dc.contributor.advisor | Wernke, Steven A. | |
dc.creator | Zimmer-Dauphinee, James Robert | |
dc.date.accessioned | 2023-08-25T17:20:30Z | |
dc.date.created | 2023-08 | |
dc.date.issued | 2023-07-18 | |
dc.date.submitted | August 2023 | |
dc.identifier.uri | http://hdl.handle.net/1803/18439 | |
dc.description.abstract | Prehispanic Andean people constructed monumental-scale agricultural infrastructure in the form of canals and agricultural terracing, transforming the steep and arid Pacific slopes of the south-central Andes into highly productive anthropogenic landscapes. The invasion of these landscapes by Spanish colonists in the 16th century and their subsequent policies of forced resettlement of indigenous Andean people into planned colonial towns threatened to disrupt the complex social and economic systems that maintained this infrastructure. Yet the effects of this disruption were not consistent across the landscape, and current understandings of how these processes varied across regions are limited by unsystematic and small sample sizes. This research develops machine learning models for semi-autonomous detection of relict architecture and agricultural features to enable very large-scale archaeological prospection using high-resolution satellite imagery. Set in the southern highlands of Peru, this study presents an architectural feature Deep Learning (DL) model that performs in a comparable and complementary fashion in detection sensitivity and specificity to a manual imagery survey of the same area. This model sets a foundation for future large-scale semi-autonomous imagery surveys of archaeological settlements. Secondly, this study examines transregional patterns of agricultural deintensification using AI-assisted satellite imagery surveys to inventory the distribution of active and abandoned agricultural fields in an ~81,000 km² region of the south-central Andean highlands. Deep Learning was used to segment and classify agricultural fields, identifying ~412,000 hectares of agricultural fields, nearly a quarter (23.6%) of which were abandoned. Statistical and geospatial modeling show substantial variation in the distribution of field abandonment and its relationship to reducción settlements, suggesting that the effects of the resettlement were strongly mediated by local social, political, and ecological contexts and decisions. | |
dc.format.mimetype | application/pdf | |
dc.language.iso | en | |
dc.subject | colonialism | |
dc.subject | agricultural deintensification | |
dc.subject | AI | |
dc.subject | terracing | |
dc.subject | Peru | |
dc.subject | Andes | |
dc.subject | remote semsing | |
dc.subject | archaeology | |
dc.subject | GIS | |
dc.subject | reducción | |
dc.subject | resettlement | |
dc.title | Trans-Regional Perspectives on Agricultural Deintensification in the Colonial Andes through Remote Sensing and AI-Assisted Archaeological Survey | |
dc.type | Thesis | |
dc.date.updated | 2023-08-25T17:20:30Z | |
dc.type.material | text | |
thesis.degree.name | PhD | |
thesis.degree.level | Doctoral | |
thesis.degree.discipline | Anthropology | |
thesis.degree.grantor | Vanderbilt University Graduate School | |
local.embargo.terms | 2024-02-01 | |
local.embargo.lift | 2024-02-01 | |
dc.creator.orcid | 0000-0003-1489-7747 | |
dc.contributor.committeeChair | Wernke, Steven A. | |