Bayesian regression versus machine learning - Rapid age estimation of archaeological features identified with LiDAR

Aug 4, 2026

LiDAR has transformed archaeology by enabling high-resolution, large-area mapping of surface features and thus documenting ancient tropical urbanism at unprecedented scale. A key limitation remains the lack of temporal depth, since traditional dating is slow and destructive. Using Angkor temples as a case study, predictive regression and machine-learning models are compared to estimate foundation dates from remote-sensing data. A Bayesian regression framework is then evaluated to better incorporate chronological uncertainties and improve interpretive value.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages15

Reading Time38 min

AccessFree to download

Bayesian regression versus machine learning

LiDAR-enabled archaeological discovery

Challenge: adding temporal depth to spatial data

Angkor case study and predictive modeling

Model comparison and uncertainty handling

Updated training data and spatiotemporal growth estimates

Implications for rapid chronology of large LiDAR datasets

[www. nature.com/scientificreports](www. nature.com/scientificreports) OPEN Bayesian regression versus machine learning for rapid age estimation of archaeological features identified with lidarat Angkor W. Christopher Carleton1*, Sarah Klassen2, Jonathan Niles‑Weed 3, Damian Evans4,9, Patrick Roberts5,6 & Huw S. Groucutt1,6,7,8 Lidar (light‑detection and ranging) has revolutionized archaeology. We are now able to produce high‑ resolution maps of archaeological surface features over vast areas, allowing us to see ancient land‑use and anthropogenic landscape modification at previously un‑imagined scales. In the tropics, this has enabled documentation of previously archaeologically unrecorded cities in various tropical regions, igniting scientific and popular interest in ancient tropical urbanism. An emerging challenge, however, is to add temporal depth to this torrent of new spatial data because traditional archaeological investigations are time consuming and inherently destructive. So far, we are aware of only one attempt to apply statistics and machine learning to remotely‑sensed data in order to add time‑depth to spatial data. Using temples atthe well‑known massive urban complex of Angkor in Cambodia asa case study, a predictive model was developed combining standard regression with novel machine learning methods to estimate temple foundation dates for undated Angkorian temples identified with remote sensing, including lidar. The model’s predictions were used to produce an historical population curve for Angkor and study urban expansion at this important ancient tropical urban centre. The approach, however, has certain limitations. Importantly, its handling of uncertainties leaves room for improvement, and like many machine learning approaches it is opaque regarding which predictor variables are most relevant. Here we describe a new study in which we investigated an alternative Bayesian regression approach applied to the same case study. We compare the two models in terms of their inner workings, results, and interpretive utility. We also use an updated database of Angkorian temples as the training dataset, allowing us to produce the most current estimate for temple foundations and historic spatiotemporal urban growth patterns at Angkor. Our results demonstrate that, in principle, predictive statistical and machine learning methods could be used to rapidly add chronological information to large lidar datasets and a Bayesian paradigm makes it possible to incorporate important uncertainties—especially chronological—into modelled temporal estimates. Lidar (light-detection and ranging) has become widely appreciated as a revolutionary new tool for archaeological discovery and heritage management1. It is a sophisticated laser scanning technology that can be used to produce 3D models of the Earth’s surface even through dense forest canopies. In 2011, a lidar surface map of an historically important Classic Maya centre in Belize called Caracol was published2. The lidar scanning covered 200 km2 and revealed as many as eleven new causeways and thousands of previously unrecorded residential 1Extreme Events Research Group, Max Planck Institutes of/for, Geoanthropology, Chemcial Ecology, and Biogeochemistry, Jena, Germany. 2Department of Anthropology, University of Toronto, Toronto, Canada. 3Courant Institute of Mathematical Sciences and Center for Data Science, New York University, New York, USA. 4École française d’Extrême-Orient, Paris, France. 5isoTROPIC Research Group, Max Planck Institute of Geoanthropology, Jena, Germany. 6Department of Archaeology, Max Planck Institute of Geoanthropology, Jena, Germany. 7Department of Classics and Archaeology, University of Malta, Msida, Malta. 8Institute of Prehistoric Archaeology, University of Cologne, Cologne, Germany. 9Damian Evans is deceased. *email: carleton@ [gea.mpg.de](gea.mpg.de) [www. nature.com/scientificreports/](www. nature.com/scientificreports/) buildings and agr

FAQs

Why is estimating the age of LiDAR-identified archaeological features challenging?

Temporal depth is difficult because conventional archaeological investigations are time-consuming and destructive. LiDAR provides detailed spatial coverage but not dates for many features.

How was the Angkor temple case study used in the models?

A predictive approach was built using temple foundation dates to train models, estimating dates for undated Angkorian temples identified via remote sensing including LiDAR, and then using predictions to infer historical growth patterns.

What advantages does the Bayesian regression approach aim to provide?

The Bayesian framework is intended to incorporate important uncertainties, especially chronological uncertainty, and to address limitations of prior methods such as opaque relevance of predictor variables.

Aug 4, 2026

Upload by Olivia Brown