Search

Remote sensing of individual tree species has many applications in resource management, biodiversity assessment, and conservation. Airborne remote sensing using light detection and ranging (LiDAR) and hyperspectral sensors has been used extensively to extract biophysical traits of vegetation and to detect species. However, its application for individual tree mapping remains limited due to the technical challenges of precise coalignment of images acquired from different sensors and accurately delineating individual tree crowns (ITCs). In this study, we developed a generic workflow to map tree species at ITC level from hyperspectral imagery and LiDAR data using a combination of well established and recently developed techniques. The workflow uses a nonparametric image registration approach to coalign images, a multiclass normalized graph cut method for ITC delineation, robust principal component analysis for feature extraction, and support vector machine for species classification. This workflow allows us to automatically map tree species at both pixel- and ITC-level. Experimental tests of the technique were conducted using ground data collected from a fully mapped temperate woodland in the UK. The overall accuracy of pixel-level classification was 91%, while that of ITC-level classification was 61%. The test results demonstrate the effectiveness of the approach, and in particular the use of robust principal component analysis to prune the hyperspectral dataset and reveal subtle difference among species.

The evaluation of accuracy is essential for assuring the reliability of ecological models. Usually, the accuracy of above-ground biomass (AGB) predictions obtained from remote sensing is assessed by the mean differences (MD), the root mean squared differences (RMSD), and the coefficient of determination (R^2) between observed and predicted values. In this article we propose a more thorough analysis of accuracy, including a hypothesis test to evaluate the agreement between observed and predicted values, and an assessment of the degree of overfitting to the sample employed for model training. Using the estimation of forest AGB from LIDAR and spectral sensors as a case study, we compared alternative prediction and variable selection methods using several statistical measures to evaluate their accuracy. We showed that the hypothesis tests provide an objective method to infer the statistical significance of agreement. We also observed that overfitting can be assessed by comparing the inflation in residual sums of squares experienced when carrying out a cross-validation. Our results suggest that this method may be more effective than analysing the deflation in R^2. We proved that overfitting needs to be specifically addressed since, in light of MD, RMSD and R^2 alone, predictions may apparently seem reliable even in clearly unrealistic circumstances, for instance when including too many predictor variables. Moreover, Theil’s partial inequality coefficients, which are employed to resolve the proportions of the total errors due to the unexplained variance, the slope and the bias, may become useful to detect averaging effects common in remote sensing predictions of AGB. We concluded that statistical measures of accuracy, precision and agreement are necessary but insufficient for model evaluation. We therefore advocate for incorporating evaluation measures specifically devoted to testing observed-versus-predicted fit, and to assessing the degree of overfitting.

© 2017 by the authors. We apply a spatially-implicit, allometry-based modelling approach to predict stem diameter distributions (SDDs) from low density airborne LiDAR data in a heterogeneous, temperate forest in Ontario, Canada. Using a recently published algorithm that relates the density, size, and species of individual trees to the height distribution of first returns, we estimated parameters that succinctly describe SDDs that are most consistent with each 0.25-ha LiDAR tile across a 30,000 ha forest landscape. Tests with independent validation plots showed that the diameter distribution of stems was predicted with reasonable accuracy in most cases (half of validation plots had R2 ≥ 0.75, and another 23% had 0.5 ≤ R2 < 0.75). The predicted frequency of larger stems was much better than that of small stems (8 ≤ x < 11 cm diameter), particularly small conifers. We used the predicted SDDs to calculate aboveground carbon density (ACD; RMSE = 21.4 Mg C/ha), quadratic mean diameter (RMSE = 3.64 cm), basal area (RMSE = 6.99 m2/ha) and stem number (RMSE = 272 stems/ha). The accuracy of our predictions compared favorably with previous studies that have generally been undertaken in simpler conifer-dominated forest types. We demonstrate the utility of our results to spatial forest management planning by mapping SDDs, the proportion of broadleaves, and ACD at a 0.25 ha resolution.

© 2017, Canadian Science Publishing. All rights reserved. Estimation of the Gini coefficient (GC) of tree sizes using airborne laser scanning (ALS) can provide maps of forest structure across the landscape, which can support sustainable forest management. A challenge arise s in determining the optimal spatial resolution that maximizes the stability and precision of GC estimates, which in turn depends on stand density or ALS scan density. By subsampling different plot sizes within large field plots, we evaluated the optimal spatial resolution by observing changes in GC estimation and in its correlation with ALS metrics. We found that plot size had greater effects than either stand density or ALS scan density on the relationship between GC and ALS metrics. Uncertainty in GC estimates fell as plot size increased. Correlation with ALS metrics showed convex curves with maxima at 250–450m 2 , which thus was considered the optimal plot size and, consequently, the optimal spatial resolution. By thinning the density of the ALS point cloud, we deduced that at least 3 points·m −2 were needed for reliable GC estimates. Many nationwide ALS scan densities are sparser than this, so may be unreliable for GC estimation. Ours is a simple approach for evaluating the optimal spatial resolution in remote sensing estimation of any forest attribute.

Understanding the causes of variation in plant functional traits is a central issue in ecology, particularly in the context of global change. Spectroscopy is increasingly used for rapid and non-destructive estimation of foliar traits, but few studies have evaluated its accuracy when assessing phenotypic variation in multiple traits. Working with 24 chemical and physical leaf traits of six European tree species growing on strongly contrasting soil types (i.e. deep alluvium versus nearby shallow chalk), we asked (i) whether variability in leaf traits is greater between tree species or soil type; and (ii) whether field spectroscopy is effective at predicting intraspecific variation in leaf traits as well as interspecific differences. Analysis of variance showed that inter-specific differences in traits were generally much stronger than intraspecific differences related to soil type, accounting for 25% versus 5% of total trait variation, respectively. Structural traits, phenolic defences and pigments were barely affected by soil type. In contrast, foliar concentrations of rock-derived nutrients did vary: P and K concentration were lower on chalk than alluvial soils, while Ca, Mg, B, Mn and Zn concentrations were all higher, consistent with the findings of previous ecological studies. Foliar traits were predicted from 400-2500 nm reflectance spectra collected by field spectroscopy using partial least square regression, a method that is commonly employed in chemometrics. Pigments were best modelled using reflectance data from the visible region (400 - 700 nm), whilst all other traits were best modelled using reflectance data from the shortwave infrared region (1100 - 2500 nm) region. Spectroscopy delivered accurate predictions of species-level variation in traits. However, it was ineffective at detecting intraspecific variation in rock-derived nutrients (with the notable exception of P). The explanation for this failure is that rock-derived elements do not have absorption features in the 400-2500 nm region, and their estimation is indirect, relying on elemental concentrations co-varying with structural traits that do have absorption features in that spectral region (“constellation effects”). Since the structural traits did not vary with soil type, it was impossible for our regression models to predict intraspecific variation in rock-derived nutrients via constellation effects. This study demonstrates the value of spectroscopy for rapid, non-destructive estimation of foliar traits across species, but highlights problems with predicting intraspecific variation indirectly. We discuss the implications of these findings for mapping functional traits by airborne imaging spectroscopy.

© 2020 The Authors. Biotropica published by Wiley Periodicals LLC on behalf of Association for Tropical Biology and Conservation The factors that limit the maximum height of trees, whether ecophysiological or mechanical, are the subject of longstanding debate. Here, we examine the role of mechanical stability in limiting tree height and focus on trees from the tallest tropical forests on Earth, in Sabah, Malaysian Borneo, including the recently discovered tallest tropical tree, a 100.8 m Shorea faguetiana named Menara. We use terrestrial laser scans, in situ strain gauge data and finite element simulations, to map the architecture of tall tropical trees and monitor their response to wind loading. We demonstrate that a tree's risk of breaking due to gravity or self-weight decreases with tree height and is much more strongly affected by tree architecture than by material properties. In contrast, wind damage risk increases with tree height despite the larger diameters of tall trees, resulting in a U-shaped curve of mechanical risk with tree height. Our results suggest that the relative rarity of extreme wind speeds in north Borneo may be the reason it is home to the tallest trees in the tropics. Abstract in MALAY is available with online material.