Accounting for twenty-first-century annual forest loss in the Atlantic Forest of Brazil using high-resolution global maps

posted on 09.02.2020 by Fernando Andreacci, Rosemeri Carvalho Marenzi

Deforestation in the Brazilian Atlantic Forest has been mapped since 1986 by SOS Mata Atlântica in partnership with the National Institute for Space Research under the project named Atlas dos Remanescentes Florestais da Mata Atlântica (SOSMA). The mapping results depict a worrying scenario of over-fragmentation and reduction of forest cover. The limited information provided, however, may underestimate deforestation in this Brazilian biome. We analysed the use of new high-resolution global deforestation datasets to quantify 10 years of deforestation in a heterogeneous landscape, where forest fragments are immersed in an agricultural and urban matrix. We set up a framework based on high-resolution satellite images available on Google Earth to visually inspect whether the areas of Global Forest Change (GFC) and SOSMA considered to be forest loss corresponded to actual deforestation. Results showed that no more than 21% of the forest loss area of the GFC were considered deforestation of Atlantic Forest remnants. More than 55% of the mapped patches (counts) were considered misclassifications. Misclassifications varied throughout the years, revealing the absence of a clear pattern, and they were considered dependent on the year. Forest plantation, forest regeneration, and grassland were the main sources of misclassifications. Despite errors, GFC accounted for significantly more deforestation than the official Brazilian mapping of the Atlantic Forest (SOSMA), showing that many small forest fragments that are important for maintaining biodiversity are being lost with little or no concern about it. We suggest that instead of sustaining an entire national programme that relies on acquiring and processing imagery, it would be better to work alongside a global programme such as the GFC by providing visual inspection, cross-validation, local knowledge, and further improvements.


This work was supported by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior.