Stefano Puliti
Forsker
Forfattere
Binbin Xiang Maciej Wielgosz Theodora Kontogianni Torben Peters Stefano Puliti Rasmus Astrup Konrad SchindlerSammendrag
Det er ikke registrert sammendrag
Sammendrag
Det er ikke registrert sammendrag
Sammendrag
Det er ikke registrert sammendrag
Forfattere
Alexander Becker Stefania Russo Stefano Puliti Nico Lang Konrad Schindler Jan Dirk WegnerSammendrag
Monitoring and managing Earth’s forests in an informed manner is an important requirement for addressing challenges like biodiversity loss and climate change. While traditional in situ or aerial campaigns for forest assessments provide accurate data for analysis at regional level, scaling them to entire countries and beyond with high temporal resolution is hardly possible. In this work, we propose a method based on deep ensembles that densely estimates forest structure variables at country-scale with 10-m resolution, using freely available satellite imagery as input. Our method jointly transforms Sentinel-2 optical images and Sentinel-1 syntheticaperture radar images into maps of five different forest structure variables: 95th height percentile, mean height, density, Gini coefficient, and fractional cover. We train and test our model on reference data from 41 airborne laser scanning missions across Norway and demonstrate that it is able to generalize to unseen test regions, achieving normalized mean absolute errors between 11% and 15%, depending on the variable. Our work is also the first to propose a variant of so-called Bayesian deep learning to densely predict multiple forest structure variables with well-calibrated uncertainty estimates from satellite imagery. The uncertainty information increases the trustworthiness of the model and its suitability for downstream tasks that require reliable confidence estimates as a basis for decision making. We present an extensive set of experiments to validate the accuracy of the predicted maps as well as the quality of the predicted uncertainties. To demonstrate scalability, we provide Norway-wide maps for the five forest structure variables.
Forfattere
Stefano PulitiSammendrag
Det er ikke registrert sammendrag
Sammendrag
Det er ikke registrert sammendrag
Forfattere
Stefano Puliti Grant Pearse Peter Surovy Luke Wallace Markus Hollaus Maciej Wielgosz Rasmus AstrupSammendrag
The challenge of accurately segmenting individual trees from laser scanning data hinders the assessment of crucial tree parameters necessary for effective forest management, impacting many downstream applications. While dense laser scanning offers detailed 3D representations, automating the segmentation of trees and their structures from point clouds remains difficult. The lack of suitable benchmark datasets and reliance on small datasets have limited method development. The emergence of deep learning models exacerbates the need for standardized benchmarks. Addressing these gaps, the FOR-instance data represent a novel benchmarking dataset to enhance forest measurement using dense airborne laser scanning data, aiding researchers in advancing segmentation methods for forested 3D scenes. In this repository, users will find forest laser scanning point clouds from unamnned aerial vehicle (using Riegl sensors) that are manually segmented according to the individual trees (1130 trees) and semantic classes. The point clouds are subdivided into five data collections representing different forests in Norway, the Czech Republic, Austria, New Zealand, and Australia. These data are meant to be used either for developement of new methods (using the dev data) or for testing of exisitng methods (test data). The data splits are provided in the data_split_metadata.csv file. A full description of the FOR-instance data can be found at http://arxiv.org/abs/2309.01279
Forfattere
Stefano Puliti Grant Pearse Peter Surovy Luke Wallace Markus Hollaus Maciej Wielgosz Rasmus AstrupSammendrag
The FOR-instance dataset (available at this https URL) addresses the challenge of accurate individual tree segmentation from laser scanning data, crucial for understanding forest ecosystems and sustainable management. Despite the growing need for detailed tree data, automating segmentation and tracking scientific progress remains difficult. Existing methodologies often overfit small datasets and lack comparability, limiting their applicability. Amid the progress triggered by the emergence of deep learning methodologies, standardized benchmarking assumes paramount importance in these research domains. This data paper introduces a benchmarking dataset for dense airborne laser scanning data, aimed at advancing instance and semantic segmentation techniques and promoting progress in 3D forest scene segmentation. The FOR-instance dataset comprises five curated and ML-ready UAV-based laser scanning data collections from diverse global locations, representing various forest types. The laser scanning data were manually annotated into individual trees (instances) and different semantic classes (e.g. stem, woody branches, live branches, terrain, low vegetation). The dataset is divided into development and test subsets, enabling method advancement and evaluation, with specific guidelines for utilization. It supports instance and semantic segmentation, offering adaptability to deep learning frameworks and diverse segmentation strategies, while the inclusion of diameter at breast height data expands its utility to the measurement of a classic tree variable. In conclusion, the FOR-instance dataset contributes to filling a gap in the 3D forest research, enhancing the development and benchmarking of segmentation algorithms for dense airborne laser scanning data.
Sammendrag
Det er ikke registrert sammendrag
Sammendrag
Det er ikke registrert sammendrag
Sammendrag
Det er ikke registrert sammendrag
Forfattere
Adrian Straker Stefano Puliti Johannes Breidenbach Christopher Kleinn Grant Pearse Rasmus Astrup Paul MagdonSammendrag
Fine-grained information on the level of individual trees constitute key components for forest observation enabling forest management practices tackling the effects of climate change and the loss of biodiversity in forest ecosystems. Such information on individual tree crowns (ITC's) can be derived from the application of ITC segmentation approaches, which utilize remotely sensed data. However, many ITC segmentation approaches require prior knowledge about forest characteristics, which is difficult to obtain for parameterization. This can be avoided by the adoption of data-driven, automated workflows based on convolutional neural networks (CNN). To contribute to the advancements of efficient ITC segmentation approaches, we present a novel ITC segmentation approach based on the YOLOv5 CNN. We analyzed the performance of this approach on a comprehensive international unmanned aerial laser scanning (UAV-LS) dataset (ForInstance), which covers a wide range of forest types. The ForInstance dataset consists of 4192 individually annotated trees in high-density point clouds with point densities ranging from 498 to 9529 points m-2 collected across 80 sites. The original dataset was split into 70% for training and validation and 30% for model performance assessment (test data). For the best performing model, we observed a F1-score of 0.74 for ITC segmentation and a tree detection rate (DET %) of 64% in the test data. This model outperformed an ITC segmentation approach, which requires prior knowledge about forest characteristics, by 41% and 33% for F1-score and DET %, respectively. Furthermore, we tested the effects of reduced point densities (498, 50 and 10 points per m-2) on ITC segmentation performance. The YOLO model exhibited promising F1-scores of 0.69 and 0.62 even at point densities of 50 and 10 points m-2, respectively, which were between 27% and 34% better than the ITC approach that requires prior knowledge. Furthermore, the areas of ITC segments resulting from the application of the best performing YOLO model were close to the reference areas (RMSE = 3.19 m-2), suggesting that the YOLO-derived ITC segments can be used to derive information on ITC level.
Sammendrag
Det er ikke registrert sammendrag
Forfattere
Stefano PulitiSammendrag
Det er ikke registrert sammendrag
Sammendrag
Inthis study, we introduce Point2Tree, a modular and versatile framework that employs a three-tiered methodology, inclusive of semantic segmentation, instance segmentation, and hyperparameter optimization analysis, designed to process laser point clouds in forestry. The semantic segmentation stage is built upon the Pointnet++ architecture and is primarily tasked with categorizing each point in the point cloud into meaningful groups or ’segments’, specifically in this context, differentiating between diverse tree parts, i.e., vegetation, stems, and coarse woody debris. The category for the ground is also provided. Semantic segmentation achieved an F1-score of 0.92, showing a high level of accuracy in classifying forest elements. In the instance segmentation stage, we further refine this process by identifying each tree as a unique entity. This process, which uses a graph-based approach, yielded an F1-score of approximately 0.6, signifying reasonable performance in delineating individual trees. The third stage involves a hyperparameter optimization analysis, conducted through a Bayesian strategy, which led to performance improvement of the overall framework by around four percentage points. Point2Tree was tested on two datasets, one from a managed boreal coniferous forest in Våler, Norway, with 16 plots chosen to cover a range of forest conditions. The modular design of the framework allows it to handle diverse pointcloud densities and types of terrestrial laser scanning data.
Sammendrag
Sustainable forest management systems require operational measures to preserve the functional design of forest roads. Frequent road data collection and analysis are essential to support target-oriented and efficient maintenance planning and operations. This study demonstrates an automated solution for monitoring forest road surface deterioration using consumer-grade optical sensors. A YOLOv5 model with StrongSORT tracking was adapted and trained to detect and track potholes in the videos captured by vehicle-mounted cameras. For model training, datasets recorded in diverse geographical regions under different weather conditions were used. The model shows a detection and tracking performance of up to a precision and recall level of 0.79 and 0.58, respectively, with 0.70 mean average precision at an intersection over union (IoU) of at least 0.5. We applied the trained model to a forest road in southern Norway, recorded with a Global Navigation Satellite System (GNSS)−fitted dashcam. GNSS-delivered geographical coordinates at 10 Hz rate were used to geolocate the detected potholes. The geolocation performance over this exemple road stretch of 1 km exhibited a root mean square deviation of about 9.7 m compared to OpenStreetMap. Finally, an exemple road deterioration map was compiled, which can be used for scheduling road maintenance operations.
Forfattere
Binbin Xiang Torben Peters Theodora Kontogianni Frawa Vetterli Stefano Puliti Rasmus Astrup Konrad SchindlerSammendrag
Panoptic segmentation is the combination of semantic and instance segmentation: assign the points in a 3D point cloud to semantic categories and partition them into distinct object instances. It has many obvious applications for outdoor scene understanding, from city mapping to forest management. Existing methods struggle to segment nearby instances of the same semantic category, like adjacent pieces of street furniture or neighbouring trees, which limits their usability for inventory- or management-type applications that rely on object instances. This study explores the steps of the panoptic segmentation pipeline concerned with clustering points into object instances, with the goal to alleviate that bottleneck. We find that a carefully designed clustering strategy, which leverages multiple types of learned point embeddings, significantly improves instance segmentation. Experiments on the NPM3D urban mobile mapping dataset and the FOR-instance forest dataset demonstrate the effectiveness and versatility of the proposed strategy.
Sammendrag
Information on tree height-growth dynamics is essential for optimizing forest management and wood procurement. Although methods to derive information on height-growth information from multi-temporal laser scanning data already exist, there is no method to derive such information from data acquired at a single point in time. Drone laser scanning data (unmanned aerial vehicles, UAV-LS) allows for the efficient collection of very dense point clouds, creating new opportunities to measure tree and branch architecture. In this study, we examine if it is possible to measure the vertical positions of branch whorls, which correspond to nodes, and thus can in turn be used to trace the height growth of individual trees. We propose a method to measure the vertical positions of whorls based on a single-acquisition of UAV-LS data coupled with deep-learning techniques. First, single-tree point clouds were converted into 2D image projections, and a YOLOv5 (you-only-look-once) convolutional neural network was trained to detect whorls based on a sample of manually annotated images. Second, the trained whorl detector was applied to a set of 39 trees that were destructively sampled after the UAV-LS data acquisition. The detected whorls were then used to estimate tree-, plot- and stand-level height-growth trajectories. The results indicated that 70 per cent (i.e. precision) of the measured whorls were correctly detected and that 63 per cent (i.e. recall) of the detected whorls were true whorls. These results translated into an overall root-mean-squared error and Bias of 8 and −5 cm for the estimated mean annual height increment. The method’s performance was consistent throughout the height of the trees and independent of tree size. As a use case, we demonstrate the possibility of developing a height-age curve, such as those that could be used for forecasting site productivity. Overall, this study provides proof of concept for new methods to analyse dense aerial point clouds based on image-based deep-learning techniques and demonstrates the potential for deriving useful analytics for forest management purposes at operationally-relevant spatial-scales.
Sammendrag
Det er ikke registrert sammendrag
Forfattere
Emily Lines Matthew Allen Carlos Cabo Kim Calders Amandine Debus Stuart Grieve Milto Miltiadou Adam Noach Harry JF Owen Stefano PulitiSammendrag
With the rise in high resolution remote sensing technologies there has been an explosion in the amount of data available for forest monitoring, and an accompanying growth in artificial intelligence applications to automatically derive forest properties of interest from these datasets. Many studies use their own data at small spatio-temporal scales, and demonstrate an application of an existing or adapted data science method for a particular task. This approach often involves intensive and time-consuming data collection and processing, but generates results restricted to specific ecosystems and sensor types. There is a lack of widespread acknowledgement of how the types and structures of data used affects performance and accuracy of analysis algorithms. To accelerate progress in the field more efficiently, benchmarking datasets upon which methods can be tested and compared are sorely needed.Here, we discuss how lack of standardisation impacts confidence in estimation of key forest properties, and how considerations of data collection need to be accounted for in assessing method performance. We present pragmatic requirements and considerations for the creation of rigorous, useful benchmarking datasets for forest monitoring applications, and discuss how tools from modern data science can improve use of existing data. We list a set of example large-scale datasets that could contribute to benchmarking, and present a vision for how community-driven, representative benchmarking initiatives could benefit the field.
Sammendrag
The assessment of forest abiotic damages such as snow breakage is important to ensure compensation to forest owners. Currently, information on the extent of snow breakage is gathered through time-consuming and potentially biased field surveys. In such situations where field surveys are still common practice, unmanned aerial vehicles (UAVs) are increasingly being used to provide a more cost-efficient and objective methods to answer forest information needs. Further, the advent of sophisticated computer vision techniques such as convolutional neural networks (CNNs) offers new ways to analyze image data more efficiently and accurately. We proposed an object detection method to automatically identify trees and classify them according to the damage by snow based on a YOLO CNN architecture. UAV imagery collected across 89 study areas and over the course of the entire year were manually annotated into a total of >55 K single trees classified as healthy, damaged, or dead. The annotated trees, along with the corresponding UAV imagery were used to train a YOLOv5 object detection model. Furthermore, we tested the effect of seasonality, and varying atmospheric and lighting conditions on the model’s performance. Based on an independent test set of data we found that the general model including all of the data (i.e. any seasons, atmospheric conditions, and time of the day) outperformed all other tested scenarios (i.e. precision = 62 %; recall = 61 %). Furthermore, we found that despite the fact that the snow damaged trees represented a minority class (i.e. 16 % of the annotated trees), they were detected with the largest precision (76 %) and recall (78 %). Finally, the general model transferred well across the variation in seasons, atmospheric and illumination conditions, making it suitable for usage for any new UAV image acquisition.
Sammendrag
Det er ikke registrert sammendrag
Forfattere
Saheba Bhatnagar Stefano Puliti Bruce Talbot Joachim Bernd Heppelmann Johannes Breidenbach Rasmus AstrupSammendrag
Wheel ruts, i.e. soil deformations caused by harvesting machines, are considered a negative environmental impact of forest operations and should be avoided or ameliorated. However, the mapping of wheel ruts that would be required to monitor harvesting operations and to plan amelioration measures is a tedious and time-consuming task. Here, we examined whether a combination of drone imagery and algorithms from the field of artificial intelligence can automate the mapping of wheel ruts. We used a deep-learning image-segmentation method (ResNet50 + UNet architecture) that was trained on drone imagery acquired shortly after harvests in Norway, where more than 160 km of wheel ruts were manually digitized. The cross-validation of the model based on 20 harvested sites resulted in F1 scores of 0.69–0.84 with an average of 0.77, and in total, 79 per cent of wheel ruts were correctly detected. The highest accuracy was obtained for severe wheel ruts (average user’s accuracy (UA) = 76 per cent), and the lowest accuracy was obtained for light wheel ruts (average UA = 67 per cent). Considering the nowadays ubiquitous availability of drones, the approach presented in our study has the potential to greatly increase the ability to effectively map and monitor the environmental impact of final felling operations with respect to wheel ruts. The automated mapping of wheel ruts may serve as an important input to soil impact analyses and thereby support measures to restore soil damages.
Sammendrag
Det er ikke registrert sammendrag
Forfattere
Benjamin Allen Michele Dalponte Hans Ole Ørka Erik Næsset Stefano Puliti Rasmus Astrup Terje GobakkenSammendrag
Numerous species of pathogenic wood decay fungi, including members of the genera Heterobasidion and Armillaria, exist in forests in the northern hemisphere. Detection of these fungi through field surveys is often difficult due to a lack of visual symptoms and is cost-prohibitive for most applications. Remotely sensed data can offer a lower-cost alternative for collecting information about vegetation health. This study used hyperspectral imagery collected from unmanned aerial vehicles (UAVs) to detect the presence of wood decay in Norway spruce (Picea abies L. Karst) at two sites in Norway. UAV-based sensors were tested as they offer flexibility and potential cost advantages for small landowners. Ground reference data regarding pathogenic wood decay were collected by harvest machine operators and field crews after harvest. Support vector machines were used to classify the presence of root, butt, and stem rot infection. Classification accuracies as high as 76% with a kappa value of 0.24 were obtained with 490-band hyperspectral imagery, while 29-band imagery provided a lower classification accuracy (~60%, kappa = 0.13).
Sammendrag
UAV, drone, ortho photo, image analysis, deep learning, wheel rut, monitoring, GIS, method
Forfattere
Stefano Puliti Grant D. Pearse Michael S. Watt Edward Mitchard Ian McNicol Magnus Bremer Martin Rutzinger Peter Surovy Luke Wallace Markus Hollaus Rasmus AstrupSammendrag
Survey-grade laser scanners suitable for drones (UAV-LS) allow the efficient collection of finely detailed three-dimensional (3D) information on tree structures allowing to resolve the complexity of the forest into discrete individual trees and species as well as into different component of the tree. Current developments are hindered by the limited availability of survey-grade UAV-LS data and by the lack of a publicly available benchmark dataset for developing and validating methods. We present a new benchmarking dataset composed of manually labelled UAV-LS data covering forests in different continents and eco-regions. Such data consists in single-tree point clouds, with each point classified as either stem, branches, and leaves. This benchmark dataset offers new possibilities to develop single-tree segmentation algorithms and validate existing ones.
Forfattere
Stefano Puliti Grant Pearse Michael Watt Edward Mitchard Ian McNicol Magnus Bremer Martin Rutzinger Peter Surovy Luke Wallace Markus Hollaus Rasmus AstrupSammendrag
Det er ikke registrert sammendrag
Forfattere
Stefano Puliti Johannes Breidenbach Johannes Schumacher Marius Hauglin Torgeir Ferdinand Klingenberg Rasmus AstrupSammendrag
This study aimed at estimating total forest above-ground net change (ΔAGB; Gg) over five years (2014–2019) based on model-assisted estimation utilizing freely available satellite imagery. The study was conducted for a boreal forest area (approx. 1.4 Mha) in Norway where bi-temporal national forest inventory (NFI), Sentinel-2, and Landsat data were available. Biomass change was modelled based on a direct approach. The precision of estimates using only the NFI data in a basic expansion estimator was compared to four different alternative model-assisted estimates using 1) Sentinel-2 or Landsat data, and 2) using bi- or uni-temporal remotely sensed data. We found that spaceborne optical data improved the precision of the purely field-based estimates by a factor of up to three. The most precise estimates were found for the model-assisted estimation using bi-temporal Sentinel-2 (standard error; SE = 1.7 Gg). However, the decrease in precision when using Landsat data was small (SE = 1.92 Gg). We also found that ΔAGB could be precisely estimated when remotely sensed data were available only at the end of the monitoring period. We conclude that satellite optical data can considerably improve ΔAGB estimates, when repeated and coincident field data are available. The free availability, global coverage, frequent update, and long-term time horizon make data from programs such as Sentinel-2 and Landsat a valuable data source for consistent and durable monitoring of forest carbon dynamics.
Sammendrag
There is a need for mapping of forest areas with young stands under regeneration in Norway, as a basis for conducting tending, or precommercial thinning (PCT), whenever necessary. The main objective of this article is to show the potential of multitemporal Sentinel-1 (S-1) and Sentinel-2 (S-2) data for characterization and detection of forest stands under regeneration. We identify the most powerful radar and optical features for discrimination of forest stands under regeneration versus other forest stands. A number of optical and radar features derived from multitemporal S-1 and S-2 data were used for the class separability and cross-correlation analysis. The analysis was performed on forest resource maps consisting of the forest development classes and age in two study sites from south-eastern Norway. Important features were used to train the classical random forest (RF) classification algorithm. A comparative study of performance of the algorithm was used in three cases: I) using only S-1 features, II) using only S-2 optical bands, and III) using combination of S-1 and S-2 features. RF classification results pointed to increased class discrimination when using S-1 and S-2 data in relation to S-1 or S-2 data only. The study shows that forest stands under regeneration in the height interval for PCT can be detected with a detection rate of 91% and F-1 score of 73.2% in case III as most accurate, while tree density and broadleaf fraction could be estimated with coefficient of determination ( R2 ) of about 0.70 and 0.80, respectively.
Forfattere
Stefano Puliti Johannes Breidenbach Johannes Schumacher Marius Hauglin Torgeir Ferdinand KlingenbergSammendrag
Det er ikke registrert sammendrag
Sammendrag
In the EU 2020 biodiversity strategy, maintaining and enhancing forest biodiversity is essential. Forest managers and technicians should include biodiversity monitoring as support for sustainible forest management and conservation issues, through the adoption of forest biodiversity indices. The present study investigates the potential of a new type of Structure from Motion (SfM) photogrammetry derived variables for modelling forest structure indicies, which do not require the availability of a digital terrain model (DTM) such as those obtainable from Airborne Laser Scanning (ALS) surveys. The DTM-independent variables were calculated using raw 3D UAV photogrammetric data for modeling eight forest structure indices which are commonly used for forest biodiversity monitoring, namely: basal area (G); quadratic mean diameter (DBHmean); the standard deviation of Diameter at Breast Height (DBHσ); DBH Gini coefficient (Gini); the standard deviation of tree heights (Hσ); dominant tree height (Hdom); Lorey’s height (Hl); and growing stock volume (V). The study included two mixed temperate forestsareas withadifferenttype ofmanagement, with onearea, left unmanagedfor thepast 50years while the other being actively managed. A total of 30 fieldsample plots were measured in the unmanaged forest, and 50 field plots were measured in the actively managed forest. The accuracy of UAV DTM-independent predictions was compared with a benchmark approach based on traditional explanatory variables calculated from ALS data. Finally, DTM-independent variables were used to produce wall-to-wall maps of the forest structure indices in the two test areas and to estimate the mean value and its uncertainty according to a model-assisted regression estimators. DTM-independent variables led to similar predictive accuracy in terms of root mean square error compared to ALS in both study areas for the eight structure indices (DTM-independent average RMSE% = 20.5 and ALS average RMSE% = 19.8). Moreover, we found that the model-assisted estimation, with both DTM-independet and ALS, obtained lower standar errors (SE) compared to the one obtained by modelbased estimation using only field plots. Relative efficiency coefficient (RE) revealed that ALS-based estimates were, on average, more efficient (average RE ALS = 3.7) than DTM-independent, (average RE DTM-independent = 3.3). However, the RE for the DTM-independent models was consistently larger than the one from theALSmodelsfortheDBH-relatedvariables(i.e.G,DBHmean,andDBHσ)andforV.Thishighlightsthepotential of DTM-independent variables, which not only can be used virtually on any forests (i.e., no need of a DTM), but also can produce as precise estimates as those from ALS data for key forest structural variables and substantially improve the efficiency of forest inventories.
Sammendrag
In this study, we aim at developing ways to directly translate raw drone data into actionable insights, thus enabling us to make management decisions directly from drone data. Drone photogrammetric data and data analytics were used to model stand-level immediate tending need and cost in regeneration forests. Field reference data were used to train and validate a logistic model for the binary classification of immediate tending need and a multiple linear regression model to predict the cost to perform the tending operation. The performance of the models derived from drone data was compared to models utilizing the following alternative data sources: airborne laser scanning data (ALS), prior information from forest management plans (Prior) and the combination of drone +Prior and ALS +Prior. The use of drone data and prior information outperformed the remaining alternatives in terms of classification of tending needs, whereas drone data alone resulted in the most accurate cost models. Our results are encouraging for further use of drones in the operational management of regeneration forests and show that drone data and data analytics are useful for deriving actionable insights. Key words: UAV, DAP, forest inventory, photogrammetry, precommercial thinning, airborne laser scanning.
Sammendrag
Laser scanning data from unmanned aerial vehicles (UAV-LS) offer new opportunities to estimate forest growing stock volume ( V ) exclusively based on the UAV-LS data. We propose a method to measure tree attributes and using these measurements to estimate V without the use of field data for calibration. The method consists of five steps: i) Using UAV-LS data, tree crowns are automatically identified and segmented wall-to-wall. ii) From all detected tree crowns, a sample is taken where diameter at breast height (DBH) can be recorded reliably as determined by visual assessment in the UAV-LS data. iii) Another sample of crowns is taken where tree species were identifiable from UAV image data. iv) DBH and tree species models are fit using the samples and applied to all detected tree crowns. v) Single tree volumes are predicted with existing allometric models using predicted species and DBH, and height directly obtained from UAV-LS. The method was applied to a Riegl-VUX data set with an average density of 1130 points m−2 and 3 cm orthomosaic acquired over an 8.8 ha managed boreal forest. The volumes of the identified trees were aggregated to estimate plot-, stand-, and forest-level volumes which were validated using 58 independently measured field plots. The root-mean-square deviance ( RMSD% ) decreased when increasing the spatial scale from the plot (32.2%) to stand (27.1%) and forest level (3.5%). The accuracy of the UAV-LS estimates varied given forest structure and was highest in open pine stands and lowest in dense birch or spruce stands. On the forest level, the estimates based on UAV-LS data were well within the 95% confidence interval of the intense field survey estimate, and both estimates had a similar precision. While the results are encouraging for further use of UAV-LS in the context of fully airborne forest inventories, future studies should confirm our findings in a variety of forest types and conditions.
Forfattere
Stefano Puliti Marius Hauglin Johannes Breidenbach P. Montesano C.S.R. Neigh Johannes Rahlf Svein Solberg T. F. Klingenberg Rasmus AstrupSammendrag
Boreal forests constitute a large portion of the global forest area, yet they are undersampled through field surveys, and only a few remotely sensed data sources provide structural information wall-to-wall throughout the boreal domain. ArcticDEM is a collection of high-resolution (2 m) space-borne stereogrammetric digital surface models (DSM) covering the entire land area north of 60° of latitude. The free-availability of ArcticDEM data offers new possibilities for aboveground biomass mapping (AGB) across boreal forests, and thus it is necessary to evaluate the potential for these data to map AGB over alternative open-data sources (i.e., Sentinel-2). This study was performed over the entire land area of Norway north of 60° of latitude, and the Norwegian national forest inventory (NFI) was used as a source of field data composed of accurately geolocated field plots (n=7710) systematically distributed across the study area. Separate random forest models were fitted using NFI data, and corresponding remotely sensed data consisting of either: i) a canopy height model (ArcticCHM) obtained by subtracting a high-quality digital terrain model (DTM) from the ArcticDEM DSM height values, ii) Sentinel-2 (S2), or iii) a combination of the two (ArcticCHM+S2). Furthermore, we assessed the effect of the forest- and terrain-specific factors on the models’ predictive accuracy. The best model (,i.e., ArcticCHM+S2) explained nearly 60% of the variance of the training set, which translated in the largest accuracy in terms of root mean square error (RMSE=41.4 t ha−1 ). This result highlights the synergy between 3D and multispectral data in AGB modelling. Furthermore, this study showed that despite the importance of ArcticCHM variables, the S2 model performed slightly better than ArcticCHM model. This finding highlights some of the limitations of ArcticDEM, which, despite the unprecedented spatial resolution, is highly heterogeneous due to the blending of multiple acquisitions across different years and seasons. We found that both forest- and terrain-specific characteristics affected the uncertainty of the ArcticCHM+S2 model and concluded that the combined use of ArcticCHM and Sentinel-2 represents a viable solution for AGB mapping across boreal forests. The synergy between the two data sources allowed for a reduction of the saturation effects typical of multispectral data while ensuring the spatial consistency in the output predictions due to the removal of artifacts and data voids present in ArcticCHM data. While the main contribution of this study is to provide the first evidence of the best-case-scenario (i.e., availability of accurate terrain models) that ArcticDEM data can provide for large-scale AGB modelling, it remains critically important for other studies to investigate how ArcticDEM may be used in areas where no DTMs are available as is the case for large portions of the boreal zone.
Forfattere
Stefano Puliti Marius Hauglin Johannes Breidenbach Paul M. Montesano C.S.R. Neigh Johannes Rahlf Svein Solberg Torgeir Ferdinand Klingenberg Rasmus AstrupSammendrag
Det er ikke registrert sammendrag
Forfattere
Johannes Breidenbach Lars T. Waser Misganu Debella-Gilo Johannes Schumacher Johannes Rahlf Marius Hauglin Stefano Puliti Rasmus AstrupSammendrag
Det er ikke registrert sammendrag
Forfattere
Johannes Breidenbach Lars T. Waser Misganu Debella-Gilo Johannes Schumacher Johannes Rahlf Marius Hauglin Stefano Puliti Rasmus AstrupSammendrag
Nation-wide Sentinel-2 mosaics were used with National Forest Inventory (NFI) plot data for modelling and subsequent mapping of spruce-, pine-, and deciduous-dominated forest in Norway at a 16 m × 16 m resolution. The accuracies of the best model ranged between 74% for spruce and 87% for deciduous forest. An overall accuracy of 90% was found on stand level using independent data from more than 42 000 stands. Errors mostly resulting from a forest mask reduced the model accuracies by ∼10%. The produced map was subsequently used to generate model-assisted (MA) and poststratified (PS) estimates of species-specific forest area. At the national level, efficiencies of the estimates increased by 20% to 50% for MA and up to 90% for PS. Greater minimum numbers of observations constrained the use of PS. For MA estimates of municipalities, efficiencies improved by up to a factor of 8 but were sometimes also less than 1. PS estimates were always equally as or more precise than direct and MA estimates but were applicable in fewer municipalities. The tree species prediction map is part of the Norwegian forest resource map and is used, among others, to improve maps of other variables of interest such as timber volume and biomass.
Sammendrag
Det er ikke registrert sammendrag
Sammendrag
Det er ikke registrert sammendrag
Sammendrag
A novel method for age-independent site index estimation is demonstrated using repeated single-tree airborne laser scanning (ALS) data. A spruce-dominated study area of 114 km2 in southern Norway was covered by single-tree ALS twice, i.e. in 2008 and 2014. We identified top height trees wall-to-wall, and for each of them we derived based on the two heights and the 6-year period length. We reconstructed past, annual height growth in a field campaign on 31 sample trees, and this showed good correspondence with ALS based heights. We found a considerable increase in site index, i.e. about 5 m in the H40 system, from the old site index values. This increase corresponded to a productivity increase of 62%. This increase appeared to mainly represent a real temporal trend caused by changing growing conditions. In addition, the increase could partly result from underestimation in old site index values. The method has the advantages of not requiring tree-age data, of representing current growing conditions, and as well that it is a cost-effective method with wall-towall coverage. In slow-growing forests and short time periods, the method is least reliable due to possible systematic differences in canopy penetration between repeated ALS scans.
Sammendrag
Global Forest Watch (GFW) provides a global map of annual forest cover loss (FCL) produced from Landsat imagery, offering a potentially powerful tool for monitoring changes in forest cover. In managed forests, FCL primarily provides information on commercial harvesting. A semi-autonomous method for providing data on the location and attributes of harvested sites at a landscape level was developed which could significantly improve the basis for catchment management, including risk mitigation. FCL in combination with aerial images was used for detecting and characterising harvested sites in a 1607 km2 mountainous boreal forest catchment in south-central Norway. Firstly, the forest cover loss map was enhanced (FCLE) by removing small isolated forest cover loss patches that had a high probability of representing commission errors. The FCLE map was then used to locate and assess sites representing annual harvesting activity over a 17-year period. Despite an overall accuracy of >98%, a kappa of 0.66 suggested only a moderate quality for detecting harvested sites. While errors of commission were negligible, errors of omission were more considerable and at least partially attributed to the presence of residual seed trees on the site after harvesting. The systematic analysis of harvested sites against aerial images showed a detection rate of 94%, but the area of the individual harvested site was underestimated by 29% on average. None of the site attributes tested, including slope, area, altitude, or site shape index, had any effect on the accuracy of the area estimate. The annual harvest estimate was 0.6% (standard error 12%) of the productive forest area. On average, 96% of the harvest was carried out on flat to moderately steep terrain (<40% slope), 3% on steep terrain (40% to 60% slope), and 1% on very steep terrain (>60% slope). The mean area of FCLE within each slope category was 1.7 ha, 0.9 ha, and 0.5 ha, respectively. The mean FCLE area increased from 1.0 ha to 3.2 ha on flat to moderate terrain over the studied period, while the frequency of harvesting increased from 249 to 495 sites per year. On the steep terrain, 35% of the harvesting was done with cable yarding, and 62% with harvester-forwarder systems. On the very steep terrain (>60% slope), 88% of the area was harvested using cable yarding technology while harvesters and forwarders were used on 12% of the area. Overall, FCL proved to be a useful dataset for the purpose of assessing harvesting activity under the given conditions.
Sammendrag
Det er ikke registrert sammendrag
Sammendrag
Root and butt-rot (RBR) has a significant impact on both the material and economic outcome of timber harvesting, and therewith on the individual forest owner and collectively on the forest and wood processing industries. An accurate recording of the presence of RBR during timber harvesting would enable a mapping of the location and extent of the problem, providing a basis for evaluating spread in a climate anticipated to enhance pathogenic growth in the future. Therefore, a system to automatically identify and detect the presence of RBR would constitute an important contribution in addressing the problem without increasing workload complexity for the machine operator. In this study we developed and evaluated an approach based on RGB images to automatically detect tree-stumps and classify them as to the absence or presence of rot. Furthermore, since knowledge of the extent of RBR is valuable in categorizing logs, we also classify stumps to three classes of infestation; rot = 0%, 0% < rot < 50% and rot >= 50%. In this work we used deep learning approaches and conventional machine learning algorithms for detection and classification tasks. The results showed that tree-stumps were detected with precision rate of 95% and recall of 80%. Using only the correct output (TP) of the stump detector, stumps without and with root and butt-rot were correctly classified with accuracy of 83.5% and 77.5%. Classifying rot to three classes resulted in 79.4%, 72.4% and 74.1% accuracy for stumps with rot = 0%, 0% < rot < 50% and rot >= 50\%, respectively. With some modifications, the algorithm developed could be used either during the harvesting operation to detect RBR regions on the tree-stumps or as a RBR detector for post-harvest assessment of tree-stumps and logs.
Sammendrag
Det er ikke registrert sammendrag
Sammendrag
Det er ikke registrert sammendrag
Sammendrag
Det er ikke registrert sammendrag
Sammendrag
Det er ikke registrert sammendrag
Sammendrag
Det er ikke registrert sammendrag
Forfattere
Jakob Iglhaut Carlos Cabo Stefano Puliti Livia Piermattei James O’Connor Jacqueline RosetteSammendrag
Purpose of Review The adoption of Structure from Motion photogrammetry (SfM) is transforming the acquisition of three-dimensional (3D) remote sensing (RS) data in forestry. SfM photogrammetry enables surveys with little cost and technical expertise. We present the theoretical principles and practical considerations of this technology and show opportunities that SfM photogrammetry offers for forest practitioners and researchers. Recent Findings Our examples of key research indicate the successful application of SfM photogrammetry in forestry, in an operational context and in research, delivering results that are comparable to LiDAR surveys. Reviewed studies have identified possibilities for the extraction of biophysical forest parameters from airborne and terrestrial SfM point clouds and derived 2D data in area-based approaches (ABA) and individual tree approaches. Additionally, increases in the spatial and spectral resolution of sensors available for SfM photogrammetry enable forest health assessment and monitoring. The presented research reveals that coherent 3D data and spectral information, as provided by the SfM workflow, promote opportunities to derive both structural and physiological attributes at the individual tree crown (ITC) as well as stand levels. Summary We highlight the potential of using unmanned aerial vehicles (UAVs) and consumer-grade cameras for terrestrial SfM-based surveys in forestry. Offering several spatial products from a single sensor, the SfM workflow enables foresters to collect their own fit-for-purpose RS data. With the broad availability of non-expert SfM software, we provide important practical considerations for the collection of quality input image data to enable successful photogrammetric surveys.
Sammendrag
The objective of this study was to assess the use of unmanned aerial vehicle (UAV) data for modelling tree density and canopy height in young boreal forests stands. The use of UAV data for such tasks can be beneficial thanks to the high resolution and reduction of the time spent in the field. This study included 29 forest stands, within which 580 clustered plots were measured in the field. An area-based approach was adopted to which random forest models were fitted using the plot data and the corresponding UAV data and then applied and validated at plot and stand level. The results were compared to those of models based on airborne laser scanning (ALS) data and those from a traditional field-assessment. The models based on UAV data showed the smallest stand-level RMSE values for mean height (0.56 m) and tree density (1175 trees ha−1 ). The RMSE of the tree density using UAV data was 50% smaller than what was obtained using ALS data (2355 trees ha−1 ). Overall, this study highlighted that the use of UAVs for the inventory of forest stands under regeneration can be beneficial both because of the high accuracy of the derived data analytics and the time saving compared to traditional field assessments.
Sammendrag
Det er ikke registrert sammendrag
Sammendrag
Det er ikke registrert sammendrag
Forfattere
Stefano PulitiSammendrag
Det er ikke registrert sammendrag
Forfattere
Francesca Giannetti Gherardo Chirici Terje Gobakken Erik Næsset Davide Travaglini Stefano PulitiSammendrag
Det er ikke registrert sammendrag
Sammendrag
Det er ikke registrert sammendrag
Forfattere
Stefano PulitiSammendrag
Det er ikke registrert sammendrag
Sammendrag
Uganda designated 16% of its land as Protected Area (PA). The original goal was natural resources, habitat and biodiversity conservation. However, PAs also offer great potential for carbon conservation in the context of climate change mitigation. Drawing on a wall-to-wall map of forest carbon change for the entire Uganda, that was developed using two Digital Elevation Model (DEM) datasets for the period 2000–2012, we (1) quantified forest carbon gain and loss within 713 PAs and their external buffer zones, (2) tested variations in forest carbon change among management categories, and (3) evaluated the effectiveness of PAs and the prevalence of local leakage in terms of forest carbon. The net annual forest carbon gain in PAs of Uganda was 0.22 ± 1.36 t/ha, but a significant proportion (63%) of the PAs exhibited a net carbon loss. Further, carbon gain and loss varied significantly among management categories. About 37% of the PAs were “effective”, i.e., gained or at least maintained forest carbon during the period. Nevertheless, carbon losses in the external buffer zones of those effective PAs significantly contrast with carbon gains inside of the PA boundaries, providing evidence of leakage and thus, isolation. The combined carbon losses inside the boundaries of a large number of PAs, together with leakage in external buffer zones suggest that PAs, regardless of the management categories, are threatened by deforestation and forest degradation. If Uganda will have to benefit from carbon conservation from its large number of PAs through climate change mitigation mechanisms such as REDD+, there is an urgent need to look into some of the current PA management approaches, and design protection strategies that account for the surrounding landscapes and communities outside of the PAs.
Sammendrag
Det er ikke registrert sammendrag
Forfattere
Stefano PulitiSammendrag
Det er ikke registrert sammendrag
Forfattere
Stefano PulitiSammendrag
Det er ikke registrert sammendrag
Forfattere
Stefano PulitiSammendrag
Det er ikke registrert sammendrag
Forfattere
Sam Steyaert Shane Frank Stefano Puliti Rudy Badia Mie P. Arnberg Jack Beardsley Asle Økelsrud Rakel BlaalidSammendrag
Cadaver decomposition islands around animal carcasses can facilitate establishment of various plant life. Facultative scavengers have great potential for endozoochory, and often aggregate around carcasses. Hence, they may disperse plant seeds that they ingest across the landscape towards cadaver decomposition islands. Here, we demonstrate this novel mechanism along a gradient of wild tundra reindeer carcasses. First, we show that the spatial distribution of scavenger faeces (birds and foxes) was concentrated around carcasses. Second, faeces of the predominant scavengers (corvids) commonly contained viable seeds of crowberry, a keystone species of the alpine tundra with predominantly vegetative reproduction. We suggest that cadaver decomposition islands function as endpoints for directed endozoochory by scavengers. Such a mechanism could be especially beneficial for species that rely on small-scale disturbances in soil and vegetation, such as several Nordic berry-producing species with cryptic generative reproduction.
Sammendrag
Unmanned aerial vehicles (UAVs) are increasingly used as tools to perform a detailed assessment of post-harvest sites. One of the potential use of UAV photogrammetric data is to obtain tree-stump information that can then be used to support more precise decisions. This study developed and tested a methodology to automatically detect, segment, classify, and measure tree-stumps. Among the potential applications for single stump data, this study assessed the possibility (1) to detect and map root- and butt-rot on the stumps using a machine learning approach, and (2) directly measure or model tree stump diameter from the UAV data. The results revealed that the tree-stumps were detected with an overall accuracy of 68–80%, and once the stump was detected, the presence of root- and butt-rot was detected with an accuracy of 82.1%. Furthermore, the root mean square error of the UAV-derived measurements or model predictions for the stump diameter was 7.5 cm and 6.4 cm, respectively, and with the former systematically under predicting the diameter by 3.3 cm. The results of this study are promising and can lead to the development of more cost-effective and comprehensive UAV post-harvest surveys.
Forfattere
Stefano Puliti Svein Solberg Erik Næsset Terje Gobakken Eliakimu Zahabu Ernest William Mauya Rogers Ernest MalimbwiSammendrag
The use of Interferometric Synthetic Aperture Radar (InSAR) data has great potential for monitoring large scale forest above ground biomass (AGB) in the tropics due to the increased ability to retrieve 3D information even under cloud cover. To date; results in tropical forests have been inconsistent and further knowledge on the accuracy of models linking AGB and InSAR height data is crucial for the development of large scale forest monitoring programs. This study provides an example of the use of TanDEM-X WorldDEM data to model AGB in Tanzanian woodlands. The primary objective was to assess the accuracy of a model linking AGB with InSAR height from WorldDEM after the subtraction of ground heights. The secondary objective was to assess the possibility of obtaining InSAR height for field plots when the terrain heights were derived from global navigation satellite systems (GNSS); i.e., as an alternative to using airborne laser scanning (ALS). The results revealed that the AGB model using InSAR height had a predictive accuracy of RMSE = 24.1 t·ha−1 ; or 38.8% of the mean AGB when terrain heights were derived from ALS. The results were similar when using terrain heights from GNSS. The accuracy of the predicted AGB was improved when compared to a previous study using TanDEM-X for a sub-area of the area of interest and was of similar magnitude to what was achieved in the same sub-area using ALS data. Overall; this study sheds new light on the opportunities that arise from the use of InSAR data for large scale AGB modelling in tropical woodlands.
Divisjon for skog og utmark
PathFinder
Towards an Integrated Consistent European LULUCF Monitoring and Policy Pathway Assessment Framework
Divisjon for skog og utmark
Lukkede hogster: Konsekvenser for produksjon, økonomi og biomangfold
Hovedmålet til prosjektet er å bygge kunnskap og kompetanse om hvordan en omlegging av norsk skogbruk, fra flatehogst til mer bruk av lukkede hogster, vil påvirke produksjon, økonomi og biologisk mangfold.