Mapping Tree Species Research Articles

Fusing multi-season UAS images with convolutional neural networks to map tree species in Amazonian forests

Remote sensing images obtained by unoccupied aircraft systems (UAS) across different seasons enabled capturing of species-specific phenological patterns of tropical trees. The application of UAS multi-season images to classify tropical tree species is still poorly understood. In this study, we used RGB images from different seasons obtained by a low-cost UAS and convolutional neural networks (CNNs) to map tree species in an Amazonian forest. Individual tree crowns (ITC) were outlined in the UAS images and identified to the species level using forest inventory data. The CNN model was trained with images obtained in February, May, August, and November. The classification accuracy in the rainy season (November and February) was higher than in the dry season (May and August). Fusing images from multiple seasons improved the average accuracy of tree species classification by up to 21.1 percentage points, reaching 90.5%. The CNN model can learn species-specific phenological characteristics that impact the classification accuracy, such as leaf fall in the dry season, which highlights its potential to discriminate species in various conditions. We produced high-quality individual tree crown maps of the species using a post-processing procedure. The combination of multi-season UAS images and CNNs has the potential to map tree species in the Amazon, providing valuable insights for forest management and conservation initiatives.

Mapping tree species proportions from satellite imagery using spectral–spatial deep learning

Remote sensing can be used to collect information related to forest management. Previous studies demonstrated the potential of using multispectral satellite imagery for classifying tree species. However, methods that can map tree species in mixed forest stands on a large scale are lacking. We propose an innovative method for mapping the proportions of tree species using Sentinel-2 imagery. A convolutional neural network was used to quantify the per-pixel basal area proportions of tree species considering the neighbouring environment (spectral–spatial deep learning). A nested U-shaped neural network (UNet++) architecture was implemented. We produced a map of the entire Wallonia Region (southern Belgium). Nine species or groups of species were considered: Spruce genus, Oak genus, Beech, Douglas fir, Pine genus, Poplar genus, Larch genus, Birch genus, and remaining species. The training dataset for the convolutional neural network model was prepared using a map of forest parcels extracted from the public forest administration’s geodatabase of Wallonia. The accuracy of the predicted map covering the region was independently assessed using data from the regional forest inventory of Wallonia. A robust assessment method for tree species proportions maps was proposed for assessing the (1) majority species, (2) species composition (presence or absence), and (3) species proportions (proportion values). The achieved value of indicator OAmaj (0.73) shows that our approach can map the majority tree species in mixed and pure forest stands. Indicators MS (0.89), MPS (0.72) and MUS (0.83) support that the model can predict the species composition in most cases in the study area. Spruce genus, Oak genus, Beech, and Douglas fir achieved the best results, with PAs and UAs close to or higher than 0.70. Particularly, high performance was achieved for detecting Oak genus and Beech in low area proportions: PAs and UAs higher than 0.70 from the 0.4 proportion. Predicted proportions had a Radj2 of 0.50. The proposed method, which uses spectral–spatial deep learning to map the proportions of tree species, is innovative because it was adapted to the complexity of mixed forests and spatial resolution of current satellite imagery. Additionally, it optimises the use of available forest data in the model conception by considering all pixels from pure stands to highly mixed forest stands. When forest inventories are available in a broad sense, that is, georeferenced areas with the proportions of tree species, this method is highly reproducible and applicable at a large scale, offering potential for use in forest management.

Mapping Dominant Tree Species of German Forests

The knowledge of tree species distribution at a national scale provides benefits for forest management practices and decision making for site-adapted tree species selection. An accurate assignment of tree species in relation to their location allows conclusions about potential resilience or vulnerability to biotic and abiotic factors. Identifying areas at risk helps the long-term strategy of forest conversion towards a natural, diverse, and climate-resilient forest. In the framework of the national forest inventory (NFI) in Germany, data on forest tree species are collected in sample plots, but there is a lack of a full coverage map of the tree species distribution. The NFI data were used to train and test a machine-learning approach that classifies a dense Sentinel-2 time series with the result of a dominant tree species map of German forests with seven main tree species classes. The test of the model’s accuracy for the forest type classification showed a weighted average F1-score for deciduous tree species (Beech, Oak, Larch, and Other Broadleaf) between 0.77 and 0.91 and for non-deciduous tree species (Spruce, Pine, and Douglas fir) between 0.85 and 0.94. Two additional plausibility checks with independent forest stand inventories and statistics from the NFI show conclusive agreement. The results are provided to the public via a web-based interactive map, in order to initiate a broad discussion about the potential and limitations of satellite-supported forest management.

Commentary to the paper by Roell, Nord-Larsen et al. ‘Mapping tree species ecograms based on soil pH and soil water availability across Denmark’

Commentary to the paper by Roell, Nord-Larsen et al. ‘Mapping tree species ecograms based on soil pH and soil water availability across Denmark’

Mapping tree species in natural and planted forests using Sentinel-2 images

ABSTRACT Using remote-sensing technology to accurately map the composition and distribution of tree species is vital for sustainable forest resource management. Sentinel-2 data with the dense time-series observations enable to identify tree species. However, few studies clarify the differences in classification using Sentinel-2 images in natural forest and planted forest. Two study areas with different forest environments (planted forest and natural forest) were selected to evaluate the potential of Sentinel-2 imagery. Our results show that red-edge band, short-wave infrared (SWIR) band and vegetation indices (VIs), such as Red-Edge Normalized Difference Vegetation Index (ReNDVI), Soil-Adjusted Vegetation Index (SAVI), and Enhanced Vegetation Index (EVI), have important effects on tree species classification, especially in the growing season. By using two machine learning algorithms (support vector machine [SVM] and random forest [RF]), the results show that the classification accuracy for planted forests (91.27% in SVM and 88.35% in RF) significantly exceeds that of natural forests (84.34% in SVM and 81.03% in RF). This accuracy difference may be related to the spatial heterogeneity inside the forest and the surrounding environmental implications. Although the multi-temporal Sentinel-2 images produce satisfactory accuracy for classifying tree species, further research is needed to improve the classification accuracy.

Mapping of Urban Vegetation with High-Resolution Remote Sensing: A Review

Green space is increasingly recognized as an important component of the urban environment. Adequate management and planning of urban green space is crucial to maximize its benefits for urban inhabitants and for the urban ecosystem in general. Inventorying urban vegetation is a costly and time-consuming process. The development of new remote sensing techniques to map and monitor vegetation has therefore become an important topic of interest to many scholars. Based on a comprehensive survey of the literature, this review article provides an overview of the main approaches proposed to map urban vegetation from high-resolution remotely sensed data. Studies are reviewed from three perspectives: (a) the vegetation typology, (b) the remote sensing data used and (c) the mapping approach applied. With regard to vegetation typology, a distinction is made between studies focusing on the mapping of functional vegetation types and studies performing mapping of lower-level taxonomic ranks, with the latter mainly focusing on urban trees. A wide variety of high-resolution imagery has been used by researchers for both types of mapping. The fusion of various types of remote sensing data, as well as the inclusion of phenological information through the use of multi-temporal imagery, prove to be the most promising avenues to improve mapping accuracy. With regard to mapping approaches, the use of deep learning is becoming more established, mostly for the mapping of tree species. Through this survey, several research gaps could be identified. Interest in the mapping of non-tree species in urban environments is still limited. The same holds for the mapping of understory species. Most studies focus on the mapping of public green spaces, while interest in the mapping of private green space is less common. The use of imagery with a high spatial and temporal resolution, enabling the retrieval of phenological information for mapping and monitoring vegetation at the species level, still proves to be limited in urban contexts. Hence, mapping approaches specifically tailored towards time-series analysis and the use of new data sources seem to hold great promise for advancing the field. Finally, unsupervised learning techniques and active learning, so far rarely applied in urban vegetation mapping, are also areas where significant progress can be expected.

Mapping forest types over large areas with Landsat imagery partially affected by clouds and SLC gaps

The ecosystem services that forests provide depend on tree species composition. Therefore, it is important to map not only forest extent and its dynamics, but also composition. Open access to Landsat has resulted in considerable improvements in remote sensing methods for mapping tree species, but most approaches fail to perform when there is a shortage of clear observations. Our main goal was to map forest composition with Landsat imagery in various data availability conditions, and to investigate how the missing data, either due to clouds or scan line problems affect classification accuracy. We tested a data driven approach that is based on multi-temporal analysis of the tree species’ spectral characteristics making it applicable to regional-scale mapping even when the gap-free imagery is not available. Our study area consisted of one Landsat footprint (26/28) located in Northern Wisconsin, USA. We selected this area because of numerous tree species (23), heterogenic composition of forests where the majority of stands are mixed, and availability of high-quality reference data. We quantified how classification accuracy at the species level was affected by a) the amount of missing data due to cloud cover and Scanning Line Corrector (SLC) gaps, b) the number of acquisitions, and c) the seasonal availability of images. We applied a decision tree classifier, capable of handling missing data to both single- and a three-year Landsat-7 and Landsat-8 observations. We classified the dominant tree species in each pixel and grouped results to forest stands to match our reference data. Our results show four major findings. First, producer’s and user’s accuracies range from 46.2% to 96.2% and from 59.9% to 93.7%, respectively for the most abundant forest types in the study area (all types covering greater than 2% of the forest area). Second, all tree species were mapped with overall accuracy above 70% even in when we restricted our data set to images having gaps larger than 30% of the study area. Third, the classification accuracy improved with more acquisitions, especially when images were available for the fall, spring, and summer. Finally, producer’s accuracies for pure-stands were higher than those for mixed stands by 10 to 30 percentage points. We conclude that inclusion of Landsat imagery with missing data allows to map forest types with accuracies that previously could be achieved only for those rare years for which several gap-free images were available. The approach presented here is directly applicable to Landsat-like observations and derived products such as seasonal composites and temporal statistics that miss 30% or more of the data for any single date to develop forest composition maps that are important for both forest management and ecology.

Assessing Landsat-8 and Sentinel-2 spectral-temporal features for mapping tree species of northern plantation forests in Heilongjiang Province, China

Accurate mapping of tree species is highly desired in the management and research of plantation forests, whose ecosystem services are currently under threats. Time-series multispectral satellite images, e.g., from Landsat-8 (L8) and Sentinel-2 (S2), have been proven useful in mapping general forest types, yet we do not know quantitatively how their spectral features (e.g., red-edge) and temporal frequency of data acquisitions (e.g., 16-day vs. 5-day) contribute to plantation forest mapping to the species level. Moreover, it is unclear to what extent the fusion of L8 and S2 will result in improvements in tree species mapping of northern plantation forests in China. We designed three sets of classification experiments (i.e., single-date, multi-date, and spectral-temporal) to evaluate the performances of L8 and S2 data for mapping keystone timber tree species in northern China. We first used seven pairs of L8 and S2 images to evaluate the performances of L8 and S2 key spectral features for separating these tree species across key growing stages. Then we extracted the spectral-temporal features from all available images of different temporal frequency of data acquisition (i.e., L8 time series, S2 time series, and fusion of L8 and S2) to assess the contribution of image temporal frequency on the accuracy of tree species mapping in the study area. 1) S2 outperformed L8 images in all classification experiments, with or without the red edge bands (0.4%–3.4% and 0.2%–4.4% higher for overall accuracy and macro-F1, respectively); 2) NDTI (the ratio of SWIR1 minus SWIR2 to SWIR1 plus SWIR2) and Tasseled Cap coefficients were most important features in all the classifications, and for time-series experiments, the spectral-temporal features of red band-related vegetation indices were most useful; 3) increasing the temporal frequency of data acquisition can improve overall accuracy of tree species mapping for up to 3.2% (from 90.1% using single-date imagery to 93.3% using S2 time-series), yet similar overall accuracies were achieved using S2 time-series (93.3%) and the fusion of S2 and L8 (93.2%). This study quantifies the contributions of L8 and S2 spectral and temporal features in mapping keystone tree species of northern plantation forests in China and suggests that for mapping tree species in China's northern plantation forests, the effects of increasing the temporal frequency of data acquisition could saturate quickly after using only two images from key phenological stages.

Machine Learning Classification of Endangered Tree Species in a Tropical Submontane Forest Using WorldView-2 Multispectral Satellite Imagery and Imbalanced Dataset

Accurate maps of the spatial distribution of tropical tree species provide valuable insights for ecologists and forest management. The discrimination of tree species for economic, ecological, and technical reasons is usually necessary for achieving promising results in tree species mapping. Most of the data used in tree species mapping normally have some degree of imbalance. This study aimed to assess the effects of imbalanced data in identifying and mapping trees species under threat in a selectively logged sub-montane heterogeneous tropical forest using random forest (RF) and support vector machine with radial basis function (RBF-SVM) kernel classifiers and WorldView-2 multispectral imagery. For comparison purposes, the original imbalanced dataset was standardized using three data sampling techniques: oversampling, undersampling, and combined oversampling and undersampling techniques in R. The combined oversampling and undersampling technique produced the best results: F1-scores of 68.56 ± 2.6% for RF and 64.64 ± 3.4% for SVM. The balanced dataset recorded improved classification accuracy compared to the original imbalanced dataset. This research observed that more separable classes recorded higher F1-scores. Among the species, Syzygium guineense and Zanthoxylum gilletii were the most accurately mapped whereas Newtonia buchananii was the least accurately mapped. The most important spectral bands with the ability to detect and distinguish between tree species as measured by random forest classifier, were the Red, Red Edge, Near Infrared 1, and Near Infrared 2.

Mapping abundance distributions of allergenic tree species in urbanized landscapes: A nation-wide study for Belgium using forest inventory and citizen science data

Mapping the distribution of allergenic plants in urbanized landscapes is of high importance to evaluate its impact on human health. However, data is not always available for the allergy-relevant species such as alder, birch, hazel, especially within cities where systematic inventories are often missing or not readily available. This research presents an approach to produce high-resolution abundance maps of allergenic tree species using existing forest inventories and opportunistic open-access citizen science data. Following a two-step approach, we first built species distribution models (SDMs) to predict species habitat suitability, using environmental characteristics as predictors. Second, we used statistical regressions to model the relationships between abundance, the habitat suitability predicted by the SDMs, and additional vegetation cover covariates. The combination of forest inventory data with citizen science data improves the accuracy of abundance distribution models of allergenic tree species. This produces a continuous, 1-hectare resolution map of alder, birch, and hazel showing spatial variations of abundance distributions both within the urban fabric and along the urban–rural gradient. Species abundance modelling can offer a better understanding of the existing and potential future allergy risk posed by green spaces and pave the way for a wide variety of applications at fine-scale, which is indispensable for evidence-based urban green space policy and planning in support of public health.

Tree Species Mapping on Sentinel-2 Satellite Imagery with Weakly Supervised Classification and Object-Wise Sampling

Information on forest composition, specifically tree types and their distribution, aids in timber stock calculation and can help to better understand the biodiversity in a particular region. Automatic satellite imagery analysis can significantly accelerate the process of tree type classification, which is traditionally carried out by ground-based observation. Although computer vision methods have proven their efficiency in remote sensing tasks, specific challenges arise in forestry applications. The forest inventory data often contain the tree type composition but do not describe their spatial distribution within each individual stand. Therefore, some pixels can be assigned a wrong label in the semantic segmentation task if we consider each stand to be homogeneously populated by its dominant species. Another challenge is the spatial distribution of individual stands within the study area. Classes are usually imbalanced and distributed nonuniformly that makes sampling choice more critical. This study aims to enhance tree species classification based on a neural network approach providing automatic markup adjustment and improving sampling technique. For forest species markup adjustment, we propose using a weakly supervised learning approach based on the knowledge of dominant species content within each stand. We also propose substituting the commonly used CNN sampling approach with the object-wise one to reduce the effect of the spatial distribution of forest stands. We consider four species commonly found in Russian boreal forests: birch, aspen, pine, and spruce. We use imagery from the Sentinel-2 satellite, which has multiple bands (in the visible and infrared spectra) and a spatial resolution of up to 10 meters. A data set of images for Leningrad Oblast of Russia is used to assess the methods. We demonstrate how to modify the training strategy to outperform a basic CNN approach from F1-score 0.68 to 0.76. This approach is promising for future studies to obtain more specific information about stands composition even using incomplete data.

The flowering of Atlantic Forest Pleroma trees

Mapping the spatial distribution of a plant is a current challenge in ecology. Here, a convolutional neural network (CNN) and 33,798 Sentinel-2 satellite images were used to detect and map forest stands dominated by trees of the genus Pleroma by their magenta-to-deep-purple blossoms in the entire Brazilian Atlantic Forest domain, from June 2016 to July 2020. The Pleroma genus, known for its pioneer behaviour, was detected in an area representing 10.8% of the Atlantic Forest, associated negatively with temperature and positively with elevation, slope, tree cover and precipitation. The detection of another genus by the model, 18% of all the detections contained only pink blooming Handroanthus trees, highlighted that botanical identification from space must be taken with caution, particularly outside the known distribution range of the species. The Pleroma blossom seasonality occurred over a period of ~5–6 months centered on the March equinox and populations with distinct blossom timings were found. Our results indicate that in the Atlantic Forest, the remaining natural forest is less diverse than expected but is at least recovering from degradation. Our study suggests a method to produce ecological-domain scale maps of tree genera and species based on their blossoms that could be used for tree studies and biodiversity assessments.

Mapping temperate forest tree species using dense Sentinel-2 time series

Precise information on tree species composition is critical for forest management and conservation, but mapping tree species with satellite data over large areas is still a challenge. Since 2017, Sentinel-2A/B provide multi-spectral time series with global coverage at an unprecedented spatial and temporal resolution. This is a new opportunity for mapping tree species over large areas that has not yet been fully explored. Because of the high spatial and temporal resolution, Sentinel-2 time series improve the characterization of vegetation phenology and canopy structure, parameters that are intrinsically linked to tree species. The objective of this study was to test the utility of a Sentinel-2 time-series based approach for mapping tree species in a temperate forest region in Central Europe. Using stand-wise forest inventory data for single species stands we assess how well main and minor tree species can be mapped, and if the addition of environmental variables and spatial texture metrics improves the classification accuracy. Our time series approach utilizes all available Sentinel-2 observations and an ensemble of radial basis convolution filters to build cloud-free 5-day time series for each spectral band. The time series are then used as input features to classify seventeen tree species.Our results show the potential of Sentinel-2 time-series based classification, but they also show the challenges associated with mapping a diverse portfolio of tree species. Accuracy of the nine main species, with an area proportion greater than 0.5%, ranged between 98.9% and 66.8%, which is promising for a large area. Adding detailed environmental data and texture metrics to the spectral model only marginally increased the accuracy of a few minor tree species. Overall, the eight minor tree species with area proportions less than 0.5% were most strongly affected by classification errors. Although the absolute mapped area of minor species correlated well with the estimated reference area, the small class areas of minor species lead to high classification errors in relative terms. Mapping minor tree species is challenging for statistical reasons (i.e., class imbalance, small sample size and class variance). Using all available Sentinel-2 data allows building dense time series at high spatial resolution that are mandatory for improved tree species mapping. We were able to show that the spectral time series is the prime explanatory information, even when complementing our analyses with texture information and various environmental data. The results suggest that with the applied data harmonization approach precise regional tree species mapping is feasible.

Mapping tree species ecograms based on soil pH and soil water availability across Denmark

Abstract A prerequisite for successful afforestation and reforestation is understanding the quality of a site before establishing a specific tree species. Ecograms have been widely applied to determine the suitability of different sites for different species by a simple assessment of nutrient and water availability. Their graphical representation of suitability into classes allows ecograms to be easily understood. However, ecograms have generally been mapped for small areas only and their validity has rarely been documented. The aim of this study is to map ecograms for five tree species across Denmark and validate the maps using forest stand production data. For this purpose, we classified the landscape into six nutrient classes and nine water classes based on four variables to generate the ecogram maps. Based on these classes, the generated maps depict if a tree species is unsuitable, suitable or optimal for a specific site. The absolute average misclassification for nutrient and water supply was 1 and 2 classes, respectively. Stepwise linear regression was implemented to determine if the four variables used to create the ecograms were able to predict production as observed from forest experiment and management planning data obtained from across the country. All five species used the full model to explain variation in production. However, the average production values per ecogram growing condition were not significantly different for all species. The range of $R$2 for the five species was 0.05–0.32, indicating that one ecogram template might not work for all species. The high-resolution national ecogram maps incorporate large-scale variables important for tree growth and will be beneficial when selecting new land for afforestation. The simplicity of ecograms allows for easy interpretation, meaning that foresters can quickly determine which regions of a landscape are suitable, saving time and resources.

Deep learning-based tree species mapping in a highly diverse tropical urban setting

Spatially explicit information on urban tree species distribution is crucial for green infrastructure management in cities. This information is usually acquired with ground-based surveys, which are time-consuming and usually cover limited spatial extents. The combination of machine learning algorithms and remote sensing images has been hailed as a promising way to map tree species over broad areas. Recently, convolutional neural networks (CNNs), a type of deep learning method, have achieved outstanding results for tree species discrimination in various remote sensing data types. However, there is a lack of studies using CNN-based methods to produce tree species composition maps, particularly for tropical urban settings. Here, we propose a multi-task CNN to map tree species in a highly diverse neighborhood in Rio de Janeiro, Brazil. Our network architecture takes an aerial photograph (RGB bands and pixel size = 0.15 m) and delivers two outputs: a semantically segmented image and a distance map transform. In the former, all pixel positions are labeled, while in the latter, each pixel position contains the Euclidean distance to the crown boundary. We developed a post-processing approach that combines the two outputs, and we classified nine and five tree species with an average F1-score of 79.3 ± 8.6% and 87.6 ± 4.4%, respectively. Moreover, our post-processing approach produced a realistic tree species composition map by labeling only pixels of the target species with high class membership probabilities. Our results show the potential of CNNs and aerial photographs to map tree species in highly diverse tropical urban settings, providing valuable insights for urban forest management and green spaces planning.

Characterizing and classifying urban tree species using bi-monthly terrestrial hyperspectral images in Hong Kong

Urban trees exhibit a wide range of ecosystem services that have long been unveiled and increasingly reported. The ability to map tree species and analyze tree health conditions would become vividly essential. Remote sensing techniques, especially hyperspectral imaging, are being evolved for species identification and vegetation monitoring from spectral reponse patterns. In this study, a hyperspectral library for urban tree species in Hong Kong was established comprising 75 urban trees belonging to 19 species. 450 bi-monthly images were acquired by a terrestrial hyperspectral camera (SPECIM-IQ) from November 2018 to October 2019. A Deep Neural Network classification model was developed to identify tree species from the hyperspectral imagery with an overall accuracy ranging from 85% to 96% among different seasons. Representative spectral reflectance curves of healthy and unhealthy conditions for each species were extracted and analyzed. The hyperspectral phenology models were developed to achieve high accuracy and optimization of data acquisition. The bi-monthly canopy signatures and vegetation indices revealed different seasonality patterns of evergreen and deciduous species in Hong Kong. We explored the utility of terrestrial hyperspectral remote sensing and Deep Neural Network for urban tree species identification and characterizing. This provides a unique baseline to understand hyperspectral characteristics and seasonality of urban tree species in Hong Kong that can also contribute to hyperspectral imaging and database development elsewhere in the world.

Classification of tree species classes in a hemi-boreal forest from multispectral airborne laser scanning data using a mini raster cell method

Classification of tree species or species classes is still a challenge for remote sensing-based forest inventory. Operational use of Airborne Laser Scanning (ALS) data for prediction of forest variables has this far been dominated by area-based methods where laser scanning data have been used for estimation of forest variables within raster cells. Classification of tree species has however not been achieved with sufficient accuracy with area-based methods using only ALS data. Furthermore, analysis of tree species at the level of raster cells with typical size of 15 m × 15 m is not ideal in the case of mixed species stands. Most ALS systems for terrestrial mapping use only one wavelength of light. New multispectral ALS systems for terrestrial mapping have recently become operational, such as the Optech Titan system with wavelengths 1550 nm, 1064 nm, and 532 nm. This study presents an alternative type of area-based method for classification of tree species classes where multispectral ALS data are used in combination with small raster cells. In this “mini raster cell method” features for classification are derived from the intensity of the different wavelengths in small raster cells using a moving window average approach to allow for a heterogeneous tree species composition. The most common tree species in the Nordic countries are Pinus sylvestris and Picea abies, constituting about 80% of the growing stock volume. The remaining 20% consists of several deciduous species, mainly Betula pendula and Betula pubescens, and often grow in mixed forest stands. Classification was done for pine (Pinus sylvestris), spruce (Picea abies), deciduous species and mixed species in middle-aged and mature stands in a study area located in hemi-boreal forest in the southwest of Sweden (N 58°27’, E 13°39’). The results were validated at plot level with the tree species composition defined as proportion of basal area of the tree species classes. The mini raster cell classification method was slightly more accurate (75% overall accuracy) than classification with a plot level area-based method (68% overall accuracy). The explanation is most likely that the mini raster cell method is successful at classifying homogenous patches of tree species classes within a field plot, while classification based on plot level analysis requires one or several heterogeneous classes of mixed species forest. The mini raster cell method also results in a high-resolution tree species map. The small raster cells can be aggregated to estimate tree species composition for arbitrary areas, for example forest stands or area units corresponding to field plots.

Prediction and model-assisted estimation of diameter distributions using Norwegian national forest inventory and airborne laser scanning data

Diameter at breast height (DBH) distributions offer valuable information for operational and strategic forest management decisions. We predicted DBH distributions using Norwegian national forest inventory and airborne laser scanning data and compared the predictive performances of linear mixed-effects (PPM), generalized linear-mixed (GLM), and k nearest-neighbor (NN) models. While GLM resulted in smaller prediction errors than PPM, both were clearly outperformed by NN. We therefore studied the ability of the NN model to improve the precision of stem frequency estimates by DBH classes in the 8.7 Mha study area using a model-assisted (MA) estimator suitable for systematic sampling. MA estimates yielded greater than or approximately equal efficiencies as direct estimates using field data only. The relative efficiencies (REs) associated with the MA estimates ranged between 0.95–1.47 and 0.96–1.67 for 2 and 6 cm DBH class widths, respectively, when dominant tree species were assumed to be known. The use of a predicted tree species map, instead of the observed information, decreased the REs by up to 10%.

Tree species classification using Sentinel-2 imagery and Bayesian inference

The increased temporal frequency of optical satellite data acquisitions provides a data stream that has the potential to improve land cover mapping, including mapping of tree species. However, for large area operational mapping, partial cloud cover and different image extents can pose challenges. Therefore, methods are needed to assimilate new images in a straightforward way without requiring a total spatial coverage for each new image. This study shows that Bayesian inference applied sequentially has the potential to solve this problem. To test Bayesian inference for tree species classification in the boreo-nemoral zone of southern Sweden, field data from the study area of Remningstorp (58°27′18.35″ N, 13°39′8.03″ E) were used. By updating class likelihood with an increasing number of combined Sentinel-2 images, a higher and more stable cross-validated overall accuracy was achieved. Based on a Mahalanobis distance, 23 images were automatically chosen from the period of 2016 to 2018 (from 142 images total). An overall accuracy of 87% (a Cohen’s kappa of 78.5%) was obtained for four tree species classes: Betula spp., Picea abies, Pinus sylvestris, and Quercus robur. This application of Bayesian inference in a boreo-nemoral forest suggests that it is a practical way to provide a high and stable classification accuracy. The method could be applied where data are not always complete for all areas. Furthermore, the method requires less reference data than if all images were used for classification simultaneously.

A New CBAM-P-Net Model for Few-Shot Forest Species Classification Using Airborne Hyperspectral Images

High-precision automatic identification and mapping of forest tree species composition is an important content of forest resource survey and monitoring. The airborne hyperspectral image contains rich spectral and spatial information, which provides the possibility of high-precision classification and mapping of forest tree species. Few-shot learning, as an application of deep learning, has become an effective method of image classification. Prototypical networks (P-Net) is a simple and practical deep learning network, which has significant advantages in solving few-shot classification problems. Considering the high band correlation and large data volume associated with airborne hyperspectral images, how to fully extract effective features, filter or reduce redundant features is the key to improving the classification accuracy of P-Net, in order to extract effective features in hyperspectral images and obtain a high-precision forest tree species classification model with limited samples. In this research, we embedded the convolutional block attention module (CBAM) between the convolution blocks of P-Net, the CBAM-P-Net was constructed, and a method to improve the feature extraction efficiency of the P-Net was proposed, although this method makes the network more complex and increases the computational cost to a certain extent. The results show that the combination strategy using Channel First for CBAM greatly improves the feature extraction efficiency of the model. In different sample windows, CBAM-P-Net has an average increase of 1.17% and 0.0129 in testing overall accuracy (OA) and kappa coefficient (Kappa). The optimal classification window is 17 × 17, the OA reaches 97.28%, and Kappa reaches 0.97, which is an increase of 1.95% and 0.0214 along with just 49 s of training time expended, respectively, compared with P-Net. Therefore, using a suitable sample window and applying the proposed CBAM-P-Net to classify airborne hyperspectral images can achieve high-precision classification and mapping of forest tree species.