Structural attributes of individual trees for identifying homogeneouspatches in a tropical rainforest.

Abstract

Mapping and monitoring tropical rainforests and quantifying their carbon stocks are important, bothfor devising strategies for their conservation and mitigating the effects of climate change. AirborneLaser Scanning (ALS) has advantages over other remote sensing techniques for describing the three-dimensional structure of forests. This study identifies forest patches using ALS-based structural attributesin a tropical rainforest in Sumatra, Indonesia. A method to group trees with similar attributes into forestpatches based on Thiessen polygons and k-medoids clustering is developed, combining the advantagesof both raster and individual tree–based methods. The structural composition of the patches could be anindicator of habitat type and quality. The patches could also be a basis for developing allometric modelsfor more accurate estimation of carbon stock than is currently possible with generalised models.