Nationwide urban tree canopy mapping and coverage assessment in Brazil from high-resolution remote sensing images using deep learning

Abstract

Urban tree canopy maps are essential for providing urban ecosystem services. The relationship between urban trees and urban climate change, air pollution, urban noise, biodiversity, urban crime, health, poverty, and social inequality provides important information for better understanding and management of cities. To better service Brazil’s urban ecosystem, this study developed a semi-supervised deep learning method, which is able to learn semantic segmentation knowledge from both labeled and unlabeled images, to robustly segment urban trees from high spatial resolution remote sensing images. Using this approach, this study created 0.5 m fine-scale tree canopy products for 472 cities in Brazil and made them freely available to the community. Results showed that the urban tree canopy coverage in Brazil is between 5% and 35%, and the average urban tree canopy cover is approximately 18.68%. The statistical results of these tree canopy maps quantify the nationwide urban tree canopy inequality problem in Brazil. Urban tree canopy coverage from 130 cities that can accommodate approximately 27.22% of the total population is greater than 20%, whereas 342 cities that can accommodate approximately 42% of the total population have tree canopy cover less than 20%. We expect that urban tree canopy maps will encourage research on Brazilian urban ecosystem services to support urban development and improve inhabitants’ quality of life to achieve the goals of the Agenda for Sustainable Development. In addition, it can serve as a benchmark dataset for other high-resolution and mid/low-resolution remote sensing images urban tree canopy mapping results assessments.