| Summary: | This thesis finalises a 5+3 PhD project within the joint doctoral programme in Digital Heritage established in collaboration between History, Archaeology and Classical Studies, Graduate School, the Faculty of Arts, Aarhus University and the University of York. The thesis deals with the overarching theme of spatial data in archaeological excavation recording. Spatial data are at the core of all archaeological observations, and are expressed in numerous ways, ranging from traditional hand drawings to digital two- and three-dimensional representations in Geographic Information Systems and proprietary 3D software. Yet, despite technological advances, state-of-the art digital spatial data are almost equally detached from textual archaeological interpretation as they were using conventional tools decades ago. The thesis presents a study of how technological advances influence archaeological excavation traditions and methodologies. Special emphasis is directed at exploring how the increased use of image-based 3D documentation may contribute to increased quality of field recording and, in particular, what theoretical conceptualisations and technical developments are needed to harness its full potential. The thesis is composed of four articles, which constitute individual chapters (2-5). Each chapter covers a theme within the underlying topic of integrating spatial data in archaeology, supplemented by an introductory chapter (1), a synthesis (6) and a conclusion (7). The first article (chapter 2) provides an introduction to the overarching research questions and their methodological and historical background. It offers some rudimentary impressions of differing excavation and recording traditions in Britain and Denmark, to critically assess the use of GIS in archaeology and the negotiation between state-of-the-art technology and archaeological practice. The article discusses how the adaptation of GIS may have contributed significantly to the detrimental effect of creating stand-alone silos of spatial data that are rarely fully integrated with non-spatial, textual data, and has acted to stifle the development of digital standards of recording by perpetuating outmoded analogue recording conventions from a previous century. The chapter outlines the potential of born-digital 3D recording technologies such as Structure From Motion (SFM), GPS, and laser scanning in current practice, while advocating for a conceptualisation of new types of data and data representation in archaeological documentation. This, however, requires changes in archaeological methodologies and workflows and that we redefine more explicitly what we actually want to do with spatial data in archaeology. The second article (chapter 3) seeks to advance the conceptual framework of 3D models within archaeological excavation recording. 3D documentation advocates for a new workflow with a more three-dimensional reasoning, allowing for the utilisation of 3D as a tool for continuous progress planning and evaluation of an excavation and its results. Just like the general use of models to form hypotheses, it is possible to use 3D models as spatial hypotheses of an ongoing excavation. This allows us to visually realise or spatially conceptualise our hypotheses as a virtual reconstruction and to combine it with our observational data. The article presents first-hand experiences of working with 3D reconstruction and visualisations during the excavations at Viking Age site Jelling, and explores how the concept of authenticity may facilitate negotiations between visualising what we know, and what we think we know. The third article (chapter 4) further addresses the challenges inherent to the integration of 3D documentation: specifically its inability to convey archaeological interpretations. Image-based 3D modelling is generally considered a superior tool for generating geometrically accurate and photo-realistic recording of an excavation, but does not immediately encourage reflexive or interpretative practice. This is a direct consequence of the technical limitations of currently available tools, but also reflects an archaeological methodology and spatial conceptualisation based on two-dimensional abstractions. Using the example of the excavations at the Iron Age site Alken Enge, this article takes a more technical approach to exploring how new tools developed for segmenting field-recorded 3D geometry allow embedding archaeological interpretations directly in the 3D model, thereby augmenting its semantic value considerably. This is considered a precondition for the successful integration of 3D models as archaeological documentation. Furthermore, the article explores how web-based 3D platforms may facilitate collaborative exchange of 3D excavation content and how the integration of spatial and attribute data into one common event-based data model may be advantageous. The event-based approach is used for conceptualising how digital spatial data are created, derived and evolve throughout the documentation and post-excavation process. This effectively means building a conceptualisation of excavation recording procedures and seeing them through to the data model implementation itself. The fourth and last article (chapter 5) further explores the technologies outlined in chapters two and four. In particular, it focuses on evaluating analytical capabilities and alternative visualisation end-goals for 3D excavation recording. The chapter presents a simple case study, demonstrating the pipeline from excavating an archaeological feature, through image-based documentation and processing, to volumetric visual representation, while exploring the potential of machine learning to aid in feature recognition and classification. Chapter six acts as a synopsis, which provides added context to the results of the preceding chapters and furthermore discusses archaeological data models in general, conceptual reference models and, finally, presents the data model and implementation developed during the research project. The research introduces several novel approaches and technical developments aimed at aggregating the fragmented excavation data throughout the archaeological sector. This includes developing software for harvesting 2D GIS data from file storage at local archaeological institutions, functions for 3D semantic segmentation, automated processes for pattern recognition (SVM), machine learning and volumetric visualisation, and database mappings to web-services such as the MUD excavation database - all of which feed into the development of the Archaeo Framework. The online database \url{www.archaeo.dk} provides an implementation of the proposed data model for complex spatial field recorded data, and demonstrates the achieved data management capabilities, analytical queries, various spatial and visual representations and data interoperability functions. The Archaeo Framework acts as a data repository for excavation data, and provides long-awaited integration of spatial and textual data in Denmark. The benefits of spatial integration are clearly evident, notably having all information in one system, available online for research, dissemination and data re-use. For the first time, it is possible to perform large-scale validation of digital excavation plans against the written record, and perform complex spatial queries at a much deeper level than merely a site on a map. This research frames the basis for further developments of dynamic data management approaches to the integration of complex spatial data in field archaeology. The data model is expected to assist archaeologists in implementing better conceptualised excavation data models, and to facilitate a better understanding and use of 3D for archaeological documentation and analysis. Ultimately, the implementation provides access to the inaccessible dimensions of archaeological recording by joining hitherto isolated and fragmentary archaeological datasets - spatial and textual. Future areas of investigation should seek to advance this further in order to facilitate the persistence of complex spatial data as integrated components of archaeological data models.
|
|---|
