In this paper, we present a two-stage approach for characterizing the structure of Pinus sylvestris L. stands in forests of central Spain. The first stage was to delimit forest stands using eCognition and a digital canopy height model (DCHM) derived from lidar data. The polygons were then clustered (fc-means algorithm) into forest structure types based on the DCHM data within forest stands. Hypsographs of each polygon and field data validated the separability of structure types. In the study area, 112 polygons of Pinus sylvestris were segmented and classified into five forest structure types, ranging from high dense forest canopy (850 trees ha 1 and Loreys height of 17.4 m) to scarce tree coverage (60 tree ha 1 and Loreys height of 9.7 m). Our results indicate that the best variables for the definition and characterization of forest structure in these forests are the median and standard deviation (S.D.), both derived from lidar data. In these forest types, lidar median height and standard deviation (S.D.) varied from 15.8 m (S.D. of 5.6 m) to 2.6 m (S.D. of 4.5 m). The present approach could have an operational application in the inventory procedure and forest management plans.

The mean and standard deviation (SD) of light detection and ranging (LiDAR)-derived canopy height are related to forest structure. However, LiDAR data typically cover a limited area and have a high economic cost compared with satellite optical imagery. Optical images may be required to extrapolate LiDAR height measurements across a broad landscape. Different spectral indices were obtained from three Landsat scenes. The mean, median, SD and coefficient of variation (CV) of LiDAR canopy height measurements were calculated in 30-m square blocks corresponding with Landsat Enhanced Thematic Mapper Plus (ETM+) pixels. Correlation and forward stepwise regression analysis was applied to these data sets. Mean and median LiDAR height versus normalized difference vegetation index (NDVI), normalized difference moisture index (NDMI), normalized burn ratio (NBR) and wetness Tasseled Cap showed the best correlation coefficients (R ² ranging between −0.62 and −0.76). Nineteen regression models were obtained (R ² = 0.65–0.70). These results show that LiDAR-derived canopy height may be associated with Landsat spectral indices. This approach is of interest in sustainable forest management, although further research is required to improve accuracy.

The Lorenz curve, as a descriptor of tree size inequality within a stand, has been suggested as a reliable means for characterizing forest structure and distinguishing even from uneven-sized areas. The aim of this study was to achieve a thorough understanding on the relations between airborne laser scanning (ALS) metrics and indicators based on Lorenz curve ordering: Gini coefficient (GC) and Lorenz asymmetry (S). Exploratory multivariate analysis was carried out using correlation tests, partial least squares (PLS), and an information-theoretic approach for multimodel inference (MMI). Best subset linear model was selected for GC and S prediction, as variable transformations yielded no improvement in the relation of ALS with the given response. Relative variable importance based on the MMI model showed that GC is best predicted by ALS metrics expressing canopy coverage, return dispersion, and low–high percentile combinations. Although ALS metrics showed no correlation with S, they did so against its consti... L'utilisation de la courbe de Lorenz en tant que descripteur de l'inégalité de la taille des arbres à l'intérieur d'un peuplement a été suggérée comme moyen fiable pour caractériser la structure forestière et pour distinguer entre les zones d'arbres de taille égale et inégale. L'objectif de cette étude est d'acquérir une connaissance approfondie des relations entre les données SLA (scanneur laser aéroporté) et les indicateurs basés sur la distribution de la courbe de Lorenz: le coefficient de Gini (CG) et l'asymétrie de Lorenz (S). Une analyse exploratoire multivariée a été réalisée à l'aide de tests de corrélation, de la méthode des moindres carrés partiels (PLS) et d'une approche théorique d'information pour l'inférence multimodèle (IMM). Le modèle linéaire basé sur le meilleur sous-ensemble a été sélectionné pour la prévision du CG et de S étant donné que les transformations de variables n'ont apporté aucune amélioration dans la relation de SLA avec la réponse donnée. L'importance relative des variable...