Search

Mapping tree crowns in arid or semi-arid areas, which cover around one-third of the Earth’s land surface, is a key methodology towards sustainable management of trees. Recent advances in deep learning have shown promising results for tree crown segmentation. However, a large amount of manually labeled data is still required. We here propose a novel method to delineate tree crowns from high resolution satellite imagery using deep learning trained with automatically generated labels from 3D radiative transfer modeling, intending to reduce human annotation significantly. The methodological steps consist of 1) simulating images with a 3D radiative transfer model, 2) image style transfer learning based on generative adversarial network (GAN) and 3) tree crown segmentation using U-net segmentation model. The delineation performances of the proposed method have been evaluated on a manually annotated dataset consisting of more than 40,000 tree crowns. Our approach, which relies solely on synthetic images, demonstrates high segmentation accuracy, with an F1 score exceeding 0.77 and an Intersection over Union (IoU) above 0.64. Particularly, it achieves impressive accuracy in extracting crown areas (r2 greater than 0.87) and crown densities (r2 greater than 0.72), comparable to that of a trained dataset with human annotations only. In this study, we demonstrated that the integration of a 3D radiative transfer model and GANs for automatically generating training labels can achieve performances comparable to human labeling, and can significantly reduce the time needed for manual labeling in remote sensing segmentation applications.

ABSTRACTThe discrepancies in data across different phases and the unexplored optimal spatial resolution present challenges when using multi-temporal canopy height models to accurately discern actual forest growth. In this study, we evaluated the reliability of bi-temporal CHMs to characterize growth changes in a boreal natural forest over a four-year period. A maximum mosaic method was introduced to construct a CHM from various flight strips, aimed at minimizing data alignment errors. Subsequently, the canopy height and height changed derived from six different height percentile metrics and five spatial resolutions were evaluated. The results showed that higher resolution (e.g., < 2 m) and lower height metrics (e.g., 85th height percentile) consistently underestimated. Tree growth correlation with individual segmentation surpassed field-measurements at all resolutions and height metrics. for the optimal resolution and height metrics, the results suggest that using a 95th height percentile the 10 m scale effectively represents both canopy height ([Formula: see text] = 0.95, [Formula: see text] = 0.88 m, [Formula: see text] = 5.83%; [Formula: see text] = 0.96, [Formula: see text] = 1.11 m, [Formula: see text] = 6.91%) and height changes ([Formula: see text] = 0.59, [Formula: see text] = 0.86 m, [Formula: see text] = 18.38%). This study demonstrates the necessity of carefully evaluating data characteristics and resolutions when employing multi-temporal CHM for forest dynamics monitoring.

Quantifying the relationship between light and stands or individual trees is of great significance in understanding tree competition, improving forest productivity, and comprehending ecological processes. However, accurately depicting the spatiotemporal variability of light under complex forest structural conditions poses a challenge, especially for precise forest management decisions that require a quantitative study of the relationship between fine-scale individual tree structure and light. 3D RTMs (3-dimensional radiative transfer models), which accurately characterize the interaction between solar radiation and detailed forest scenes, provide a reliable means for depicting such relationships. This study employs a 3D RTM and LiDAR (light detection and ranging) data to characterize the light environment of larch plantations at a fine spatiotemporal scale, further investigating the relationship between absorbed photosynthetically active radiation (APAR) and forest structures. The impact of specific tree structural parameters, such as crown diameter, crown area, crown length, crown ratio, crown volume, tree height, leaf area index, and a distance parameter assessing tree competition, on the daily-scale cumulative APAR per tree was investigated using a partial least squares regression (PLSR) model. Furthermore, variable importance in projection (VIP) was also calculated from the PLSR. The results indicate that among the individual tree structure parameters, crown volume is the most important one in explaining individual tree APAR (VIP = 4.19), while the competition from surrounding trees still plays a role in explaining individual tree APAR to some extent (VIP = 0.15), and crown ratio contributes the least (VIP = 0.03). Regarding the spatial distribution of trees, the average cumulative APAR per tree of larch plots does not increase with an increase in canopy gap fraction. Tree density and average cumulative APAR per tree were fitted using a natural exponential equation, with a coefficient of determination (R2 = 0.89), and a small mean absolute percentage error (MAPE = 0.03). This study demonstrates the potential of combining 3D RTM with LiDAR data to quantify fine-scale APAR in plantations, providing insights for optimizing forest structure, enhancing forest quality, and implementing precise forest management practices, such as selective breeding for superior tree species.

Pine shoot beetle (PSB) is one of the most damaging forest insects of Yunnan pine plantations in southwest China. However, the subtle symptoms of heterogeneous tree crowns make it difficult to accurately detect at early stage of PSB attack. Here, we evaluated the potential of a combination of plant traits (PTs) and vegetation indices (VIs) to distinguish different levels of tree damage by integrating hyperspectral, thermal imagery, and light detection and ranging (lidar) data based on unmanned airborne vehicle (UAV) systems. A voxelization method was used to fuse hyperspectral reflectance, temperature, and lidar point cloud data. Subsequently, PTs such as pigments (i.e., chlorophyll, carotenoid, and anthocyanin contents) were retrieved from a radiative transfer model inversion, and structural, fluorescence, and thermal traits, as well as VIs, were derived from the fusion data. We developed a novel analytical approach using random forest (RF) algorithm with predictors from different spatial distributions (i.e., horizontal directions, vertical layers, and vertical clusters) to compare the performance of tree severity classification. The results showed that the difference of both PTs and VIs in vertical layers between different severity levels are more than that in horizontal directions. The performance of RF model with predictors of the vertical layers (OA = 74 %, kappa = 0.65) was better than that using the predictive variables in horizontal directions (OA = 69 % and kappa = 0.60) of tree crowns. Using the vertical clustering features RF model increased the accuracy (OA = 78 % and kappa = 0.70), especially for slightly and moderately damaged trees, with improvements of 10 % and 12 %, respectively. Among all variables analyzed, chlorophylls were the most important predictor, followed by photochemical reflectance index and structural traits. Our work demonstrates the effectiveness of using fused UAV-based multi-sensor data for early detection of PSB attack, and can be applied other potential forest diseases and insect monitoring.

Plant area density (PAD) of individual trees is an important structural indicator related to tree growth status, stress levels due to pests and diseases, photosynthesis potential, and evapotranspiration. Airborne laser scanning (ALS) provides unprecedented 3D information for mapping forest canopy parameters. Previous studies mainly focused on mapping stand-level and 2D leaf area index. This study proposes a method to estimate PAD from discrete and multiple return ALS data at individual tree scales. The proposed method uses path length distribution to eliminate crown-shape-induced clumping, as well as intensity information to estimate crown transmittance from relative low-density points. The path length distribution is derived from the 3D crown boundary contours created by an alpha shape algorithm, which explicitly considers the non-uniform LiDAR pulse penetration distances. Pulse intensity is calibrated with the nearest pure-ground pulse to mitigate the need for prior leaf and ground reflectance information, which can be used in areas with a heterogeneous background. The proposed method was evaluated both in virtual experiments as well as with terrestrial laser scanning (TLS) data. The virtual experiments used the large-scale remote sensing data and image simulation model (LESS) to simulate virtual ALS scanning data based on abstract and realistic canopies. Results showed that the ALS-derived PAD is highly accurate, with RMSE less than 0.02 and R2 > 0.99 for the abstract sphere and cube crowns, and RMSE = 0.19 and R2 = 0.578 for the realistic crowns. The comparison with TLS of a birch plot shows that the ALS-derived PAD is consistent with those derived from TLS, with RMSE = 0.14 and R2 = 0.46. This study demonstrated that using the full intensity and geometry information of a point cloud is capable of generating high-resolution forest parameters from ALS data.

Generating canopy-reflectance datasets using radiative transfer models under various leaf and soil optical property combinations is important for remote sensing retrieval of vegetation parameters. One-dimensional radiative transfer models have been frequently used. However, three-dimensional (3D) models usually require detailed 3D information that is difficult to obtain and long model execution time, limiting their use in remote sensing applications. This study aims to address these limitations for practical use of 3D models, proposing a semi-empirical speed-up method for canopy-reflectance simulation based on a LargE-Scale remote sensing data and image Simulation model (LESS), called Semi-LESS. The speed-up method is coupled with 3D LESS to describe the dependency of canopy reflectance on the wavelength, leaf, soil, and branch optical properties for a scene with fixed 3D structures and observation/illumination configurations, allowing fast generating accurate reflectance images under various wavelength-dependent optical parameters. The precomputed dataset stores simulated multispectral coefficient images under few predefined soil, branch, and leaf optical properties for each RAdiation transfer Model Intercomparison-V scene, which can then be used alone to compute reflectance images on the fly without the participation of LESS. Semi-LESS has been validated with full 3D radiative-transfer-simulated images, showing very high accuracy (root mean square error < 0.0003). The generation of images using Semi-LESS is much more efficient than full LESS simulations with an acceleration of more than 320 times. This study is a step further to promote 3D radiative transfer models in practical remote sensing applications such as vegetation parameter inversions.

To improve the simulation accuracy of vegetation canopy reflectance in optical bands, the Radiosity Applicable to Porous IndiviDual objects (RAPID) model has been upgraded to better deal with branches in the latest RAPID4. Previous versions of RAPID (RAPID1 and RAPID3) neglected branches in porous objects in optical bands, while RAPID2 emphasized them in microwave bands. This inconsistency needed to be addressed to establish a unified radiosity-based simulation framework. By incorporating branches in RAPID4, we have improved several aspects of the model, including the random dynamic projection process, the equivalent reflectance or transmittance, the single scattering estimation, the multiple scattering solution, and the bidirectional reflectance factor (BRF) calculation. Three-dimensional trees from the fifth RAdiation transfer Model Intercomparison (RAMI-V) have been used to test the contribution of branches on BRF. Comparisons with a ray-tracing-based LESS model (the LargE-Scale remote sensing data and image Simulation framework) on RAMI-V scenes show a general agreement on BRF (R2 ≥ 0.96 and root mean square error ranging from 0.014 to 0.054). The major biases occur in a realistic scene (i.e., HET51_WWO_TLS) created from terrestrial laser scanning data. Sensitivity analysis has been conducted to show the branch contribution on BRF in optical domain. Without considering dense branches, the BRF error can exceed 0.1.

Pine wilt disease (PWD), caused by pine wood nematode (PWN), poses a tremendous threat to global pine forests because it can result in rapid and widespread infestations within months, leading to large-scale tree mortality. Therefore, the implementation of preventive measures relies on early detection of PWD. Unmanned aerial vehicle (UAV)-based hyperspectral images (HSI) can detect tree-level changes and are thus an effective tool for forest change detection. However, previous studies mainly used single-date UAV-based HSI data, which could not monitor the temporal changes of disease distribution and determine the optimal detection period. To achieve these purposes, multi-temporal data is required. In this study, Pinus koraiensis stands were surveyed in the field from May to October during an outbreak of PWD. Concurrently, multi-temporal UAV-based red, green, and blue bands (RGB) and HSI data were also obtained. During the survey, 59 trees were confirmed to be infested with PWD, and 59 non-infested trees were used as control. Spectral features of each tree crown, such as spectral reflectance, first and second-order spectral derivatives, and vegetation indices (VIs), were analyzed to identify those useful for early monitoring of PWD. The Random Forest (RF) classification algorithm was used to examine the separability between the two groups of trees (control and infested trees). The results showed that: (1) the responses of the tree crown spectral features to PWD infestation could be detected before symptoms were noticeable in RGB data and field surveys; (2) the spectral derivatives were the most discriminable variables, followed by spectral reflectance and VIs; (3) based on the HSI data from July to October, the two groups of trees were successfully separated using the RF classifier, with an overall classification accuracy of 0.75–0.95. Our results illustrate the potential of UAV-based HSI for PWD early monitoring.

Understanding post-fire forest recovery is critical to the study of forest carbon dynamics. Many previous studies have used multispectral imagery to estimate post-fire recovery, yet post-fire forest structural development has rarely been evaluated in the Great Xing’an Mountain. In this study, we extracted the historical fire events from 1987 to 2019 based on a classification of Landsat imagery and assessed post-fire forest structure for these burned patches using Global Ecosystem Dynamics Investigation (GEDI)-derived metrics from 2019 to 2021. Two drivers were assessed for the influence on post-fire structure recovery, these being pre-fire canopy cover (i.e., dense forest and open forest) and burn severity levels (i.e., low, moderate, and high). We used these burnt patches to establish a 25-year chronosequence of forest structural succession by a space-for-time substitution method. Our result showed that the structural indices suggested delayed recovery following the fire, indicating a successional process from the decomposition of residual structures to the regeneration of new tree species in the post-fire forest. Across the past 25-years, the dense forest tends toward greater recovery than open forest, and the recovery rate was faster for low severity, followed by moderate severity and high severity. Specifically, in the recovery trajectory, the recovery indices were 21.7% and 17.4% for dense forest and open forest, and were 27.1%, 25.8%, and 25.4% for low, moderate, and high burn severity, respectively. Additionally, a different response to the fire was found in the canopy structure and height structure since total canopy cover (TCC) and plant area index (PAI) recovered faster than relative height (i.e., RH75 and RH95). Our results provide valuable information on forest structural restoration status, that can be used to support the formulation of post-fire forest management strategies in Great Xing’an Mountain.

The Yellow River Basin serves as a crucial ecological barrier in China, emphasizing the importance of accurately examining the spatial distribution of forest carbon stocks and enhancing carbon sequestration in order to attain “carbon peaking and carbon neutrality”. Forest patches have complex interactions that impact ecosystem services. To our knowledge, very few studies have explored the connection between these interactions and carbon stock. This study addressed this gap by utilizing complex network theory to establish a forest ecospatial network (ForEcoNet) in the Yellow River Basin in which forest patches are represented as nodes (sources) and their interactions as edges (corridors). Our objective was to optimize the ForEcoNet’s structure and enhance forest carbon stocks. First, we employed downscaling technology to allocate the forest carbon stocks of the 69 cities in the study area to grid cells, generating a spatial distribution map of forest carbon density in the Yellow River Basin. Next, we conducted morphological spatial pattern analysis (MSPA) and used the minimum cumulative resistance model (MCR) to extract the ForEcoNet in the basin. Finally, we proposed optimization of the ForEcoNet based on the coupling coordination between the node carbon stock and topological structure. The results showed that: (1) the forest carbon stocks of the upper, middle, and lower reaches accounted for 42.35%, 54.28%, and 3.37% of the total, respectively, (2) the ForEcoNet exhibited characteristics of both a random network and a scale-free network and demonstrated poor network stability, and (3) through the introduction of 51 sources and 46 corridors, we optimized the network and significantly improved its robustness. These findings provide scientific recommendations for the optimization of forest allocation in the Yellow River Basin and achieving the goal of increasing the forest carbon stock.

Forest canopy cover (FCC) plays an important role in many ecological, hydrological and forestry applications. For large-scale applications, FCC is usually estimated from remotely sensed data by inverting radiative transfer models (RTMs) or using data-driven regressions. In this article, we proposed a hybrid model, which combines a 3-D RTM and transfer learning-based convolutional neural network (T-CNN), to estimate FCC from very high-resolution satellite images (e.g., Chinese GaoFen-2, 1 m resolution with 4 bands). Unlike common hybrid models that are purely trained with simulation data, T-CNN combines simulation data-based pre-training and actual data-based transfer learning, which is a widely used technique in artificial intelligence for fine-tuning models. The performance of T-CNN was compared with a random forest (RF) model and two general CNN models, including CNN trained with actual dataset only (data-CNN) and CNN trained with RTM simulation data only (RTM-CNN). Results on the independent validation dataset (not used in training stage) showed that T-CNN had higher accuracy (RMSE = 0.121, R2 = 0.83), compared with RF (RMSE = 0.26, R2 = 0.61), Data-CNN (RMSE = 0.142, R2 = 0.81), and RTM-CNN (RMSE = 0.144, R2 = 0.73), which indicates that T-CNN has a strong transferability. Tests on different training sizes showed that T-CNN (0.084 < RMSE < 0.108) provided constantly better performances than RF (0.116 < RMSE < 0.122) and data-CNN (0.103 < RMSE < 0.128), which demonstrates the potential of T-CNN as an alternative to RTM-based inversion and data-driven regressions to estimate FCC, especially when training data is imbalanced and inadequate.

Accurate wall-to-wall estimation of forest crown cover is critical for a wide range of ecological studies. Notwithstanding the increasing use of UAVs in forest canopy mapping, the ultrahigh-resolution UAV imagery requires an appropriate procedure to separate the contribution of understorey from overstorey vegetation, which is complicated by the spectral similarity between the two forest components and the illumination environment. In this study, we investigated the integration of deep learning and the combined data of imagery and photogrammetric point clouds for boreal forest canopy mapping. The procedure enables the automatic creation of training sets of tree crown (overstorey) and background (understorey) data via the combination of UAV images and their associated photogrammetric point clouds and expands the applicability of deep learning models with self-supervision.Based on the UAV images with different overlap levels of 12 conifer forest plots that are categorized into “I”, “II” and “III” complexity levels according to illumination environment, we compared the self-supervised deep learning-predicted canopy maps from original images with manual delineation data and found an average intersection of union (IoU) larger than 0.9 for “complexity I” and “complexity II” plots and larger than 0.75 for “complexity III” plots. The proposed method was then compared with three classical image segmentation methods (i.e., maximum likelihood, Kmeans, and Otsu) in the plot-level crown cover estimation, showing outperformance in overstorey canopy extraction against other methods. The proposed method was also validated against wall-to-wall and pointwise crown cover estimates using UAV LiDAR and in situ digital cover photography (DCP) benchmarking methods. The results showed that the model-predicted crown cover was in line with the UAV LiDAR method (RMSE of 0.06) and deviate from the DCP method (RMSE of 0.18). We subsequently compared the new method and the commonly used UAV structure-from-motion (SfM) method at varying forward and lateral overlaps over all plots and a rugged terrain region, yielding results showing that the method-predicted crown cover was relatively insensitive to varying overlap (largest bias of less than 0.15), whereas the UAV SfM-estimated crown cover was seriously affected by overlap and decreased with decreasing overlap. In addition, canopy mapping over rugged terrain verified the merits of the new method, with no need for a detailed digital terrain model (DTM). The new method is recommended to be used in various image overlaps, illuminations, and terrains due to its robustness and high accuracy. This study offers opportunities to promote forest ecological applications (e.g., leaf area index estimation) and sustainable management (e.g., deforestation).

Abstract Background The assessment of change in forest ecosystems, especially the change of canopy heights, is essential for improving global carbon estimates and understanding effects of climate change. Spaceborne lidar systems provide a unique opportunity to monitor changes in the vertical structure of forests. NASA’s Ice, Cloud and Land Elevation Satellites, ICESat-1 for the period 2003 to 2009, and ICESat-2 (available since 2018), have collected elevation data over the Earth’s surface with a time interval of 10 years. In this study, we tried to discover forest canopy changes by utilizing the global forest canopy height map of 2005 (complete global coverage with 1 km resolution) derived from ICESat-1 data and the ATL08 land and vegetation products of 2019 (sampling footprints with 17 m diameter) from ICESat-2. Results Our study revealed a significant increase in forest canopy heights of China’s Beijing-Tianjin-Hebei region. Evaluations of unchanging areas for data consistency of two products show that the bias values decreased significantly from line-transect-level (− 8.0 to 6.2 m) to site-level (− 1.5 to 1.1 m), while RMSE values are still relatively high (6.1 to 15.2 m, 10.2 to 12.0 m). Additionally, 58% of ATL08 data are located in ‘0 m’ pixels with an average height of 7.9 m, which are likely to reflect the ambitious tree planting programs in China. Conclusions Our study shows that it is possible, with proper calibrations, to use ICESat-1 and -2 products to detect forest canopy height changes in a regional context. We expect that the approach presented in this study is potentially suitable to derive a fine-scale map of global forest change.

Tree species information is an important factor in forest resource surveys, and light detection and ranging (LiDAR), as a new technical tool for forest resource surveys, can quickly obtain the 3D structural information of trees. In particular, the rapid and accurate classification and identification of tree species information from individual tree point clouds using deep learning methods is a new development direction for LiDAR technology in forest applications. In this study, mobile laser scanning (MLS) data collected in the field are first pre-processed to extract individual tree point clouds. Two downsampling methods, non-uniform grid and farthest point sampling, are combined to process the point cloud data, and the obtained sample data are more conducive to the deep learning model for extracting classification features. Finally, four different types of point cloud deep learning models, including pointwise multi-layer perceptron (MLP) (PointNet, PointNet++, PointMLP), convolution-based (PointConv), graph-based (DGCNN), and attention-based (PCT) models, are used to classify and identify the individual tree point clouds of eight tree species. The results show that the classification accuracy of all models (except for PointNet) exceeded 0.90, where the PointConv model achieved the highest classification accuracy for tree species classification. The streamlined PointMLP model can still achieve high classification accuracy, while the PCT model did not achieve good accuracy in the tree species classification experiment, likely due to the small sample size. We compare the training process and final classification accuracy of the different types of point cloud deep learning models in tree species classification experiments, further demonstrating the advantages of deep learning techniques in tree species recognition and providing experimental reference for related research and technological development.

Tree species is a critical factor in the practice of forest resource field sample surveys. Light detection and ranging (LiDAR) can obtain three-dimensional structural information about forests and trees and is increasingly being used in forest resource surveys. We used three pointwise multi-layer perceptron (MLP)-based deep learning methods (PointNet, PointNet++, and PointMLP) to identify individual tree point clouds of seven different tree species to explore the effectiveness of point cloud deep learning in classifying individual tree point clouds. Experiment results were extremely exciting. Higher classification accuracy can be attained in trials utilizing 2048 points. The tree classification accuracies of PointMLP and PointNet++ on the test set were 0.9474 and 0.9483, respectively, in classification experiments with a balanced sample size. PointMLP, the current state-of-the-art pointwise MLP-based model, is faster to train and performs better.

Three dimensional (3D) green volume is an important tree factor used in forest surveys as a prerequisite for estimating aboveground biomass (AGB). In this study, we developed a method for accurately calculating the 3D green volume of single trees from unmanned aerial vehicle laser scanner (ULS) data, using a voxel coupling convex hull by slices algorithm, and compared the results using voxel coupling convex hull by slices algorithm with traditional 3D green volume algorithms (3D convex hull, 3D concave hull (alpha shape), convex hull by slices, voxel and voxel coupling convex hull by slices algorithms) to estimate AGB. Our results showed the following: (1) The voxel coupling convex hull by slices algorithm can accurately estimate the 3D green volume of a single ginkgo tree (RMSE = 11.17 m3); (2) Point cloud density can significantly affect the extraction of 3D green volume; (3) The addition of the 3D green volume parameter can significantly improve the accuracy of the model to estimate AGB, where the highest accuracy was obtained by the voxel coupling convex hull by slices algorithm (CV-R2 = 0.85, RMSE = 11.29 kg, and nRMSE = 15.12%). These results indicate that the voxel coupling convex hull by slices algorithms can more effectively calculate the 3D green volume of a single tree from ULS data. Moreover, our study provides a comprehensive evaluation of the use of ULS 3D green volume for AGB estimation and could significantly improve the estimation accuracy of AGB.

Live fuel moisture content (LFMC) is an important index used to evaluate the wildfire risk and fire spread rate. In order to further improve the retrieval accuracy, two ensemble models combining deep learning models were proposed. One is a stacking ensemble model based on LSTM, TCN and LSTM-TCN models, and the other is an Adaboost ensemble model based on the LSTM-TCN model. Measured LFMC data, MODIS, Landsat-8, Sentinel-1 remote sensing data and auxiliary data such as canopy height and land cover of the forest-fire-prone areas in the Western United States, were selected for our study, and the retrieval results of different models with different groups of remote sensing data were compared. The results show that using multi-source data can integrate the advantages of different types of remote sensing data, resulting in higher accuracy of LFMC retrieval than that of single-source remote sensing data. The ensemble models can better extract the nonlinear relationship between LFMC and remote sensing data, and the stacking ensemble model with all the MODIS, Landsat-8 and Sentinel-1 remote sensing data achieved the best LFMC retrieval results, with R2 = 0.85, RMSE = 18.88 and ubRMSE = 17.99. The proposed stacking ensemble model is more suitable for LFMC retrieval than the existing method.

Brandt et al. reported the results of their analysis from high-resolution satellite images, covering more than 1.3 million square kilometers of the Western Sahara and Sahel in West Africa. They mapped the locations and sizes of approximately 1.8 billion trees. Prior to this, scientists had never made such a detailed map of trees in such a large area. Commercial satellites have begun to collect data and can detect small ground objects that are 1 square meter or less in size. Therefore, the field of terrestrial remote sensing may have a significant advance from mainly a comprehensive landscape-scale measurement to mapping the position and canopy size of each tree at a regional or even global scale. This progress will revolutionize how we think, monitor, simulate, and manage the global terrestrial ecosystem.①① Original source in Chinese: Huaguao Huang, Using satellites to map trees, Bulletin of National Natural Science Foundation of China. 35 (2) (2021) 239-240.

To investigate forest resources, it is necessary to identify the tree species. However, it is a challenge to identify tree species using 3D point clouds of trees collected by light detection and ranging (LiDAR). PointNet++, a point cloud deep learning network, can effectively classify 3D objects. It is important to establish high-quality individual tree point cloud datasets when applying PointNet++ to identifying tree species. However, there are different data processing methods to produce sample datasets, and the processes are tedious. In this study, we suggest how to select the appropriate method by designing comparative experiments. We used the backpack laser scanning (BLS) system to collect point cloud data for a total of eight tree species in three regions. We explored the effect of tree height on the classification accuracy of tree species by using different point cloud normalization methods and analyzed the effect of leaf point clouds on classification accuracy by separating the leaves and wood of individual tree point clouds. Five downsampling methods were used: farthest point sampling (FPS), K-means, random, grid average sampling, and nonuniform grid sampling (NGS). Data with different sampling points were designed for the experiments. The results show that the tree height feature is unimportant when using point cloud deep learning methods for tree species classification. For data collected in a single season, the leaf point cloud has little effect on the classification accuracy. The two suitable point cloud downsampling methods we screened were FPS and NGS, and the deep learning network could provide the most accurate tree species classification when the number of individual tree point clouds was in the range of 2048–5120. Our study further illustrates that point-based end-to-end deep learning methods can be used to classify tree species and identify individual tree point clouds. Combined with the low-cost and high-efficiency BLS system, it can effectively improve the efficiency of forest resource surveys.

Outbreaks of pine shoot beetles (Tomicus spp.) have caused widespread tree mortality in Southwest China. However, the understanding of the role of climatic drivers in pine shoot beetle outbreaks is limited. This study aimed to characterize the relationships between climate variables and pine shoot beetle outbreaks in the forests of Yunnan pine (Pinus yunnanensis Franch) in Southwest China. The pine shoot beetle-infested total area from 2000 to 2017 was extracted from multi-data Landsat images and obtained from field survey plots. A temporal prediction model was developed by partial least squares regression. The results indicated that multi consecutive year droughts was the strongest predictor, as such a condition greatly reduced the tree resistance to the beetles. The beetle-infested total area increased with spring temperature, associated with a higher success rate of trunk colonization and accelerated larval development. Warmer temperatures and longer solar radiation duration promoted flight activity during the trunk transfer to the shoot period and allowed the completion of sister broods. Multi consecutive year droughts combined with the warmer temperatures and higher solar radiation duration could provide favorable conditions for shoot beetle outbreaks. Generally, identifying the climate variables that drive pine shoot beetle outbreaks could help improve current strategies for outbreak control.

Infestations of Tomicus spp. have caused the deaths of millions of Yunnan pine forests in Southwest China; consequently, accurate monitoring methods are required to assess the damage caused by these pest insects at an early stage. Considering the limited sensitivity of optical reflectance on the early stage of beetle stress, the potential of thermal infrared (TIR) can be exploited for monitoring forest health on the basis of the change of canopy surface temperature (CST). However, few studies have investigated the impact of the leaf area index (LAI) on the accuracy of TIR data-based SDR assessments. Therefore, the current study used unmanned airborne vehicle (UAV)-based TIR and light detection and ranging (LiDAR) data to assess the capacity of determining the potential for using TIR data for determining SDR under different LAI conditions. The feasibility of using TIR for monitoring SDRs at the tree level and plot scales were analyzed using the relationship between SDR and canopy temperature. Results revealed that: (1) prediction accuracy of SDR from CST is promising at high LAI values and decreases quickly with LAI, and is higher at the single tree scale (R2 = 0.7890) than at the plot scale (R2 = 0.5532); (2) at either single tree or plot scale, a significant negative correlation can be found between CST and LAI (−0.9121 at tree scale and −0.5902 at plot scale); (3) LAI affects the transmission paths of sunlight and sensor, which mainly disturbs the relationship between CST and SDR. This article evaluated the high possibility of using TIR data to monitor SDRs at both tree and plot levels and assessed the negative impact of a low LAI (

Pine wilt disease (PWD) is a global destructive threat to forests which has been widely spread and has caused severe tree mortality all over the world. It is important to establish an effective method for forest managers to detect the infected area in a large region. Remote sensing is a feasible tool to detect PWD, but the traditional empirical methods lack the ability to explain the signals and can hardly be extended to large scales. The studies using physically-based models either ignore the within-canopy heterogeneity or rely too much on prior knowledge. In this study, we propose an approach to retrieve PWD infected areas from medium-resolution satellite images of two phases based on the simulations of an extended stochastic radiative transfer model for forests infected by pests (SRTP). A small amount of prior knowledge was used, and a change of background soil was considered in this approach. The performance was evaluated in different study sites. The inversion method performs best in the three-dimensional model LESS simulation sample plots (R2 = 0.88, RMSE = 0.059), and the inversion accuracy decreases in the real forest sample plots. For Jiangxi masson pine stand with large coverage and serious damage, R2 = 0.57, RMSE = 0.074; and for Shandong black pine stand with sparse and a small number of single plant damage, R2 = 0.48, RMSE = 0.063. This study indicates that the SRTP model is more feasible for pest damage inversion over different regions compared with empirical methods. The stochastic radiative transfer theory provides a potential approach for future monitoring of terrestrial vegetation parameters.

Achieving carbon neutrality is a necessary effort to rid humanity of a catastrophic climate and is a goal for China in the future. Ecological space plays an important role in the realization of carbon neutrality, but the relationship between the structure of vegetation ecological space and vegetation carbon sequestration capacity has been the focus of research. In this study, we extracted the base data from MODIS products and other remote sensing products, and then combined them with the MCR model to construct a vegetation ecospatial network in the Yellow River Basin in 2018. Afterward, we calculated the topological indicators of ecological nodes in the network and analyzed the relationship between the carbon sequestration capacity (net biome productivity) of ecological nodes and these topological indicators in combination with the Biome-BGC model. The results showed that there was a negative linear correlation between the betweenness centrality of forest nodes and their carbon sequestration capacity in the Yellow River Basin (p < 0.05, R2 = 0.59). On the other hand, there was a positive linear correlation between the clustering coefficient of grassland nodes and their carbon sequestration capacity (p < 0.01, R2 = 0.49). In addition, we briefly evaluated the vegetation ecospatial network in the Yellow River BASIN and suggested its optimization direction under the background of carbon neutrality in the future. Increasing the carbon sequestration capacity of vegetation through the construction of national ecological projects is one of the ways to achieve carbon neutrality, and this study provides a reference for the planning of future national ecological projects in the Yellow River Basin. Furthermore, this is also a case study of the application of remote sensing in vegetation carbon budgeting.

As one of the most devastating disasters to pine forests, pine wilt disease (PWD) has caused tremendous ecological and economic losses in China. An effective way to prevent large-scale PWD outbreaks is to detect and remove the damaged pine trees at the early stage of PWD infection. However, early infected pine trees do not show obvious changes in morphology or color in the visible wavelength range, making early detection of PWD tricky. Unmanned aerial vehicle (UAV)-based hyperspectral imagery (HI) has great potential for early detection of PWD. However, the commonly used methods, such as the two-dimensional convolutional neural network (2D-CNN), fail to simultaneously extract and fully utilize the spatial and spectral information, whereas the three-dimensional convolutional neural network (3D-CNN) is able to collect this information from raw hyperspectral data. In this paper, we applied the residual block to 3D-CNN and constructed a 3D-Res CNN model, the performance of which was then compared with that of 3D-CNN, 2D-CNN, and 2D-Res CNN in identifying PWD-infected pine trees from the hyperspectral images. The 3D-Res CNN model outperformed the other models, achieving an overall accuracy (OA) of 88.11% and an accuracy of 72.86% for detecting early infected pine trees (EIPs). Using only 20% of the training samples, the OA and EIP accuracy of 3D-Res CNN can still achieve 81.06% and 51.97%, which is superior to the state-of-the-art method in the early detection of PWD based on hyperspectral images. Collectively, 3D-Res CNN was more accurate and effective in early detection of PWD. In conclusion, 3D-Res CNN is proposed for early detection of PWD in this paper, making the prediction and control of PWD more accurate and effective. This model can also be applied to detect pine trees damaged by other diseases or insect pests in the forest.

Climate change has become a global concern for scientists as it is affecting almost every ecosystem. Larix gmelinii and Betula platyphylla are native and dominant forest species in the Daxing’anling Mountains of Inner Mongolia, playing a major role in carbon sequestration of this region. This study was carried out to assess the effect of climate variables including precipitation and temperature on the biomass of Larix gmelinii and Betula platyphylla forests. For this purpose, we used the climate-sensitive stem biomass allometric model for both species separately to find out accurate stem biomass along with climatic factors from 1950 to 2016. A total of 66 random plots were taken to attain the data from this study area. Larix gmelinii and Betula platyphylla stem biomass have a strong correlation with annual precipitation (R2 = 0.86, R2 = 0.71, R2 = 0.79, and R2 = 0.59) and maximum temperature (R2 = 0.76, R2 = 0.64, R2 = 0.67, and R2 = 0.52). However, annual minimum temperature (R2 = 0.58, R2 = 0.43, R2 = 0.55, and R2 = 0.46) and annual mean temperature (R2 = 0.40, R2 = 0.22, R2 = 0.36, and R2 = 0.19) have a relatively negative impact on tree biomass. Therefore, we suggest that both species have a very strong adaptive nature with climatic variation and hence can survive under drought and high-temperature stress climatic conditions.

Profile radar allows direct characterization of the vertical forest structure. Short-wavelength, such as Ku or X band, microwave data provide opportunities to detect the foliage. In order to exploit the potential of radar technology in forestry applications, a helicopter-borne Ku-band profile radar system, named Tomoradar, has been developed by the Finnish Geospatial Research Institute. However, how to use the profile radar waveforms to assess forest canopy parameters remains a challenge. In this study, we proposed a method by matching Tomoradar waveforms with simulated ones to estimate forest canopy leaf area index (LAI). Simulations were conducted by linking an individual tree-based forest gap model ZELIG and a three-dimension (3D) profile radar simulation model RAPID2. The ZELIG model simulated the parameters of potential local forest succession scene, and the RAPID2 model utilized the parameters to generate 3D virtual scenes and simulate waveforms based on Tomoradar configuration. The direct comparison of simulated and collected waveforms from Tomoradar could be carried out, which enabled the derivation of possible canopy LAI distribution corresponding to the Tomoradar waveform. A 600-m stripe of Tomoradar data (HH polarization) collected in the boreal forest at Evo in Finland was used as a test, which was divided into 60 plots with an interval of 10 m along the trajectory. The average waveform of each plot was employed to estimate the canopy LAI. Good results have been found in the waveform matching and the uncertainty of canopy LAI estimation. There were 95% of the plots with the mean relative overlapping rate (RO) above 0.7. The coefficients of variation of canopy LAI estimates were less than 0.20 in 80% of the plots. Compared to lidar-derived canopy effective LAI estimation, the coefficient of determination was 0.46, and the root mean square error (RMSE) was 1.81. This study established a bridge between the Ku band profile radar waveform and the forest canopy LAI by linking the RAPID2 and ZELIG model, presenting the uncertainty of forest canopy LAI estimation using Tomoradar. It is worth noting that since the difference of backscattering contribution is caused by both canopy structure and tree species, similar waveforms may correspond to different canopy LAI, inducing the uncertainty of canopy LAI estimation, which should be noticed in forest parameters estimation with empirical methods.

The influence of leaf temperature on transpiration, photosynthesis, respiration, and other metabolic activities is critical to plant growth, development, production and distribution. However, traditional measurement of canopy temperature by thermocouples or thermal infrared thermometers is laborious and difficult, especially for tall trees. The recent development of a handheld thermal infrared imager has made it possible to perform high temporal and spatial canopy temperature measurements efficiently. However, the signal recorded by the sensor is often a mixture of radiation from the target and the atmosphere, which must be corrected to get the true temperature. In this study, we propose a ground-based indirect measurement method of canopy temperature by a handheld thermal infrared imager through upward observation. Visible and thermal images are combined to distinguish the canopy pixels and sky pixels. To remove the atmospheric radiation from the sky, an empirical atmospheric model is established, which can perform atmospheric correction accurately and efficiently. To validate the proposed method, we collected canopy temperatures of 36 species of trees with a FLIR T420 thermal infrared imager and compared the estimated temperatures with those directly measured by thermocouples. The accuracy of the corrected canopy temperature has been significantly improved with mean absolute error reduced from 3.73 °C to 0.64 °C. This proposed canopy temperature measurement method can be used to various applications in remote sensing product validation, and ecosystem and forestry studies.

The authors wish to make the following corrections to this paper [...]

Quercus aquifolioides is one of the most representative broad-leaved plants in Qinghai-Tibet Plateau with important ecological status. So far, understanding how to quickly estimate the chlorophyll content of plants in plateau areas is still an urgent problem. Field Spec 3 spectrometer was used to measure hyperspectral reflectance data of Quercus aquifolioides leaves at different altitudes, and CCI (chlorophyll relative content) of corresponding leaves was measured by a chlorophyll meter. The correlation and univariate linear fitting analysis techniques were used to establish their relationship models. The results showed that: (1) Chlorophyll relative content of Quercus aquifolioides, under different altitude gradients, were significantly different. From 2905 m to 3500 m, chlorophyll relative content increased first and then decreased. Altitude 3300 m was the most suitable growth area. (2) In 350~550 nm, the spectral reflectance was 3500 m > 3300 m > 2905 m. In 750~1100 nm, the spectral reflectivity was 2905 m > 3500 m > 3300 m. (3) There were 4 main reflection peaks and 5 main absorption valleys in the leaf surface spectral reflection curve. While, 750~1400 nm was the sensitive range of leaf spectral response of Quercus aquifolioides. (4) The red edge position and red valley position moved to short wave direction with the increase of altitude, while the yellow edge position and green peak position moved to long wave direction first and then to short wave direction. (5) The correlation curve between the original spectrum and the CCI value was the best between the wavelengths 509~650 nm. The correlation between the first derivative spectrum and CCI value was the best and most stable at 450~500 nm. The green peak reflectance was most sensitive to the relative chlorophyll content of Quercus aquifolioides. The estimation model R2 of green peak reflectance was the highest (y = 206.98e−10.85x, R2 = 0.8523), and the prediction accuracy was 95.85%. The research results can provide some technical and theoretical support for the protection of natural Quercus aquifolioides forests in Tibet.

Hyperspectral light detection and ranging (LiDAR) (HSL) combines the characteristics of hyperspectral imaging and LiDAR techniques into a single instrument without any data registration. It provides more information than hyperspectral imaging or LiDAR alone in the extraction of vegetation physiological and biochemical parameters. However, the laser pulse intensity is affected by the incident angle, and its effect on HSL has not yet been fully explored. It is important for employing HSL to investigate vegetation properties. The aim of this paper is to study the incident angle effect of leaf reflectance with HSL and build a model about this impact. In this paper, we studied the angle effect of leaf reflectance from indoor HSL measurements of individual leaves from four typical tree species in Beijing. We observed that (a) the increasing of incident angle decreases the leaf reflectance; (b) the leaf spectrum observed by HSL from 650 to 1000 nm with 10 nm spectral resolution (36 channels) are consistent with those that measured by Analytica Spectra Devices (ASD) spectrometer (R2 = 0.9472 ~ 0.9897); (c) the specular reflection is significant in the red bands, and clear non-Lambertian characteristics are observed. In the near-infrared, there is little specular reflection, but it follows the Lambert-scattering law. We divided the whole band (650−1000 nm) into six bands and established an empirical model to correct the influence of angle effect on the reflectance of the leaf for HSL applications. In the future, the calibration of HSL measurements applied for other targets will be studied by rigorous experiments and modelling.

Similar to light detection and ranging (lidar), profile radar can detect forest vertical structure directly. Recently, the first Ku-band profile radar system designed for forest applications, called Tomoradar, has been developed and evaluated in boreal forest. However, the physical relationships between the waveform and forest structure parameters such as height, leaf area index (LAI), and aboveground biomass are still unclear, which limits later forestry applications. Therefore, it is necessary to develop a theoretical model to simulate the relationship and interpret the mechanism behind. In this study, we extend the Radiosity Applicable to Porous IndiviDual objects (RAPID2) model to simulate the profile radar waveform of forest stands. The basic assumption is that the scattering functions of major components within forest canopy are similar between profile radar and the side-looking radar implemented in RAPID2, except several modifications. These modifications of RAPID2 mainly include: (a) changing the observation angle from side-looking to nadir-looking; (b) enhancing the ground specular scattering in normal direction using Fresnel coefficient; (c) increasing the timing resolution and recording waveform. The simulated waveforms were evaluated using two plots of Tomoradar waveforms at co- and cross- polarizations, which are collected in thin and dense forest stands respectively. There is a good agreement (R2 ≥ 0.80) between the model results and experimental waveforms in HH and HV polarization modes and two forest scenes. After validation, the extended RAPID2 model was used to explore the sensitivity of the stem density, single tree LAI, crown shape, and twig density on the penetration depth in the Ku-band. Results indicate that the backscattering of the profile radar penetrates deeper than previous studies of synthetic aperture radar (SAR), and the penetration depth tends to be several meters in Ku-band. With the increasing of the needle and twig density in the microwave propagation path, the penetration depth decreases gradually. It is worth noting that variation of stem density seems to have the least effect on the penetration depth, when there is no overlapping between the single tree crowns.

In recent years, the outbreak of the pine shoot beetle (PSB), Tomicus spp., has caused serious shoots damage and the death of millions of trees in Yunnan pine forests in southwestern China. It is urgent to develop a convincing approach to accurately assess the shoot damage ratio (SDR) for monitoring the PSB insects at an early stage. Unmanned airborne vehicles (UAV)-based sensors, including hyperspectral imaging (HI) and lidar, have very high spatial and spectral resolutions, which are very useful to detect forest health. However, very few studies have utilized HI and lidar data to estimate SDRs and compare the predictive power for mapping PSB damage at the individual tree level. Additionally, the data fusion of HI and lidar may improve the detection accuracy, but it has not been well studied. In this study, UAV-based HI and lidar data were fused to detect PSB. We systematically evaluated the potential of a hyperspectral approach (only-HI data), a lidar approach (only-lidar data), and a combined approach (HI plus lidar data) to characterize PSB damage of individual trees using the Random Forest (RF) algorithm, separately. The most innovative point is the proposed new method to extract the three dimensional (3D) shadow distribution of each tree crown based on a lidar point cloud and the 3D radiative transfer model RAPID. The results show that: (1) for the accuracy of estimating the SDR of individual trees, the lidar approach (R2 = 0.69, RMSE = 12.28%) performed better than hyperspectral approach (R2 = 0.67, RMSE = 15.87%), and in addition, it was useful to detect dead trees with an accuracy of 70%; (2) the combined approach has the highest accuracy (R2 = 0.83, RMSE = 9.93%) for mapping PSB damage degrees; and (3) when combining HI and lidar data to predict SDRs, two variables have the most contributions, which are the leaf chlorophyll content (Cab) derived from hyperspectral data and the return intensity of the top of shaded crown (Int_Shd_top) from lidar metrics. This study confirms the high possibility to accurately predict SDRs at individual tree level if combining HI and lidar data. The 3D radiative transfer model can determine the 3D crown shadows from lidar, which is a key information to combine HI and lidar. Therefore, our study provided a guidance to combine the advantages of hyperspectral and lidar data to accurately measure the health of individual trees, enabling us to prioritize areas for forest health promotion. This method may also be used for other 3D land surfaces, like urban areas.

To enhance the capability of three-dimensional (3D) radiative transfer models at the kilometer scale (km-scale), the radiosity applicable to porous individual objects (RAPID) model has been upgraded to RAPID3. The major innovation is that the homogeneous porous object concept (HOMOBJ) used for a tree crown scale is extended to a heterogeneous porous object (HETOBJ) for a forest plot scale. Correspondingly, the radiosity-graphics-combined method has been extended from HOMOBJ to HETOBJ, including the random dynamic projection algorithm, the updated modules of view factors, the single scattering estimation, the multiple scattering solutions, and the bidirectional reflectance factor (BRF) calculations. Five cases of the third radiation transfer model intercomparison (RAMI-3) have been used to verify RAPID3 by the RAMI-3 online checker. Seven scenes with different degrees of topography (valleys and hills) at 500 m size have also been simulated. Using a personal computer (CPU 2.5 GHz, memory 4 GB), the computation time of BRF at 500 m is only approximately 13 min per scene. The mean root mean square error is 0.015. RAPID3 simulated the enhanced contrast of BRF between backward and forward directions due to topography. RAPID3 has been integrated into the free RAPID platform, which should be very useful for the remote sensing community. In addition, the HETOBJ concept may also be useful for the speedup of ray tracing models.

Yunnan pine shoot beetles (PSB), Tomicus yunnanensis and Tomicus minor have spread through southwestern China in the last five years, leading to millions of hectares of forest being damaged. Thus, there is an urgent need to develop an effective approach for accurate early warning and damage assessment of PSB outbreaks. Remote sensing is one of the most efficient methods for this purpose. Despite many studies existing on the mountain pine beetle (MPB), very little work has been undertaken on assessing PSB stress using remote sensing. The objective of this paper was to develop a spectral linear mixing model aided by radiative transfer (RT) and a new Yellow Index (YI) to simulate the reflectance of heterogeneous canopies containing damaged needles and quantitatively inverse their PSB stress. The YI, the fraction of dead needles, is a physically-explicit stress indicator that represents the plot shoots damage ratio (plot SDR). The major steps of this methods include: (1) LIBERTY2 was developed to simulate the reflectance of damaged needles using YI to linearly mix the green needle spectra with the dead needle spectra; (2) LIBERTY2 was coupled with the INFORM model to scale the needle spectra to the canopy scale; and (3) a look-up table (LUT) was created against Sentinel 2 (S2) imagery and inversed leaf chlorophyll content (LCC), green leaf area index (LAI) and plot SDR. The results show that (1) LIBERTY2 effectively simulated the reflectance spectral values on infested needles (mean relative error (MRE) = 1.4–18%), and the YI can indicate the degrees of needles damage; (2) the coupled LIBERTY2-INFORM model is suitable to estimate LAI (R2 = 0.73, RMSE = 0.17 m m−2, NRMSE = 11.41% and the index of agreement (IOA) = 0.92) and LCC (R2 = 0.49, RMSE = 56.24 mg m−2, NRMSE = 25.22% and IOA = 0.72), and is better than the original LIBERTY model (LAI: R2 = 0.38, RMSE = 0.43 m m−2, NRMSE = 28.85% and IOA = 0.68; LCC: R2 = 0.34, RMSE = 76.44 mg m−2, NRMSE = 34.23% and IOA = 0.57); and (3) the inversed YI is positively correlated with the measured plot SDR (R2 = 0.40, RMSE = 0.15). We conclude that the LIBERTY2 model improved the reflectance simulation accuracy of both the needles and canopies, making it suitable for assessing PSB stress. The YI has the potential to assess PSB damage.

Land surface temperatures (LSTs) obtained from remote sensing data are crucial in monitoring the conditions of crops and urban heat islands. However, since retrieved LSTs represent only the average temperature states of pixels, the distributions of temperatures within individual pixels remain unknown. Such data cannot satisfy the requirements of applications such as precision agriculture. Therefore, in this paper, we propose a model that combines a fast radiosity model, the Radiosity Applicable to Porous IndiviDual Objects (RAPID) model, and energy budget methods to dynamically simulate brightness temperatures (BTs) over complex surfaces. This model represents a model-based tool that can be used to estimate temperature distributions using fine-scale visible as well as near-infrared (VNIR) data and temporal variations in meteorological conditions. The proposed model is tested over a study area in an artificial oasis in Northwestern China. The simulated BTs agree well with those measured with the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER). The results reflect root mean squared errors (RMSEs) less than 1.6 °C and coefficients of determination (R2) greater than 0.7. In addition, compared to the leaf area index (LAI), this model displays high sensitivity to wind speed during validation. Although simplifications may be adopted for use in specific simulations, this proposed model can be used to support in situ measurements and to provide reference data over heterogeneous vegetation surfaces.

Background Monitoring forest health and biomass for changes over time in the global environment requires the provision of continuous satellite images. However, optical images of land surfaces are generally contaminated when clouds are present or rain occurs. Methods To estimate the actual reflectance of land surfaces masked by clouds and potential rain, 3D simulations by the RAPID radiative transfer model were proposed and conducted on a forest farm dominated by birch and larch in Genhe City, DaXing’AnLing Mountain in Inner Mongolia, China. The canopy height model (CHM) from lidar data were used to extract individual tree structures (location, height, crown width). Field measurements related tree height to diameter of breast height (DBH), lowest branch height and leaf area index (LAI). Series of Landsat images were used to classify tree species and land cover. MODIS LAI products were used to estimate the LAI of individual trees. Combining all these input variables to drive RAPID, high-resolution optical remote sensing images were simulated and validated with available satellite images. Results Evaluations on spatial texture, spectral values and directional reflectance were conducted to show comparable results. Conclusions The study provides a proof-of-concept approach to link lidar and MODIS data in the parameterization of RAPID models for high temporal and spatial resolutions of image reconstruction in forest dominated areas.

The Complicate Observations and Multi-Parameter Land Information Constructions on Allied Telemetry Experiment (COMPLICATE) comprises a network of remote sensing experiments designed to enhance the dynamic analysis and modeling of remotely sensed information for complex land surfaces. Two types of experimental campaigns were established under the framework of COMPLICATE. The first was designed for continuous and elaborate experiments. The experimental strategy helps enhance our understanding of the radiative and scattering mechanisms of soil and vegetation and modeling of remotely sensed information for complex land surfaces. To validate the methodologies and models for dynamic analyses of remote sensing for complex land surfaces, the second campaign consisted of simultaneous satellite-borne, airborne, and ground-based experiments. During field campaigns, several continuous and intensive observations were obtained. Measurements were undertaken to answer key scientific issues, as follows: 1) Determine the characteristics of spatial heterogeneity and the radiative and scattering mechanisms of remote sensing on complex land surfaces. 2) Determine the mechanisms of spatial and temporal scale extensions for remote sensing on complex land surfaces. 3) Determine synergist inversion mechanisms for soil and vegetation parameters using multi-mode remote sensing on complex land surfaces. Here, we introduce the background, the objectives, the experimental designs, the observations and measurements, and the overall advances of COMPLICATE. As a result of the implementation of COMLICATE and for the next several years, we expect to contribute to quantitative remote sensing science and Earth observation techniques.