. 2017 Sep;61:70–80. doi: 10.1016/j.jag.2017.05.001

Table 2.

Segmentation parameters used in this research.

Forest mask	Segmentation algorithm	Scale	Color/shape smoothness/compactness	Layers	Objectives
Base year	Multiresolution	10	0.8/0.2	Bands 3, 4, 5	Identify forest and non-forest areas
Base year	Multiresolution	10	0.5/0.5	Bands 3, 4, 5	Identify forest and non-forest areas
Other years	Chessboard	1000	–	Thematic (previous years’ forest mask)	Incorporate forest mask from previous year
	Multiresolution	50	0.9/0.1	Bands 3, 4, 5	Identify new clearings
	Multiresolution	50	0.5/0.5	Bands 3, 4, 5	Identify new clearings