Landscape as Palimpsest

Bill Beaver
3 days ago
10 min read

Codex Ephraemi Rescriptus - Greek manuscript of the Bible from the 5th century [Llywrch 2005] — *Codex Ephraemi Rescriptus -* Greek manuscript of the Bible from the 5th century [**Llywrch 2005**]

"In the spring or warmer weather when the snow thaws in the woods the tracks of winter reappear on slender pedestals and the snow reveals in palimpsest old buried wanderings, struggles, scenes of death. Tales of winter brought to light again like time turned back upon itself."

[Cormac McCarthy 2010 “Child of God”, p.138]

“... everyone started to draw, all the while discoursing on ancient things ...”

[Da Sangallo, Letters on Familiar Matters 6.2, in Barkan, 1999, p. 3]

This is the start of a series of what will be more like a research notebook than long articles. Much shorter and more focused. Firstly, it will be about a possible pilgrimage route through Ohio during the Middle Woodland period (Ohio Hopewell) between 100 BC and 700 CE. This route is between the present Ohio cities of Chillicothe and Newark. Enjoy.

Palimpsest

A palimpsest is a surface that has been marked upon, erased, and marked again. The erasing process is not complete; fragments of older markings remain. A palimpsest is a powerful analogy for an archaeological landscape. Changes in a landscape over time sort into two main categories: geological and human. Geological change is slow and gradual, punctuated by long-term climate changes, such as glaciation, and sudden catastrophes, including floods, volcanoes, or earthquakes. Human changes are faster than geological changes and have built up to a rapid pace over the last two centuries.

Maps

A map is a way of representing a landscape on a two-dimensional surface. Maps were once considered a Western, European invention, but this is no longer true; all humans can mentally project a 3D surface onto a flat surface and recreate some representation of a known landscape, the so-called 'God Trick.' [Haraway 1990] We can also move quickly between the representation and what we see in the world. Maps were scratched on rocks or painted, or drawn. Maps can be internalized in memory with enough practice. In the 1990s, a new method for representing maps was developed. This is called GIS (Geographic Information Systems). This is essentially using computers to draw maps. Since maps can be digitally represented, they become data objects that can be analysed. Although GIS is a process and a map is an object created by this process, most use the terms interchangeably. What a map means to archaeology and specifically to archaeological theory is, I believe, still an open question. Much theoretical work in the 1990s and early 2000s focused on actual and perceived biases in the process of map representation. [Ullah et al. 2024] Then, the availability of mapping software opened up two decades of active usage. The success of GIS still leaves room for archaeological critique. One critique refers back to the palimpsest; [Bailey 2007] modern mapping software, for the most part, doesn't do changes over time. A map must be more than a better (or worse) alternative to a table or graph, a passive object; it needs to be an analytical tool in itself. There is also much discussion on the importance and usage of 2D versus 3D representations.[Gupta & Devillers 2017] For this article, I will be referring to 2D representations only.

Lidar

Lidar stands for Light Detection and Ranging. A remote sensing technology, laser light hits an object, bounces back, and is detected by a sensor. Landscape lidar uses satellites, airplanes, or drones to scan a region. Lidar gives accurate measurements of the ground and objects on the ground. Most importantly, it can penetrate forest cover and thick vegetation. The resulting data is called a point cloud, a set of 3d points where the laser light struck something and bounced back to the sensor. The point cloud is separated into surface and terrain, anything not surface, and the terrain is then classified into categories like vegetation and structures. Each point cloud is then converted into an image, also called a raster. Each square pixel in the image is a certain size, called the raster's resolution. The pixel also contains at least one attribute, its elevation. The general name for this raster is Digital Elevation Model or DEM. For archaeological purposes, especially in Europe, the terms DSM for Digital Surface Model and DTM for Digital Terrain Model are used. In addition, an archaeological elevation model needs to include features that might be in the DTM, called a Digital Feature Model (DFM). [Štular et al. 2021a]

DFM, DTM, and DSM [Štular et al. 2021a, p. 1] — DFM, DTM, and DSM [**Štular et al. 2021a, p. 1**]

The source of lidar data can be a flight specifically conducted for a project, or it can be from national or regional data sources. One advantage of in-house data is that the point cloud can be processed with archaeology in mind. Lozić & Štular propose an 18-step workflow with about half the steps dealing with converting the point cloud into a DFM. [Lozić & Štular 2021] [Štular, et al. 2021b] One issue is that the points are not distributed evenly once they are projected onto a 2D surface; bare land contains fewer points than, say, dense woods. Another way to measure resolution is by points per pixel; a value of less than or greater than 1 indicates that the elevation value for the pixel was extrapolated. There are different methods of extrapolation. The point density of a DFM is used to generate a secondary data object, a confidence model. [Lozić & Štular 2021, p. 8] The documentation and creation of data objects related to the process of transforming data are referred to as paradata. [Lozić & Štular 2021, p. 2]

Visualization

A DFM is a representation of a landscape. In science, a representation represents what is called a target, a question, or a hypothesis. [Frigg & Nguyen 2021] [Štular et al. 2012] With a palimpsest, our target here is the traces of something that has been erased. Thus, the DFM requires another round of processing, a visualization. [Opgenhaffen 2021] Visualization proceeds using a dizzying set of image processing filters and procedures that reveal the subtleties of the landscape. [Kokalj et al. 2011] [Orengo & Petrie 2018] [Zakšek et al. 2011] There is no recipe for this; all are very context-dependent, and each one distorts the raster to some extent. No single one will do, and the recommended practice is to use several layers of processes and blend them into a single optimal image. [Kokalj & Somrak 2019] Additionally, grayscale is recommended, with color used only to represent controlled categories. An open-source software product called the Relief Visualization Toolbox (RVT) [Kokalj & Zakšek 2014] is currently popular.

Documentation

Documenting a visualization includes not only metadata and paradata but the archiving all of the data objects created as well. The goal here is for anyone wanting to replicate the results, a way is available. I am a big proponent of open source, but I realize that for some archaeological discoveries, the unknown location of a fragile site is the only thing protecting it. I have found papers where images are missing and or parts of images are blacked out. Some level of security must be devised and access provided for all types of interest.

Automated Classification

Once a landscape has been visualized, the next step is to categorize what is seen. This is usually done by human judgment, but for the last fifty years, there has been a great deal of research in automating the process. In the last ten years, there has been increasing success. [Casana et al. 2021] [De Matos‐Machado 2019] [Menéndez Blanco 2020] There are two main methods based on what is loosely called 'machine learning.' [Davis 2020] One is oriented more toward shapes or objects and generally involves a neural net. [Sherba et al. 2012] [Somrak et al. 2020] Others are purely statistical and generalized. These statistical methods, along with recent breakthroughs in what are called 'diffusion models,' are what is currently being hyped as AI. Machine learning has been successful and allows for previously unknown regional views of a landscape. There are caevats. [Kadhim & Abed 2023] These processes must be trained with known data, sometimes a considerable amount. Results sometimes have to be tweaked and rerun. Success is measured in the percentage of true positives and true negatives found. Whether a visualization that is optimal for a human is also optimal for machine learning is an open question. Again, there is no plug-and-play; all settings are heterogeneous to the landscape being studied. [Berganzo-Besga et al. 2021] [Bonhage et al. 2021] [Guyot et al. 2020] [Vletter 2014]

Thick Description

The idea of thick description was first conceived in two essays by the British philosopher Gilbert Ryle in 1968. Thinking and Reflecting [Ryle 1968a] and The Thinking of Thoughts. [Ryle 1968b] This is basically about the difficulty one human has in describing and understanding the thoughts and behaviors of another. [Tanney 2022] This was expanded by the anthropologist Clifford Geertz as a way for ethnographic description to explain culture. [Geertz 1973] Christopher Carr expands this concept into archaeology as 'thick prehistory.' [Carr & Chase 2005] Out of many definitions and comments on thick prehistory that Carr makes in the book, I find this the most interesting:

"thick prehistory uses diverse theories, generalizations, and analogs, with their diverse assumptions about humans, in an exploratory manner to generate insights into past human situations" [Carr & Chase 2005, p. 50]

So what is a 'thick landscape'? Again, the analogy of a palimpsest. A thick landscape is a dynamic one, an attempt to peel back the layers of erasure to see what lies underneath. This includes old aerial photographs, maps, and records, along with assumptions about how humans have reconfigured the landscape, past and present.

Bailey, Geoff. 2007. “Time Perspectives, Palimpsests and the Archaeology of Time.” Journal of Anthropological Archaeology 26 (2): 198–223. https://doi.org/10.1016/j.jaa.2006.08.002.
Berganzo-Besga, Iban, Hector A. Orengo, Felipe Lumbreras, Miguel Carrero-Pazos, João Fonte, and Benito Vilas-Estévez. 2021. “Hybrid MSRM-Based Deep Learning and Multitemporal Sentinel 2-Based Machine Learning Algorithm Detects Near 10k Archaeological Tumuli in North-Western Iberia.” Remote Sensing 13 (20): 20. https://doi.org/10.3390/rs13204181.
Bonhage, Alexander, Mahmoud Eltaher, Thomas Raab, Michael Breuß, Alexandra Raab, and Anna Schneider. 2021. “A Modified Mask Region-Based Convolutional Neural Network Approach for the Automated Detection of Archaeological Sites on High-Resolution Light Detection and Ranging-Derived Digital Elevation Models in the North German Lowland.” Archaeological Prospection 28 (2): 177–86. https://doi.org/10.1002/arp.1806.
Carr, Christopher, and D. Troy Case, eds. 2005. Gathering Hopewell: Society, Ritual, and Ritual Interaction. Interdisciplinary Contributions to Archaeology. Kluwer Academic/Plenum Publishers.
Casana, Jesse, Elise J. Laugier, Austin Chad Hill, et al. 2021. “Exploring Archaeological Landscapes Using Drone-Acquired Lidar: Case Studies from Hawai’i, Colorado, and New Hampshire, USA.” Journal of Archaeological Science: Reports 39 (October): 103133. https://doi.org/10.1016/j.jasrep.2021.103133.
Davis, Dylan S. 2020. “Defining What We Study: The Contribution of Machine Automation in Archaeological Research.” Digital Applications in Archaeology and Cultural Heritage 18 (September): e00152. https://doi.org/10.1016/j.daach.2020.e00152.
De Matos‐Machado, Rémi, Jean‐Pierre Toumazet, Jean‐Claude Bergès, et al. 2019. “War Landform Mapping and Classification on the Verdun Battlefield (France) Using Airborne LiDAR and Multivariate Analysis.” Earth Surface Processes and Landforms 44 (7): 1430–48. https://doi.org/10.1002/esp.4586.
Frigg, Roman, and James Nguyen. 2021. “Scientific Representation.” In The Stanford Encyclopedia of Philosophy, Winter 2021, edited by Edward N. Zalta. Metaphysics Research Lab, Stanford University. https://plato.stanford.edu/archives/win2021/entries/scientific-representation/.
Geertz, Clifford. 1973. “Thick Description: Toward an Interpretive Theory of Culture.” In The Interpretation of Cultures: Selected Essays. Basic Books.
Gupta, Neha, and Rodolphe Devillers. 2017. “Geographic Visualization in Archaeology.” Journal of Archaeological Method and Theory 24 (3): 852–85. https://doi.org/10.1007/s10816-016-9298-7.
Guyot, Alexandre, Marc Lennon, and Laurence Hubert-Moy. 2021. “Objective Comparison of Relief Visualization Techniques with Deep CNN for Archaeology.” Journal of Archaeological Science: Reports 38 (August): 103027. https://doi.org/10.1016/j.jasrep.2021.103027.
Haraway, Donna. Simians, Cyborgs, and Women: The Reinvention of Nature. New York: Routledge, 1990. https://doi.org/10.4324/9780203873106.
Kadhim, Israa, and Fanar M. Abed. 2023. “A Critical Review of Remote Sensing Approaches and Deep Learning Techniques in Archaeology.” Sensors 23 (6): 6. https://doi.org/10.3390/s23062918.
Kokalj, Žiga, and Klemen Zakšek. 2014. Relief Visualization Toolbox (RVT). V. 2.2.1. Released. https://www.zrc-sazu.si/en/rvt.
Kokalj, Žiga, and Maja Somrak. 2019. “Why Not a Single Image? Combining Visualizations to Facilitate Fieldwork and On-Screen Mapping.” Remote Sensing 11 (7). https://www.mdpi.com/2072-4292/11/7/747.
Kokalj, Žiga, Klemen Zakšek, and Krištof Oštir. 2011. “Application of Sky-View Factor for the Visualisation of Historic Landscape Features in Lidar-Derived Relief Models.” Antiquity 85 (327): 263–73. https://doi.org/10.1017/S0003598X00067594.
Lozić, Edisa, and Benjamin Štular. 2021. “Documentation of Archaeology-Specific Workflow for Airborne LiDAR Data Processing.” Geosciences 11 (1): 1. https://doi.org/10.3390/geosciences11010026.
Llywrch. 2005. Codex Ephraemi Rescriptus. National Library in Paris. https://commons.wikimedia.org/wiki/File:Codex_ephremi_(The_S.S._Teacher%27s_Edition-The_Holy_Bible_-_Plate_XXIV).jpg.
Menéndez Blanco, Andrés, Jesús García Sánchez, José Manuel Costa-García, João Fonte, David González-Álvarez, and Víctor Vicente García. 2020. “Following the Roman Army between the Southern Foothills of the Cantabrian Mountains and the Northern Plains of Castile and León (North of Spain): Archaeological Applications of Remote Sensing and Geospatial Tools.” Geosciences 10 (12): 12. https://doi.org/10.3390/geosciences10120485.
Opgenhaffen, Loes. 2021. “Visualizing Archaeologists: A Reflexive History of Visualization Practice in Archaeology.” Open Archaeology 7 (1): 353–77. https://doi.org/10.1515/opar-2020-0138.
Orengo, Hector A., and Cameron A. Petrie. 2018. “Multi-Scale Relief Model (MSRM): A New Algorithm for the Visualization of Subtle Topographic Change of Variable Size in Digital Elevation Models.” Earth Surface Processes and Landforms 43 (6): 1361–69. https://doi.org/10.1002/esp.4317.
Ryle, Gilbert. 1968a. “Thinking and Reflecting.” In The Human Agent. Palgrave Macmillan, London. https://doi.org/10.1007/978-1-349-27908-1_12.
Ryle, Gilbert. 1968b. The Thinking of Thoughts. University of Saskatchewan.
Sherba, Jason, Leonhard Blesius, and Jerry Davis. 2014. “Object-Based Classification of Abandoned Logging Roads under Heavy Canopy Using LiDAR.” Remote Sensing 6 (5): 5. https://doi.org/10.3390/rs6054043.
Somrak, Maja, Sašo Džeroski, and Žiga Kokalj. 2020. “Learning to Classify Structures in ALS-Derived Visualizations of Ancient Maya Settlements with CNN.” Remote Sensing 12 (14): 14. https://doi.org/10.3390/rs12142215.
Štular, Benjamin, Žiga Kokalj, Krištof Oštir, and Laure Nuninger. 2012. “Visualization of Lidar-Derived Relief Models for Detection of Archaeological Features.” Journal of Archaeological Science 39 (11): 3354–60. https://doi.org/10.1016/j.jas.2012.05.029.
Štular, Benjamin, Edisa Lozić, and Stefan Eichert. 2021a. “Airborne LiDAR-Derived Digital Elevation Model for Archaeology.” Remote Sensing 13 (9): 9. https://doi.org/10.3390/rs13091855.
Štular, Benjamin, Stefan Eichert, and Edisa Lozić. 2021b. “Airborne LiDAR Point Cloud Processing for Archaeology. Pipeline and QGIS Toolbox.” Remote Sensing 13 (16): 16. https://doi.org/10.3390/rs13163225.
Tanney, Julia. 2022. “Gilbert Ryle.” In The Stanford Encyclopedia of Philosophy, Summer 2022, edited by Edward N. Zalta. Metaphysics Research Lab, Stanford University. https://plato.stanford.edu/archives/sum2022/entries/ryle/.
Ullah, Isaac I., Zachery Clow, and Juliette Meling. 2024. “Paradigm or Practice? Situating GIS in Contemporary Archaeological Method and Theory.” Journal of Archaeological Method and Theory 31 (3): 1185–231. https://doi.org/10.1007/s10816-023-09638-1.
Vletter, Willem. F. 2014. “(Semi) Automatic Extraction from Airborne Laser Scan Data of Roads and Paths in Forested Areas.” Second International Conference on Remote Sensing and Geoinformation of the Environment. https://www-spiedigitallibrary-org.ezproxy4.library.arizona.edu/conference-proceedings-of-spie/9229/1/Semi-automatic-extraction-from-airborne-laser-scan-data-of-roads/10.1117/12.2069709.full.
Zakšek, Klemen, Kristof Oštir, and Žiga Kokalj. 2011. “Sky-View Factor as a Relief Visualization Technique.” Remote Sensing 3 (2): 2. https://doi.org/10.3390/rs3020398.