ID: 1966
Presenting Author: Roberto Leal Rosillo
Session: 689 - Earth Observation for Impact Assessment
Status: pending
Study using satellite and drone data with machine learning to map carbon stocks and guide reforestation to restore Rapa Nui’s landscape and resilience.
This presentation shares the results of a study conducted to support our client in efforts to understand and reverse the loss of vegetation cover on the island of RapaNui. The Inter-American Development Bank (IDB) applied advanced Earth Observation tools to quantify vegetation loss and estimate the carbon stocks affected by development projects. In parallel, drone imagery was used to establish environmental baselines for areas targeted for revegetation.
The study integrated high-resolution satellite imagery, drone orthomosaics, and machine learning analysis of land cover and land use (LCLU) to produce a spatially explicit model of carbon storage. The workflow involved: (1) acquisition and preprocessing of multi-temporal optical satellite data; (2) compilation of average carbon stock values for each LCLU (3) machine learning classification of LCLU; (4) calculation and mapping of total carbon stocks; and (5) synthesis of results in a report evaluating methodologies, uncertainties, and alternative carbon-stock estimation approaches.
The resulting carbon-stock map provides a robust evidence base for environmental impact assessment, guiding the design of a mitigation and reforestation plan that seeks to restore ancestral landscapes and strengthen climate resilience. By combining Earth Observation data, drone-based monitoring, and computational modeling, the study demonstrates how quantitative carbon accounting can inform data-driven decision-making, support landscape restoration, and operationalize climate adaptation strategies in small-island environments such as Rapa Nui
Senior Environmental Specialist at the IDB, Washington DC, Roberto implements environmental safeguards, monitors GHG emissions, and leads the division’s initiatives in drone imagery applications
Coauthor 1: Victoria Laporte