Forest Parameters Research Articles

Spatiotemporal Landscape Pattern Analyses Enhanced by an Integrated Index: A Study of the Changbai Mountain National Nature Reserve

The analysis of spatiotemporal changes of landscape patterns is of great significance for forest protection. However, the selection of landscape metrics is often subjective, and existing composite landscape metrics rarely consider the effects of spatial correlation. A more objective approach to formulating composite landscape metrics involves proper weighting that incorporates spatial structure information into integrating individual conventional metrics selected for building a composite metric. This paper proposes an integrated spatial landscape index (ISLI) based on variogram modeling and entropy weighting. It was tested through a case study, which sought to analyze spatiotemporal changes in the landscape pattern in the Changbai Mountains over 30 years based on six global land-cover products with a fine classification system at 30 m resolution (GLC_FCS30). The test results confirm: (1) spatial structure information is useful for weighting conventional landscape pattern metrics when constructing ISLI as validated by correlation analysis between the incorporated conventional metrics and their variogram ranges. In terms of the range parameters of different land cover types, broadleaf forest and needleleaf forest have much larger range values than those of other land cover types; (2) DIVISION and PLAND, two of the conventional landscape metrics considered for constructing ISLI, were assigned the greatest weights in computing ISLI for this study; and (3) ISLI values can be used to determine the dominant landscape types. For the study area, ISLI values of broadleaf forests remained the largest until 2020, indicating that forest landscape characteristics were the most prominent during that period. After 2020, the dominance of needleleaf forest gradually increased, with its ISLI value reaching a maximum of 0.91 in 2025. Therefore, the proposed ISLI not only functions as an extension and complement to conventional landscape metrics but also provides more comprehensive information concerning landscape pattern dynamics.

Comparing mobile and terrestrial laser scanning for measuring and modelling tree stem taper

Abstract Measuring and modelling the shape of tree stems is a fundamental component of forest inventory systems for both commercial and biological purposes. The change in diameter of the stem along its length (a.k.a. 'taper') is one of the most important and widely used means of predicting tree stem volume. Until recently, the options for obtaining accurate estimates of stem taper and developing stem taper models have been limited to measurements of felled trees or the use of optical dendrometers on standing live trees. Here, we tested both a tripod-mounted terrestrial laser scanner (TLS; a Focus 3D 120 of FARO Technologies, Inc., Lake Mary, FL, USA), and a mobile laser scanner (MLS; the ZEB1 of the GeoSLAM Ltd, Nottingham, UK) to measure tree diameters at various heights along the stem of 20 destructively harvested broadleaf and needleleaf species using the outer hull modelling method, for the purpose of developing individual-tree and species-specific taper models. Laser scanner specifications were a major factor determining stem taper measurement accuracy. The longer-range, low beam divergence TLS could estimate stem diameter to an average of 15.7 m above ground (about 79 per cent of the canopy height), while the shorter-range high beam divergence MLS could estimate an average of 11.5 m above ground (about 45 per cent of the canopy height). Stem taper error increased with respect to height above ground, with the TLS providing more consistent and reliable diameter measurements (root mean square error (RMSE) = 1.93 cm; 9.57 per cent) compared with the MLS (RMSE = 2.59 cm; 12.84 per cent), but both methods were nearly unbiased. We attribute ~60 per cent of the uncertainty in stem measurements to laser beam diameter and point density, showing positive and negative correlations, respectively. MLS was unable to converge on the two tested taper models but was found to be an efficient means of easily sampling diameters at breast height (DBH) and reconstructing stem maps in simple forest stands with trees greater than ~10 cm DBH. TLS provided precision stem diameter measurements that allowed for the creation of similar taper models for three out of the four study species. Future work should focus on evaluating MLS systems with improved specifications (e.g. beam divergence and range), since these instruments will likely lead to dramatic improvements in reliable estimates of forest inventory parameters, in line with the current TLS technology.

Forest fire susceptibility mapping with sensitivity and uncertainty analysis using machine learning and deep learning algorithms

In the hilly region of the Western Himalayas, forest fires play a crucial role in forest destruction and biodiversity loss. Therefore, addressing the problem of forest fires is an immediate priority to mitigate the impact of forest fires. Therefore, the objective of this research is to assess and map the susceptibility zones of forest fire in the hilly state of Uttarakhand using geospatial techniques and machine learning algorithms (MLA) and deep Learning based sensitivity and uncertainty analysis of the forest fire ignition parameters. Thirteen forest fire ignition parameters were used to map forest fire susceptibility. Further, we used five distinct MLAs, including artificial neural networks (ANN), random forest (RF), logistic regression (LR), support vector machine (SVM) and an ensemble machine learning model. The generated models were validated using the receiver operating characteristic (ROC). The result shows that the ''very high'' and ''high'' zones for forest fires cover almost 55% of the total area of Uttarakhand. Validation of the models used indicated that all models produced most significant results. The ensemble machine learning had the highest accuracy (AUC = 0.977), followed by RF (0.973) and ANN (0.972). In addition, a Deep Neural Networks (DNN) based sensitivity approach is proposed to perform sensitivity and uncertainty analysis of the ignition parameters. For this, we have selected the best predictive model (Ensemble Machine Learning) to evaluate the sensitivity and uncertainty analysis. The sensitivity analysis shows that evapotranspiration has the highly influencing parameter to forest fires, followed by annual rainfall, wind speed, distance to agriculture, distance to tourist spots, distance to built-up and distance to road. However, aspect, temperature and distance to religious centers have less influence on susceptibility to forest fires in this study area. This study is unique because it provides new insights into the construction of an ensemble machine learning model and the integration of DNN-based sensitivity and uncertainty for robust and scientific modelling of forest fire susceptibility. The contribution of parameters demonstrates a novel approach in understanding the impact of a single parameter sample on forest fire susceptibility, which will help in developing prevention strategies. A similar approach may be used in other parts of the world with similar natural and anthropogenic conditions for mapping forest fire susceptibility.

A New Strategy for Individual Tree Detection and Segmentation from Leaf-on and Leaf-off UAV-LiDAR Point Clouds Based on Automatic Detection of Seed Points

Accurate and efficient estimation of forest volume or biomass is critical for carbon cycles, forest management, and the timber industry. Individual tree detection and segmentation (ITDS) is the first and key step to ensure the accurate extraction of detailed forest structure parameters from LiDAR (light detection and ranging). However, ITDS is still a challenge to achieve using UAV-LiDAR (LiDAR from Unmanned Aerial Vehicles) in broadleaved forests due to the irregular and overlapped canopies. We developed an efficient and accurate ITDS framework for broadleaved forests based on UAV-LiDAR point clouds. It involves ITD (individual tree detection) from point clouds taken during the leaf-off season, initial ITS (individual tree segmentation) based on the seed points from ITD, and improvement of initial ITS through a refining process. The results indicate that this new proposed strategy efficiently provides accurate results for ITDS. We show the following: (1) point-cloud-based ITD methods, especially the Mean Shift, perform better for seed point selection than CHM-based (Canopy Height Model) ITD methods on the point clouds from leaf-off seasons; (2) seed points significantly improved the accuracy and efficiency of ITS algorithms; (3) the refining process using DBSCAN (density-based spatial clustering of applications with noise) and kNN (k-Nearest Neighbor classifier) classification significantly reduced edge errors in ITS results. Our study developed a novel ITDS strategy for UAV-LiDAR point clouds that demonstrates proficiency in dense deciduous broadleaved forests, and this proposed ITDS framework could be applied to single-phase point clouds instead of the multi-temporal LiDAR data in the future if the point clouds have detailed tree trunk points.

Large-Eddy Simulation of Wind Turbine Wakes in Forest Terrain

In this study, large-eddy simulation was employed to investigate the influence of the forest canopy on wind turbine wakes. Nine forest case studies were carried out with different vertical distributions of leaf area density (LAD) and values of leaf area index (LAI). It was found that the wake in forest canopies recovers at a faster rate when compared with the flat terrain. An interesting observation was the significant reduction in turbulence kinetic energy (TKE) in the lower part of the wake above the forest in comparison with the inflow TKE, which occurred for a wide range of turbine downstream positions. The increase of TKE, on the other hand, was mainly located in the region around the top tip. Analyses of the power spectral density showed that the increase in TKE happened at a certain range of frequencies for the forest canopy cases and at all the examined frequencies for the flat case. Wake meandering was also examined and was found to be of a higher amplitude and a lower dominant frequency for the forest cases compared with the flat case. In terms of the influence of forest canopy parameters, the LAI was found to have an impact greater than the vertical distribution of LAD. Specifically, the wake-added TKE and wake-added Reynolds shear stress were found to be approximately the same for cases with the same LAI, regardless of the vertical distribution of LAD.

Comparison of Errors Produced by ABA and ITC Methods for the Estimation of Forest Inventory Attributes at Stand and Tree Level in Pinus radiata Plantations in Chile

Airborne laser scanning (ALS) technology is fully implemented in forest resource assessment processes, providing highly accurate and spatially continuous results throughout the area of interest, thus reducing inventory costs when compared with traditional sampling inventories. Several approaches have been employed to estimate forest parameters using ALS data, such as the Area-Based Approach (ABA) and Individual Tree Crown (ITC). These two methodologies use different information processing and field data collection approaches; thus, it is important to have a selection criterion for the method to be used based on the expected results and admissible errors. The objective of this study was to compare the prediction errors of forest inventory attributes in the functioning of ABA and ITC approaches. A plantation of 500 ha of Pinus radiata (400–600 trees ha−1) in Chile was selected; a forest inventory was conducted using the ABA and ITC methods and the accuracy of both methods was analyzed. The ITC models performed better than the ABA models at low tree densities for all forest inventory attributes (15% MAPE in tree density—N—and 11% in volume—V). There was no significant difference in precision regarding the volume and basal area (G) estimations at medium densities, although ITC obtained better results for density and dominant height (Ho). At high densities, ABA performed better for all the attributes except for height (6.5% MAPE in N, 8.7% in G, and 8.9% in V). Our results showed that the precision of forest inventories based on ALS data can be adjusted depending on tree density to optimize the selected approach (ABA and ITC), thus reducing the inventory costs. Hence, field efforts can be greatly decreased while achieving better prediction accuracies.

Evaluating the Sensitivity of Polarimetric Features Related to Rotation Domain and Mapping Chinese Fir AGB Using Quad-Polarimetric SAR Images

Unaffected by cloud cover and solar illumination, synthetic aperture radar (SAR) images coupled with quad-polarimetric techniques have significant potential for mapping forest aboveground biomass (AGB) in the mountains of southern China. To improve the accuracy of mapping forest AGB, it is necessary to accurately interpret and evaluate the sensitivity of polarimetric features related to polarimetric response in complex forests. In this study, several rotated polarimetric features were extracted from L-band quad-polarimetric ALOS PALSAR-2 images based on uniform polarimetric matrix rotation theory. In addition, the sensitivity of rotated polarimetric features with forest parameters was evaluated by the Pearson correlation coefficient, sensitivity index (SI), and saturation levels. Ultimately, the forest AGB was mapped with various combinatorial feature sets by a proposed feature selection method based on the sensitivity index. The results illustrated that rotated polarimetric features extracted from the rotational domain have higher sensitivity with various forest parameters and higher saturation levels for mapping forests than other traditional features. After using the proposed feature selection method and combinatorial feature sets, the rRMSE of mapped forest AGB ranged from 22.5% to 33.9% for two acquired images, and the best result was obtained from the combination of three types of polarimetric features (BC + C4 + Ro). It is also confirmed that different types of features extracted from quad-polarimetric SAR images have better compensation effects and the accuracy of mapped forest AGB is significantly improved.

Comparison of GEDI LiDAR Data Capability for Forest Canopy Height Estimation over Broadleaf and Needleleaf Forests

The GEDI LiDAR system was specifically designed to detect vegetation structure and has proven to be a suitable tool for estimating forest biophysical parameters, especially canopy height, at a global scale. This study compares the GEDI relative height metric (RH100) over different forest types, especially deciduous broadleaf and evergreen coniferous located in Thuringia, Germany, to understand how the forest structural differences affect the GEDI height estimation. A canopy height model that was produced using digital terrain and surface models (DTM and DSM) derived from airborne laser scanning data is used as the reference data. Based on the result, GEDI canopy height over needleleaf forest is slightly more accurate (RMSE = 6.61 m) than that over broadleaf (RMSE = 8.30 m) and mixed (RMSE = 7.94 m) forest. Evaluation of the GEDI acquisition parameters shows that differences in beam type, sensitivity, and acquisition time do not significantly affect the accuracy of canopy heights, especially over needleleaf forests. Considering foliage condition impacts on canopy height estimation, the contrasting result is observed in the broadleaf and needleleaf forests. The GEDI dataset acquired during the winter when deciduous species shed their leaves (the so-called leaf-off dataset), outperforms the leaf-on dataset in the broadleaf forest but shows less accurate results for the needleleaf forest. Considering the effect of the plant area index (PAI) on the accuracy of the GEDI canopy height, the GEDI dataset is divided into two sets with low and high PAI values with a threshold of median PAI = 2. The results show that the low PAI dataset (median PAI < 2) corresponds to the non-growing season (autumn and winter) in the broadleaf forest. The slightly better performance of GEDI using the non-growing dataset (RMSE = 7.40 m) compared to the growing dataset (RMSE = 8.44 m) in the deciduous broadleaf forest and vice versa, the slightly better result using the growing dataset (RMSE = 6.38 m) compared to the non-growing dataset (RMSE = 7.24 m) in the evergreen needleleaf forest is in line with the results using the leaf-off/leaf-on season dataset. Although a slight improvement in canopy height estimation was observed using either the leaf-off or non-growing season dataset for broadleaf forest, and either the leaf-on or growing season dataset for needleleaf forest, the approach of filtering GEDI data based on such seasonal acquisition time is recommended when retrieving canopy height over pure stands of broadleaf or needleleaf species, and the sufficient dataset is available.

Mapping the Distribution and Dynamics of Coniferous Forests in Large Areas from 1985 to 2020 Combining Deep Learning and Google Earth Engine

Mapping the distribution of coniferous forests is of great importance to the sustainable management of forests and government decision-making. The development of remote sensing, cloud computing and deep learning has provided the support of data, computing power and algorithms for obtaining large-scale forest parameters. However, few studies have used deep learning algorithms combined with Google Earth Engine (GEE) to extract coniferous forests in large areas and the performance remains unknown. In this study, we thus propose a cloud-enabled deep-learning approach using long-time series Landsat remote sensing images to map the distribution and obtain information on the dynamics of coniferous forests over 35 years (1985–2020) in the northwest of Liaoning, China, through the combination of GEE and U2-Net. Firstly, to assess the reliability of the proposed method, the U2-Net model was compared with three Unet variants (i.e., Resnet50-Unet, Mobile-Unet and U-Net) in coniferous forest extraction. Secondly, we evaluated U2-Net’s temporal transferability of remote sensing images from Landsat-5 TM, Landsat-7 ETM+ and Landsat-8 OLI. Finally, we compared the results obtained by the proposed approach with three publicly available datasets, namely GlobeLand30-2010, GLC_FCS30-2010 and FROM_GLC30-2010. The results show that (1) the cloud-enabled deep-learning approach proposed in this paper that combines GEE and U2-Net achieves a high performance in coniferous forest extraction with an F1 score, overall accuracy (OA), precision, recall and kappa of 95.4%, 94.2%, 96.6%, 95.5% and 94.0%, respectively, outperforming the other three Unet variants; (2) the proposed model trained by the sample blocks collected from a specific time can be applied to predict the coniferous forests in different years with satisfactory precision; (3) Compared with three global land-cover products, the distribution of coniferous forests extracted by U2-Net was most similar to that of actual coniferous forests; (4) The area of coniferous forests in Northwestern Liaoning showed an upward trend in the past 35 years. The area of coniferous forests has grown from 945.64 km2 in 1985 to 6084.55 km2 in 2020 with a growth rate of 543.43%. This study indicates that the proposed approach combining GEE and U2-Net can extract coniferous forests quickly and accurately, which helps obtain dynamic information and assists scientists in developing sustainable strategies for forest management.

ICESat-2 data classification and estimation of terrain height and canopy height

ICESat-2 (Ice, Cloud, and Land Elevation Satellite-2) was launched in 2018 with a photon-counting LiDAR (Light Detection and Ranging) system, ATLAS (Advanced Topographic Laser Altimeter System). It is collecting massive earth elevation data all over the world, which has shown the potential of large-scale forest monitoring. However, the energy emitted by the LiDAR system is low, and the received signals are easily affected by noise. Accurate classification of photons is an important step for forest parameter retrieval. Given the limitations of existing photon classification algorithms in areas with complex terrain, we proposed an improved local outlier factor algorithm with a rotating search area (LOFR). First, photons are transformed to the along-track direction, and noise photons are preliminarily filtered out by using the elevation histogram and elevation statistical methods. Next, ground photons are extracted by using the LOF (Local Outlier Factor) with the horizontal ellipse search area algorithm (LOFE) during an initial classification stage to filter photons that are far away from the ground. During the refined classification stage, which is the core of the algorithm, the terrain slope is calculated according to the ground photons extracted during initial classification. The elliptic search area is then rotated to align its long axis with the terrain slope. Finally, the LOFR scores of the photons are calculated to remove noise photons and signal photons are classified into top-of-canopy photons, canopy photons, and ground photons. The results show that the algorithm can effectively classify photons. Both estimated terrain height and canopy height derived using the classified photons are in good agreement with airborne LiDAR data. The mean absolute error (MAE) of the estimated terrain height relative to airborne data was 1.45 m and the root mean square error (RMSE) was 2.82 m. For canopy height validation, the correlation coefficient (R2), MAE, and RMSE at the best study scale (80 m) were 0.86, 1.82 m, and 2.72 m, respectively. These results demonstrated that the proposed LOFR algorithm can improve photon classification over complex terrain without prior knowledge of the terrain. Therefore, it could provide a robust approach for large-scale ATLAS data processing.

Constrained optimization of objective functions determined from random forests

In this paper, we examine a data‐driven optimization approach to making optimal decisions as evaluated by a trained random forest, where these decisions can be constrained by an arbitrary polyhedral set. We model this optimization problem as a mixed‐integer linear program. We show this model can be solved to optimality efficiently using pareto‐optimal Benders cuts for ensembles containing a modest number of trees. We consider a random forest approximation that consists of sampling a subset of trees and establish that this gives rise to near‐optimal solutions by proving analytical guarantees. In particular, for axis‐aligned trees, we show that the number of trees we need to sample is sublinear in the size of the forest being approximated. Motivated by this result, we propose heuristics inspired by cross‐validation that optimize over smaller forests rather than one large forest and assess their performance on synthetic datasets. We present two case studies on a property investment problem and a jury selection problem. We show this approach performs well against other benchmarks while providing insights into the sensitivity of the algorithm's performance for different parameters of the random forest.

Retrieval of Mangrove leaf area index and its response to typhoon based on WorldView-3 image

In recent decades, mangroves have become one of the world's most threatened and vulnerable ecosystems due to human disturbance and climate change. Remote sensing techniques have helped to obtain biophysical parameters of forests on a large scale. However, to our knowledge, mangrove leaf area index (LAI) inversion using texture features and its scale dependency is rarely discussed. Whether the dynamics of remotely sensed LAI could reveal the responses of mangrove species to typhoon disturbance remain unknown. In this study, we classified mangrove species, estimated the LAI of each species based on high-resolution WorldView-3 (WV-3) images, and evaluated the performance of different object-based classifiers and the effects of texture features and scale dependency on the LAI model. We characterized typhoon disturbance on different species using interannual LAI variation. Random forest (RF) has the highest accuracy in mangrove species classification, with an overall accuracy of 82.29% after post-classification, compared with naive Bayes (NB), classification and regression trees (CART), k-nearest neighbor (KNN), and support vector machines (SVM). Texture features significantly increased the accuracy of LAI estimates, and the optimal prediction result was generated at a 2-m resampling rate (R2 = 0.52; MSE = 0.56). The mean LAIs of B. gymnorrhiza and S. apetala assemblages were significantly higher than those of K. obovata, S. caseolaris, and A. marina. LAI variation demonstrates a disturbance pattern caused by Typhoon Mangkhut. This typhoon had the greatest impact on K. obovata, while S. apetala and S. caseolaris showed good resistance to wind damage. We concluded that RF was the best classifier for mapping mangrove species distribution using object-based classification algorithms with WV-3 image. The texture features play a key role in constructing the estimation model of mangrove LAI with scale dependency.

Blows or Falls? Distinction by Random Forest Classification.

In this study, we propose a classification method between falls and blows using random forests. In total, 400 anonymized patients presenting with fractures from falls or blows aged between 20 and 49 years old were used. There were 549 types of fractures for 57 bones and 12 anatomical regions observed. We first tested various models according to the sensibility of random forest parameters and their effects on model accuracies. The best model was based on the binary coding of 12 anatomical regions or 28 bones with or without baseline (age and sex). Our method achieved the highest accuracy rate of 83% in the distinction between falls and blows. Our findings pave the way for applications to help forensic experts and archaeologists.

Protection of Coastal Shelter Forests Using UAVs: Individual Tree and Tree-Height Detection in Casuarina equisetifolia L. Forests

Casuarina equisetifolia L. plays a significant role in sandy, coastal regions for sand stabilization and windbreaks. However, C. equisetifolia forests are susceptible to plant diseases and insect pests, resulting in mortality due to pure stands and a harsh natural environment. Mapping the distribution of C. equisetifolia and detecting its height can inform forest-management decisions. Unmanned aerial vehicle (UAV) imagery, coupled with the classical detection method, can provide accurate information on tree-level forest parameters. Considering that the accuracy of a forest-parameter estimation is impacted by various flight altitudes and extraction parameters, the purpose of this study is to determine the appropriate flight altitude and extraction parameters for mapping C. equisetifolia using UAV imagery and the local maxima algorithm in order to monitor C. equisetifolia more accurately. A total of 11 different flight altitudes and 36 combinations of circular smoothing window size (CSWS) and fixed circular window size (FCWS) were tested, and 796 trees with corresponding positions in the UAV image and ground–tree heights were used as reference. The results show that the combination of a 0.1 m CSWS and a 0.8 m FCWS for individual tree detection (ITD) and tree-height detection achieved excellent accuracy (with an F1 score of 91.44% for ITD and an estimation accuracy (EA) of 79.49% for tree-height detection). A lower flight altitude did not indicate a higher accuracy for individual tree and tree-height detection. The UAV image obtained within a flight altitude of 60 m–80 m can meet the accuracy requirements for the identification of C. equisetifolia tree-height estimation (F1 score > 85% for ITD; EA > 75% for tree-height estimation). This study provides a foundation for monitoring C. equisetifolia by using UAV imagery and applying the local maxima algorithm, which may help forestry practitioners detect C. equisetifolia trees and tree heights more accurately, providing more information on C. equisetifolia growth status.

Assessing the Vertical Structure of Forests Using Airborne and Spaceborne LiDAR Data in the Austrian Alps

Vertical structure is an important parameter not only for assessment of the naturalness of a forest and several functional parameters, such as biodiversity or protection from avalanches or rockfall, but also for estimating biomass/carbon content. This study analyses the options for assessing vertical forest structure by using airborne (ALS) and spaceborne LiDAR data (GEDI) in a mountainous near-natural forest in the Austrian Alps. Use of the GEDI waveform data (L1B) is still heavily underexploited for vertical forest structure assessments. Two indicators for explaining forest vertical structure are investigated in this study: foliage height diversity (FHD) and number of layers (NoL). For estimation of NoL, two different approaches were tested: break-detection algorithm (BDA) and expert-based assessment (EBA). The results showed that FHD can be used to separate three structural classes; separability is only slightly better for ALS than for GEDI data on a 25 m diameter plot level. For NoL, EBA clearly outperformed BDA in terms of overall accuracy (OA) by almost 20%. A better OA for NoL was achieved using ALS (49.5%) rather than GEDI data (44.2%). In general, OA is limited by difficult terrain and near-natural forests with high vertical structure. The usability of waveform-based structure parameters is, nonetheless, promising and should be further tested on larger areas, including managed forests and simpler stands.

Monitoring of Monthly Height Growth of Individual Trees in a Subtropical Mixed Plantation Using UAV Data

The assessment of changes in the height growth of trees can serve as an accurate basis for the simulation of various ecological processes. However, most studies conducted on changes in the height growth of trees are on an annual scale. This makes it difficult to obtain basic data for correcting time differences in the height growth estimates of trees within a year. In this study, the digital elevation models (DEMs) were produced based on stereo images and light detection and ranging (LiDAR) data obtained by unmanned aerial vehicles (UAVs). Individual tree crowns were segmented by employing the watershed segmentation algorithm and the maximum value within each crown was extracted as the height of each tree. Subsequently, the height growth of each tree on a monthly-scale time series was extracted to simulate the time difference correction of regional tree height estimates within a year. This was used to verify the feasibility of the time difference correction method on a monthly scale. It is evident from the results that the DEM based on UAV stereo images was closely related to the DEM based on UAV LiDAR, with correlation coefficients of R2 = 0.96 and RMSE = 0.28 m. There was a close correlation between the tree height extracted from canopy height models (CHMs) based on UAV images and the measured tree height, with correlation coefficients of R2 = 0.99, and RMSE = 0.36 m. Regardless of the tree species, the total height growth in each month throughout the year was 46.53 cm. The most significant changes in the height growth of trees occurred in May (14.26 cm) and June (14.67 cm). In the case of the Liriodendron chinense tree species, the annual height growth was the highest (58.64 cm) while that of the Osmanthus fragrans tree species was the lowest (34.00 cm). By analyzing the height growth estimates of trees each month, it was concluded that there were significant differences among various tree species. In the case of the Liriodendron chinense tree species, the growth season occurred primarily from April to July. During this season, 56.92 cm of growth was recorded, which accounted for 97.08% of the annual growth. In the case of the Ficus concinna tree species, the tree height was in a state of growth during each month of the year. The changes in the height growth estimates of the tree were higher from May to August (44.24 cm of growth, accounting for 77.09% of the annual growth). After applying the time difference correction to the regional tree growth estimates, the extraction results of the changes in the height growth estimates of the tree (based on a monthly scale) were correlated with the height of the UAV image-derived tree. The correlation coefficients of R2 = 0.99 and RMSE = 0.26 m were obtained. The results demonstrate that changes in the height growth estimates on a monthly scale can be accurately determined by employing UAV stereo images. Furthermore, the results can provide basic data for the correction of the time differences in the growth of regional trees and further provide technical and methodological guidance for regional time difference correction of other forest structure parameters.

Effect of forest structure and management on the functional diversity and composition of understorey plant communities

AbstractQuestionsDo forest structural parameters related to stand heterogeneity enhance functional diversity (FD) of understorey plant communities? Do FD and functional composition of understorey plant communities vary between high‐forest (HF) and old coppices‐with‐standard (oldCWS) management types? Are HF stands characterized by a higher FD than oldCWS?LocationSubmediterranean beech forests of Montagne della Duchessa Reserve (central Italy).MethodsWe sampled 57 (20 m × 20 m) forest plots, 29 oldCWS and 28 HF stands, where we recorded plant species cover abundance. We used Leaf–Height–Seed and clonal traits, and run multiple separate linear models to quantify the effect of forest structural parameters and management types on FD and functional composition of understorey plant communities.ResultsWe found that increasing lying deadwood enhanced FD of specific leaf area (SLA) through micro‐environmental heterogeneity of soil fertility regardless of management type. We also found that increased horizontal stratification filtered the range of plant sizes, probably through light reduction. HF management enhanced FD of SLA and clonal traits through micro‐environmental heterogeneity, while reducing the FD of plant size and seed mass. HF tended to select plant communities characterized by high resource acquisition strategies but lower persistence between ramets, possibly as an effect of more mature forest conditions.ConclusionOur study suggests that understorey plant community diversity and composition change in response to forest structure and management. Combining Leaf–Height–Seed with clonal traits offers a promising framework for understanding and predicting plant response to management practices.

A Photon Cloud Filtering Method in Forested Areas Considering the Density Difference Between Canopy Photons and Ground Photons

Photon cloud data filtering is crucial when obtaining forest vertical structure parameters from photon-counting LiDAR data. The proposed method, for the first time, takes into account the influence of the density difference between canopy photons and ground photons. A moving overlapping window approach is introduced to reduce the impact of an uneven background noise environment first. In each window, a modified elevation histogram statistics vector in the elevation direction is proposed to increase the density difference between signal and noise photons while also reducing the density difference between canopy and ground photons. The filtering results show that the average overall accuracy (OA) and standard deviation of the proposed method reach almost 0.99 and 0.01, respectively, which are much better results than those of the other existing filtering methods. Specifically, with the increase in the ratio of canopy photons to ground photons, the F-measure value of the proposed method reaches almost 0.99, and is also stable, which demonstrates that the proposed approach can almost completely eliminate the influence of the density difference between canopy photons and ground photons on the filtering results. In addition, the forest canopy heights obtained based on the proposed filtering method achieve the lowest root-mean-square error (RMSE) value of 3.18 m, compared to the other filtering methods. In summary, the proposed photon cloud data filtering method can retrieve reliable forest canopy height information from photon cloud data, and outperforms the other compared filtering methods in the given test site.

Model-Based Retrieval of Forest Parameters From Sentinel-1 Coherence and Backscatter Time Series

This letter describes a model-based algorithm for estimating tree height and other bio-physical land parameters from time series of synthetic aperture radar (SAR) interferometric coherence and backscatter supported by sparse lidar data. The random-motion-over-ground model (RMoG) is extended to time series and revisited to capture the short- and long-term temporal coherence variability caused by motion of the scatterers and changes in the soil and canopy backscatter. The proposed retrieval algorithm estimates first the spatially slow-varying RMoG model parameters using sparse lidar data, and subsequently the spatially fast-varying model parameters such as tree height. The recently published global Sentinel-1 (S-1) interferometric coherence and backscatter data set and sparse spaceborne GEDI lidar data are used to illustrate the algorithm. Results obtained for a small region over Spain show that the temporal coherence and backscatter time series have the potential to be used for global, model-based land parameter estimation.

Array 3-D SAR Tomography Using Robust Gridless Compressed Sensing

Tomographic synthetic aperture radar (TomoSAR), which can provide three-dimensional (3-D) image of the observed scenes, has become an important technology for topographic mapping, forest parameter estimation, urban buildings modeling and etc. Recently, the developed compressed sensing (CS) and other similar methods have been widely applied for the achievement of super-resolution SAR tomography. However, there always exists inevitable model errors during the mining of scene information, such as discrete gridding on used dictionary and outliers among independent identically distribution (IID) samples, which tends to dramatically degrade the TomoSAR inversion. In this paper, a novel robust gridless CS (RGLCS) algorithm is proposed for high-resolution 3-D imaging of array TomoSAR. In the scheme, the atomic norm minimization (ANM) is used to model the joint-sparsity pattern on elevation distribution between adjacent pixels, which can be treated as gridless CS to avoid the discrete error of the dictionary. Meanwhile, the outliers and disturbances not satisfying the IID elevation distribution are modelled as sparsely distributed spike-noise in the image domain. The proposed RGLCS algorithm has the capability of perfectly separating the outliers and maintaining high-precision height resolution. For efficient solution, a fast alternative optimization is used to solve the objective function to effectively reduce the computational complexity. Next, the post-processing, including point cloud clustering and double-bounce scattering detection & eliminating, are studied to obtain high-resolution 3-D point cloud image. Finally, the experimental analysis using both simulated and measured data are performed to verify the effectiveness of the proposed algorithm. In particular, a practical demonstration using measured airborne array TomoSAR data is presented for urban mapping.