Forest variable estimation using photogrammetric matching of digital aerial images in combination with a high-resolution DEM

Abstract

The rapid development in aerial digital cameras in combination with the increased availability of high-resolution Digital Elevation Models (DEMs) provides a renaissance for photogrammetry in forest management planning. Tree height, stem volume, and basal area were estimated for forest stands using canopy height, density, and texture metrics derived from photogrammetric matching of digital aerial images and a high-resolution DEM. The study was conducted at a coniferous hemi-boreal site in southern Sweden. Three different data-sets of digital aerial images were used to test the effects of flight altitude and stereo overlap on an area-based estimation of forest variables. Metrics were calculated for 344 field plots (10 m radius) from point cloud data and used in regression analysis. Stand level accuracy was evaluated using leave-one-out cross validation of 24 stands. For these stands the tree height ranged from 4.8 to 26.9 m (17.8 m mean), stem volume 13.3 to 455 m³ ha^?1 (250 m³ ha^?1 mean), and basal area from 4.1 to 42.9 m² ha^?1 (27.1 m² ha^?1 mean) with mean stand size of 2.8 ha. The results showed small differences in estimation accuracy of forest variables between the data-sets. The data-set of digital aerial images corresponding to the standard acquisition of the Swedish National Land Survey (Lantmäteriet), showed Root Mean Square Errors (in percent of the surveyed stand mean) of 8.8% for tree height, 13.1% for stem volume and 14.9% for basal area. The results imply that photogrammetric matching of digital aerial images has significant potential for operational use in forestry.     

Practical large-scale forest stand inventory using a small-footprint airborne scanning laser

Mean tree height, dominant height, mean diameter, stem number, basal area and timber volume of 116 georeferenced field sample plots were estimated from various canopy height and canopy density metrics derived by means of a small-footprint laser scanner over young and mature forest stands using regression analysis. The sample plots were distributed systematically throughout a 6500 ha study area, and the size of each plot was 232.9 m². Regressions for coniferous forest explained 60–97% of the variability in ground reference values of the six studied characteristics. A proposed practical two-phase procedure for prediction of corresponding characteristics of entire forest stands was tested. Fifty-seven test plots within the study area with a size of approximately 3740 m² each were divided into 232.9 m² regular grid cells. The six examined characteristics were predicted for each grid cell from the corresponding laser data using the estimated regression equations. Average values for each test plot were computed and compared with ground-based estimates measured over the entire plot. The bias and standard deviations of the differences between predicted and ground reference values (in parentheses) of mean height, dominant height, mean diameter, stem number, basal area and volume were ?0.58 to ?0.85 m (0.64–1.01 m), ?0.60 to ?0.99 m (0.67–0.84 m), 0.15–0.74 cm (1.33–2.42 cm), 34–108 ha^?1 (97–466 ha^?1), 0.43–2.51 m² ha^?1 (1.83–3.94 m² ha^?1) and 5.9–16.1 m³ ha^?1 (15.1–35.1 m³ ha^?1), respectively.     

Using canopy heights from digital aerial photogrammetry to enable spatial transfer of forest attribute models: a case study in central Europe

This paper describes a workflow utilizing detailed canopy height information derived from digital airphotos combined with ground inventory information gathered in state-owned forests and regression modelling techniques to quantify forest-growing stocks in private woodlands, for which little information is generally available. Random forest models were trained to predict three different variables at the plot level: quadratic mean diameter of the 100 largest trees (d₁₀₀), basal area weighted mean height of the 100 largest trees (h₁₀₀), and gross volume (V). Two separate models were created – one for a spruce- and one for a beech-dominated test site. We examined the spatial portability of the models by using them to predict the aforementioned variables at actual inventory plots in nearby forests, in which simultaneous ground sampling took place. When data from the full set of available plots were used for training, the predictions for d₁₀₀, h₁₀₀, and V achieved out-of-bag model accuracies (scaled RMSEs) of 15.1%, 10.1%, and 35.3% for the spruce- and 15.9%, 9.7%, and 32.1% for the beech-dominated forest, respectively. The corresponding independent RMSEs for the nearby forests were 15.2%, 10.5%, and 33.6% for the spruce- and 15.5%, 8.9%, and 33.7% for the beech-dominated test site, respectively.     

Estimation of above ground forest biomass from airborne discrete return laser scanner data using canopy-based quantile estimators

A conceptual model describing why laser height metrics derived from airborne discrete return laser scanner data are highly correlated with above ground biomass is proposed. Following from this conceptual model, the concept of canopy-based quantile estimators of above ground forest biomass is introduced and applied to an uneven-aged, mature to overmature, tolerant hardwood forest. Results from using the 0th, 25th, 50th, 75th and 100th percentiles of the distributions of laser canopy heights to estimate above ground biomass are reported. A comparison of the five models for each dependent variable group did not reveal any overt differences between models with respect to their predictive capabilities. The coefficient of determination (r ² ) for each model is greater than 0.80 and any two models may differ at most by up to 9%. Differences in root-mean-square error (RMSE) between models for above ground total, stem wood, stem bark, live branch and foliage biomass were 8.1, 5.1, 2.9, 2.1 and 1.1 Mg ha^?1, respectively.     

Accuracy of forest inventory using airborne laser scanning: evaluating the first nordic full-scale operational project

This research reports the major results from an evaluation of the first Nordic operational stand-based forest inventory using airborne laser scanner data. Laser data from a forest area of 250 km² were used to predict six biophysical stand variables used in forest planning. The predictions were based on regression equations estimated from 250 m² field training plots distributed systematically throughout the forest area. Test plots with an approximate size of 0.1–0.4 ha were used for validation. The testing revealed standard deviations between ground-truth values and predicted values of 0.36–1.37 m (1.9–7.6%) for mean height, 0.70–1.55 m (3.0–7.6%) for dominant height, 2.38–4.88 m² ha^?1 (7.8–14.2%) for basal area and 13.9–45.9 m³ ha^?1 (6.5–13.4%) for stand volume. No serious bias was detected.     

0.5M分辨率航摄数字影像数据在森林资源规划设计调查中的应用

以0.5 M分辨率航摄数字影像数据为基础,结合以往调查设计及经营资料,对森林资源规划设计调查小班的部分因子进行判读,探讨高分辨率航摄数字影像数据在森林资源规划设计调查中的应用,结果表明:利用0.5M分辨率航摄数字影像数据进行小班因子判读,可以作为小班因子调查的辅助手段,并可显著提高小班调查质量及效率。     

Estimating single-tree branch biomass of Norway spruce with terrestrial laser scanning using voxel-based and crown dimension features

Abstract

Many remote sensing-based methods estimating forest biomass rely on allometric biomass models for field reference data. Terrestrial laser scanning (TLS) has emerged as a tool for detailed data collection in forestry applications, and the methods have been proposed to derive, e.g. tree position, diameter-at-breast-height, and stem volume from TLS data. In this study, TLS-derived features were related to destructively sampled branch biomass of Norway spruce at the single-tree level, and the results were compared to conventional allometric models with field measured diameter and height. TLS features were derived following two approaches: one voxel-based approach with a detailed analysis of the interaction between individual voxels and each laser beam. The features were derived using voxels of size 0.1, 0.2, and 0.4 m, and the effect of the voxel size was assessed. The voxel-derived features were compared to features derived from crown dimension measurements in the unified TLS point cloud data. TLS-derived variables were used in regression models, and prediction accuracies were assessed through a Monte Carlo cross-validation procedure. The model based on 0.4 m voxel data yielded the best prediction accuracy, with a root mean square error (RMSE) of 32%. The accuracy was found to decrease with an increase in voxel size, i.e. the model based on the 0.1 m voxel yielded the lowest accuracy. The model based on crown measurements had an RMSE of 34%. The accuracies of the predictions from the TLS-based models were found to be higher than from conventional allometric models, but the improvement was relatively small.     

Laser scanning of forest resources: the nordic experience

This article reviews the research and application of airborne laser scanning for forest inventory in Finland, Norway and Sweden. The first experiments with scanning lasers for forest inventory were conducted in 1991 using the FLASH system, a full-waveform experimental laser developed by the Swedish Defence Research Institute. In Finland at the same time, the HUTSCAT profiling radar provided experiences that inspired the following laser scanning research. Since 1995, data from commercially operated time-of-flight scanning lasers (e.g. TopEye, Optech ALTM and TopoSys) have been used. Especially in Norway, the main objective has been to develop methods that are directly suited for practical forest inventory at the stand level. Mean tree height, stand volume and basal area have been the most important forest mensurational parameters of interest. Laser data have been related to field training plot measurements using regression techniques, and these relationships have been used to predict corresponding properties in all forest stands in an area. Experiences from Finland, Norway and Sweden show that retrieval of stem volume and mean tree height on a stand level from laser scanner data performs as well as, or better than, photogrammetric methods, and better than other remote sensing methods. Laser scanning is, therefore, now beginning to be used operationally in large-area forest inventories. In Finland and Sweden, research has also been done into the identification of single trees and estimation of single-tree properties, such as tree position, tree height, crown width, stem diameter and tree species. In coniferous stands, up to 90% of the trees represented by stem volume have been correctly identified from canopy height models, and the tree height has been estimated with a root mean square error of around 0.6 m. It is significantly more difficult to identify suppressed trees than dominant trees. Spruce and pine have been discriminated on a single-tree level with 95% accuracy. The application of densely sampled laser scanner data to change detection, such as growth and cutting, has also been demonstrated.     

Prediction of tree height,basal area and stem volume in forest stands using airborne laser scanning

The aim of this study was to develop prediction models using laser scanning for estimation of forest variables at plot level, validate the estimations at stand level (area 0.64 ha) and test the effect of different laser measurement densities on the estimation errors. The predictions were validated using 29 forest stands (80×80 m²), each containing 16 field plots with a 10 m radius. For the best tested case, mean tree height, basal area and stem volume were predicted with a root mean square error of 0.59 m (3% of average value), 2.7 m² ha^?1 (10% of average value) and 31 m³ ha^?1 (11% of average value), respectively, at stand level. There were small differences in terms of prediction errors for different measuring densities. The results indicate that mean tree height, basal area and stem volume can be estimated in small stands with low laser measurement densities producing accuracies similar to traditional field inventories.     

Identification of boreal forest stands with high herbaceous plant diversity using airborne laser scanning

Boreal forest stands with high herbaceous plant species diversity have been found to be one of the main habitats for many endangered species, but the locations and sizes of these herb-rich forest stands are not well known in many areas. Better identification of the stands could improve both their conservation and management. A new approach is proposed here for locating the mature herb-rich forest stands using airborne laser scanner (ALS) data and logistic regression, or the k-NN classifier. We show that ALS technology is capable of distinguishing the ecologically important herb-rich forests from those growing on less fertile site types, mainly on the basis of unique but quantifiable crown structure and vertical profile that characterise forests on high fertility sites. The study site, Koli National Park, is located on the border of the southern and middle boreal vegetation zones in Finland, and includes 63 herb-rich forest stands of varying sizes. The model and test data comprised 274 forest stands belonging to five forest site types varying from very fertile to poor. The best overall classification accuracy achieved with the k-NN method was 88.9%, the herb-rich forests being classified correctly in 65.0% of cases and the other forest site types in 95.7%. The best overall classification accuracy achieved with logistic regression was 85.6%, being 55.0% for the herb-rich forests and 94.3% for the other forest site types. Both methods demonstrated promising potential for separating herb-rich forests from other forest site types, although slightly better results were obtained with the non-parametric k-NN method, which was capable of utilising a higher number of explanatory variables. It is concluded that ALS-based data analysis techniques are applicable to the detection of mature boreal herb-rich forests in large-scale forest inventories.     

Estimating forest carbon fluxes for large regions based on process-based modelling, NFI data and Landsat satellite images

The aim of this study was to develop and evaluate a new approach for estimating forest carbon fluxes for large regions based on climate-sensitive process-based model, national forest inventory (NFI) data and satellite images. The approach was tested for Central Finland and Lapland with NFI field data and daily weather data from 2004 to 2008.The approach combines (1) a light use efficiency (LUE) model, (2) a process-based summary model for estimating gross primary production (GPP) and net primary production (NPP), and (3) the Yasso07 soil carbon model, which together allow the estimation of net ecosystem exchange (NEE). Landsat TM 5 satellite images were utilized to generalize the carbon fluxes obtained for field sample plots for all forested areas using the k-NN imputation method. The accuracy of the imputations was examined by leave-one-out cross validation and by comparing the imputed and simulated values with Eddy covariance (EC) measurements.RMSE of the k-NN imputations was slightly better in Central Finland than in Lapland, the bias staying at a similar level. Based on the EC comparisons, the approach seemed to work rather well with GPP estimates in both areas, but in the north the NEE estimates were remarkably biased. The main advantages of the approach include its applicability to basic NFI data and a high output resolution (30 m).The method proved to be a promising way to produce carbon flux estimates based on large-scale forest inventory data and could therefore be easily applied to the whole of Northern Europe. However, there are still drawbacks to the approach, such as lacking parameters for peat lands. One of the future goals is to integrate the approach with an interactive mapping framework, which could thereafter be utilized, for example, in climate change research.     

Comparison of estimators and feature selection procedures in forest inventory based on airborne laser scanning and digital aerial imagery

Digital maps of forest resources are a crucial factor in successful forestry applications. Since manual measurement of this data on large areas is infeasible, maps must be constructed using a sample field data set and a prediction model constructed from remote sensing materials, of which airborne laser scanning (ALS) data and aerial images are currently widely used in management planning inventories. ALS data is suitable for the prediction of variables related to the size and volume of trees, whereas optical imagery helps in improving distinction between tree species. We studied the prediction of forest attributes using field data from National Forest Inventory complemented with ad hoc field plots in combination with ALS and aerial imagery data in Aland province, Finland. We applied feature selection with genetic algorithm and greedy forward selection and compared multiple linear and nonlinear estimators. Maximally around 40 features from a total of 154 were required to achieve the best prediction performances. Tree height was predicted with normalized root mean squared error value of 0.1 and tree volume with a value around 0.25. Predicting the volumes of spruce and broadleaved trees was the most challenging due to small proportions of these tree species in the study area.     

Process-based approach to automated classification of forest structures using medium format digital aerial photos and ancillary GIS information

The methods of forest inventory data acquisition, based on the analysis of remotely sensed images have been well tested and implemented during the last decade. The predominant visual interpretation and pixel-based automated techniques are now being gradually replaced by the object-based image classification at multiple levels. This paper describes an experiment using medium-format digital aerial imagery for the purpose of automated updating of the existing GIS forest management database (LHPO). The method emphasises the pre-processing phase, where various image transforms and additional channels i.e. spectral ratios and vegetation indices (NDVI), low-pass filters, Sobel edge and GLCM (grey level co-occurrence matrix) texture measures are derived from the original dataset. The layer stack is then transferred into the object-oriented classification environment together with the existing thematic vector layer, and analysed on three hierarchical object levels. The classification involves the recognition of the successional stage of forest compartments and the estimation of tree species composition in terms of area coverage. In addition, age information on the GIS forestry management map can be updated and the spatial distribution of classes corrected using the multi-scale object relations of the former analysis. The advances of the automated procedure based on sequential processing of image objects are partially covered. Moreover, aspects of utilisation of the medium-format colour infra-red images (CIR) as an alternative to traditional aerial photos and very high resolution (VHR) satellite data, were considered.     

Comparison of field sampling methods for assessing coarse woody debris and use of airborne laser scanning as auxiliary information

Coarse woody debris (CWD) has been recognized as one of the strongest indicators of forest biodiversity and its assessment has been emphasized in the development of new inventory methods. In this study, the most commonly referenced probability sampling methods were tested in a field area of 305.8 ha to gain comparative information on their performance and efficiency. Simple random sampling (SRS), systematic sampling and cluster sampling with fixed sized circular sample plots were tested, as well as strip sampling, transect relascope sampling and adaptive cluster sampling (ACS). Point relascope sampling and line intersect sampling were also tested for inventories of downed dead wood volumes. In addition, the amount of standing dead wood was assessed by means of traditional small angle relascope sampling. In general, the use of additional information in the inventory process has shown promising results. A new method for using data derived from airborne laser scanning (ALS) as a source of auxiliary information in the assessment of CWD volumes is presented, using probability proportional to size (PPS) sampling for the selection of the first-stage sample units in ACS (ACS_PPS) and for the placement of fixed sized plots (PLOT_PPS). The sampling methods were compared in terms of the cost-effectiveness. Point relascope sampling proved the most efficient sampling method for inventorying CWD volumes. PLOT_PPS and ACS_PPS were more efficient than the inventory of fixed sized plots (PLOT_SRS) and ACS (ACS_SRS) where sample units were selected with SRS. However, these methods could not achieve the same efficiency as relascope samplings. Nevertheless, the use of probability layers derived from ALS data gave promising results and offers new possibilities for inventorying CWD volumes more efficiently.     

Reconstructing forest canopy from the 3D triangulations of airborne laser scanning point data for the visualization and planning of forested landscapes

Key message

We present a data-driven technique to visualize forest landscapes and simulate their future development according to alternative management scenarios. Gentle harvesting intensities were preferred for maintaining scenic values in a test of eliciting public’s preferences based on the simulated landscapes.

Context

Visualizations of future forest landscapes according to alternative management scenarios are useful for eliciting stakeholders’ preferences on the alternatives. However, conventional computer visualizations require laborious tree-wise measurements or simulators to generate these observations.

Aims

We describe and evaluate an alternative approach, in which the visualization is based on reconstructing forest canopy from sparse density, leaf-off airborne laser scanning data.

Methods

Computational geometry was employed to generate filtrations, i.e., ordered sets of simplices belonging to the three-dimensional triangulations of the point data. An appropriate degree of filtering was determined by analyzing the topological persistence of the filtrations. The topology was further utilized to simulate changes to canopy biomass, resembling harvests with varying retention levels. Relative priorities of recreational and scenic values of the harvests were estimated based on pairwise comparisons and analytic hierarchy process (AHP).

Results

The canopy elements were co-located with the tree stems measured in the field, and the visualizations derived from the entire landscape showed reasonably realistic, despite a low numerical correspondence with plot-level forest attributes. The potential and limitations to improve the proposed parameterization are discussed.

Conclusion

Although the criteria to evaluate the landscape visualization and simulation models were not conclusive, the results suggest that forest scenes may be feasibly reconstructed based on data already covering broad areas and readily available for practical applications.

     

地面三维激光扫描技术在林业中的应用与展望

以地面三维激光扫描技术的测量原理、技术特征为切入点, 重点论述了该技术在国内外林业领域的应用现状、不足及前景。与传统研究手段相比, 地面三维激光扫描技术在测树效率、三维建模、模型化精度等方面具有较大的优势, 显示出更广阔的应用前景。     

Automatic recognition and measurement of single trees based on data from airborne laser scanning over the richly structured natural forests of the Bavarian Forest National Park

The purpose of this study was to test a method for delineating individual tree crowns based on a fully automated recognition methodology. The study material included small-footprint time-of-flight laser scanner data acquired in the spring and summer of 2002. The data were collected with a Toposys II airborne laser system flown over the Norway spruce (Picea abies) and European beech (Fagus sylvatica) dominated forests of the Bavarian Forest National Park, Germany. The applied algorithm, which earlier had been validated for Swedish forest conditions, is a watershed algorithm that is based on the use of laser scanning data. 2584 trees in a total of 28 representative reference stands, each 0.1–0.25 ha in area, were included in the investigation. With the algorithm, 76.9% of the trees in the upper layer could be recognised. This corresponds to 85.2% of the timber volume determined by ground measurements. The results for conifers were more accurate in this respect than for deciduous trees. A negative aspect was the number of falsely identified trees, the percentage of which was 5.4%.     

Modeling individual tree height growth of Norway spruce and Scots pine from national forest inventory data in Norway

We developed individual tree height growth models for Norway spruce (Picea abies (L.) Karst.) and Scots pine (Pinus sylvestris L.) in Norway based on national forest inventory data. Potential height growth is based on existing dominant height growth models and reduced due to competition by functions developed in this study. Three spatially explicit and two spatially non-explicit competition indices were tested. Distance effects and diameter ratio effects were estimated from the data simultaneously with parameters of the potential modifier functions. Large height measurement errors in the national forest inventory data caused large residual variation of the models. However, the effects of competition on height growth were significant and plausible. The potential modifier functions show that height growth of dominant trees is largely unaffected by competition. Only at higher levels of competition, height growth is reduced as a consequence of competition. However, Scots pine also reduced height growth at very low levels of competition. Distance effects in the spatially explicit competition indices indicated that the closest neighbors are most important for height growth. However, for Scots pine also competitors at larger distance affected height growth. The five competition indices tested in this study explained similar proportions of the variation in relative height growth. Given that unbiased predictions can only be expected for the same plot size, we recommend a spatially explicit index, which describes the distance function with a negative exponential, for use in growth simulators.