Mapping tree species proportions from satellite imagery using spectral–spatial deep learning

Remote sensing can be used to collect information for forests management. Previous studies have shown the great potential of multispectral satellite imagery for tree species classification. However, methods capable of mapping tree species in mixed forest stands on a large scale are lacking. This study proposes an innovative method to map the proportions of tree species from Sentinel-2 imagery. A convolution neural network (CNN) was used to quantify the per pixel basal area proportions of tree species considering the neighbouring environment (spectral-spatial deep learning). We produced a map for the entire Wallonia region. Nine species or groups of species were considered: Spruce genus, Oak genus, Beech, Douglas fir, Pine genus, Poplar genus, Larch genus, Birch genus, and remaining species. The training dataset for the CNN model was prepared using a map of forest parcels extracted from the public forest administration’s geodatabase of Wallonia. The accuracy of the predicted map covering the whole region was independently assessed using data from the regional forest inventory of Wallonia. In the case where forest inventories in the broad sense are available, i.e. georeferenced areas with proportions of tree species, this method is highly reproducible and is applicable at large scale, offering valuable potential for forest management.

BOLYN Corentin;  LEJEUNE Philippe;  MICHEZ Adrien;  LATTE Nicolas; 

2022-08-22

ELSEVIER SCIENCE INC

JRC129170

0034-4257 (online),   

https://www.sciencedirect.com/science/article/pii/S0034425722003145,    https://publications.jrc.ec.europa.eu/repository/handle/JRC129170,   

10.1016/j.rse.2022.113205 (online),