Advancing Machine Learning Approaches to Identifying Charcoal Morphologies and Fuels for Sedimentary Charcoal Analysis

Summary

This is an abstract from the "Practice, Theory, and Ethics of Machine Learning in Archaeology" session, at the 90th annual meeting of the Society for American Archaeology.

Differentiating between natural and anthropogenic fire in the past remains one of the principal challenges in interpreting paleo-charcoal records and has implications for contextualizing changing fire regimes in our world today. During the Holocene, cultural burning practices throughout the globe were motivated by diverse social institutions, values, and economics; however, the frequency, seasonality, spatial distribution, and ecological severity of anthropogenic fire likely differed enough from natural fire to generate lasting ecological effects. Similarly, prescribed fire operations conducted by land managers in the United States are modern examples of cultural burning to achieve desired ecological outcomes. Nonetheless, relatively few studies have utilized prescribed fires as laboratories for testing methods to interpret anthropogenic fire activity in charcoal records. We present the results of a comprehensive study of charcoal production and morphology collected during a series of highly instrumented prescribed fires that occurred during March 2021 within the Hitchiti Experimental Forest, Georgia, USA. We utilize a machine-learning classification approach to relate field-collected and lab-created charcoal datasets to pre-/post-burn vegetation inventories and radiometric measurement of energy release collected at 12 study plots throughout the burn area. Using this approach, we seek to expand the interpretive potential of paleo-charcoal records for identifying past anthropogenic burning.

This Resource is Part of the Following Collections

• Society for American Archaeology 90th Annual Meeting, Denver, CO (2025)   •
• Practice, Theory, and Ethics of Machine Learning in Archaeology

Keywords

Record Identifiers

Abstract Id(s): 51933