TY - JOUR
T1 - Tree species classification using plant functional traits from LiDAR and hyperspectral data
AU - Shi, Yifang
AU - Skidmore, Andrew K.
AU - Wang, Tiejun
AU - Holzwarth, Stefanie
AU - Heiden, Uta
AU - Pinnel, Nicole
AU - Zhu, Xi
AU - Heurich, Marco
PY - 2018/12
Y1 - 2018/12
N2 - Plant functional traits have been extensively used to describe, rank and discriminate species according to their variability between species in classical plant taxonomy. However, the utility of plant functional traits for tree species classification from remote sensing data in natural forests has not been clearly established. In this study, we integrated three selected plant functional traits (i.e. equivalent water thickness (Cw), leaf mass per area (Cm) and leaf chlorophyll (Cab)) retrieved from hyperspectral data with hyperspectral derived spectral features and airborne LiDAR derived metrics for mapping five tree species in a natural forest in Germany. Our results showed that when plant functional traits were combined with spectral features and LiDAR metrics, an overall accuracy of 83.7% was obtained, which was statistically significantly higher than using LiDAR (65.1%) or hyperspectral (69.3%) data alone. The results of our study demonstrate that plant functional traits retrieved from hyperspectral data using radiative transfer models can be used in conjunction with hyperspectral features and LiDAR metrics to further improve individual tree species classification in a mixed temperate forest.
AB - Plant functional traits have been extensively used to describe, rank and discriminate species according to their variability between species in classical plant taxonomy. However, the utility of plant functional traits for tree species classification from remote sensing data in natural forests has not been clearly established. In this study, we integrated three selected plant functional traits (i.e. equivalent water thickness (Cw), leaf mass per area (Cm) and leaf chlorophyll (Cab)) retrieved from hyperspectral data with hyperspectral derived spectral features and airborne LiDAR derived metrics for mapping five tree species in a natural forest in Germany. Our results showed that when plant functional traits were combined with spectral features and LiDAR metrics, an overall accuracy of 83.7% was obtained, which was statistically significantly higher than using LiDAR (65.1%) or hyperspectral (69.3%) data alone. The results of our study demonstrate that plant functional traits retrieved from hyperspectral data using radiative transfer models can be used in conjunction with hyperspectral features and LiDAR metrics to further improve individual tree species classification in a mixed temperate forest.
KW - Tree species classification
KW - Plant functional traits
KW - Airborne LiDAR
KW - Airborne hyperspectral
KW - Natural forest
UR - http://www.scopus.com/inward/record.url?scp=85059451051&partnerID=8YFLogxK
U2 - 10.1016/j.jag.2018.06.018
DO - 10.1016/j.jag.2018.06.018
M3 - Article
SN - 1569-8432
VL - 73
SP - 207
EP - 219
JO - International Journal of Applied Earth Observation and Geoinformation
JF - International Journal of Applied Earth Observation and Geoinformation
ER -