Aboveground forest carbon mapping in São Paulo by fusing GEDI spaceborne LiDAR and Planet Scope satellite images

• Grant number: 24/08886-2
• Support Opportunities: Scholarships abroad - Research Internship - Doctorate
• Start date: November 10, 2024
• End date: November 09, 2025
• Field of knowledge: Agronomical Sciences - Forestry Resources and Forestry Engineering - Nature Conservation
• Principal Investigator: Paulo Guilherme Molin
• Grantee: Giulio Brossi Santoro
• Supervisor: Carlos Alberto Silva
• Host Institution: Escola Superior de Agricultura Luiz de Queiroz (ESALQ). Universidade de São Paulo (USP). Piracicaba , SP, Brazil
• Institution abroad: University of Florida, Gainesville (UF), United States
• Associated to the scholarship: 23/00241-0 - Modeling aboveground biomass in São Paulo State Atlantic Forest: an upscalling approach, BP.DR

Abstract

The substantial loss of forests, coupled with human reliance on fossil fuels, is driving irreversible changes in global climate dynamics. In Brazil, the Atlantic Forest suffered intense deforestation over the years compromising ecosystem services such as climate regulation. Multiple national and international forest restoration commitments have been established seeking to minimize the impacts of climate change and global warming. Thus, forest monitoring is fundamental not only to ensure the success of restoration trajectory but also to measure the increment of carbon stocks through forest biomass. The main goal of this research proposal is to improve GEDI-derived aboveground carbon (AGC) estimates for São Paulo state native forests using a 3-dimensional (3D) scene regression convolutional neural network (CNN) model. Methods include i) a random sampling of Santoro et al., (2024) GEDI-derived AGC estimates; ii) acquisition and compiling of multitemporal and multispectral Planet Scope imagery; iii) training and evaluating the model through its loss function (Mean Squared Error) and other key metrics; iv) running the 3D-CNN model to predict AGC for native forest cover; and finally iv) validating predicted values using pre-existing GEDI and ALS-derived AGC estimates. We expect to generate a reliable high resolution wall-to-wall AGC map which can be used to estimate the carbon baseline across São Paulo state, contributing to multiple restoration commitments and initiatives especially focused on climate regulation.