Mapping Human Influence

Bayesian modelling of urban sediments at Koh Ker, Cambodia

W. Christopher Carleton | 2026-09-21

Introduction

SOME CONTENT

Based on Carleton et al. (2026), Bayesian modelling of the latent human influence field in archaeological urban sediments at Koh Ker, Cambodia. doi:10.1016/j.jas.2026.106663

Open with the challenge of reconstructing a city’s changing activity across a large landscape. Replace the placeholder with a Koh Ker photograph.

Overview

• The urban sampling problem.
• Koh Ker as a pilot study.
• From sediment proxies to a latent field.
• Results and prospects for larger surveys.

THE URBAN SAMPLING PROBLEM

Space and Time

Lidar and survey

• Extensive spatial coverage.
• Limited chronological depth.

Excavation

• Temporal depth.
• Limited spatial coverage.

How can we reconstruct an urban trajectory between these scales?

Different Proxies, Different Measurements

• Proxy records differ in resolution, distribution, and measurement structure.
• Joint inference needs a model linking observations to a shared process.

Use an example contrasting a continuous measurement with an artefact count. Introduce the distinction between what we observe and what we want to infer.

Research Aim

Estimate past human influence by combining heterogeneous sediment proxies.

Introduce the latent human influence field (LHIF) as the organising concept.

KOH KER

Site and Context

Add the paper’s site map. Outline the historical context and distinguish political history from the sedimentary evidence for local activity.

Pilot Study

• Eight test trenches at Koh Ker, Cambodia.
• Paired magnetic susceptibility and ceramic abundance.
• Rapid, low-impact excavation as a potential survey strategy.

Two Observation Channels

Magnetic susceptibility

Ceramic abundance

Show paired observations from one trench before introducing the model. Explain how each channel was recorded and its depth resolution.

THE LATENT HUMAN INFLUENCE FIELD

What Is a Latent Field?

• Latent: human influence is inferred from its traces.
• Field: influence can vary across the sediment archive.
• Both proxies are noisy observations of a shared LHIF.

The model connects archaeological observations to an explicit theoretical quantity.

Model Structure

• Bayesian hierarchical latent-variable model.
• Proxy channels linked within a causal framework.
• Depth-wise reconstructions for the pilot trenches.

Insert the model diagram. Walk through the latent process, channel-specific observation models, and hierarchical structure. Add equations here if appropriate for the audience.

From Observations to Inference

Field sampling → proxy measurements → joint model → estimated influence

Explain what a reconstructed profile means and how to read its uncertainty. Use the same trench as on the observation slide.

RESULTS

Reconstructing Human Influence

Insert the principal reconstruction figure. Highlight the main depth-wise patterns and differences between trenches. Add the figure’s specific result as a short takeaway.

Magnetic Susceptibility and Archaeological Horizons

• Magnetic susceptibility tracks ceramic-rich anthropogenic horizons.

Use the paper’s comparison to explain the practical value of susceptibility measurements. Discuss what the pilot establishes about this proxy at Koh Ker.

From Depth to Urban Time

A spacetime reconstruction requires sufficient sampling and temporal coregistration.

Add the chronological evidence and explain how depth profiles could be aligned in time. Separate the pilot reconstructions from the proposed city-wide application.

IMPLICATIONS

Towards Larger Surveys

• Low-impact trenching may enable scalable coring surveys.
• Additional sedimentary proxies can extend the framework.

Sketch a future sampling design. Discuss spatial coverage, dating, and the additional information different proxy channels could contribute.

Model Thinking in the Field

What do we want to infer?

Which observations would constrain it?

How should we collect those observations?

Return to the connection between theory, statistical modelling, and field strategy. Develop this as the broader methodological message of the talk.

Take-Home Points

• A latent field provides a common target for heterogeneous proxies.
• Koh Ker demonstrates the framework through a trench-based pilot.
• Larger urban reconstructions depend on sampling and chronology.

Reference

Carleton, W. Christopher, Thon Tho, David Brotherson, Samnang Phin, Huw Groucutt, Jürgen Renn, and Patrick Roberts. 2026. “Bayesian modelling of the latent human influence field in archaeological urban sediments at Koh Ker, Cambodia.” Journal of Archaeological Science 193: 106663. https://doi.org/10.1016/j.jas.2026.106663

Discussion

How could this framework change the way we sample ancient cities?