

the full range into the appropriate box. 

> For processing in SMT Kingdom: go to Tools > TracePAK > Process Multiple 

Traces > Select the lines and data type > select the processes you want to 

perform 

> Import a Geotiff from ArcGIS into SMT Kingdom. First export the required map 

in ArcGIS using the following workflow: file > export map (make sure to tick the 

box that says ‘save world file (tfw)’). Import into SMT Kingdom using: Culture > 

import > culture group. 

[ 45 ]


Figure 5.16 Processing steps performed to enhance the seismic lines and aid 

interpretation. 

[ 46 ]



The first issue was the lack of corrected navigational data written into the header 

of the seismic data. On further examination, it appeared that the older data were 

acquired with GPS and that a layback correction would not account for the fact 

that the horizontal positional accuracy would at best be ±30m. The more recent 

data, which have DGPS navigation accuracy (±3m), were acquired with the Chirp 

system towed relatively close to the survey vessel and thus, the positional 

accuracy should be within a few metres. When comparing features on the seismic 

lines with the multibeam data, the horizontal accuracy of the seismics is within 

the accuracies predicted for the navigational system used. 

The tidal height issue was tackled using the interactive vertical shift tool in SMT 

Kingdom. However, the fact that there is a navigational error and the velocity of 

sound through the water is not accurately known (this is needed to convert the 

measured two-way travel time into a depth measurement), means that the final 

result is very much an approximation. Nonetheless, by using the seismic data in 

association with the multibeam data, it is possible to determine a rough 

calculation for the depth of buried strata (this in turn will depend on the velocity 

of sound through the sediments). 

The third issue encountered had to do with a limitation of the SMT Kingdom 

software which is mainly geared towards the oil and gas industry, thus dealing 

with lower frequency seismic data used for imaging deeper structures. When 

processing needs to be performed, SMT Kingdom is only able to cope with data 

that have a sampling rate of >0.05 ms (= Nyquist frequency of 10000 Hz). The 

raw Chirp data have a sampling rate of 0.031 ms (= Nyquist frequency of 16129 

Hz) and resampling to 0.05ms would mean losing all data above 10 kHz. 

Although we currently use a bandpass filter to eliminate all frequencies above 7.5 

kHz, we might still need the higher frequencies for future analysis. Therefore, the 

sampling rate was manually overwritten in SMT Kingdom during data import from 

0.031ms to 0.031 s. This means that, instead of the data being in the 2-12 kHz 

range, the software thinks the data are in the 2-12 Hz range with a Nyquist 

frequency of 16.129 Hz. In other words, kHz becomes Hz and milliseconds 

become seconds, allowing us to perform any processing we wish without any 

need for resampling. 

A final issue relates to a problem with some of the data for interpretation. Firstly, 

much of data have been acquired in areas with a gravelly or coarse sandy 

seabed. The high frequency signal of the Chirp system is known to be subject to 

scattering and high attenuation in these sedimentary types, thus not imaging the 

deeper structures (Figure 5.17). Secondly, on some of the survey days there 

must have been a great deal of swell/strong waves. This makes it difficult to 

interpret the subsurface unless the seabed can be artificially flattened in the 

software (Figure 5.18). Although this is possible in SMT Kingdom, it also means 

that all sedimentary features (e.g. sand ripples, sand waves) are flattened. 

Without any metadata about the sea-state during acquisition, it is also impossible 

to know which features are caused by surface waves and which are actual 

sedimentary features. 

[ 47 ]


Figure 5.17 Example of Chirp profile acquired over gravelly seafloor (typically 

characterized by the diffraction hyperbolae) – note that there is no further 

penetration through the seabed. 

Figure 5.18a) Example of Chirp profile acquired during adverse weather 

conditions (waves). b) The same profile but with seabed flattened reveals a more 

complex infill feature. 

[ 48 ]


5.5 GIS – spatial integration of data sets 

Three designated GIS projects have been developed - one project for shipwreck 

investigation, one for investigation of submerged landscapes and one with the 

results of the backscatter segmentation. 

For all three projects, the quadrants with processed 1m multibeam and 

backscatter data were exported from IVS Fledermaus in ascii raster format. These 

were then converted into ERDAS Imagine grids (float data type) using the 

ArcCatalog ASCII to raster conversion tool. All grids were georeferenced and 

projected in UTM Zone 29N. Subsequently, a mosaic was created in ArcCatalog 

using the Data Management Tool (raster > mosaic to new raster) with 

parameters set to 32-bit float for the pixel type and choosing the LAST mosaic 

method for the bathymetry mosaic and the BLEND mosaic method for the 

backscatter mosaic (Figures 5.19 and 5.20). As the backscatter was acquired with 

five different systems, five backscatter mosaics were created. In ArcMAP the 

colour histograms of the backscatter data were manually manipulated to make 

the transition between the datasets as seamless as possible. In order to do that, 

all backscatter data were normalised, with a value of 0 indicating a high 

backscatter value (i.e. hard substrate) and a value of 1 indicating a low 

backscatter value (i.e. soft substrate). However, due to the mode changes (and 

associated changes in the frequencies of the emitted pulses) of the EM710 and 

compensation problems with the Reson7125, it has been impossible to create a 

single backscatter map on which the boundaries between the different datasets 

disappear. As said before, caution must be taken when interpreting these maps 

without groundtruthing information. 

The 1m bathymetric grid was post-processed using Spatial Analyst within ArcGIS 

to derive a hillshaded image, bathymetric contour lines and slope angles (Figure 

5.21). 

Figure 5.19 Overview map of the JIBS bathymetric data gridded to 1m and 

entered into a GIS environment. 

[ 49 ]


Figure 5.20 Overview map of the JIBS backscatter data gridded to 1m and 

entered into a GIS environment. 

Figure 5.21 Slope map for the JIBS 1m bathymetric dataset, created in ArcGIS. 

Ancillary and updated datasets that have been integrated in the GIS during Phase 

2 include: 

[ 50 ]


• (Terrestrial) Sites and Monuments Records (SMRs). These record data on 

3,129 archaeological sites and historic monuments for Co. Donegal and 

16,020 sites for Co.’s Derry and Antrim, ranging from prehistoric tombs to 

post-medieval settlements and were provided to us by the Department of 


(available on http://www.archaeology.ie/smrmapviewer/mapviewer.aspx ) 

and by the Northern Ireland Environment Agency (NIEA) for the Counties 

in Northern Ireland. 

• Maritime Sites and Monuments Records (MSMRs) for Counties Donegal, 

Derry and Antrim. These were provided by the DEHLG for Co. Donegal and 

by the NIEA for Co.’s Derry and Antrim and were updated in 2009 (Figure 

5.22). 

• Shuttle Radar Topography Mission (SRTM) data. This data is freely 

downloadable through the USGS “The National Map Seamless Server” 

(http://seamless.usgs.gov/ or ftp://e0srp01u.ecs.nasa.gov/srtm/ ). SRTM 

is an international effort to obtain digital elevation models on a global 

scale using radar interferometry. Data used for this study consist of 

SRTM3 data with a sample spacing of 3-arc seconds. These data were 

used to link the terrestrial topography to the acquired marine bathymetry 

(Figures 5.22 and 5.23). 

• Most up-to-date UKHO wreck record, received July 2009 (Figure 5.22). 

• Raster and vector versions of all Admiralty Charts for the north coast. 

• Navigational data for the acquired Chirp sub-bottom profiler lines. 

• Sport diver wreck records for the Counties Donegal, Derry and Antrim. 

These are publically available on http://www.irishwrecksonline.net/. 

• Data interpretations (anomalies, features, archaeological potential) 

detected during Phase 1 of the project 

Figure 5.22 Overview map of the JIBS bathymetric data gridded to 1m, showing 

documented shipwrecks off the north coast of Ireland. 

[ 51 ]


Figure 5.23 Overview map of the JIBS bathymetric data gridded to 1m, showing 

the three identified anomaly classes from Phase 1 of the project. 

In addition to the processing work, all datasets discussed above have been 

updated such that all associated metadata have been entered using ISO standard 

19115. 

[ 52 ]


6. RESULTS 

The results described below concentrate on the new results developed during 

Phase 2 of the project. For details on the results from Phase 1, please refer to the 

2008 report. The results section below is structured around the project 

objectives. 

6.1 Refining palaeo-geographies – The Skerries case study 

The Skerries are a chain of islands formed by a dolerite sill and extend off the 

coast of Co. Antrim, east of Portrush (Figures 6.1 and 6.2). They enclose an area 

of c. 2 km 2 and were identified in Phase 1 of the project as a region of high 

archaeological potential for the preservation of submerged prehistoric material. 

This interpretation was derived using the lower resolution 4m and 6m multibeam 

bathymetric datasets (see Phase 1 report). The following sections provide an 

updated interpretation of the palaeo-geography and archaeological potential of 

this area using combined higher (1m) resolution multibeam bathymetry and 

seismic profiles from Chirp sub-bottom surveys. 

Figure 6.1 Overview map of part of the Northern Ireland JIBS bathymetric data 

gridded to 1m, showing location of the Skerries. 

[ 53 ]


Figure 6.2 Photo showing the Skerries and mainland coast looking to the west. 

Palaeo-geographic reconstruction 

The changing palaeo-geography of the Skerries at selected time intervals is 

shown in Figure 6.3. At 14,000 BP, based on the Glacio-Isostatic Adjustment 

(GIA) model of Brooks et al. (2008), relative sea-level (RSL) in this area was 

approximately 13.5m below present (Figure 6.3a). Although there is currently no 

archaeological evidence for an occupation of Ireland at this time, recent evidence 

which pushes the occupation of Scotland (visible from north coast of Ireland) 

back to 14,000 BP (Saville et al., 2007) shows that there is no reason why an 

earlier occupation should not be considered. Based on the palaeo-geographic 

reconstruction, the coastline extended out by several hundred metres up to a 

kilometre and with the Skerries themselves directly linked to the mainland 

forming a sheltered embayment with water depths of c. 5 to 10m. The next 

timestep modelled is 12,000 BP (Figure 6.3b). Again, there is no recorded Irish 

occupation, but the comments made above in reference to the Scottish situation 

still apply. RSL was still broadly similar to the previous timestep (-14m vs. - 

13.5m) though the Brooks et al. (2008) model predicts that RSL may have risen 

by several metres and then fallen again in the intervening period (Figure 6.3f). By 

10,000 BP (Figure 6.3c), the period of the first known Mesolithic occupation, RSL 

was rising and had reached -9.5m. The sheltering embayment was now flooded, 

and the Skerries separated from the mainland. The mainland palaeo-coastline 

was open and linear and still extended out from the modern beach by several 

hundred metres. Although the embayment had gone, the Skerries were still 

extended by lowered sea-levels and may have afforded more shelter to the 

coastline than at present. RSL rose rapidly over the next thousand years, 

reaching -5m by 9000 BP (Figure 6.3d). The palaeo-shoreline was now a strip 

extending the modern shoreline by several tens to low hundred metres. By 8000 

BP, RSL had broadly reached present levels (-1.5m) and the palaeo-coastline was 

situated in the vicinity of the modern low tide mark (Figure 6.3e) 

[ 54 ]


Figure 6.3 Palaeo-geographic reconstruction based on JIBS 1m multibeam 

bathymetry. Terrestrial DEM is provided by the SRTM dataset. Boundary between 

the blue and light green areas shows the palaeo-shoreline, areas shaded 

turquoise show the position of the maximum lowstand thus providing an 

indication of the degree of land loss. A) 14,000 BP, RSL at -13.5m. B) 12,000 BP, 

RSL at -14m. C) 10,000 BP, RSL at -9.5m. D) 9,000 BP, RSL at -5m. E) 8,000 BP, 

RSL at -1.5m. F) GIA modelled curve from Brooks et al. (2008) showing position 

of modelled timesteps along the RSL curve. 

[ 55 ]


Seismic stratigraphy 

An extensive seismic (CHIRP sub-bottom) survey of the Skerries study area was 

undertaken in 1997 (Figure 6.4). The following sections discuss the seismic 

stratigraphy of the Skerries and its relationship to RSL and palaeo-geographic 

change by reference to a profile taken from the western side of the Skerries (see 

Figure 6.4 for profile location). At least seven stratigraphic units can be observed 

in this profile, indicating a complex pattern of sedimentation and erosion beneath 

the modern seabed (Figure 6.5). 

Figure 6.4 Tracklines of the Chirp survey undertaken in 1997 overlain in SMT 

Kingdom onto A) JIBS 1m bathymetry geotiff. B) JIBS 1m backscatter. Red lines 

shows position of seismic section discussed in the text. 

Interpretation of the profile is based on the sedimentary sequence observed at 

Portballintrae (c.5 km to the east; McCabe et al., 1994), and previous research 

[ 56 ]


on the offshore stratigraphy of the Skerries area (Cooper et al., 2002). The 

sequence seen in Figure 6.5 is interpreted as follows and summarised in Table 

6.1. 

Acoustic basement (Unit B) is interpreted as bedrock. This is indicated by a lack 

of seismic penetration and relief that is similar to known rocky outcrops imaged 

on other seismic lines. It is uncertain whether this acoustic basement relates to 

the dolerite sill forming the Skerries or a basalt or chalk outcrop similar to those 

presently seen onshore (see Figure 6.2). 

Unit D overlying Unit B is interpreted as diamicton, deposited in a glacio-marine 

setting by a retreating ice sheet. This unit can be seen at Portballintrae where it 

comprises of massive muds with dispersed pebbles, cobbles and boulders. Unit D 

can be divided into two units with the lower unit clearly showing point reflectors 

and chaotic bedding and the upper unit characterized by greater acoustic 

transparency with some point reflectors. Based on the acoustic signature these 

units were interpreted as lodgement till deposited under the ice (higher 

proportion of coarse material) draping the bedrock and grading into waterlain till 

(higher proportion of fines with occasional boulder) deposited at the ice front as it 

retreated. 

Overlying Unit D is Unit SM, a coarsening upwards sequence of muds, silts and 

sands deposited in a shallow marine setting following further ice retreat. Such a 

sequence can be seen in the exposed section at Portballintrae overlying the 

diamicton (McCabe et al., 1994). Interpretation of a coarsening upward sequence 

is based on the transition from largely acoustic transparent deposits at the base 

of the unit to medium high amplitude sub parallel reflectors and then to gravel 

rich deposits where the unit outcrops on the seabed (as shown by high amplitude 

point reflectors). The presence of gravel on the seabed in this area is also 

supported from by the high values shown in the backscatter data. The coarsening 

upwards nature of Unit SM is created by the decrease in water levels as isostatic 

rebound resulted in RSL fall. This brought the seabed increasingly into mean 

wave base resulting in winnowing of the finer fractions and then a transition to 

sand and gravel rich shoreface deposits. The higher concentration of gravel where 

this deposit is exposed at the seabed is also a function of winnowing by modern 

currents. Whether the sequence exhibits the same level of preservation as the 

Portballintrae sequence, where rhythmically bedded sands and muds 

corresponding to episodic storm events can be observed, is uncertain. The 

uppermost boundary of this unit forms an unconformity possibly created by wave 

action as RSL fell. 

Unit SM is covered, in the southern section of the profile by unit RS. This is 

interpreted as a series of regressing sand deposits that formed as RSL fell (see 

Cooper et al., 2002). With continued RSL fall to the lowstand, the sand sheet was 

subaerially exposed and eroded (unconformity) and possibly overlain by a thin 

veneer of terrestrial sediment (soil formation). Consequently, the bright reflector 

above Unit SM is classified as Unit T. In other seismic profiles within the study 

area channel features can be seen cutting though Unit T. 

As RSL rose again following the lowstand, the terrestrial surface was probably, in 

some areas, eroded by wave action or in other cases, covered by beach 

sediment, as represented by Unit BS. The acoustic signature of Unit BS is similar 

to its overlying deposit, Unit MS. This latter unit comprises modern marine sand, 

the source of which is reworking of Units SM and RS and can be observed over 

large parts of the study area. 

[ 57 ]


Within the seismic profile, it is clear that Units RS, T, BS and MS are restricted to 

the southern half, with the northern half represented by the winnowed gravelly 

surface of Unit SM. The high gravel content of this area may be explained by 

transgressive and modern erosion removing the finer sediment from Unit SM and 

carrying it onshore by wave and tidal action to bury Units RS, T and BS. The 

pattern of erosion (erosion in the northern half, deposition in the southern half) is 

explained by proximity to a gap between Ramore Head and the westernmost 

island of the Skerries. This narrow (c. 350m) gap funnels strong tidal currents 

into the Skerries resulting on localized erosion (see backscatter plot Figure 6.4b). 

Figure 6.5 Seismic profile from the Skerries showing A) uninterpreted data and 

B) interpreted stratigraphy (see text and table 6.1 for description and Figure 6.4 

for profile location). 

[ 58 ]


Table 6.1 Summary table showing seismic sequence from profile in the 

Skerries, stratigraphic interpretation and RSL change. 

Archaeological implications 

1) Palaeo-geographic reconstruction 

Analysis of seismic data shows that considerable variations in sedimentation have 

taken place since the ice retreated. Consequently, the exact geometry of the 

palaeo-shoreline represented in Figure 6.3 should still be taken as a first order 

approximation despite the use of higher resolution bathymetry. It is clear from 

the seismic data, that much of the modern seabed is comprised of sand reworked 

[ 59 ]


from earlier deposits. In some instance, this sand has buried pre-existing 

deposits while in other instances, the pre-existing deposits have completely 

eroded. This can be illustrated by reference to the channel feature shown in 

Figure 6.3, the presence of which controls the existence of the reconstructed 

lowstand embayment. In this example, it is highly probably that Units RS and T 

were once more laterally extensive, stretching out across the area now occupied 

by the channel. However, with RSL rise breaching the Skerries and the funnelling 

of strong tidal currents into the gap between Ramore Head and the westernmost 

island of the Skerries, localized erosion resulted in reworking of Units RS, T and 

the uppermost levels of SM such that finer sediments (sands) are removed and 

deposited onshore to form Unit MS. The coarser deposits are left behind in the 

channel to form the gravel rich outcrop at the seabed. Effectively, the sides of the 

channel are higher than during the lowstand, while its base is lower and it is likely 

that the lowstand Skerries may have been characterised by a more extensive 

level surface rather than the reconstructed embayment which may have formed 

later (i.e. during the period of RSL rise). 

2) Preservation of archaeological deposits 

In phase 1, the entirety of the Skerries was defined as an area of high 

archaeological potential. However, integration of high-resolution bathymetry with 

backscatter data and seismic profiles allows further refinement of this 

classification. Archaeological deposits are most likely located on or within Unit T, 

which may be preserved where it is buried beneath Units MS and BS. The 

winnowed surface of Unit SM, which is potentially exposed over large areas within 

the Skerries (see Figure 6.5) is of low archaeological potential, though it is 

possible that lithics in secondary context may still be present as a lag deposit. 

3) Targets for ground-truthing and archaeological sampling 

Validation of the above preliminary interpretation requires ground-truthing of the 

seismic profiles by sediment sampling. Priority ground-truthing targets include 

grabs of Unit SM where it lies exposed on the seabed, and Units RS and T where 

they pinch out close to the seabed. Coring of Units T and RS where they lie close 

to the seabed surface is also a priority as they offer the highest potential for 

containing archaeological remains and associated palaeo-environmental evidence. 

6.2 Target sites identified for direct sampling 

2009 Marine Institute ship-time funding 

In 2009, the investigators were successful in securing ship-time on the state 

Research Vessel Celtic Explorer (see section 7 for further details) to continue to 

acquire data and samples within the JIBS study area. The cruise, operating from 

16 th to 21 st December 2009, will concentrate efforts on acquiring additional 

acoustic, video and direct sample data from study sites deemed of highest 

archaeological potential. The total value of this cruise is 112,000, funded 

through the auspices of the Marine Institute’ ship-time programme. 

Three main phases of data acquisition are planned for this cruise: 

[1] Seismic profiles will be acquired around Rathlin Island and through the length 

of Lough Foyle to aid detailed palaeo-geographic reconstructions of these highpotential 

areas. 

[ 60 ]


[2] Time-lapse multi-beam bathymetric surveys will be conducted over two 

shipwreck sites to capture short-term change as an aid to producing time-stepped 

wreck site formation models. 

[3] Sediment samples will be acquired at high-potential sites with a view towards 

particle size analyses and dating of any organic material sampled from the grabs 

and cores. These data have the potential to constrain sediment mobility models 

and provide sea-level index points. Additionally, optical data will be acquired at 

each sample station to reinforce geological and archaeological interpretations. 

2010 initiatives 

In addition to the 2009 cruise, the investigators have submitted a 2010 proposal 

to the Marine Institute ship-time programme, with a view to returning to the 

study area next year to collect more acoustic and geological samples to further 

constrain interpretations. 

Furthermore, we are currently completing a NERC proposal for submission to the 

National Facility for Scientific Diving (NFSD) for support to ground-truth 

archaeological interpretations of the JIBS data. This proposal follows the remit of 

the NERC Science Based Archaeology panel, namely to: 

[1] Assess marine archaeological prospecting techniques with a view to 

identifying and investigating submerged landscapes; and 

[2] To investigate processes affecting the underwater archaeological record, 

namely how contemporary and past processes impact the geological signatures of 

inundated archaeological landscapes and how contemporary physical processes 

act to impart primary control on shipwreck site formation. 

Ground-truthing will take the form of visual inspection and sampling on targeted 

sites, assessing the range of materials exposed on the seafloor (lithics etc.) and 

sampling dateable material (peat etc.). 

6.3 Refining shipwreck site formation models and site IDs 

6.3.1 Results from automated segmentation of backscatter 

The results of each classification are presented independently before being 

integrated as a composite classified dataset. The coverage of each respective 

system (EM3002D and EM710 [modes 01; 02 and 03]) is demonstrated in Figure 

6.6. As the binary mask polygons were created from the bounds of each 

individual classification, there is some overlap at the edges of each area. This 

extended area is due to the interpolation process, based on the classified vectors. 

The level of transparency has been set to 50% for each layer, and the order of 

display has been set to show the most resolute data first in order to enhance 

visualization of the data. 

[ 61 ]


Figure 6.6 Study site for backscatter segmentation. The different coloured 

polygons show changes in acquisition system and within system variations in 

operational frequency related to pulse length. The position in the key is indicative 

of the drawing order in the map. 

EM3002D (Jetstream) 

A selection of results from the EM3002 (293-307 kHz) classification are presented 

in Figure 6.7. Figure 6.7a shows the overall coverage of the EM3002D data in the 

context of mainland North Antrim and Rathlin Island. This classification, based on 

sampling at a scale of 257x17 pixels, resulted in 679,826 vectors being submitted 

to the classification process. Of these, a random 40,000 records were subjected 

to the clustering algorithm over 5 iterations, from which QTC-Multiview defined 

15 as the optimum number of classes. The results of the second method of 

unsupervised classification (Erdas Imagine v. 9.2: Isodata) are presented in 

Figure 6.7a*. This shows a very coherent linear identification of many of the 

principal features evident in the backscatter mosaics, and is based on the 1m 

gridded backscatter imagery. 

The classification process has identified 3 principal classes, and several 

gradational transitions between them. The colour of the classes is indicative of 

their acoustic similarity (for example; classes 8, 10, 13 and 14 [pink] are 

identified by the software as being acoustically similar, as are classes 2, 9 and 12 

[teal]; and 3, 4, 5 and 15 [blue]). The geographic distributions of the classes are 

presented pre- and post-interpolation in Figure 6.7 (a, a* and e). From this, it is 

evident that the main occurrence of classes 8, 10, 13 and 14 (pink) is restricted 

to Church Bay (Figure 6.7a inset detail) and offshore from Ballycastle. There are 

[ 62 ]


other significant instances of these classes, but they do not form large 

aggregations. Many of these are lost after interpolation (Figure 6.7e). The 

detailed area in Figure 6.7a shows the classified vector data in Church Bay. Note 

the presence of strong linear artefacts aligned northwest-southeast. These are 

related to poor range compensation and follow the orientation of the vessel 

heading during acquisition. 

The teal classes (2, 9, 12) cover the majority of the central portion of Rathlin 

Sound, and show a marked gradation approaching the shore of both Rathlin and 

the mainland. The blue classes (3, 4, 5 and 11) are concentrated around the 

north and southeast sides of the Island at Kinramer, and around Fair Head on the 

mainland. Figures 6.7 (b, c and d) show the inverse distance weighted 

interpolations of the principal components values Q1, Q2 and Q3 from the 

software QTC-Multiview. These are the values most responsible for the variation 

in the acoustic data as described by the software; as such they form the basis for 

the classifications. Several of the classes (6 and 7) did not survive the 

interpolation process (Figure 6.7e), although these are present in the classified 

vector datasets (Figure 6.7a). Figures 6.7 (f and g) show the interpolated 

confidence and probability values associated with the classification procedure. 

Note the geographic variation in these values and the corresponding classes with 

which they occur (Figure 6.7a and e). 

A comparison of the different classification approaches to a known wreck site is 

summarised in Figure 6.8. Figure 6.8a shows the location of the remains of HMS 

Drake in Church Bay, Rathlin Island. The highlighted red box is a 0.8km 2 area 

centred on the wreck, and is the focal area for the remainder of the figures (6.8be). 

Figure 6.8b shows the wreck site and context as evidenced by the backscatter 

mosaic. The classified vector data set has identified something different from the 

surrounding area, but has classified this as similar to the teal classes (2, 9, 12) 

closer to the eastern edge of the image. Notice that there are also data gaps 

around the wreck site. This may be due to the application of a threshold to the 

data, whereby data anomalous to their surrounding neighbours may have been 

omitted from the classification process. The results of the unsupervised 

classification process conducted in Erdas Imagine have identified the wreck site 

as being distinct from both the surrounding areas, and the teal classes in the east 

of the image. 

[ 63 ]


Figure 6.7 a) QTC-Multiview classification results for Jetstream EM3002D 

[Classified vector dataset]. Inset detail of Church Bay, Rathlin Island (position 

indicated by red box). a*) Results of manual segmentation of the Church Bay 

area. b) Inverse distance weighted interpolation of Q1 values at 20m cell size. c) 

Inverse distance weighted interpolation of Q2 values at 20m cell size. d) Inverse 

distance weighted interpolation of Q3 values at 20m cell size. e) QTC-Clams 

categorical interpolation of classified vector dataset at 20m cell size. f) Inverse 

distance weighted interpolation of confidence at 20m cell size. g) Inverse distance 

weighted interpolation of probability at 20m cell size. 

[ 64 ]


Figure 6.8 a) Location map of the HMS Drake, Church Bay, Rathlin Island. b) 

Backscatter response in the vicinity of the HMS Drake compensated using FM 

Geocoder (area displayed 0.8 km 2 ). c) QTC-Multiview classified vector dataset for 

the 0.8 km2 study area. d) QTC-Clams categorically interpolated raster dataset of 

the same area. e) Erdas Imagine v.9.2 unsupervised classification of the 

backscatter mosaic in same area. Colour map is approximate to the QTC scale to 

facilitate comparison. 

EM710 (Victor Hensen) 

The data from the EM710 were classified in three independent stages owing to 

the aforementioned pulse length variations. Within the total study area for the 

segmentation (sub-areas A1_5; A1_6; A2_5; A2_6 and A3_6), there were a total 

of 268 lines of EM710 data, of which all were successfully incorporated into the 

classification process. The results are presented below. 

[ 65 ]


Mode 01 

The results of the classification of the EM710 mode 01 data are presented in 

Figure 6.9 (a-g). The data coverage using this system was spatially 

heterogeneous, and produced varied results. The relative coverage is indicated on 

Figure 6.6 (blue transparency). The results were produced using the same 

settings as the EM3002 to enhance continuity between classifications. In this 

instance, QTC-Multiview identified 14 classes based on a 40,000 record subset of 

289, 459 vectors over 5 iterations. The geographic distribution of the 14 classes 

is presented in Figure 6.9a (pre-interpolation), along with the principal 

components responsible for the classification (Figure 6.9 b, c, and d showing 

interpolated plots of Q1, Q2 and Q3 respectively). The interpolation process 

(Figure 6.9e) has been successful in suppressing the range dependent artefacts 

from classes 4 and 10 in the south-eastern portion of the classified vector dataset 

(Figure 6.9a). It is also clear that the detailed area highlighted in the red box is 

relatively complex, containing most of the variation in the remainder of the 

classification summarised in the same small area. This pattern is repeated in the 

extreme north west of the site which has another instance of mode 01 data. 

Figures 6.9 (f and g) show the interpolated confidence and probability values 

associated with the classification procedure. 

Mode 02 

The results of the mode 02 classification are presented in Figure 6.10 (a-g). This 

classification represented the most spatially comprehensive and homogeneous of 

the EM710 results. The coverage is indicated on Figure 6.6 (pink transparency). 

The results were also processed using the same rectangular patch dimensions as 

the two previous classifications (257x17 pixels). On this occasion, QTC-Multiview 

identified 14 as the optimum number of classes based on a 40,000 subset of 

439,842 vectors over 5 iterations. The geographic distribution of these classes 

and their constituent ‘Q’ values are presented in the following Figures (6.10a-g). 

Figure 6.10a shows the distribution of the classified vector dataset, preinterpolation. 

Class 4 (mauve) is the dominant class in this classification, covering 

a large geographic area. This is intersected in the area to the west of Rathlin 

Island and close to the East lighthouse by classes 3 and 6 (pink), and in the 

south east and areas of Rathlin Sound by class 1 (green). Classes 7, 8, 12 and 14 

are also introduced in the area to the north of the Island, and parallel to the 

entrance to the North Channel. The presence of all these classes are significantly 

reduced by the process of interpolation as class 4 is prevalent throughout all of 

these areas. The most obvious result of the interpolation process is the reduction 

in extent of the areas of classes 2, 5, 11, and 13 (blue). Closer inspection of the 

vector dataset shows that many of these classes adhere to particular points in the 

range of the sonar swath, and while they may be identifying coherent changes in 

the along track domain, there is undoubtedly a more complex situation than is 

being evidenced by the classification process. Interestingly, the same detail area 

as presented in the mode 01 classification (red box) has sufficient coverage post 

interpolation to facilitate some comparison between the classifications of the two 

modes, although this has not been possible at this stage of the analysis. 

Figures 6.10 (f and g) show the interpolated confidence and probability values 

associated with the classification procedure. There is markedly less variation in 

the mode 02 confidence and probability values than the previous classifications 

(Figures 6.7 and 6.9). 

[ 66 ]


Figure 6.9 a) QTC-Multiview classification results for Victor Hensen mode 01 - 

206 s [Classified vector dataset]. Inset detail of North Shore, Rathlin Island 

(position indicated by red box). b) Inverse distance weighted interpolation of Q1 

values at 20m cell size. c) Inverse distance weighted interpolation of Q2 values at 

20m cell size. d) Inverse distance weighted interpolation of Q3 values at 20m cell 

size. e) QTC-Clams categorical interpolation of classified vector dataset at 20m 

cell size. f) Inverse distance weighted interpolation of confidence at 20m cell size. 

g) Inverse distance weighted interpolation of probability at 20m cell size. 

[ 67 ]



500 s [Classified vector dataset]. Inset detail of North Shore, Rathlin Island 

(position indicated by red box). b) Inverse distance weighted interpolation of Q1 

values at 20m cell size. c) Inverse distance weighted interpolation of Q2 values at 

20m cell size. d) Inverse distance weighted interpolation of Q3 values at 20m cell 

size. e) QTC-Clams categorical interpolation of classified vector dataset at 20m 

cell size. f) Inverse distance weighted interpolation of confidence at 20m cell size. 

g) Inverse distance weighted interpolation of probability at 20m cell size. 

[ 68 ]


Mode 03 

The results of the mode 03 classification are presented in Figure 6.11 (a-g). The 

classification represents the smallest physical area covered by the EM 710, off the 

north shore of Rathlin Island. The coverage is indicated on Figure 6.6 by the 

yellow transparency. These results were obtained by processing in QTC-Multiview 

at the same parameters as observed in the previous classifications (257x17). For 

this classification, the sampling of the image-based data yielded 14,612 vectors 

which were subjected to the clustering process. On this basis, the software 

defined 7 as being the optimum number of classes in the mode 03 dataset. The 

geographic distribution of the classified vector dataset, pre-interpolation is 

presented in Figure 6.11a. Similarly to the mode 02 classification results (Figure 

6.10), this classification is dominated by a single class (2) There are more subtle 

differences evident in the interpolated Q values (Figures 6.11 b,c and d), most 

particularly Figure 6.11b, although this variation has not come through in the 

final classification. The blue classes (1 and 3) exhibit a strong range dependence 

in both the pre- and post-interpolated phases (Figure 6.11a and e). Figures 6.11 

(f and g) show the interpolated confidence and probability values associated with 

the classification procedure, both of which show a similar degree of uniformity 

across the mode 03 area. 

[ 69 ]



2000 s [Classified vector dataset]. b) Inverse distance weighted interpolation of 

Q1 values at 20m cell size. c) Inverse distance weighted interpolation of Q2 

values at 20m cell size. d) Inverse distance weighted interpolation of Q3 values at 

20m cell size. e) QTC-Clams categorical interpolation of classified vector dataset 

at 20m cell size. f) Inverse distance weighted interpolation of confidence at 20m 

cell size. g) Inverse distance weighted interpolation of probability at 20m cell 

size. 

[ 70 ]


Composite Classification 

The results of the categorically interpolated classifications for the EM 3002D and 

EM 710 data are presented summarily in Figure 6.12. The most spatially resolute 

data has kept the highest drawing order in the figure presented, as in Figure 6.6. 

The differences in colour are indicative of similarity only within the individual 

classification, not between classifications. It would be necessary to train the 

classifications around the boundaries in order to get independent classifications to 

adhere to a common colour scheme. 

Figure 6.12 Results of the QTC-Multiview classification in JIBS areas A1_5; 

A1_6; A2_5; A2_6 and A3_6. The boundary between adjacent systems is 

demarcated by solid black line. Drawing order is the same as Figure 6.6 and the 

figure key. 


The results from the automated segmentation of the backscatter data form 

important inputs into models of shipwreck site formation as they allow for 

derivation of robust substrate maps and quantification of sediment budgets and 

contemporary sedimentary processes. In addition, for the fist time, this treatment 

of the backscatter data indicates that automated and semi-automated 

segmentation of backscatter data can identify shipwreck sites. This result alone is 

highly significant and will result in the compilation of a paper for submission to 

the Journal of Archaeological Science in 2010. 

6.3.2 Re-assessment of anomalies identified in Phase 1 

The higher resolution DEM (1m versus 4m and 6m used during Phase 1) and 

availability of backscatter data have allowed a re-assessment of the anomalies 

[ 71 ]


detected during Phase 1 of the project. Figure 6.13 shows the significant 

improvement in the amount of detail and clarity of the bathymetric image in 

comparison to the data used during Phase 1 of this project (6m and 4m grid). An 

example of the added value that backscatter can deliver for the detection of 

shipwrecks is shown in Figure 6.14. 

Figure 6.13 Comparison between bathymetric data acquired over the S.S. 

Lochgarry site, gridded to 6m, 4m and 1m. 

[ 72 ]


Figure 6.14 JIBS (A) high resolution (1m) multibeam and (B) backscatter data 

acquired over the dispersed remains of the armoured cruiser HMS Drake 

(torpedoed in 1917) and steam trawler Ella Hewett (collided with remains of HMS 

Drake in 1962). Note the importance of the backscatter data in providing a 

clearer image of the wreck remains and the bedforms (eg. to the SW) in addition 

to giving information on sediment types (dark areas = gravel; light areas = sand) 

- such data are important for the study of site formation processes. 

Case study: Rathlin Island 

Due to the revised and shortened timescale of this project and the time taken to 

process the vast quantity of data, a complete reassessment of all shipwreck 

anomalies was not feasible. It was therefore decided, in the first instance, to 

focus on a single case study. The study area chosen was the Rathlin Island 

Special Area of Conservation (SAC). This SAC, designated on the basis of rare 

marine habitats, entirely encompasses Rathlin Island and its adjacent waters. 

From an archaeological perspective, Rathlin Island has been a shipping hazard for 

centuries, with 81 documented wrecking incidences falling within the SAC 

boundaries (Figure 6.15). 

[ 73 ]


Figure 6.15 Overview map showing Rathlin Island, the SAC boundaries (black 

box) and all documented shipwrecks within the SAC from the Northern Ireland 

MSMR (red circles). 

In Phase 1 of the shipwreck analysis, 44 anomalies were detected that fell within 

the SAC boundaries (Figure 6.16 and Table 6.2). Of these, 3 were designated as 

class 1 (most probably a shipwreck), 16 were designated as class 2 (most 

probably a geological feature or shipwreck) and 25 were designated as class 3 

(most probably a processing error or a shipwreck). 

[ 74 ]


Figure 6.16 Distribution and classification of Phase 1 anomalies falling within the 

Rathlin Island SAC. 

As part of the Phase 2 analysis, each of these anomalies was visually reassessed 

using the JIBS 1m DEM and the associated backscatter data. If necessary, the 

anomalies were reclassified. Further, new anomalies not visible on the Phase 1 

datasets were also added to the database. In addition to the classes described 

above, the following new classes were used: 

• Class 4 – anomaly detected in Phase 1. No longer visible on 1m 

data, definite processing error. 

• Class 5 - anomaly detected in Phase 1. Visible on 1m data, but now 

confirmed geological feature. 

The results of the reclassification are summarized in Table 6.2 and Figure 6.17. 

The most noticeable outcome of the re-analysis was the reclassification of all but 

three Class 3 anomalies into Class 4. Effectively, these were possible processing 

errors/wrecks that were confirmed to be processing errors. The three exceptions 

were found to be geological features, with two confirmed (i.e. Class 5) and one 

possible geological/wreck (i.e. Class 2). The three most probable wreck anomalies 

were confirmed as wrecks, hence the lack of change in class 1. Of the Class 2 

anomalies, 12 were reinterpreted as confirmed geological features. This was 

particularly clear for the north side of Rathlin Island where a number of anomalies 

were confirmed to be slump deposits. Finally, 13 new anomalies were detected. 

These were all small (


Table 6.2Anomaly classes within the Rathlin SAC as identified during Phase 1 

and Phase 2 of the project 

Phase 1 analysis (n) 

Phase 2 re-analysis 

Class 1 3 3 

Class 2 16 17 

Class 3 25 0 

Class 4 - 24 

Class 5 - 14 

Total 44 58 

Figure 6.17 Distribution and classification of anomalies falling within the Rathlin 

Island SAC following the Phase 2 reclassification 

Revised contact sheets 

In addition to re-assessing the anomalies within the Rathlin Island SAC 

boundaries, a revised contact sheet for each known and definite shipwreck was 

created. This led to the removal of four such sheets created during Phase 1, 

which had initially been interpreted as shipwreck but were now found to be 

processing errors. Three new wrecks were added to the contact sheets; two 

(Towey and Templemore) represented by very low (


Vessel Name 

Geophysical data 

Position 

Geographical location 

Water Depth (m) 

Andania? 

JIBS multibeam 1m resolution 

Latitude-Longitude 1 6° 12’ 47’’W 55° 19’ 33’’N 

UTM (zone 29) 

676817E 6134602N 

2km north of Rathlin Island 

195m 

Anomaly description 

Seabed description 

Inferred Energy 

Conditions 

Interpretation 

Notes 

Appearance 

Dimensions 

Max. height above 

seabed 

3m high sand waves 

E-W linear mound 

123x21m 

14m 

Moderate to high energy, formation of bedforms 

possible despite great depth 

Largely intact shipwreck 

The anomaly is listed in the UKHO database as the Andania, an armed liner 

torpedoed in 1918. The MSMR record for the Andania is 16km E of this anomaly 

while a dive record is 12.5km to the NW and lists her as lying in 123m water. 

Anomaly dimensions do fit historical dimensions (158x19m). Wreck is upright and 

lies in a field of bedforms (sandwaves?). Bathymetry and backscatter suggest the 

bedforms are lapping onto the wreck and there may be scour on its southerm side. 

It is worth noting that another vessel of similar size, the Calgarian, is also listed as 

being torpedoed 2miles N of Rathlin in 1918. Given the discrepancies with the dive 

record, it is possible that this wreck could be the anomaly or the dive record is 

actually that of the Calgarian. 

Associated Imagery 

Top: JIBS 1m multibeam bathymetry showing the Andania. View from directly 

overhead 

Middle: JIBS backscatter image of the Andania and surrounding seabed. View from 

directly overhead 

Bottom: JIBS backscatter draped over 1m multibeam bathymetry. Perspective (3D) 

view looking northwest 

1 

Latitude Longitude in WGS-84 

[ 77 ]


[ 78 ]


Vessel Name 


S.S. Castle Eden 


Position 


UTM (zone 29) 

623429E 6132276N 

Geographical location 5.2km NNW of Ballymagaraghy Point, Co. Donegal 

Water Depth (m) 30-31m 


Appearance 

E-W collection of upstanding 

anomalies 

Dimensions 

85x16m 

Max. height above 2.5m 

seabed 

Seabed description Hummocky terrain: look like glacial features but could 

be sand/gravel waves 


? medium energy ? 

Conditions 


Partially buried (?) remains of a shipwreck 

Notes 

MSMR 420m to NW of anomaly: S.S. Castle Eden (square). Dive record 470m to NW 

of anomaly: S.S. Castle Eden (diamond). Dimensions similar to historical information 

(86.25x12.21x5.61m). Steel steam collier, torpedoed in 1918. Dive record notes 

seabed as coarse gravel. Backscatter shows relatively little difference between the 

wreck and the surrounding seabed, possibly indicating a hard substrate. 


Top: JIBS 1m multibeam bathymetry showing the Castle Eden. View from directly 

overhead 

Middle: JIBS backscatter image of the Castle Eden and surrounding seabed. View 

from directly overhead 


view looking NE. Image has been 2 times vertically exaggerated 

2 


[ 79 ]


[ 80 ]


Vessel Name 


H.M.T. Corienties 


Position 


UTM (zone 29) 

615968E 6137154N 

Geographical location 3.3km N of Glengad Head, Co. Donegal 



Appearance 

Dimensions 

2 sections (one highly 

protruding, one elongate) 

with developed scour pit 

High anomaly: 15x6m; 

elongated anomaly: 20x6m 


seabed 

Seabed description Sand/gravel waves or glacial features 


? medium energy? 

Conditions 


Remains of a shipwreck with scour, broken up in at 

least 2 sections 

Notes 

Dive record 800m to NW of anomaly: H.M.T Corienties. MSMR 1900m to ENE of 

anomaly: H.M.T Corienties. Dimensions close to historical information (40.1x6.9m). 

Steel admiralty trawler, mined in 1917. Dive record notes seabed as coarse shell 

gravel, backscatter suggests hard undifferentiated substrate with scour pits 

surrounding the wreck. 


Top: JIBS 1m multibeam bathymetry showing the Corientes. View from directly 

overhead 

Middle: JIBS backscatter image of the Corientes and surrounding seabed. View from 



view looking SE. 

3 


[ 81 ]


[ 82 ]


Vessel Name 


Position 



H.M.S. Drake 



UTM (zone 29) 

677272E 6130079N 

West off Rathlin; Church Bay 

15m 


Appearance 

NW-SW elongated anomaly 

with scour pit 

Dimensions 

150x27m 

Max. height above 4m 

seabed 

Seabed description Fairly featureless 


No clear bedforms around wreck, small ripples to 

Conditions 

southwest. Low energy? 


Remains of a shipwreck, not intact 

Notes 

There is a MSMR record 520m to N of anomaly: H.M.S. Drake (square). There is a 

dive record 210m to NW of anomaly: H.M.S. Drake (diamond). Position of H.M.S. 

Drake noted on Admiralty Chart and UKHO database in location of anomaly. 

Dimension fit historical information (152x22x8m). Armour plating. Torpedoed in 

1917. Dive record noted seabed to be sand over clay. Backscatter data indicates 

sand around wreck is featureless. Wreck is very low-lying and broken up, result of 

demolition work following sinking. Protrusion from NW section of anomaly is the 

wreck of Ella Hewitt, a trawler which collided with the wreck and sank in 1962. 


Top: JIBS 1m multibeam bathymetry showing the Drake. View from directly 

overhead 

Middle: JIBS backscatter image of the Drake and surrounding seabed. View from 



view looking SSE. Image has been 2 times vertically exaggerated 

4 


[ 83 ]


[ 84 ]


Vessel Name 


H.M.S. Laurentic 


Position 


UTM (zone 29) 

589379E 6129571N 

Geographical location Mouth Lough Swilly, 3.7km NE of Fanad Head, Co. 

Donegal 


Anomaly description Appearance 

Similar to bedrock outcrop 

but more distinct upstanding 

features 

Dimensions 

125x36m 


seabed 

Seabed description Some bedrock outcrop to W; sand/gravel waves or 

glacial features 


? medium energy ? 

Conditions 


Partially broken remains of a shipwreck 

Notes 

MSMR close to anomaly: S.S. Laurentic (square). Wreck indicated on Admiralty 

Chart and dive records in position of the anomaly. Dimensions similar to historical 

information (167.6x20.4x12.5m). Part of vessel appears to be broken off. Steel liner, 

mined in 1917. Dive record notes seabed as stones and shale. Backscatter suggest 

chaotic bedforms at the southern end of the anomaly and a scour pit running along 

the eastern side of the wreck. 


Top: JIBS 1m multibeam bathymetry showing the Laurentic. View from directly 

overhead 

Middle: JIBS backscatter image of the Laurentic and surrounding seabed. View from 



view looking SW 

5 


[ 85 ]


[ 86 ]


Vessel Name 


S.S. Lochgarry 


Position 


UTM (zone 29) 

679549E 6128026N 

Geographical location East off Rathlin; 1.4km NE of Rue Point 



NW-SE elongated mound 

Dimensions 

84x14m 


seabed 

Seabed description Bedrock coastline 


Conditions 

No clear bedforms, prob. very strong currents resulting 

in high erosion 


Largely intact shipwreck 

Notes 

Dive record 50m to the NE (diamond): S.S. Lochgarry, steel, struck rocks, lost in 

1942. MSMR record for Lochgarry is located 1km to the SE of the anomaly. Location 

of wreck indicated on the Admiralty Chart and UKHO database. Dimensions agree 

with historical information (80.8x10.2x4.7m). Seabed noted as stones and shale in 

dive record, based on current strength, this is probably a thin lag over bedrock. 


Top: JIBS 1m multibeam bathymetry showing the Lochgarry. View from directly 

overhead 

Middle: JIBS backscatter image of the Lochgarry and surrounding seabed. View from 



view looking north 

6 


[ 87 ]


[ 88 ]


Vessel Name 


S.S. Lugano 


Position 


UTM (zone 29) 

671912E 6129316N 

Geographical location 1.5km SSW off Bull Point, Rathlin Island 



Appearance 

E-W mound, collection of 

irregular anomalies 

Dimensions 

107x18m 


seabed 

Seabed description Fairly featureless 


No clear bedforms: very low (bedforms do not form) or 

Conditions 

very high currents (everything eroded) ? 


Intact shipwreck 

Notes 

There is a dive record 140 to ESE of anomaly: S.S. Lugano (diamond).There is an 

MSMR record 1600m to SW of anomaly: S.S. Lugano (square). Wreck indicated on 

Admiralty Chart and UKHO database. Dimensions close to historical information 

(107x16x7m). S.S. Lugano was steel cargo steamer, torpedoed in 1917. Dive record 

describes seabed as rocky, backscatter indicates uniform seabed substrate and 

possible associated wreckage located c. 25m off the south side of the wreck 


Top: JIBS 1m multibeam bathymetry showing the Lugano. View from directly 

overhead 

Middle: JIBS backscatter image of the Lugano and surrounding seabed. View from 



view looking SE. 

7 


[ 89 ]


[ 90 ]


Vessel Name 


S.S. Santa Maria 


Position 

Latitude-Longitude 8 

UTM (zone 29) 

Geographical location 1.4km NW off Fair Head 


6° 9’ 47’’W 55° 14’ 11’’N 

680389E 6124766N 


High E-W mound 

Dimensions 

90x14m 


seabed 

Seabed description Hummocky terrain: look like glacial features but could 

be sand/gravel waves, some sand ripples 


Medium energy (eddies noted on Admiralty Chart) 

Conditions 


Largely intact shipwreck, possibly some broken 

fragments to the E 

Notes 

MSMR 1040m to E (square): S.S. Santa Maria, steel, lost in 1918. Dive record 530m 

to the NNW (diamond): S.S. Santa Maria, steel, lost in 1918. A wreck is noted on the 

Admiralty Chart and UKHO database in this location. Dive records indicate that S.S. 

Santa Maria is broken into two sections, with one section on a ledge at 30m water 

depth and another at the bottom of a cliff at 60m water depth. The multibeam does 

not show such a steep drop-off. However, the dimensions of the anomaly fit those of 

the dive record. Therefore, this wreck has been interpreted as S.S. Santa Maria. 

Steel steam tanker, torpedoed in 1918. Dive record notes seabed as rocky, 

backscatter suggests hard substrate, with possible scour or additional wreckage at 

the eastern end of the anomaly. 


Top: JIBS 1m multibeam bathymetry showing the Santa Maria. View from directly 

overhead 

Middle: JIBS backscatter image of the Santa Maria and surrounding seabed. View 



view looking NE. Note that lighter areas in this image show where there is no 

backscatter coverage 

8 


[ 91 ]


[ 92 ]


Vessel Name 


Templemore 


Position 


UTM (zone 29) 

676120E 6121798N 

Geographical location 500m NW from Ballycastle 



Appearance 

Irregular linear anomaly, 

with peaks at its northern 

and southern end 

Dimensions 

40x11m 


seabed 

Seabed description Fairly featureless seabed 


Probable high energy 

Conditions 


Shipwreck, broken up 

Notes 

MSMR 440m to the SW, diver record (diamond) is 50m to the NW. Adjacent MSMR 

(square) is from another vessel. Wreck is listed in the UKHO database as 

Templemore, an iron coaster lost in 1911. Anomaly dimensions are close to historical 

dimensions (48x8m). Wreck lies low to the seabed, and is heavily broken up and 

hence shows as a small anomaly on the backscatter. Diver records report seabed as 

rock and sand, backscatter indicates a featureless seabed. Both records combined 

indicate high energy conditions. 


Top: JIBS 1m multibeam bathymetry showing the Templemore. View from directly 

overhead 

Middle: JIBS backscatter image of the Templemore and surrounding seabed. View 



view looking SE. Image has been 2 times vertically exaggerated 

9 


[ 93 ]


[ 94 ]


Vessel Name 


Towey 


Position 


UTM (zone 29) 

647371E 6119638N 

Geographical location 1.7km ESE of Portrush Harbour 



Appearance 

Two isolated anomalies 

aligned W-E 

Dimensions 

45x11m 


seabed 

Seabed description Fairly featureless seabed, rocky outcrops to S 


Probable high energy 

Conditions 


Shipwreck, broken up 

Notes 

MSMR record is 460m East, diver record is 880m to the NW. Wreck is listed in the 

UKHO database as Towey, a steel coaster lost in 1930. Anomaly dimensions are 

close to historical dimensions (34x7m) and include scour feature surrounding the 

two parts of the wreck. Wreck lies low to the seabed, and is heavily broken up and 

hence shows as a small anomaly on the backscatter. Diver records report seabed as 

sand and shingle, backscatter indicates a featureless seabed (with exception of rocky 

outcrops to S). Both records combined indicate high energy conditions. 


Top: JIBS 1m multibeam bathymetry showing the Towey. View from directly 

overhead 

Middle: JIBS backscatter image of the Towey and surrounding seabed. View from 



view looking SE. Image has been 2 times vertically exaggerated 

10 


[ 95 ]


[ 96 ]


Vessel Name 


U-1003? 


Position 


UTM (zone 29) 

643328E 6130321N 

Geographical location 14km N of Portstewart strand 



Long E-W anomaly 

Dimensions 

65x9m 


seabed 

Seabed description Fairly featureless, some sand waves to the South 


Low – medium energy 

Conditions 


Shipwreck, largely intact 

Notes 

Dive record (diamond) almost exactly on location of the anomaly. Position of a wreck 

is indicated on the Admiralty chart to the NW of the anomaly. Wreck has been 

identified by divers as a steel U-Boat (U-1003), wrecked after collision in 1945. Dive 

record notes the seabed as sand and rocks. There may be some uncertainty with 

respect to which U-boat it is. The UKHO database lists U-1003 as being lost 16km 

NW. Backscatter shows a relatively featureless seabed, with a possible scour tail 

extending from the W end of the wreck. 


Top: JIBS 1m multibeam bathymetry showing the U-1003. View from directly 

overhead 

Middle: JIBS backscatter image of the U-1003 and surrounding seabed. View from 



view looking NE. Lighter north of the anomaly and black area show where there is no 

backscatter coverage 

11 


[ 97 ]


[ 98 ]


Vessel Name 


Position 



S.F.V. William Mannel 



UTM (zone 29) 

622165E 6130734N 

2.6km NNE of Rubonid Point, Co. Donegal 

26.5-27.5m 


Appearance 

Irregular long upstanding 

anomaly, NE-SW, large 

scour pit to SW, debris to NE 

Dimensions 

53x11m 


seabed 

Seabed description Hummocky terrain: sand waves or glacial features. Poss 

small sand waves/ripples around the wreck 


Medium energy 

Conditions 


Remains of a shipwreck with scour pits surrounding the 

wreck 

Notes 

Dive record 33m to SE of anomaly: S.F.V. William Mannel. MSMR and indication on 

Admiralty Chart 355m to NW of anomaly: S.F.V. William Mannel (square). Iron 

fishing trawler, sank in 1946 while under tow after she struck rocks near Glengad 

Head. Anomaly slightly bigger than historic dimensions (38.3x7.1x3.9m), but 

measured dimensions include scour features. 


Top: JIBS 1m multibeam bathymetry showing the William Mannel. View from directly 

overhead 

Middle: JIBS backscatter image of the William Mannel and surrounding seabed. View 



view looking WSW. Image has been two times vertically exaggerated 

12 


[ 99 ]


[ 100 ]



1) The fact that the data has gone through more rigorous cleaning means 

that all anomalies that were previously classified as ‘processing errors or 

possible wreck (class 3)’ have now been eliminated around Rathlin Island. 

This implies that the same will be true for the wider dataset, though this 

will have to be confirmed with future re-analysis. 

2) Owing to the higher resolution of the DEM, it is easier to confirm the 

nature of features on the seabed as being geological. It is anticipated that, 

when analyzing the full dataset, a large number of class 2 (possible 

geological feature or possible shipwreck) will be found to be geological. 

3) However, the higher resolution of the data also means that more smaller 

anomalies are now being detected. This will necessitate an examination of 

the full 1m data rather than simply re-assessing the previously picked 

anomalies. 

4) Addition of the backscatter has proven to be a great aid in determining the 

nature of the anomalies. Further studies on backscatter should 

demonstrate the possibility to use such data for the characterization of the 

archaeological material. 

5) Overall, we are confident that we should be able to reduce the amount of 

potential archaeological anomalies to a realistic number for 

groundtruthing. 

6.4 Dissemination of results 

6.4.1 Web-based dissemination 

Web-based dissemination occurs through the dedicated project web-site (Figure 

6.18): http://www.science.ulster.ac.uk/cma/instar/ 

6.4.2 Research output 

Two manuscripts resulting from Phase 1 and early Phase 2 of the project have been 

submitted for review to the International Journal of Nautical Archaeology .We will 

continue to publish the results from the project in international peer-reviewed 

journals and make these outputs available in PDF format on the CMA web-site. 

Additionally, a news article detailing preliminary results of the JIBS-INSTAR work 

was published in the Northern Ireland Environment Agency’s ‘Coast’ magazine: 

Plets R & Westley K. 2009. The seabed revealed. Northern Ireland Environment 

Agency Coast Magazine 5:10-13. [Appendix B] 

[ 101 ]


Figure 6.18 Front page of the dedicated project web-site, launched in August 

2008. 

6.4.3 Conferences and public seminars 

Results from Phase 1 and early 2 of the project have been presented at a series of 

national and international conferences and seminars: 

Forsythe, W. and Westley, K., 2009, Shipwrecks and Settlements on the North 

Coast. Invited Seminar as part of the Autumn Talks Series, Northern Ireland 

Environment Agency Coastal Zone centre, Portrush (12th November 2009). 

[Appendix B] 

Plets, R., Bell, T., Quinn, R., Westley, K., O’ Sullivan, A. and Renouf, M.A.P., 2009, 

A research strategy for mapping prehistoric submerged archaeological landscapes 

on the seabed off Ireland and Newfoundland. Poster presented at the European 

Association of Archaeologists Conference, Riva del Garda (15-20 th September 2009). 

[Appendix B] 

Quinn, R., Forsythe, W., Plets, R., Westley, K., Clements, A., Benetti, S., Bell, T., 

McGrath, F. and Robinson, R., 2009, Archaeological applications of the Joint Irish 

Bathymetric (JIBS) data. Poster presented at the Irish Shelf Petroleum Studies 

Group Atlantic Ireland 2009 Conference, IMI Conference Centre, Dublin (19-20 

October 2009) [Appendix B] 

Quinn, R., Plets, R., Westley, K., Clements, A., Forsythe, W., Benetti, S., Bell, T., 

McGrath, F. and Robinson, R., 2009, Methodologies to Assess Shipwreck Sites and 

Submerged Landscapes. Invited Seminar at Near Surface 2009 – the 15th European 

Meeting of Environmental and Engineering Geophysics of the Near Surface, 

[ 102 ]


Geoscience Division, European Association of Geophysicists and Engineers, Dublin 

Castle, Ireland (7-9 th September 2009). [Appendix B] 

Westley, K., Quinn, R., Forsythe, W., Plets, R. and Bell, T., 2009, Archaeological 

applications of the JIBS dataset: Submerged landscape mapping andr econstruction 

off the north coast of Ireland. Paper presented at the European Association of 

Archaeologists Conference, Riva del Garda (15-20 th September 2009). [Appendix B] 

[ 103 ]


7. IMPACT OF THE PROJECT 

Provision of the JIBS data is transforming geo-archaeological research and teaching 

at the University of Ulster. These data represent an unrivalled opportunity to 

contribute to the understanding of post-glacial sea-level change and colonization of 

Ireland and afford the opportunity to produce models for shipwreck site formation at 

a resolution previously unachievable. 

7.1 Teaching 

JIBS case studies are now being used in the delivery of undergraduate and 

postgraduate modules, providing the next generation of earth- and archaeologicalscientists 

graduating from the University of Ulster with data processing and 

interpretation skills to meet the demands of a range of employers within the marine 

sector in Ireland. 

For example, new lectures and practical classes involving JIBS data were introduced 

into Modules EGM310 (Introduction to Remote Sensing and GIS) and EGM502 

(Seafloor Mapping) in 2009 [Appendix C]. The undergraduate students response to 

these data has been resoundingly positive. 

7.2 Research 

In 2009, the investigators were successful in securing ship time on the state 

Research Vessel Celtic Explorer (Figure 7.1) to continue to acquire data and 

samples within the JIBS study area. This research cruise will serve the dual purpose 

of acquiring new data for research and also act as a student training cruise to train 

undergraduate students in the methods to acquire, process, integrate and interpret 

marine geospatial data. 

Figure 7.1 The Marine Institute’s RV Celtic Explorer 

7.3 Practise 

Two main groups will benefit from this project: 

[ 104 ]


7.3.1 Policy makers and heritage bodies 

Managing cultural heritage is particularly difficult when objects are found on, or 

buried within, the seabed. Heritage bodies are generally faced with two major 

issues: (i) how to build up the underwater archaeological record (i.e. how to find 

sites) and (ii) how to manage a site once it has been discovered. Results from this 

study will address both of these issues ensuring that the Northern Ireland 

Environment Agency (NIEA) and the Republic of Ireland's Department of 

Environment, Heritage and Local Government (DEHLG) will be able to supplement 

the known archaeological record, prioritize which sites need immediate attention 

and aid the development of longer term management plans through the adaptation 

of site formation models. 

Current policies and laws in place for shipwreck finds (e.g. Merchant Shipping Act 

1995 for Northern Ireland and the National Monuments Act 1994 for the Republic of 

Ireland), are often difficult to enforce. While any wreck over 100 years old is 

automatically protected in the Republic of Ireland, in Northern Ireland, there is only 

a single wreck designated under the Protection of Wrecks Act 1973 (Spanish 

Armada site of La Girona). This project will potentially highlight more candidates for 

designation and steer future heritage policy. 

Although the case studies in this project focus on the maritime archaeology of the 

northern Irish coast, there is wider global applicability as heritage agencies from 

different nations should benefit from a critique of the methodology followed by this 

project for the assessment of acoustic data for archaeological purposes. This will be 

written up and published in international peer-reviewed journals (e.g. International 

Journal of Nautical Archaeology or Journal of Archaeological Science). 

Furthermore, the study of offshore hydrodynamic processes as part of the site 

formation processes study provides invaluable information for coastal managers, 

particularly in the study of coastal vulnerability and the impact of rising sea-levels. 

7.3.2 The commercial/private sector 

Commercial seabed developers are increasingly required to ensure that they cause 

minimal disturbance to the underwater archaeological record. EU Council Directives 

require that environmental assessments are undertaken in the face of development 

applications, including an assessment of the archaeological impact. Licenses are 

required to dredge and dispose of material and other substances at sea, and also for 

work that involves building below the average spring high-water level. At present, in 

Northern Ireland, the CMA acts on behalf of the NIEA in assessing whether proposed 

work will affect the archaeological record, whether a licence should be granted or 

whether further mitigation measures are required. 

Information derived from the archaeological analysis of the JIBS and 

acoustic/geophysical data (e.g. accurate positions of submerged sites) will speed up 

the licensing and assessment process by ensuring a faster turnaround of the license 

application's approval or suggestions for amendments to the proposed 

development. Further, it will allow the CMA and DEHLG to advise developers during 

the survey planning stage by identifying the location of archaeologically sensitive 

areas or data gaps. 

[ 105 ]


8. SCHEDULE OF EXPENDITURE 

Below is the schedule of expenditure from 01.09.09 to 04.12.09. Please refer to the 

interim report for the schedule of expenditure for the earlier part of the project. 

Item Amount ( ) 

Salaries 19,188.40 

Travel 1,669.77 

Consumables and lab supplies 2,078.96 

TOTAL 22,937.13 

[ 106 ]


9. REFERENCES 

Bamford, D.B. and Woodman, P., 2004. Tool hoards and Neolithic use of the 

landscape in North-Eastern Ireland. Oxford Journal of Archaeology, 23(1): 21-44. 

Bell, T. et al., 2008. A research strategy for mapping prehistoric archaeological 

potential on the seabed off Newfoundland and Ireland, World Archaeological 

Congress 6, Dublin. 

Bell, T., O' Sullivan, A. and Quinn, R., 2006, Discovering ancient landscapes under 

the sea, Archaeology Ireland, 20 (2): 12-17. 

Breen, C., 1996. Maritime archaeology in Northern Ireland: An interim statement. 

International Journal of Nautical Archaeology, 25(1): 55-65. 

Breen, C. and Forsythe, W., 2001. Management and protection of the maritime 

cultural resource in Ireland. Coastal Management, 29(1): 41-51. 

Breen, C. and Forsythe, W., 2004. Boats and Shipwrecks of Ireland. Tempus 

Publishing Ltd., Stroud, Gloucestershire, 192 pp. 

Breen, C., Quinn, R. and Forsythe, W., 2007. A preliminary analysis of historic 

shipwrecks in northern Ireland. Historical Archaeology, 41(3): 4-8. 

Brooks, A. and Edwards, R., 2006. The Development of a Sea-Level Database for 

Ireland. Irish Journal of Earth Sciences, 24: 13-27. 

Brooks, A.J., 2007. Late Devensian and Holocene Relative Sea-Level Change 

around Ireland, Unpublished PhD Thesis. Dept. of Geography, Trinity College, 

Dublin. 

Brooks, A.J., Bradley, S.L., Edwards, R.J., Milne, G.A., Horton, B.P., Shennan, I. 

2008. Post-Glacial Relative Sea-Level Observations from Ireland and their Role in 

Glacial Rebound Modelling. Journal of Quaternary Science 23(2), 175-192. 

Bull, J.M., Quinn, R. and Dix, J.K., 1998, Reflection coefficient calculation from 

marine high-resolution seismic reflection (Chirp) data and application to an 

archaeological case study. Marine Geophysical Researches, 20: 1-11. 

Cooper, J. A. G., J. T. Kelley, D. F. Belknap, R. Quinn & J. McKenna, 2002. Inner 

shelf seismic stratigraphy off the north coast of Northern Ireland: new data on the 

depth of the Holocene lowstand. Marine Geology, 186, 369-87. 

Edwards, R., A. J. Brooks, I. Shennan, G. A. Milne & S. L. Bradley, 2008. Reply: 

Postglacial relative sea-level observations from Ireland and their role in glacial 

rebound modelling. A. J. Brooks, S. L. Bradley, R. J. Edwards, G. A. Milne, B. 

Horton and I. Shennan (2008). Journal of Quaternary Science 23: 175-192. 

Journal of Quaternary Science, 23(8), 821-5. 

Evans, 2005. Computer software: grid convert version 1.0. URL Available from: 

http://www.geospatialdesigns.com/gridconvert.htm (2005) (accessed 13.10.09) 

Flemming, N.C., 1998. Archaeological evidence for vertical movement on the 

continental shelf during the Palaeolithic, Neolithic and Bronze Age periods. In: I. 

[ 107 ]


Stewart and C. Vita-Finzi (Editors), Coastal Tectonics. Geological Society Special 

Publications. Geological Society, London pp. 129-146. 

Forsythe, W., Breen, C., Callaghan, C. & McConkey, R., 2000, Historic storms and 

shipwrecks in Ireland - a preliminary survey of severe synoptic conditions as a 

causal factor in underwater archaeology. International Journal of Nautical 

Archaeology 29, 2, 247-259. 

International Hydrographic Bureau, 1998. IHO Standards for Hydrographic 

Surveys - Special Publication No. 44. 

Kelley, J.T., Cooper, J.A.G., Jackson, D.W.T., Belknap, D.F. and Quinn, R., 2006, 

Sea-level change and inner shelf stratigraphy off Northern Ireland. Marine Geology, 

232: 1-15. 

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: 437-448. 

Lambeck, K. and Chappell, J., 2001. Sea level change through the Last Glacial 

cycle. Science, 292: 679-686. 

Lambeck, K. and Purcell, T., 2001. Sea-level change in the Irish Sea since the 

Last Glacial Maximum: constraints from isostatic modelling. Journal of Quaternary 

Science, 16(5): 497-506. 

McCabe, A.M., Carter, R.W.G. and Haynes, J.R., 1994. A shallow marine 

emergent sequence from the northwestern sector of the last British ice sheet, 

Portballintrae, Northern Ireland. Marine Geology, 117: 19-34. 

McCabe, A.M., Cooper, J.A.G. and Kelley, J.T., 2007. Relative sea-level changes 

from NE Ireland during the last glacial termination. Journal of the Geological 

Society, London,, 164: 1059-1063. 

McDowell, J.L., Knight, J. and Quinn, R., 2005. High-resolution geophysical 

investigations seaward of the Bann Estuary, Northern Ireland coast. In: D.M. 

FitzGerald and J. Knight (Editors), High Resolution Morphodynamics and 

Sedimentary Evolution of Estuaries. Coastal Systems and Continental Margins. 

Springer, pp. 11-31. 

McGonigle, C., Brown, C., Quinn, R and Grabowski, J. 2009. Evaluation of imagebased 

multibeam sonar backscatter classification for benthic habitat discrimination 

and mapping at Stanton Banks, UK. Estuarine, Coastal and Shelf Science, 81: 423- 

437 

McGonigle, C., Brown, C. and Quinn, R. in press. Operational parameters, data 

density and benthic ecology: considerations for image-based classification of MBES 

backscatter, Marine Geodesy 

O' Sullivan, A., 2001, Foragers, farmers and fishers in a coastal landscape: an 

intertidal archaeological survey of the Shannon estuary (Discovery Programme 

Monograph 5), Royal Irish Academy, Dublin, Ireland. 

Peltier, W.R., 1994. Ice Age paleotopography. Science, 265(5169): 195-201. 

[ 108 ]


Peltier, W.R. and Fairbanks, R.G., 2006. Global glacial ice volume and Last Glacial 

Maximum duration from an extended Barbados sea level record. Quaternary 

Science Reviews, 25: 3322-3337. 

Preston, J. 2009. Automated acoustic seabed classification of multibeam images of 

Stanton Banks. Applied Acoustics. 70, 10: 1277-1287 

QTC, 2005 Quester Tangent Corporation, 2005. QTC MULTIVIEW Acoustic Seabed 

Classification for Multibeam Sonar, User Manual and Reference. Version 3. Quester 

Tangent, 201–9865 West Saanich Road, Sidney, B.C., Canada V8L 5Y8 

Quinn, R., 2006, The role of scour in shipwreck site formation processes and the 

preservation of wreck-associated scour signatures in the sedimentary record. 

Journal of Archaeological Science, 33 (10): 1419-1432. 

Quinn, R., Adams, J.R., Dix, J.K. and Bull, J.M., 1998a, The Invincible (1758) site – 

an integrated geophysical assessment. International Journal of Nautical 

Archaeology, 27.3: 126-138. 

Quinn, R., Bull, J.M. and Dix, J.K., 1998b, Optimal processing of marine highresolution 

seismic reflection (Chirp) data. Marine Geophysical Researches, 20: 13- 

20. 

Quinn, R., Cooper, JAG and Williams, B., 2000, Marine geophysical investigation of 

the inshore coastal waters of Northern Ireland, International Journal of Nautical 

Archaeology, 29.2: 294-298. 

Quinn, R., Breen, C., Forsythe, W., Barton, K., Rooney, S. and O' Hara, D., 2002a, 

Integrated Geophysical Surveys of The French Frigate La Surveillante (1797), 

Bantry Bay, Co. Cork, Ireland, Journal of Archaeological Science, 29: 413-422. 

Quinn, R., Forsythe, W., Breen, C., Dean, M., Lawrence, M. and Liscoe, S., 2002b, 

Comparison of the Maritime Sites and Monuments Record with side-scan sonar and 

diver surveys: A case study from Rathlin Island, Ireland. Geoarchaeology, 17 (5): 

441-451. 

Quinn, R., Dean, M., Lawrence, M., Liscoe, S. and Boland, D., 2005, Backscatter 

responses and resolution considerations in archaeological side-scan sonar surveys: 

a control experiment, Journal of Archaeological Science, 32: 1252-1264. 

Quinn, R., Forsythe, W., Breen, C., Boland, D., Lane, P. and Lali Omar, A., 2007, 

Process-based models for port evolution and wreck site formation at Mombasa, 

Kenya. Journal of Archaeological Science, 34 (9): 1449-1460. 

Robidoux, L., Fonseca, L. and Wyatt, G. 2008. A Qualitative assessment of two 

multibeam echo sunder (MBES) backscatter analysis approach, Canadian 

Hydrographic Conference. Victoria, British Columbia, Canada, 5 - 8 May, pp. 1 - 1. 

Conference Proceeding 

Rochon, A., de Vernal, A., Sejrup, H.P. and Haflidason, H., 1998. Palynological 

evidence of climatic and oceanographic changes in the North Sea during the last 

deglaciation. Quaternary Research, 49: 197-207. 

Rowley-Conwy, C., 2004. How the West was lost. A reconsideration of agricultural 

origins in Britain, Ireland, and Southern Scandinavia. Current Anthropology, 45: 

S83-S113. 

[ 109 ]


Saville, A., Ballin, T.B. and Ward, T., 2007, Howburn, near Biggar, South 

Lanarkshire: Preliminary notice of a Scottish inland early Holocene lithic 

assemblage, Journal of the Lithic Studies Society, 28, 41-48 

Sejrup, H.P. et al., 2005. Pleistocene glacial history of the NW European 

continental margin. Marine and Petroleum Geology, 22: 1111-1129. 

Shennan, I. et al., 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): 585-599. 

Smith, C., 1992. Late Stone Age hunters of the British Isles. Routledge, London. 

Van Wijngaarden-Bakker, L.H., 1990. Faunal remains and the Irish Mesolithic. In: 

C. Bonsall (Editor), The Mesolithic in Europe. Papers presented at the Third 

International Symposium Edinburgh 1985. John Donald Publishers Ltd, 

Edinburgh, pp. 125-133. 

Wheeler, A.J., 2002. Environmental Controls on Shipwreck Preservation: The Irish 

Context. Journal of Archaeological Science, 29: 1149-1159. 

Woodman, P., 1990. The Mesolithic of Munster: a preliminary assessment. In: C. 

Bonsall (Editor), The Mesolithic in Europe. Papers presented at the Third 

International Symposium Edinburgh 1985. John Donald Publishers Ltd, 

Edinburgh, pp. 116-124. 

Woodman, P., 2000. Getting back to basics: transitions to farming in Ireland. In: 

T.D. Price (Editor), Europe's First Farmers. Cambridge University Press, 

Cambridge, pp. 219-259. 

Woodman, P., McCarthy, M. and Monaghan, N., 1997. The Irish Quaternary fauna 

project. Quaternary Science Reviews, 16: 129-159. 

[ 110 ]

Archaeological_Applications_of_JIBS_Data_Progress_Report_09.pdf

Create successful ePaper yourself

Delete template?

Save as template?