Tomorrow, Hèctor A. Orengo (ICREA Research Professor at ICAC), will be presenting in the workshop “Geomatics Methodologies in Archaeology and Cultural Heritage Research“, organised by the Universidad Internacional Menéndez Pelayo (UIMP), on the 14th and 15th September 2023.
Current advances in geospatial analysis for archaeology and Cultural Heritage
This presentation will discuss current development strategies, theoretical approaches, and practical workflows related to geospatial analysis in use or under development at the Landscape Archaeology Research Group (GIAP) of the Catalan Institute of Classical Archaeology (ICAC).
To do so a series of case studies will be employed. These will be structured following the sections exposed below:
1. Large scale modelling of movement potential in archaeology. Current route modelling approaches usually employ a single cost factor and are deterministic in their delineation of a single best route. In contrast with these, we are developing a new probabilistic method based on the use of: multi-factor modelling beyond slope, machine learning-based classification and regression to create new data types and to extrapolate costs, multi-temporal modelling, probabilistic approach using multiple origins/destinations and attractors, convolutional calculations to accommodate the combination of cost factors and to include different types of transport, and network science for the exploration of results.
2. Advanced remote sensing technologies for the study of past societies within their environment. This section will deal with novel approaches to remote sensing in archaeology. It will highlight the need to employ multi-temporal (including seasonal) and multi-source approaches facilitated through data fusion methods and the need to develop specific algorithms for the treatment and analysis of data adapted to specific case studies, which includes methods that are usually associated to 3D or GIS-based data.
3. New approaches to large-scale site detection. These include the integration of different data types (not just multi-temporal and multi-source data fusion) and the use (and combination) of diverse types of machine learning and deep learning within a probabilistic approach. A particular point is the need to develop approaches that are useful by reducing the presence of false positives and complex data validation approaches using multiple types of filtering, data augmentation and curriculum learning.
4. Site monitoring and risk detection. The methods presented above show the potential to develop large databases of archaeological sites. They also highlight associated new problems with the management of large site catalogues as they cannot be monitored using traditional methods such as field visits. This section will highlight some examples of automated algorithms designed for this purpose employing satellite and aerial datasets.
5. Recording and analysis of 3D items/events. With the increased availability and accuracy of 3D data collection methods the need to move from 3D recording to treatment and analysis is becoming pressing. In this section will demonstrate some of the methods we have developed to increase accuracy in model development, to integrate 3D models within their environment (including skyscapes) and to meaningfully analyse 3D shapes to extract environmental and cultural information.
6. Machine/deep learning applied to the detection and extraction of information from digital datasets. Besides previously commented examples of machine learning-based analyses, this section will comment on the utility of ML within processes that involve large investments of time / resources, analysis that involve not visible of difficult to quantify data, analysis that are subjective and need integration, and the correlation of multiple types of data that is not possible to do manually.
All the previous sections coincide in showing the need to move beyond the use of single static data sources and simple visual interpretative approaches. Given the advances on computation and data analysis made during the last years, we can now incorporate big data even if these data are the result of increasing the complexity of our sources and parameters. As a result, our research needs to embrace the uncertainty linked to complexity by becoming probabilistic and exploratory. Proactive and conscious research method development where techniques are adapted to the specific research questions, instead of generic out of the box approaches, have the potential to fully integrate computational archaeology and geospatial analyses within the wider field of archaeological research.
Hèctor A. Orengo is an ICREA Research Professor at GIAP (ICAC), co-director of GIAP and research coordinator at ICAC.