Predicting stem diameters and aboveground biomass of individual trees using remote sensing data

Abstract

Airborne laser scanner (ALS) data and hyperspectral (HS) data have become standard data sources in studies related to forest ecology and environmental mapping. Nowadays it is possible to have combined acquisitions of ALS and HS data, and also to have very high point density ALS data that allows us to study forests at individual tree level. In this paper, a study on the combination of HS and ALS data with models for diameter at breast height (DBH) and aboveground biomass (AGB) prediction at individual tree crown (ITC) level is presented. ALS data were used to delineate ITCs and to estimate height and crown diameter of each ITC, whereas HS data were used to identify the tree species. The proposed approach consists in developing DBH and AGB models using field data of an area and to apply these models to areas with similar characteristics. Two datasets were considered: the first dataset is located in boreal forests and it is composed by three separate study areas (Aurskog-Høland, Hadeland, and Våler); the second dataset is located in temperate forests and it is composed by two separate study areas (Val di Sella, and Pellizzano). Aurskog-Høland and Val di Sella study areas were used only to develop species-specific and non-species specific models for DBH and AGB in the two biomes, while the other study areas were used for validation at plot level with remote sensing data. Local models were also developed using field data and ALS and HS data, and they were compared to the ones developed on Aurskog-Høland and Val di Sella study areas. The results show that the proposed approach provides high accuracies in both biomes, even if problems related to underestimation exists. Predictions with a general non-species specific model developed on one study area did not differ significantly from predictions obtained by using local models