As digital recording technology advances, the quality and quantity of data that can be captured increases exponentially. For archaeological survey, this means masses of high-resolution 3D data showing the topography of the Earth’s surface, created by technologies such as Airborne Laser Scanning (also known as lidar).
Certain types of lumps and bumps shown in this topographic data can be evidence of archaeological remains. But accurately spotting and interpreting archaeology in this way is a time-consuming process and requires high levels of skill. To manually assess all of the currently available data would take decades, leaving the majority of the archaeology in the landscape unrecorded and unprotected.
For example, in Scotland (where this project is based), we know that focused field survey of specific areas increases the quantity of known archaeology by anything from 50%–200%. Considering this, an expedited analysis of the currently available ALS data alone would represent a potentially significant increase in knowledge about Scotland’s historic environment.
Recent research has shown promising initial results in using Machine Learning and Computer Vision approaches to do this expedited analysis, but there’s much work still to be done. This project aims to understand the impacts of these new methods on the complex practice of archaeological survey—which relies as much on subjective and intuitive routines of practice as it does on cutting-edge recording technologies.