Automated Identification of Archaeological Features in a Regional Lidar Dataset from Southeastern New Mexico

Summary

This is an abstract from the "SAA 2021: General Sessions" session, at the 86th annual meeting of the Society for American Archaeology.

In 2014, the Carlsbad Field Office of the Bureau of Land Management acquired 372 square miles of high resolution lidar data in an experimental attempt to map archaeological features over a wide area of southeastern New Mexico. The features of interest were burned rock middens with a distinctive topographic signature. If successful, this effort would have had significant applications for the study and management of cultural resources in the region. A GIS-based "digital survey" approach to feature identification was attempted using human operators and false color visualizations. This was successful but was too labor-intensive to apply to a broad area. In the end only 51 square miles were studied, about 14% of the lidar data acquired. This presentation reports on a successful effort to train convolutional neural networks to accomplish feature extraction at a regional scale with little labor input.

This Resource is Part of the Following Collections

• Society for American Archaeology 86th Annual Meeting, Online (2021)   •
• SAA 2021: General Sessions

Keywords

Spatial Coverage

Record Identifiers

Abstract Id(s): 32119