When Computers Dream of Charcoal

Benjamin P. Carter,Jeff H. Blackadar,Weston L. A. Conner

doi:10.1017/aap.2021.17

Abstract

AbstractThis research employs machine learning (Mask Region-Based Convolutional Neural Networks [Mask R-CNN]) and cluster analysis (Density-based spatial clustering of applications with noise [DBSCAN]) to identify more than 20,000 relict charcoal hearths (RCHs) organized in large “fields” within and around State Game Lands (SGLs) in Pennsylvania. This research has two important threads that we hope will advance the archaeological study of landscapes. The first is the significant historical impact of charcoal production, a poorly understood industry of the late eighteenth to early twentieth century, on the historic and present landscape of the United States. Although this research focuses on charcoal production in Pennsylvania, it has broad application for both identifying and contextualizing historical charcoal production throughout the world and for better understanding modern charcoal production. The second thread is the use of open data, open source, and open access tools to conduct this analysis, as well as the open publication of the resultant data. Not only does this research demonstrate the significance of open access tools and data but the open publication of our code as well as our data allow others to replicate our work, to tweak our code and protocols for their own work, and reuse our results.

Highlights

This research employs machine learning (Mask Region-Based Convolutional Neural Networks [Mask R-CNN]) and cluster analysis (Densitybased spatial clustering of applications with noise [DBSCAN]) to identify more than 20,000 relict charcoal hearths (RCHs) organized in large “fields” within and around State Game Lands (SGLs) in Pennsylvania
Our research focuses on identification of relict charcoal hearths and ignores other components
Our purpose is to assess charcoal production across Pennsylvania, rather than including the entire state of Pennsylvania in our sample, we focused our attention on State Game Lands (SGLs)

Summary

A DEEP-LEARNING APPROACH TO RCH IDENTIFICATION

Our approach here employs these lidar derivatives on a large scale. The rich detail across a large region (ca. 37,000 km2) makes it difficult, if not impossible, to identify these RCHs manually. In order to address how well DBSCAN effectively distinguished true and false positives, we reviewed the predicted RCHs in and around SGL 43 using the methods discussed above We chose this area because it is located near Hopewell Furnace, a National Historic Site as well as one of the best-known and wellresearched iron furnaces in the country (e.g., Kemper 1941; Straka and Ramer 2010; Walker 1966), and we have begun to conduct fieldwork there. We have only confirmed a few of the RCHs in the field at SGL 43, the ground truthing discussed above suggests that these methods are likely quite accurate on slopes, we are less confident in our identification of RCHs on flat terrain This is important because Williams (2020) has identified 279 colliers in the 1850 census for Clarion County (in the northwest of Pennsylvania). That the techniques described are able to identify all RCH fields—only that, compared to previous attempts, they move us significantly in the right direction

CONCLUSION

Findings

Data Availability Statement