59243

The potential of landscape metrics for assessing the impacts of selective logging in the Brazilian Amazon

Favoritar este trabalho

This paper describes the application of spatial pattern analysis to assess the impacts of selective logging in the Brazilian Amazon based on forest/non-forest maps produced from a combination of object-based and pixel-based image classification approaches. Our test area is located in the central-northern part of Mato Grosso State, which has been heavily impacted by deforestation and selective logging activities. The research encompassed a stepwise approach for producing two different types of forest/non-forest maps (with and without selective logging) for the years 2000 and 2015. Then we applied different landscape analysis schemes from the free software GuidosToolbox and compared the impact on results when integrating information on logging areas. Our results indicate a reduction in the Core forest areas from 85% to 56% of the total area (edge width 300 m) from 2000 to 2015 due to deforestation. The results are considerably impacted when selective logging is added to the analysis, leading to an increased reduction in Core forest areas and increased fragmentation. As a complementary analysis, we tested the potential of Sentinel-2 images for improved mapping of logging areas. Sentinel-2 images allow for a better delineation of selective logging areas and thus are more suited to represent the impact and extent of logging activity.