Determinação da acurácia dos polígonos de desflorestamento da base de dados do CAR no estado do Acre

Abstract

Amazonian forests are globally important for their abundance of resources as well as for their role in climate regulation. As a result, there is a need to map deforestation, which has had record increases in the last few years. There are several initiatives to monitor the deforestation, such as the Amazon Deforestation Monitoring Project (PRODES), MapBiomas, Global Forest Change (GFC) and Rural Environmental Registry (CAR) that follow distinct methodologies and, therefore, present different results from each other. The result of the monitoring of deforestation through these different databases allows us to understand the progress of this issue, by identifying more critical areas and assisting in decision making when it comes to inspections and prioritization of the most vulnerable areas for forest restoration. Thus, it is paramount to acknowledge the thematic and positional accuracy of data provided by the current monitoring bases used by the authorities. The present study aimed at evaluating the accuracy in detecting deforestation by CAR through the object-oriented analysis and to carry out a comparison between CAR, GFC, and PRODES databases. The analysis was performed for polygons of up to 10 hectares divided into 4 sized-classes: class 1 - polygons of up to 1 hectare; class 2 - polygons between 1 and 3 hectares; class 3 - polygons between 3 and 6 hectares; class 4 - polygons from 6 to 10 hectares. The STEP method was used and similarity indices for shape, theme, edge and position were generated from that. The results show that for all classes, the similarity index for the theme showed the best result whereas the edge index was the least expressive for all classes. In the comparison of the bases, CAR had superior results in all similarity indices, followed by GFC and PRODES. PRODES, in turn, outperformed GFC only in the similarity index of the theme. These results are important to support the selection of the most reliable monitoring base for carrying out projects, public policies and data dissemination. Just as it is important to identify key elements for the improvement and the enhancement of each monitoring base.