Detecting floristic diversity of forests in restoration using remote sensing and machine learning

• Grant number: 23/08556-0
• Support Opportunities: Scholarships abroad - Research Internship - Scientific Initiation
• Effective date (Start): September 01, 2023
• Effective date (End): December 31, 2023
• Field of knowledge: Biological Sciences - Ecology - Applied Ecology
• Principal Investigator: Pedro Henrique Santin Brancalion
• Grantee: Gustavo Fiedler Rossi
• Supervisor: Thiago Sanna Freire Silva
• Host Institution: Escola Superior de Agricultura Luiz de Queiroz (ESALQ). Universidade de São Paulo (USP). Piracicaba , SP, Brazil
• Research place: University of Stirling, Scotland
• Associated to the scholarship: 22/16261-7 - Analysis of floristic diversity of restored forests using remote sensing, BP.IC

Abstract

Forest restoration is considered one of the best options to mitigate the effects of climate change and reduce biodiversity loss. Several forest restoration techniques were developed in the State of São Paulo, playing a crucial role in the protection of the Atlantic Forest. However, the methods for monitoring the success of restoration mostly depend on field trips to analyze different factors, such as diversity and floristic composition, a time-consuming and costly process. With the use of remote sensing techniques, we can reduce the cost of monitoring by increasing the spatial and temporal scale and combining different sensors to detect various vegetation characteristics. The Sentinel 1, Sentinel 2 and the Planet satellite constellations image in different bands of the electromagnetic spectrum, ranging from visible to microwave , detecting different aspects of the water content and vegetation composition and structure. In addition, images derived from LiDAR sensor data, which is capable of providing a 3D view of forests, are useful for estimating different structural variables, which can be related to floristic variables. The focus of this project is to use remote sensing images (Sentinel 1, Sentinel 2, Planet and LiDAR), combined with forest inventory data obtained in the field (Newfor project - #FAPESP 2018/18416-2) and Machine Learning techniques to analyze the diversity and floristic composition of semideciduous seasonal forests under restoration in São Paulo state. The identification of floristic variables measured in the field with the potential to be monitored by remote sensing will allow us to optimize the process of biodiversity study and monitoring of forest restoration. (AU)