Tikal Research

Spatial Analysis of Vegetation in the Maya Biosphere Reserve

May 2026

Conducted jointly with my research partner and friend Sarah Troup-Galligan for an independent study under Professor Charlie Bettigole, this project investigated forest loss and vegetation structure in Guatemala's Maya Biosphere Reserve, home to the ancient Maya sites of San Bartolo and Xultún. Building on Bettigole's remote sensing and archaeological mapping work, our research aimed to strengthen the case for elevating the reserve's conservation status by documenting threats beyond looting, namely deforestation and encroaching oil palm plantations.

We used the Hansen (2013) Global Forest Change dataset to map cumulative and yearly deforestation from 2000 to 2024 across both Guatemala and the reserve, and used Google's new Satellite Embeddings dataset, a 64-dimensional land representation combining imagery, weather, and radar data, to classify oil palm plantation extent at a level of detail not previously possible. To forecast future forest loss, we built two models in Python via Google Colab: an ARIMA model for statistical time-series projection, and a Cellular Automata x Markov Chain model to spatially project deforestation scenarios at varying severities. Separately, we used ArcGIS Pro to analyze LiDAR tiles, identifying hidden Maya structures beneath the forest canopy and building a 3D canopy model to compare tree height and density over archaeological structures versus undeveloped uplands.

We found that although the reserve has experienced less cumulative deforestation than the country as a whole, it is not exempt from deforestation risk, and oil palm plantations already border the reserve's fringes. Our ARIMA model projected continued deforestation across the concessions region over the next 20 years with 95% confidence, while the Cellular Automata model showed that under every tested severity scenario, the forest edge is closing in on the San Bartolo site. Our LiDAR-based canopy analysis revealed that canopy density differs measurably between areas with Maya structures and undeveloped uplands, varying by tree height class, suggesting archaeological features may shape microhabitat structure in ways worth further study. 

Sarah and I presented our research for the Geography department with the presentation below, along with the written report.