High Density LiDAR Data Research Articles

Evaluating the applicability of high-density UAV LiDAR data for monitoring tundra grassland vegetation

ABSTRACT Grasslands are one of the most widespread biomes on Earth. The interaction of UAV LiDAR with grasses and herbaceous plants is an infrequently covered area of research, apart from its utilisation for the estimation of above ground biomass. To evaluate the applicability of LiDAR for monitoring grassland vegetation, two model plots in the arctic-alpine tundra (Krkonoše Mountains, Czech Republic) were selected. Throughout the growing season, UAV LiDAR point clouds (800 points/m2) and multispectral imagery (9 cm) were acquired in monthly intervals, along with reference botanical and terrestrial LiDAR data. The study provides insight into the analysis of compact low-lying vegetation at the species level. A set of experiments was conducted focusing on the analysis of LiDAR information loss, vertical strata, and structural metrics computed over the grass species/communities. Random forest was used to determine the importance of metrics by out-of-bag permutation of predictors and to classify vegetation species using UAV LiDAR-based metrics, as well as image-based digital surface models alone and in fusion with multispectral data. The vertical distribution of the UAV LiDAR points varied significantly between species and throughout the growing season. Loss at the canopy bottoms was apparent, with the lowest points corresponding to dry grass matter rather than relief. Grasslands had the highest penetration capability at the start of the growing season. In terms of metrics, maximum canopy height was the most important. The multitemporal LiDAR-derived structural metrics were able to differentiate (F-1 score above 90%) all the shrubs/trees and dominant grass Calamagrostis villosa. Mixed and low-abundance species were indistinguishable. The overall accuracy scores in a 9 (27) cm grid reached 75.1% (78.1) and 63.5% (66.4) for Bílá louka meadow and Úpské rašeliniště bog, respectively. Fusing the LiDAR-derived features with multispectral imagery did not enhance the classification results apart from the delineation of shrubs and trees.

POINT-WISE CLASSIFICATION OF HIGH-DENSITY UAV-LIDAR DATA USING GRADIENT BOOSTING MACHINES

Abstract. Point-wise classification of 3D point clouds is a challenging task in point cloud processing, whereas, in particular, its application to high-density point clouds needs special attention because a large number of point clouds affect computational efficiency negatively. Although deep learning based models have been gaining popularity in recent years and have reached state-of-the-art results in accuracy for point-wise classification, their requirements of the high number of training samples and computational resources make those models inefficient for high-density 3D point clouds. However, traditional machine learning classifiers require less training samples, so they are capable of reducing computational requirements, even considering the latest machine learning classifiers, particularly in ensemble learning of gradient boosting machines, the results can compete with deep learning models. In this study, we are studying the point-wise classification of high-density UAV LiDAR data and focusing on efficient feature extraction and a recent state-of-the-art gradient boosting machine learning classifier, LightGBM. Our proposed framework includes the following steps: at first, we are using point cloud sampling for creating sub-sampled point clouds, then we are calculating the features based on those scales implemented on GPU. Finally, we are using the LightGBM classifier for training and testing. For the evaluation of our framework, we used a publicly available benchmark dataset, Hessigheim 3D. According to the results, we achieved an overall accuracy of 87.59% and an average F1 score of 75.92%. Our framework has promising results and scores closer to deep learning models. However, more distinctive features are required to obtain more accurate results.

Updating forest road networks using single photon LiDAR in northern Forest environments

Abstract Knowledge about the condition and location of forest roads is important for forest management. Coupling accurate forest road information with planning and conservation strategies supports forest resource management. In Canada, spatial data of forestry road networks are available provincially; however, they lack spatial accuracy, and up-to-date information on key attributes such as road width is missing. In this study, we apply a novel approach to update forest road networks and characterize road conditions in Ontario’s Boreal and Great Lakes—St. Lawrence (GLSL) Forest regions. We use airborne laser scanning (ALS), to facilitate the identification of forest roads across densely forested landscapes. We categorized roads into four classes based on driveable width, edge vegetation, as well as surface and edge degradation as derived from high-density Single Photon LiDAR (SPL) data. Using a novel road extraction method, we produced a road probability raster and map road centerlines. We validated road location and attribute information using Global Navigation Satellite System (GNSS) ground truth data in two Ontario forest management units, in the boreal forest and the GLSL. Road segments in some regions have been altered to account for land cover changes, such as flooding or fallen trees. In other situations, the road path may deviate from the planned layout of the road, which is not always followed in the field. Our results highlight inaccuracies in the existing road networks, with 30 per cent of ‘Full access’ roads and 29 per cent of ‘Partial access’ roads being undriveable by standard vehicles and 45 per cent of ‘Status unknown’ roads, which make up 48 per cent of the pre-existing network, being driveable by standard vehicles. Results show that the average positional accuracy of updated road centerlines is 0.4 m, and the average road width error is 2 m. The production of spatially accurate forest road networks and road attribute information is important for characterizing large road networks for which often minimal information is available.

Evaluating the Combined Use of the NDVI and High-Density Lidar Data to Assess the Natural Regeneration of P. pinaster after a High-Severity Fire in NW Spain

Pinus pinaster Ait. is an important timber species in NW Spain and is affected by forest fires every year. The persistence of this species after fire mainly depends on natural regeneration, which is very variable. In this study, we evaluated the combined use of the NDVI and LiDAR data for assessing P. pinaster regeneration success after fire in terms of density, cover and height. For this purpose, we selected a P. pinaster stand affected by a high-severity wildfire in October 2017. Field surveys and remotely piloted aircraft flights (with a high-density LiDAR sensor and multispectral camera) were conducted four years after the fire (October 2021). The study area is characterized as being particularly complex terrain, with a combination of pine trees and a high density of scrub and low vegetation. Field measurements were made in 16 study plots distributed over the burned area. Two different types of software and data processing methods were used to calculate the LiDAR-derived metrics. For pine variables, the LiDAR-based estimates of structural characteristics calculated with both data processing methods proved inadequate and were very poorly correlated with the field-measured data, while for shrubland the estimates proved to be more comparable to the field measurements. The inability of the laser pulses to reach the ground due to the complexity of the area/vegetation could lead to loss of information, calling into question the accuracy of LiDAR data in this type of scenario. LiDAR technology continues to expand in different areas and applications, and in forestry, future studies should focus on application in more complex terrain.

Evaluation of Forest Features Determining GNSS Positioning Accuracy of a Novel Low-Cost, Mobile RTK System Using LiDAR and TreeNet

Accurate positioning is one of the main components and challenges for precision forestry. This study was established to test the feasibility of a low-cost GNSS receiver, u-blox ZED-F9P, in movable RTK mode with features that determine its positioning accuracy following logging trails in the forest environment. The accuracy of the low-cost receiver was controlled via a geodetic-grade receiver and high-density LiDAR data. The features of nearby logging trails were extracted from the LiDAR data in three main categories: tree characteristics; ground-surface conditions; and crown-surface conditions. An object-based TreeNet approach was used to explore the influential features of the receiver’s positioning accuracy. The results of the TreeNet model indicated that tree height, ground elevation, aspect, canopy-surface elevation, and tree density were the top influencing features. The partial dependence plots showed that tree height above 14 m, ground elevation above 134 m, western direction, canopy-surface elevation above 138 m, and tree density above 30% significantly increased positioning errors by the low-cost receiver over southern Finland. Overall, the low-cost receiver showed high performance in acquiring reliable and consistent positions, when integrated with LiDAR data. The system has a strong potential for navigating machinery in the pathway of precision harvesting in commercial forests.

Using UAV Survey, High-Density LiDAR Data and Automated Relief Analysis for Habitation Practices Characterization during the Late Bronze Age in NE Romania

The characterization of prehistoric human behavior in terms of habitation practices using GIS cartography methods is an important aspect of any modern geoarchaeological approach. Furthermore, using unmanned aerial vehicle (UAV) surveys to identify archaeological sites with temporal resolution during the spring agro-technical works and automated mapping of the geomorphological features based on LiDAR-derived DEM can provide valuable information about the human–landscape relationships and lead to accurate archaeological and cartographic products. In this study, we applied a GIS-based landform classification method to relief characterization of 362 Late Bronze Age (LBA) settlements belonging to Noua Culture (NC) (cal. 1500/1450-1100 BCE) located in the Jijia catchment (NE Romania). For this purpose, we used an adapted version of Topographic Position Index (TPI) methodology, abbreviated DEV, which consists of: (1) application of standard deviation of TPI for the mean elevation (DEV) around each analyzed LBA site (1000 m buffer zone); (2) classification of the archaeological site’s location using six slope position classes (first method), or ten morphological classes by combining the parameters from two small-DEV and large-DEV neighborhood sizes (second method). The results indicate that the populations belonging to Noua Culture preferred to place their settlements on hilltops but close to the steep slope and on the small hills/local ridges in large valleys. From a geoarchaeological perspective, the outcomes indicate a close connection between occupied landform patterns and habitation practices during the Late Bronze Age and contribute to archaeological predictive modelling in the Jijia catchment (NE Romania).

Applying High-Resolution UAV-LiDAR and Quantitative Structure Modelling for Estimating Tree Attributes in a Crop-Livestock-Forest System

Individual tree attributes, such as stem volume and biomass, are usually predicted by using traditional field-derived allometric models. However, these models are derived from data collected from small areas and lack a level of detail of tree components (e.g., stem, branches, and leaves). Remote sensing techniques such as the Quantitative Structure Modelling (QSM) applied on high-density LiDAR data emerge as a promising solution for obtaining extensive and detailed tree attribute estimates. We used a high-density LiDAR data on board of a Unmanned Aerial Vehicle (UAV) to evaluate the performance of the QSM approach in estimating field-derived individual tree attributes such as the diameter at breast height (dbh), tree height (ht), and volume (v), as well as the stem (SAGB), branch (BAGB), and total (TAGB) aboveground biomass of eucalyptus trees. QSM was used in two different approaches: (i) using dbh and h derived from QSM and then applied into the field-based equations for estimation of volume and (ii) deriving tree volume directly from QSM. In general, all fitted models using the QSM approach were satisfactory, but with a slight tendency of over-estimation of dbh (9.33%), ht (12.40%), v-QSM1 (26.35%), v-QSM2 (26.66%), TAGB (27.08%), SAGB (25.57%), and BAGB (20.08%). Non-significant differences were noticed when estimating the dbh, tree volume, stem, and aboveground biomass. Despite the overestimation, this study indicates that using the QSM approach to estimate individual tree attributes from UAV-LiDAR is a promising alternative to support the decision-making process regarding forest management activities, especially when considering tree architecture and biomass components.

Applying a Robust Empirical Method for Comparing Repeated LiDAR Data with Different Point Density

A key aspect of vegetation monitoring from LiDAR is concerned with the use of comparable data acquired from multitemporal surveys and from different sensors. Accurate digital elevation models (DEMs) to derive vegetation products, are required to make comparisons among repeated LiDAR data. Here, we aimed to apply an improved empirical method based on DEMs of difference, that adjust the ground elevation of a low-density LiDAR dataset to that of a high-density LiDAR one for ensuring credible vegetation changes. The study areas are a collection of six sites over the Sierra de Gredos in Central Spain. The methodology consisted of producing “the best DEM of difference” between low- and high-density LiDAR data (using the classification filter, the interpolation method and the spatial resolution with the lowest vertical error) to generate a local “pseudo-geoid” (i.e., continuous surfaces of elevation differences) that was used to correct raw low-density LiDAR ground points. The vertical error of DEMs was estimated by the 50th percentile (P50), the normalized median absolute deviation (NMAD) and the root mean square error (RMSE) of elevation differences. In addition, we analyzed the effects of site-properties (elevation, slope, vegetation height and distance to the nearest geoid point) on DEMs accuracy. Finally, we assessed if vegetation height changes were related to the ground elevation differences between low- and high-density LiDAR datasets. Before correction and aggregating by sites, the vertical error of DEMs ranged from 0.02 to −2.09 m (P50), from 0.39 to 0.85 m (NMDA) and from 0.54 to 2.5 m (RMSE). The segmented-based filter algorithm (CSF) showed the highest error, but there were not significant differences among interpolation methods or spatial resolutions. After correction and aggregating by sites, the vertical error of DEMs dropped significantly: from −0.004 to −0.016 m (P50), from 0.10 to 0.06 m (NMDA) and from 0.28 to 0.46 m (RMSE); and the CSF filter algorithm continued showing the greatest vertical error. The terrain slope and the distance to the nearest geoid point were the most important variables for explaining vertical accuracy. After corrections, changes in vegetation height were decoupled from vertical errors of DEMs. This work showed that using continuous surfaces with the lowest elevation differences (i.e., the best DEM of difference) the raw elevation of low-density LiDAR was better adjusted to that of a benchmark for being adapted to site-specific conditions. This method improved the vertical accuracy of low-density LiDAR elevation data, minimizing the random nature of vertical errors and decoupling vegetation changes from those errors.

A Triangulation-Based Technique for Tree-Top Detection in Heterogeneous Forest Structures Using High Density LiDAR Data

A Triangulation-Based Technique for Tree-Top Detection in Heterogeneous Forest Structures Using High Density LiDAR Data

Estimation of Vertical Fuel Layers in Tree Crowns Using High Density LiDAR Data

The accurate prediction and mitigation of wildfire behaviour relies on accurate estimations of forest canopy fuels. New techniques to collect LiDAR point clouds from remotely piloted aerial systems (RPAS) allow for the prediction of forest fuels at extremely fine scales. This study uses a new method to examine the ability of such point clouds to characterize the vertical arrangement and volume of crown fuels from within individual trees. This method uses the density and vertical arrangement of LiDAR points to automatically extract and measure the dimensions of each cluster of vertical fuel. The amount and dimensions of these extracted clusters were compared against manually measured clusters that were collected through the manual measurement of over 100 trees. This validation dataset was composed of manual point cloud measurements for all portions of living crown fuel for each tree. The point clouds used for this were ground-based LiDAR point clouds that were ~80 times denser than the RPAS LiDAR point clouds. Over 96% of the extracted clusters were successfully matched to a manually measured cluster, representing ~97% of the extracted volume. A smaller percentage of the manually measured clusters (~79%) were matched to an extracted cluster, although these represented ~99% of the total measured volume. The vertical arrangement and dimensions of the matched clusters corresponded strongly to one another, although the automated method generally overpredicted each cluster’s lower boundary. Tree-level volumes and crown width were, respectively, predicted with R-squared values of 0.9111 and 0.7984 and RMSE values of 44.36 m2 and 0.53 m. Weaker relationships were observed for tree-level metrics that relied on the extraction of lower crown features (live crown length, live crown base height, lowest live branch height). These metrics were predicted with R-squared values of 0.5568, 0.3120, and 0.2011 and RMSE values of 3.53 m, 3.55 m, and 3.66 m. Overall, this study highlights strengths and weaknesses of the developed method and the utility of RPAS LiDAR point clouds relative to ground-based point clouds.

The Penetration Analysis of Airborne Ku-Band Radar Versus Satellite Infrared Lidar Based on the Height and Energy Percentiles in the Boreal Forest

The paper investigates the penetration properties of an airborne Ku-band frequency modulated continuous waveform (FMCW) profiling radar named Tomoradar and a satellite near-infrared lidar into the boreal forest of Finland. We achieve the accumulative energy distributions based on the Tomoradar waveforms and the satellite lidar waveforms generated from the high-density airborne lidar data within Tomoradar footprints. By comparing two groups of the height percentiles and energy percentiles derived from the accumulative energy distributions, we evaluate the relationship of penetrations between the Ku-band microwave and near-infrared laser according to the coefficients of the determination (COD), and the root mean square errors (RMSE) of linear regression analyses. The quantitative analysis results demonstrate that the height and energy percentiles derived from Tomoradar waveforms correlate well with those from satellite lidar waveforms with the mean correlation coefficients of more than 0.78 and 0.85. The linear regression models for the height and energy percentile produce excellent fits with the mean CODs of 0.95 and 0.90 and the mean RMSEs of 1.25 m and 0.03, respectively. Less than 15% of height percentiles and 87.54% of the energy percentiles in the sixth stratum near the ground derived from Tomoradar waveforms surpass those from satellite lidar waveforms. Hence, the Ku-band microwave can penetrate deeper into the forest than the near-infrared laser at the same spatial scale. In addition, quadratic fitting models are established to describe the differences of the height percentile (DHP) and the energy percentile (DEP) to expound the canopy height and closure contributions numerically. The facts that the CODs of the DHP and DEP individually are more than 0.96 and 0.89 and the fitting residual histograms approximate to normal distributions reveal the reliabilities of the proposed fitting models. Thus, the penetration analyses are valid for the explorations on the FMCW radar applications and the data fusion of the Ku-band radar and near-infrared lidar in the forest investigations.

IDENTIFICATION OF PEACH TREE TRUNKS FROM LASER SCANNING DATA OBTAINED WITH SMALL UNMANNED AERIAL SYSTEM

Abstract. Agricultural robotics rely on digital tools and sensor integration in order to improve efficiency and sustainability of cultivations. One part of orchard inventory is the identification of a tree trunk i.e. localization and diameter determination. However, this is a challenging task, due to thin trunks, presence of leaves and low branches. In this paper we present a case study for determining these parameters using the example of peach orchard, for which a high-density LiDAR data (over 3000 points/m2) was obtained with a small unmanned aerial system (UAS) during a leafy and leafless season. We applied point thresholding by height and by components of normal vector, in order to identify points reflected from trunks. Alpha-shape algorithm was used to aggregate together points, that belong to the same trunk and their centroid determined the trunk location. Trunk diameters were calculated using two alternative approaches: the Principal Component Analysis (PCA) and circle fit. For the leafy season trunk identification is challenging. Omission errors were caused due to few reflections from trunks and commission errors occurred because of the unfiltered reflections from low branches and young twigs oriented towards the ground. All 194 trunks were identified from data collected during the leafless season. The accuracy of tree location was 0.27 m and the accuracy of diameter determination using PCA was 0.03 m.

DecHPoints: A New Tool for Improving LiDAR Data Filtering in Urban Areas

Identifying ground points from LiDAR data remains a challenge more than 2 decades after automatic filtering methods were first developed. The efficacy of filtering methods depends on both the physical characteristics of the environment and on the quality of the data used. Other limitations, affecting accessibility and usability, include the choice of filter and identification of optimal parameter values. To address these problems, the most recent filters have increased their level of complexity combining different strategies, so-called hybrid methods. In this study, two tools are proposed to improve the previous filters: a decimation tool for non-ground points and a densification process. Our main improvement is to combine these tools and a filter, in this case the Iterative Robust Interpolation Filter (IRI) (Kraus and Pfeifer in ISPRS J Photogramm Remote Sens 53(4):193–203. https://doi.org/10.1016/S0924-2716(98)00009-4, http://www.sciencedirect.com/science/article/pii/S0924271698000094, 1998), to (1) improve the filtering results in urban areas by removing buildings prior to filtering, which enables a downsizing of cells used for the selection of ground points and (2) to reduce the influence of parameters on the filtering accuracy. We used two LiDAR data sets: the reference data were acquired from the International Society of Photogrammetry and Remote Sensing (ISPRS) and the high density LiDAR data. In the first case, the results obtained are compared with those obtained in previous studies, using the metrics proposed by Sithole and Vosselman (ISPRS J Photogramm Remote Sens 59(1–2):85–101, https://doi.org/10.1016/j.isprsjprs.2004.05.004, http://www.sciencedirect.com/science/article/pii/S0924271604000140, 2004). For urban samples, the proposed hybrid method provided better results than the IRI algorithm, yielding a Kappa coefficient of 91.5%. The proposed method is one of the most accurate filters that has been tested with the ISPRS data. Finally, the results obtained on the basis of the high density LiDAR data reinforced the previous results and showed the potential usefulness of the proposed hybrid method.

Topological-based roof modeling from 3D point clouds

Automatic extraction of building roofs from remote sensing data is important for many applications including 3D city modeling, urban planning, disaster management, and simulations. In this paper, we propose an automatic workflow for roof reconstruction by polygonal models from classified high-density LIDAR data. Roof planes are initially delineated by a segmentation algorithm combining a robust Hough-based normal estimator and a region growing strategy. Then, each roof is modeled by a 2D α-shape mesh which is used to discover not only building outline but also all ridges defined by intersecting roof planes, without any geometrical calculations. The mesh directly encodes the topological relations between neighboring planes which allows us to build the final polygonal model straightforwardly. This topological approach makes our solution more simple and robust than existing methods which mostly extract the intersection lines by means of geometrical computations. Experimental results show that the proposed workflow offers a high success rate for extraction at plane level (94% completeness, 92.7% correctness, 90.8% quality) when LIDAR point density is sufficiently high.

A Local Projection-Based Approach to Individual Tree Detection and 3-D Crown Delineation in Multistoried Coniferous Forests Using High-Density Airborne LiDAR Data

Accurate crown detection and delineation of dominant and subdominant trees are crucial for accurate inventorying of forests at the individual tree level. The state-of-the-art tree detection and crown delineation methods have good performance mostly with dominant trees, whereas exhibits a reduced accuracy when dealing with subdominant trees. In this paper, we propose a novel approach to accurately detect and delineate both the dominant and subdominant tree crowns in conifer-dominated multistoried forests using small footprint high-density airborne Light Detection and Ranging data. Here, 3-D candidate cloud segments delineated using a canopy height model segmentation technique are projected onto a novel 3-D space where both the dominant and subdominant tree crowns can be accurately detected and delineated. Tree crowns are detected using 2-D features derived from the projected data. The delineation of the crown is performed at the voxel level with the help of both the 2-D features and 3-D texture information derived from the cloud segment. The texture information is modeled by using 3-D Gray Level Co-occurrence Matrix. The performance evaluation was done on a set of six circular plots for which reference data are available. The high detection and delineation accuracies obtained over the state of the art prove the performance of the proposed method.

DETERMINING GEOMETRIC PARAMETERS OF AGRICULTURAL TREES FROM LASER SCANNING DATA OBTAINED WITH UNMANNED AERIAL VEHICLE

Abstract. The estimation of dendrometric parameters has become an important issue for agriculture planning and for the efficient management of orchards. Airborne Laser Scanning (ALS) data is widely used in forestry and many algorithms for automatic estimation of dendrometric parameters of individual forest trees were developed. Unfortunately, due to significant differences between forest and fruit trees, some contradictions exist against adopting the achievements of forestry science to agricultural studies indiscriminately. In this study we present the methodology to identify individual trees in apple orchard and estimate heights of individual trees, using high-density LiDAR data (3200 points/m2) obtained with Unmanned Aerial Vehicle (UAV) equipped with Velodyne HDL32-E sensor. The processing strategy combines the alpha-shape algorithm, principal component analysis (PCA) and detection of local minima. The alpha-shape algorithm is used to separate tree rows. In order to separate trees in a single row, we detect local minima on the canopy profile and slice polygons from alpha-shape results. We successfully separated 92 % of trees in the test area. 6 % of trees in orchard were not separated from each other and 2 % were sliced into two polygons. The RMSE of tree heights determined from the point clouds compared to field measurements was equal to 0.09 m, and the correlation coefficient was equal to 0.96. The results confirm the usefulness of LiDAR data from UAV platform in orchard inventory.

Single-tree detection in high-density LiDAR data from UAV-based survey

ABSTRACTUnmanned aerial vehicle-based LiDAR survey provides very-high-density point clouds, which involve very rich information about forest detailed structure, allowing for detection of individual trees, as well as demanding high computational load. Single-tree detection is of great interest for forest management and ecology purposes, and the task is relatively well solved for forests made of single or largely dominant species, and trees having a very evident pointed shape in the upper part of the canopy (in particular conifers). Most authors proposed methods based totally or partially on search of local maxima in the canopy, which has poor performance for species that have flat or irregular upper canopy, and for mixed forests, especially where taller trees hide smaller ones. Such considerations apply in particular to Mediterranean hardwood forests. In such context, it is imperative to use the whole volume of the point cloud, however keeping computational load tractable. The authors propose the use of a methodology based on modelling the 3D-shape of the tree, which improves performance with respect to maxima-based models. A case study, performed on a hazel grove, is provided to document performance improvement on a relatively simple, but significant, case.

Ensemble classification of individual Pinus crowns from multispectral satellite imagery and airborne LiDAR

Distinguishing tree species is relevant in many contexts of remote sensing assisted forest inventory. Accurate tree species maps support management and conservation planning, pest and disease control and biomass estimation. This study evaluated the performance of applying ensemble techniques with the goal of automatically distinguishing Pinus sylvestris L. and Pinus uncinata Mill. Ex Mirb within a 1.3km2 mountainous area in Barcelonnette (France). Three modelling schemes were examined, based on: (1) high-density LiDAR data (160 returns m−2), (2) Worldview-2 multispectral imagery, and (3) Worldview-2 and LiDAR in combination. Variables related to the crown structure and height of individual trees were extracted from the normalized LiDAR point cloud at individual-tree level, after performing individual tree crown (ITC) delineation. Vegetation indices and the Haralick texture indices were derived from Worldview-2 images and served as independent spectral variables. Selection of the best predictor subset was done after a comparison of three variable selection procedures: (1) Random Forests with cross validation (AUCRFcv), (2) Akaike Information Criterion (AIC) and (3) Bayesian Information Criterion (BIC). To classify the species, 9 regression techniques were combined using ensemble models. Predictions were evaluated using cross validation and an independent dataset. Integration of datasets and models improved individual tree species classification (True Skills Statistic, TSS; from 0.67 to 0.81) over individual techniques and maintained strong predictive power (Relative Operating Characteristic, ROC=0.91). Assemblage of regression models and integration of the datasets provided more reliable species distribution maps and associated tree-scale mapping uncertainties. Our study highlights the potential of model and data assemblage at improving species classifications needed in present-day forest planning and management.

Calculation of coseismic displacement from lidar data in the 2016 Kumamoto, Japan, earthquake

Abstract. The spatial distribution of the coseismic displacements that occurred along the Futagawa fault during the 2016 Kumamoto earthquake of Mw 7.0 was estimated using airborne light detection and ranging (lidar) data. In this study, a pair of digital surface models (DSMs) obtained from the high-density lidar data before and after the mainshock on 16 April 2016 were used. A window matching search approach based on the correlation coefficient between the two DSMs was used to estimate the geodetic displacement in the near-field region. The results showed good agreements with the geodetic displacements calculated from strong-motion acceleration records and coincided with the fault line surveyed by the Geological Survey of Japan.

Maximizing the Diversity of Ensemble Random Forests for Tree Genera Classification Using High Density LiDAR Data

Recent research into improving the effectiveness of forest inventory management using airborne LiDAR data has focused on developing advanced theories in data analytics. Furthermore, supervised learning as a predictive model for classifying tree genera (and species, where possible) has been gaining popularity in order to minimize this labor-intensive task. However, bottlenecks remain that hinder the immediate adoption of supervised learning methods. With supervised classification, training samples are required for learning the parameters that govern the performance of a classifier, yet the selection of training data is often subjective and the quality of such samples is critically important. For LiDAR scanning in forest environments, the quantification of data quality is somewhat abstract, normally referring to some metric related to the completeness of individual tree crowns; however, this is not an issue that has received much attention in the literature. Intuitively the choice of training samples having varying quality will affect classification accuracy. In this paper a Diversity Index (DI) is proposed that characterizes the diversity of data quality (Qi) among selected training samples required for constructing a classification model of tree genera. The training sample is diversified in terms of data quality as opposed to the number of samples per class. The diversified training sample allows the classifier to better learn the positive and negative instances and; therefore; has a higher classification accuracy in discriminating the “unknown” class samples from the “known” samples. Our algorithm is implemented within the Random Forests base classifiers with six derived geometric features from LiDAR data. The training sample contains three tree genera (pine; poplar; and maple) and the validation samples contains four labels (pine; poplar; maple; and “unknown”). Classification accuracy improved from 72.8%; when training samples were selected randomly (with stratified sample size); to 93.8%; when samples were selected with additional criteria; and from 88.4% to 93.8% when an ensemble method was used.