arXiv subjects

Johannes Breidenbach

Publications and source records attributed to Johannes Breidenbach.

10 recordsLinked to original sources

arXiv2026

Enhancing Tree Species Classification: Insights from YOLOv8 and Explainable AI Applied to TLS Point Cloud Projections

Aiming to advance research in the field of interpretability of deep learning models for tree species classification using TLS 3D point clouds we present insights in the classification abilities of YOLOv8 through a new framework which enables systematic analysis of saliency maps derived from CAM (Class Activation Mapping). To investigate the contribution of structural tree features to the classification decisions of the models, we link regions with high saliency derived from the application of Finer-CAM to segments of 2D side-view images that correspond to structural tree features. Using TLS 3D point clouds from 2445 trees across seven European tree species, we trained five YOLOv8 models with cross-validation, reaching a mean accuracy of 96% (SD = 0.24%) when applied to the test data. Our results demonstrate that Finer-CAM can be considered faithful in identifying discriminative regions that discriminate target tree species. This renders Finer-CAM suitable for enhancing the interpretability of the tree species classification models. Analysis of 630 saliency maps indicate that the models primarily rely on image regions associated with tree crowns for species classification. While this result is pronounced in Silver Birch, European Beech, English oak, and Norway Spruce, image regions associated with stems contribute more frequently to the differentiation of European ash, Scots pine, and Douglas-fir. We demonstrate that the visibility of detailed structural tree features in the 2D side-view images enhances the discriminative performances of the models, indicating YOLOv8`s abilities to leverage detailed point cloud representations. Our results represent a first step toward enhancing the understanding of the classification decision processes of tree species classification models, aiding in the identification of data set and model limitations, and building confidence in model predictions.

cs.CV↗

arXiv2023

Lidar-based Norwegian tree species detection using deep learning

Background: The mapping of tree species within Norwegian forests is a time-consuming process, involving forest associations relying on manual labeling by experts. The process can involve both aerial imagery, personal familiarity, or on-scene references, and remote sensing data. The state-of-the-art methods usually use high resolution aerial imagery with semantic segmentation methods. Methods: We present a deep learning based tree species classification model utilizing only lidar (Light Detection And Ranging) data. The lidar images are segmented into four classes (Norway Spruce, Scots Pine, Birch, background) with a U-Net based network. The model is trained with focal loss over partial weak labels. A major benefit of the approach is that both the lidar imagery and the base map for the labels have free and open access. Results: Our tree species classification model achieves a macro-averaged F1 score of 0.70 on an independent validation with National Forest Inventory (NFI) in-situ sample plots. That is close to, but below the performance of aerial, or aerial and lidar combined models.

cs.CV↗

arXiv2022

Assessing and mitigating systematic errors in forest attribute maps utilizing harvester and airborne laser scanning data

Cut-to-length harvesters collect useful information for modeling relationships between forest attributes and airborne laser scanning (ALS) data. However, harvesters operate in mature forests, which may introduce selection biases that can result in systematic errors in harvester data-based forest attribute maps. We fitted regression models (harvester models) for volume (V), height (HL), stem frequency (N), above-ground biomass, basal area, and quadratic mean diameter (QMD) using harvester and ALS data. Performances of the harvester models were evaluated using national forest inventory plots in an 8.7 Mha study area. We estimated biases of large-area synthetic estimators and compared efficiencies of model-assisted (MA) estimators with field data-based direct estimators. The harvester models performed better in productive than unproductive forests, but systematic errors occurred in both. The use of MA estimators resulted in efficiency gains that were largest for HL (relative efficiency, RE=6.0) and smallest for QMD (RE=1.5). The bias of the synthetic estimator was largest for N (39%) and smallest for V (1%). The latter was due to an overestimation of deciduous and an underestimation of spruce forests that by chance balanced. We conclude that a probability sample of reference observations may be required to ensure the unbiasedness of estimators utilizing harvester data.

stat.AP↗

arXiv2021

Prediction of butt rot volume in Norway spruce forest stands using harvester, remotely sensed and environmental data

Butt rot (BR) damages associated with Norway spruce (Picea abies [L.] Karst.) account for considerable economic losses in timber production across the northern hemisphere. While information on BR damages is critical for optimal decision-making in forest management, the maps of BR damages are typically lacking in forest information systems. We predicted timber volume damaged by BR at the stand-level in Norway using harvester information of 186,026 stems (clear-cuts), remotely sensed, and environmental data (e.g. climate and terrain characteristics). We utilized random forest (RF) models with two sets of predictor variables: (1) predictor variables available after harvest (theoretical case) and (2) predictor variables available prior to harvest (mapping case). We found that forest attributes characterizing the maturity of forest, such as remote sensing-based height, harvested timber volume and quadratic mean diameter at breast height, were among the most important predictor variables. Remotely sensed predictor variables obtained from airborne laser scanning data and Sentinel-2 imagery were more important than the environmental variables. The theoretical case with a leave-stand-out cross-validation achieved an RMSE of 11.4 $m^3ha^{-1}$ (pseudo $R^2$: 0.66) whereas the mapping case resulted in a pseudo $R^2$ of 0.60. When the spatially distinct k-means clusters of harvested forest stands were used as units in the cross-validation, the RMSE value and pseudo $R^2$ associated with the mapping case were 15.6 $m^3ha^{-1}$ and 0.37, respectively. This indicates that the knowledge about the BR status of spatially close stands is of high importance for obtaining satisfactory error rates in the mapping of BR damages.

stat.AP↗

arXiv2021

Timber Volume Estimation Based on Airborne Laser Scanning -- Comparing the Use of National Forest Inventory and Forest Management Inventory Data

Large-scale forest resource maps based on national forest inventory (NFI) data and airborne laser scanning may facilitate synergies between NFIs and forest management inventories (FMIs). A comparison of models used in such a NFI-based map and a FMI indicate that NFI-based maps can directly be used in FMIs to estimate timber volume of mature spruce forests. Traditionally, FMIs and NFIs have been separate activities. The increasing availability of detailed NFI-based forest resource maps provides the possibility to eliminate or reduce the need of field sample plot measurements in FMIs if their accuracy is similar. We aim to 1) compare a timber volume model used in a NFI-based map and models used in a FMI, and 2) evaluate utilizing additional local sample plots in the model of the NFI-based map. Accuracies of timber volume estimates using models from an existing NFI-based map and a FMI were compared at plot and stand level. Estimates from the NFI-based map were similar to or more accurate than the FMI. The addition of local plots to the modeling data did not clearly improve the model of the NFI-based map.The comparison indicates that NFI-based maps can directly be used in FMIs for timber volume estimation in mature spruce stands, leading to potentially large cost savings.

stat.AP↗

arXiv2021

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%.

stat.AP↗

arXiv2021

Improving living biomass C-stock loss estimates by combining optical satellite, airborne laser scanning, and NFI data

Policy measures and management decisions aiming at enhancing the role of forests in mitigating climate-change require reliable estimates of C-stock dynamics in greenhouse gas inventories (GHGIs). Aim of this study was to assemble design-based estimators to provide estimates relevant for GHGIs using national forest inventory (NFI) data. We improve basic expansion (BE) estimates of living-biomass C-stock loss using field-data only, by leveraging with remotely-sensed auxiliary data in model-assisted (MA) estimates. Our case studies from Norway, Sweden, Denmark, and Latvia covered an area of >70 Mha. Landsat-based Forest Cover Loss (FCL) and one-time wall-to-wall airborne laser scanning (ALS) data served as auxiliary data. ALS provided information on the C-stock before a potential disturbance indicated by FCL. The use of FCL in MA estimators resulted in considerable efficiency gains which in most cases were further increased by using ALS in addition. A doubling of efficiency was possible for national estimates and even larger efficiencies were observed at the sub-national level. Average annual estimates were considerably more precise than pooled estimates using NFI data from all years at once. The combination of remotely-sensed with NFI field data yields reliable estimates which is not necessarily the case when using remotely-sensed data without reference observations.

stat.AP↗

arXiv2020

Above-ground biomass change estimation using national forest inventory data with Sentinel-2 and Landsat 8

This study aimed at estimating total forest above-ground net change (Delta AGB, Mt) 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 Mill hectares) 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 were 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 the use of remotely sensed 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 Mt). However, the decrease in precision when using Landsat data was small (SE= 1.92 Mt). In addition, we found that Delta AGB could be precisely estimated also when remotely sensed data were available only at the end of the monitoring period. We conclude that satellite optical data can considerably improve Delta AGB estimates, even in those cases where repeated and coincident NFI 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 a consistent and durable monitoring of forest carbon dynamics.

stat.AP↗

arXiv2020

Mapping forest age using National Forest Inventory, airborne laser scanning, and Sentinel-2 data

The age of forest stands is critical information for many aspects of forest management and conservation but area-wide information about forest stand age often does not exist. In this study, we developed regression models for large-scale area-wide prediction of age in Norwegian forests. For model development we used more than 4800 plots of the Norwegian National Forest Inventory (NFI) distributed over Norway between 58° and 65° northern latitude in a 181,773 km2 study area. Predictor variables were based on airborne laser scanning (ALS), Sentinel-2, and existing public map data. We performed model validation on an independent data set consisting of 63 spruce stands with known age. The best modelling strategy was to fit independent linear regression models to each observed site index (SI) level and using a SI prediction map in the application of the models. The most important predictor variable was an upper percentile of the ALS heights, and root-mean-squared-errors (RMSE) ranged between 3 and 31 years (6% to 26%) for SI-specific models, and 21 years (25%) on average. Mean deviance (MD) ranged between -1 and 3 years. The models improved with increasing SI and the RMSE were largest for low SI stands older than 100 years. Using a mapped SI, which is required for practical applications, RMSE and MD on plot-level ranged from 19 to 56 years (29% to 53%), and 5 to 37 years (5% to 31%), respectively. For the validation stands, the RMSE and MD were 12 (22%) and 2 years (3%). Tree height estimated from airborne laser scanning and predicted site index were the most important variables in the models describing age. Overall, we obtained good results, especially for stands with high SI, that could be considered for practical applications but see considerable potential for improvements, if better SI maps were available.

stat.AP↗

arXiv2020

National mapping and estimation of forest area by dominant tree species using Sentinel-2 data

Nation-wide Sentinel-2 mosaics were used with National Forest Inventory (NFI) data for modelling and subsequent mapping of spruce, pine and deciduous forest in Norway in 16 m $\times$ 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 approximately 10%. Nonetheless, efficiencies of national forest area estimates increased by 20% to 50% for model-assisted (MA) estimates and up to 90% for poststratification (PS). Greater minimum number of observations, however, constrained the use of PS. For estimates of sub-populations (municipalities), efficiencies improved by up to a factor of 8. However, even for municipalities with a decent number of NFI plots, direct NFI estimates were sometimes more precise than MA estimates. PS estimates were always more precise than direct and MA estimates but were applicable in fewer municipalities. The tree species prediction map is freely available as 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.

stat.AP↗