A comparison between Landsat ETM+ and MODIS/TERRA images for detecting deforestation increments in the Eastern Acre region

Abstract

In this paper we evaluated the capacity of MODIS images with 250 m spatial resolution to detect and map deforestation increments relative to the capacity of Landsat ETM+ images. This study was conducted in the state of Acre covering an area of 34,225 km². Our results show that the 250 m MODIS images are highly accurate to detected deforestation larger than 3 ha. Additionally, MODIS can be used to estimate the area subject to deforestation with increments larger than 70 ha with an error of less than 5% relative to the estimates obtained with Landsat images. Finally, we estimated that 86% of the total area deforested in the state of Acre between 2002 and 2003 can be detected and mapped with MODIS. Therefore, the 250 m-pixel MODIS images can be useful to monitor the forest changes due to deforestation in the Brazilian Amazon in more regular basis, overcoming the low temporal coverage of Landsat due to frequent cloud cover in the region.

Keywords

• Remote sensing
• deforestation
• Amazon
• MODIS
• Landsat.