Properties Of Individual Trees Research Articles

Large Scale Prediction with Decision Trees

This article shows that decision trees constructed with Classification and Regression Trees (CART) and C4.5 methodology are consistent for regression and classification tasks, even when the number of predictor variables grows sub-exponentially with the sample size, under natural 0-norm and 1-norm sparsity constraints. The theory applies to a wide range of models, including (ordinary or logistic) additive regression models with component functions that are continuous, of bounded variation, or, more generally, Borel measurable. Consistency holds for arbitrary joint distributions of the predictor variables, thereby accommodating continuous, discrete, and/or dependent data. Finally, we show that these qualitative properties of individual trees are inherited by Breiman’s random forests. A key step in the analysis is the establishment of an oracle inequality, which allows for a precise characterization of the goodness of fit and complexity tradeoff for a mis-specified model. Supplementary materials for this article are available online.

Quantifying vegetation indices using terrestrial laser scanning: methodological complexities and ecological insights from a Mediterranean forest

Abstract. Accurate measurement of vegetation density metrics including plant, wood and leaf area indices (PAI, WAI and LAI) is key to monitoring and modelling carbon storage and uptake in forests. Traditional passive sensor approaches, such as digital hemispherical photography (DHP), cannot separate leaf and wood material, nor individual trees, and require many assumptions in processing. Terrestrial laser scanning (TLS) data offer new opportunities to improve understanding of tree and canopy structure. Multiple methods have been developed to derive PAI and LAI from TLS data, but there is little consensus on the best approach, nor are methods benchmarked as standard. Using TLS data collected in 33 plots containing 2472 trees of 5 species in Mediterranean forests, we compare three TLS methods (lidar pulse, 2D intensity image and voxel-based) to derive PAI and compare with co-located DHP. We then separate leaf and wood in individual tree point clouds to calculate the ratio of wood to total plant area (α), a metric to correct for non-photosynthetic material in LAI estimates. We use individual tree TLS point clouds to estimate how α varies with species, tree height and stand density. We find the lidar pulse method agrees most closely with DHP, but it is limited to single-scan data, so it cannot determine individual tree properties, including α. The voxel-based method shows promise for ecological studies as it can be applied to individual tree point clouds. Using the voxel-based method, we show that species explain some variation in α; however, height and plot density were better predictors. Our findings highlight the value of TLS data to improve fundamental understanding of tree form and function as well as the importance of rigorous testing of TLS data processing methods at a time when new approaches are being rapidly developed. New algorithms need to be compared against traditional methods and existing algorithms, using common reference data. Whilst promising, our results show that metrics derived from TLS data are not yet reliably calibrated and validated to the extent they are ready to replace traditional approaches for large-scale monitoring of PAI and LAI.

Sampling forests with terrestrial laser scanning.

Background and AimsTerrestrial laser scanners (TLSs) have successfully captured various properties of individual trees and have potential to further increase the quality and efficiency of forest surveys. However, TLSs are limited to line of sight observations, and forests are complex structural environments that can occlude TLS beams and thereby cause incomplete TLS samples. We evaluate the prevalence and sources of occlusion that limit line of sight to forest stems for TLS scans, assess the impacts of TLS sample incompleteness, and evaluate sampling strategies and data analysis techniques aimed at improving sample quality and representativeness.MethodsWe use a large number of TLS scans (761), taken across a 255 650-m2 area of forest with detailed field survey data: the Harvard Forest Global Earth Observatory (ForestGEO) (MA, USA). Sets of TLS returns are matched to stem positions in the field surveys to derive TLS-observed stem sets, which are compared with two additional stem sets derived solely from the field survey data: a set of stems within a fixed range from the TLS and a set of stems based on 2-D modelling of line of sight. Stem counts and densities are compared between the stem sets, and four alternative derivations of area to correct stem densities for the effects of occlusion are evaluated. Representation of diameter at breast height and species, drawn from the field survey data, are also compared between the stem sets.Key ResultsOcclusion from non-stem sources was the major influence on TLS line of sight. Transect and point TLS samples demonstrated better representativeness of some stem properties than did plots. Deriving sampled area from TLS scans improved estimates of stem density.ConclusionsTLS sampling efforts should consider alternative sampling strategies and move towards in-progress assessment of sample quality and dynamic adaptation of sampling.

Non-forested vs forest environments: The effect of habitat conditions on host tree parameters and the occurrence of associated epiphytic lichens

The study compares the epiphytic lichen communities of two different ecological systems of northern Poland, a non-forested area and a natural lowland deciduous forest corresponding to the potential vegetation flora of a large part of Central Europe, in relation to the same tree species. The impact of habitat type and tree species on individual tree properties and the association between tree size, bark pH, conductivity, water-holding capacity, bark splitting, light intensity at tree trunks and the composition of lichens were analysed. Three sets of lichens were identified: confined to non-forested areas (40 species), confined to forests (61 species), and non-specific mutual species (53 species). The results indicate that most species demonstrate specificity to habitat type rather than to tree species. Factors related to tree species and bark properties gain in importance at the level of a specific habitat. Bark pH and water-holding capacity are decisive factors for lichen occurrence in both habitat types; additionally, in forests, tree diameter is important. The paper provides potentially useful ecological data for campaigns aimed at lichen conservation and at shaping the environment with concern for biodiversity. • Epiphytic lichens are more strongly habitat-specific than host tree-specific. • Factors related to tree bark gain in importance at the specific habitat level. • Water-holding capacity and pH of bark are decisive factors for epiphytic lichens.

Intraspecific Competition Affects Crown and Stem Characteristics of Non-Native Quercus rubra L. Stands in Germany

Accurate guidelines for silvicultural management of exotic tree species in Germany are sparse. For example, northern red oak (Quercus rubra L.) is the most commonly planted exotic deciduous tree species in Germany, but its response to varying levels of competition intensity has not yet been adequately explored. Here, we used terrestrial laser scanning to non-destructively examine the responses of stem and crown characteristics of Quercus rubra to intraspecific competition. A total of 100 dominant red oak trees were investigated in ten pure red oak stands, located in five federal states of Germany. The external stem quality characteristics namely stem non-circularity and bark anomalies decreased with increasing tree competition. Also, the crown characteristics crown volume, crown surface area, maximum crown area, crown length, and branch length declined by the degree of individual tree competition. We conclude that individual tree properties can be controlled by competition intensity, resulting in improved timber quality as shown for other tree species.

DETERMINING MORPHOMETRIC PROPERTIES OF RADIATA PINE USING LONG WAVE INFRARED SENSING AND BIOLOGICALLY-INSPIRED VISION

Abstract. A miniaturised, high resolution visible and long wave infrared (LWIR) sensor was carried onboard an unmanned aerial vehicle to observe sections of radiata pine forests. The raw irradiance measurements were temporally pre-processed using a biologically-inspired vision (BIV) model to allow information within and across the images to be normalised. This permitted a larger, denser, and more tailored set of key points within the 2D image stacks to be corresponded, thereby improving 3D reconstructions of individual trees derived using structure from motion (SfM). The BIV model comprises multiple layers of processing derived from measured or assumed responses of the photoreceptor cells in the hoverfly. Its pre-processing expands the range of input signal obtained from the LWIR sensor and enhances foreground-background contrast. Morphological image processing techniques were also applied to enhance key image features before structure from motion is applied. The result allows structural properties of individual trees to be characterised in terms of their potential volume and quality; and contrasted with the point clouds obtained from the visible imagery that only depicts the tree canopies.

Projected drought effects on the demography of Ashe juniper populations inferred from remote measurements of tree canopies

Tree mortality from drought is anticipated to increase as climate change promotes more frequent or severe water limitation. Ecosystem impacts of woody mortality depend on both the number and sizes of trees that die, but a limited capacity to predict mortality risk for individual trees hinders the capacity to forecast drought effects on tree population demography and ecosystem processes. We remotely measured leaf area of living Ashe juniper trees at three savanna sites in central Texas, USA to characterize the frequency-size distribution (FSD) of juniper populations and evaluate mortality risk from drought as a function of tree size. Mortality risk of individuals was assessed from the deviation in leaf area per tree from that of a similarly sized individual with near maximal leaf area using correlations among leaf area, growth rate, and mortality measured during a prior drought. We found that the FSD of juniper trees is bell-shaped at each site. Mortality risk from drought exceeded 25% of emergent (> 4 m height) trees in savanna juniper populations, but was highest for largest trees. Mortality risk was greatest at a grazed savanna, exceeding 50% of trees with projected canopy area > 20 m2. Results imply that severe drought could kill a large fraction (18–85%) of intermediate- to large-sized Ashe juniper trees in central Texas savannas. Our analysis demonstrates a novel use of remote measurements of canopy foliation to link mortality risk from drought to the demography of Ashe juniper populations through properties of individual trees.

Tree species diversity does not compromise stem quality in major European forest types

Mixed-species forests can have higher productivity, in terms of wood volume, than monospecific forests. In addition, higher tree species richness has been found to positively correlate with multiple ecosystem services and functions. Surprisingly, stem quality as one of the most important factors regarding the economic value of forests has rarely been formally studied in diverse forests. This paper aims at investigating how tree species richness influences stem quality and which factors may drive quality development in these stands. Stem quality, understood here essentially as the suitability of a particular stem for particular end-uses, is influenced by a tree's ability to capture sufficient resources for growth and is influenced by neighbouring trees, e.g. through shading and physical crown interactions. We collected data on crown size, stem form and tree health for over 12,000 trees in 209 study plots in six European regions (Finland, Germany, Poland, Romania, Italy and Spain) within naturally diverse forests to assess the impact of tree species richness on these characteristics. Results showed that quality variability between regions, stands and individual trees was high across species. At the stand level, there was a slight tendency towards lower stem quality with increasing diversity. However, individual trees of high quality were present at all diversity levels and for all target species. Tree species richness could not be confirmed as a primary influence on stem quality at the stand level. Rather, stand and individual tree properties such as structural composition, competition, tree size and crown characteristics were identified as the main factors for stem quality development, even in mixed stands. Many of the factors identified in this study can be directly or indirectly influenced by forest management strategies tailored to produce high-quality timber in mixed-species forests. Our findings suggest that diverse stands are not inferior regarding stem quality, while at the same time being able to provide various other ecosystem services.

The relative importance of tree and stand properties in susceptibility to spruce beetle outbreak in the mid‐20th century

AbstractTree susceptibility to potentially lethal agents is determined not only by attributes of individual trees, but also by neighborhood effects at a range of scales. For example, effects of disturbances on individual trees are often contingent on the size, configuration, and other properties of neighboring trees. Wildfires can modify postfire properties of individual trees as well as of entire forest stands, both of which can affect subsequent ecological processes, including subsequent disturbances. In recent years, much has been learned about how disturbances interact, but numerous questions concerning underlying mechanisms remain unresolved. For example, the relative importance of forest properties at different spatial scales in determining how fires affect forest susceptibility to subsequent disturbances is not well understood. This study explicitly compares the relative importance of tree vs. fine‐scale neighborhood effects (e.g., stand properties at <7 m radii), on susceptibility to a 1940s' spruce beetle outbreak. Attributes of individual trees and of stand structure were spatially reconstructed at five 250‐m2 sites that were partly burned in the late 19th century and then affected by spruce beetle outbreak in the 1940s. Random Forest models and classification trees were used to compare the relative importance of variables for susceptibility to spruce beetle attack. Individual tree properties (diameter at breast height and age) were the most important predictors of susceptibility to the outbreak across all sites combined and at each of the sites individually. In contrast, neighborhood effects were poor predictors of susceptibility. This study suggests wildfires reduce susceptibility to outbreaks primarily by reducing the size of postfire live trees and only secondarily by modifying stand structure. One implication of this is that management strategies that aim to modify stand structure over large areas in order to reduce susceptibility to spruce beetle outbreaks may be unnecessarily intensive.

Individual tree crown delineation using localized contour tree method and airborne LiDAR data in coniferous forests

Individual tree crown delineation is of great importance for forest inventory and management. The increasing availability of high-resolution airborne light detection and ranging (LiDAR) data makes it possible to delineate the crown structure of individual trees and deduce their geometric properties with high accuracy. In this study, we developed an automated segmentation method that is able to fully utilize high-resolution LiDAR data for detecting, extracting, and characterizing individual tree crowns with a multitude of geometric and topological properties. The proposed approach captures topological structure of forest and quantifies topological relationships of tree crowns by using a graph theory-based localized contour tree method, and finally segments individual tree crowns by analogy of recognizing hills from a topographic map. This approach consists of five key technical components: (1) derivation of canopy height model from airborne LiDAR data; (2) generation of contours based on the canopy height model; (3) extraction of hierarchical structures of tree crowns using the localized contour tree method; (4) delineation of individual tree crowns by segmenting hierarchical crown structure; and (5) calculation of geometric and topological properties of individual trees. We applied our new method to the Medicine Bow National Forest in the southwest of Laramie, Wyoming and the HJ Andrews Experimental Forest in the central portion of the Cascade Range of Oregon, U.S. The results reveal that the overall accuracy of individual tree crown delineation for the two study areas achieved 94.21% and 75.07%, respectively. Our method holds great potential for segmenting individual tree crowns under various forest conditions. Furthermore, the geometric and topological attributes derived from our method provide comprehensive and essential information for forest management.

Parameterization of Aerodynamic Roughness Length and Zero Plane Displacement Over Tropical Region Using Airborne LiDAR Data

Airborne LiDAR data has been one of the reliable data for individual tree properties estimation. High density airborne LiDAR data has been used previously for detailed reconstruction of tree geometry. The aim of this study is to estimate aerodynamic roughness over specific height (Zo/H) and zero plane displacement (do) over forest area using airborne LiDAR data. The results of this study will be very useful as a main guideline for related applications to understand the role of carbon and hydrological cycles, land cover and land use change, habitat fragmentation, and biogeographical modeling. The airborne LiDAR data is first classified into ground and non-ground classes. The ground points are interpolated for digital terrain model (DTM) generation and the non-ground points are used to generate digital surface model (DSM). Canopy height model (CHM) is then generated by subtracting DTM from DSM. Individual tree delineation is carried out on the CHM and individual tree height is used together with allometric equation in estimating height to crown base (HCB) and diameter at breast height (DBH). Tree crown delineation is carried out using the Inverse Watershed segmentation approach. Crown diameter, HBC and DBH are used to estimate individual tree frontal area and the total frontal area over a specific ground surface is further calculated by subtracting the intersected crowns and trunks from the total area of tree crowns and trunks. The considered ground area i.e. plants area determined the final spatial resolution of the Zo/H and do. Both parameters are calculated for different wind directions that were assumed to be originated from North/South and East/West. The results show that the estimated Zo/H and do have similar pattern and values with previous studies over vegetated area.

Detection of 3-D Individual Trees in Urban Areas by Combining Airborne LiDAR Data and Imagery

An automated approach to extracting 3-D individual trees in urban areas is developed based on jointly analyzing airborne LiDAR data and imagery. First, the spectral, geometric, and spatial context attributes are defined and integrated at the LiDAR point level. Then, a binary AdaBoost classifier is used to separate points belonging to trees from other urban objects. Once the classification is completed, a spectral clustering method by applying the normalized cuts to a graph structure of point clouds of the vegetation class is performed to segment single trees. The geometric and spectral attributes play an important role in establishing the weight matrix, which measures the similarity between every two graph nodes and determines the cut function. The performance of the approach is validated by real urban data sets, which were acquired over two European cities. The results show that 3-D individual trees can be detected with mean accuracy of up to 0.65 and 0.12 m for tree position and height. Based on the results of this work, geometric and biophysical properties of individual trees can be further retrieved.

Simulation of stand transpiration based on a xylem water flow model for individual trees

Quantifying the water exchange between a forest stand and the atmosphere is of major interest for the prediction of future growth conditions and the planning of silvicultural treatments. In the present study, we address (i) the uncertainties of sap flow estimations at the tree level and (ii) the performance of the simulation of stand transpiration. Terrestrial laser scan images (TLS) of a mature beech stand (Fagus sylvatica L.) in Southwestern Germany serve as input data for a representation of the aboveground tree architecture of the study stand. In the single-tree xylem water flow model (XWF) used here, 98 beech trees are represented by 3D graphs of connected cylinders with explicit orientation and size. Beech-specific hydraulic parameters and physical properties of individual trees determine the physiological response of the tree model to environmental conditions.The XWF simulations are performed without further calibration to sap flow measurements. The simulations reliably match up with sap flow estimates derived from sap flow density measurements. The density measurements strongly depend on individual sapwood area estimates and the characterization of radial sap flow density gradients with xylem depth. Although the observed pure beech stand is even-aged, we observe a high variability in sap flow rates among the individual trees. Simulations of the individual sap flow rates show a corresponding variability due to the distribution of the crown projection area in the canopy and the different proportions of sapwood area.Stand transpiration is obtained by taking the sum of 98 single-tree simulations and the corresponding sap flow estimations, which are then compared with the stand-level root water uptake model (RWU model) simulation. Using the RWU model results in a 35% higher simulation of seasonal stand transpiration relative to the XWF model. These findings demonstrate the importance of individual tree dimensions and stand heterogeneity assessments in estimating stand water use. As a consequence of species-specific model parameterization and precise TLS-based stand characterization, the XWF model is applicable to various sites and tree species and is a promising tool for predicting the possible water supply limitations of pure and mixed forest stands.

Estimation of stem attributes using a combination of terrestrial and airborne laser scanning

Properties of individual trees can be estimated from airborne laser scanning (ALS) data provided that the scanning is dense enough and the positions of field-measured trees are available as training data. However, such detailed manual field measurements are laborious. This paper presents new methods to use terrestrial laser scanning (TLS) for automatic measurements of tree stems and to further link these ground measurements to ALS data analyzed at the single tree level. The methods have been validated in six 80 × 80 m field plots in spruce-dominated forest (lat. 58°N, long. 13°E). In a first step, individual tree stems were automatically detected from TLS data. The root mean square error (RMSE) for DBH was 38.0 mm (13.1 %), and the bias was 1.6 mm (0.5 %). In a second step, trees detected from the TLS data were automatically co-registered and linked with the corresponding trees detected from the ALS data. In a third step, tree level regression models were created for stem attributes derived from the TLS data using independent variables derived from trees detected from the ALS data. Leave-one-out cross-validation for one field plot at a time provided an RMSE for tree level ALS estimates trained with TLS data of 46.0 mm (15.4 %) for DBH, 9.4 dm (3.7 %) for tree height, and 197.4 dm3 (34.0 %) for stem volume, which was nearly as accurate as when data from manual field inventory were used for training.

Effects of different sensors and leaf-on and leaf-off canopy conditions on echo distributions and individual tree properties derived from airborne laser scanning

The objectives of this study were to quantify and analyze differences in laser height and laser intensity distributions of individual trees obtained from airborne laser scanner (ALS) data for different canopy conditions (leaf-on vs. leaf-off) and sensors. It was also assessed how estimated tree height, stem diameter, and tree species were influenced by these differences. The study was based on 412 trees from a boreal forest reserve in Norway. Three different ALS acquisitions were carried out. Leaf-on and leaf-off data were acquired with the Optech ALTM 3100 sensor, and an additional leaf-on dataset was acquired using the Optech ALTM 1233 sensor. Laser echoes located within the vertical projection of the tree crowns were attributed to different echo categories (“first echoes of many”, “single echoes”, “last echoes of many”) and analyzed. The most pronounced changes in laser height distribution from leaf-on to leaf-off were found for the echo categories denoted as “single” and “last echoes of many” where the distributions were shifted towards the ground under leaf-off conditions. The most pronounced change in the intensity distribution was found for “first echoes of many” where the distribution was extremely skewed towards the lower values under leaf-off conditions compared to leaf-on. Furthermore, the echo height and intensity distributions obtained for the two different sensors also differed significantly. Individual tree properties were estimated fairly accurately in all acquisitions with RMSE ranging from 0.76 to 0.84 m for tree height and from 3.10 to 3.17 cm for stem diameter. It was revealed that tree species was an important model term in both and tree height and stem diameter models. A significantly higher overall accuracy of tree species classification was obtained using the leaf-off acquisition (90 vs. 98%) whereas classification accuracy did not differ much between sensors (90 vs. 93%).

Do vine species in neotropical forests see the forest or the trees?

AbstractQuestions: Is the occurrence of vine species in neotropical rain forests primarily determined by properties of the forest (environmental factors), by properties of the trees (tree species or tree size) or are vines randomly distributed?Location: Maya Biosphere Reserve, Guatemala.Methods: In five 1‐ha plots that span variation from unlogged forest to forest impacted by recurrent human disturbance we recorded the presence of all climbing vine species on every tree. The presence of all free standing vine species and 11 environmental variables were recorded in 100‐m2 subplots. The relationship of host tree diameter and host tree identity on single tree vine species richness was investigated by GLM modelling. Partial redundancy analyses were used to partition the variation in vine species composition on two sources: environmental factors and tree species identity.Results: Single tree vine richness increased with increasing host tree DBH and differed significantly among host species. For climbing vines, the ratio of variation in subplot presence explained by tree species and by environmental variables was ca. 4:1 (in the most disturbed logged plots slightly lower), for free standing vines this ratio varied from 1:2 in the most disturbed logged plots to 9:1 in reserve plots, while a ratio of ca. 1:1 was found for all plots analysed together.Conclusion: Different tree species have different probabilities of being infested by vines. Vines see both the forest and the trees; the environment is more important in earlier developmental stages, properties of individual trees become more important from the time vines start to climb.

Do vine species in neotropical forests see the forest or the trees?

Questions: Is the occurrence of vine species in neotropical rain forests primarily determined by properties of the forest (environmental factors), by properties of the trees (tree species or tree size) or are vines randomly distributed? Location: Maya Biosphere Reserve, Guatemala. Methods: In five 1-ha plots that span variation from unlogged forest to forest impacted by recurrent human disturbance we recorded the presence of all climbing vine species on every tree. The presence of all free standing vine species and 11 environmental variables were recorded in 100-m2 subplots. The relationship of host tree diameter and host tree identity on single tree vine species richness was investigated by GLM modelling. Partial redundancy analyses were used to partition the variation in vine species composition on two sources: environmental factors and tree species identity. Results: Single tree vine richness increased with increasing host tree DBH and differed significantly among host species. For climbing vines, the ratio of variation in subplot presence explained by tree species and by environmental variables was ca. 4:1 (in the most disturbed logged plots slightly lower), for free standing vines this ratio varied from 1:2 in the most disturbed logged plots to 9:1 in reserve plots, while a ratio of ca. 1:1 was found for all plots analysed together. Conclusion: Different tree species have different probabilities of being infested by vines. Vines see both the forest and the trees; the environment is more important in earlier developmental stages, properties of individual trees become more important from the time vines start to climb.

Storm disturbance in forest ecosystems in Estonia

Several storms have damaged Estonian forests in recent years. Individual tree properties such as diameter at breast height (dbh) and tree height affect the type of damage (stem breakage or uprooting) and influence the formation of postdisturbance forest structure. The aim of this study was to analyse how the risk of uprooting and stem breakage varied with respect to tree diameter, height, species and soil conditions. The data were collected from plots located in storm-damaged forests in eastern Estonia. The probability of uprooting increased with increasing diameter for Norway spruce, European aspen and birch. Black alders with large stem diameter usually suffered stem breakage. A high proportion of stem breakage also occurred among small Norway spruce in the understorey, probably damaged by falling canopy trees. Soil conditions influenced the ratio between uprooting and stem breakage of Norway spruce, the ratio being much higher on gley soils than on podzolic soils. The root plate size of Norway spruce was only slightly correlated with dbh, tree height and volume.