Search

In remote sensing-based forest inventories 3D point cloud data, such as acquired from airborne laser scanning, are well suited for estimating the volume of growing stock and stand height, but tree species recognition often requires additional optical imagery. A combination of 3D data and optical imagery can be acquired based on aerial imaging only, by using stereo photogrammetric 3D canopy modeling. The use of aerial imagery is well suited for large-area forest inventories, due to low costs, good area coverage and temporally rapid cycle of data acquisition. Stereo-photogrammetric canopy modeling can also be applied to previously acquired imagery, such as for aerial ortho-mosaic production, assuming that the imagery has sufficient stereo overlap. In this study we compared two stereo-photogrammetric canopy models combined with contemporary satellite imagery in forest inventory. One canopy model was based on standard archived imagery acquired primarily for ortho-mosaic production, and another was based on aerial imagery whose acquisition parameters were better oriented for stereo-photogrammetric canopy modeling, including higher imaging resolution and greater stereo-coverage. Aerial and satellite data were tested in the estimation of growing stock volume, volumes of main tree species, basal area and diameter and height. Despite the better quality of the latter canopy model, the difference of the accuracy of the forest estimates based on the two different data sets was relatively small for most variables (differences in RMSEs were 0â20%, depending on variable). However, the estimates based on stereo-photogrammetrically oriented aerial data retained better the original variation of the forest variables present in the study area.

There is growing interest in the use of Landsat data to enable forest monitoring over large areas. Free and open data access combined with high performance computing have enabled new approaches to Landsat data analysis that use the best observation for any given pixel to generate an annual, cloud-free, gap-free, surface reflectance image composite. Finland has a long history of incorporating Landsat data into its National Forest Inventory to produce forest information in the form of thematic maps and small area statistics on a variety of forest attributes. Herein we explore the spatial and temporal characteristics of the Landsat archive in the context of forest monitoring in Finland. The United States Geological Survey Landsat archive holds a total of 30â076 images (1972â2017) for 66 scenes (each 185 km by 185 km in size) representing the terrestrial area of Finland, of which 93.6% were acquired since 1984 with a spatial resolution of 30 m. Approximately 16.3% of the archived images have desired compositing characteristics (acquired within August 1 ±30 days,

Optical 2D remote sensing techniques such as aerial photographing and satellite imaging have been used in forest inventory for a long time. During the last 15 years, airborne laser scanning (ALS) has been adopted in many countries for the estimation of forest attributes at stand and sub-stand levels. Compared to optical remote sensing data sources, ALS data are particularly well-suited for the estimation of forest attributes related to the physical dimensions of trees due to its 3D information. Similar to ALS, it is possible to derive a 3D forest canopy model based on aerial imagery using digital aerial photogrammetry. In this study, we compared the accuracy and spatial characteristics of 2D satellite and aerial imagery as well as 3D ALS and photogrammetric remote sensing data in the estimation of forest inventory variables using k-NN imputation and 2469 National Forest Inventory (NFI) sample plots in a study area covering approximately 5800 km. Both 2D data were very close to each other in terms of accuracy, as were both the 3D materials. On the other hand, the difference between the 2D and 3D materials was very clear. The 3D data produce a map where the hotspots of volume, for instance, are much clearer than with 2D remote sensing imagery. The spatial correlation in the map produced with 2D data shows a lower short-range correlation, but the correlations approach the same level after 200 meters. The difference may be of importance, for instance, when analyzing the efficiency of different sampling designs and when estimating harvesting potential.2

Remote sensing using unmanned aerial vehicle (UAV) -borne sensors is currently a highly interesting approach for the estimation of forest characteristics. 3D remote sensing data from airborne laser scanning or digital stereo photogrammetry enable highly accurate estimation of forest variables related to the volume of growing stock and dimension of the trees, whereas recognition of tree species dominance and proportion of different tree species has been a major complication in remote sensing-based estimation of stand variables. In this study the use of UAV-borne hyperspectral imagery was examined in combination with a high-resolution photogrammetric canopy height model in estimating forest variables of 298 sample plots. Data were captured from eleven separate test sites under weather conditions varying from sunny to cloudy and partially cloudy. Both calibrated hyperspectral reflectance images and uncalibrated imagery were tested in combination with a canopy height model based on RGB camera imagery using the -nearest neighbour estimation method. The results indicate that this data combination allows accurate estimation of stand volume, mean height and diameter: the best relative RMSE values for those variables were 22.7%, 7.4% and 14.7%, respectively. In estimating volume and dimension-related variables, the use of a calibrated image mosaic did not bring significant improvement in the results. In estimating the volumes of individual tree species, the use of calibrated hyperspectral imagery generally brought marked improvement in the estimation accuracy; the best relative RMSE values for the volumes for pine, spruce, larch and broadleaved trees were 34.5%, 57.2%, 45.7% and 42.0%, respectively.k

In this paper we examine the feasibility of data from unmanned aerial vehicle (UAV)-borne aerial imagery in stand-level forest inventory. As airborne sensor platforms, UAVs offer advantages cost and flexibility over traditional manned aircraft in forest remote sensing applications in small areas, but they lack range and endurance in larger areas. On the other hand, advances in the processing of digital stereo photography make it possible to produce three-dimensional (3D) forest canopy data on the basis of images acquired using simple lightweight digital camera sensors. In this study, an aerial image orthomosaic and 3D photogrammetric canopy height data were derived from the images acquired by a UAV-borne camera sensor. Laser-based digital terrain model was applied for estimating ground elevation. Features extracted from orthoimages and 3D canopy height data were used to estimate forest variables of sample plots. -nearest neighbor method was used in the estimation, and a genetic algorithm was applied for selecting an appropriate set of features for the estimation task. Among the selected features, 3D canopy features were given the greatest weight in the estimation supplemented by textural image features. Spectral aerial photograph features were given very low weight in the selected feature set. The accuracy of the forest estimates based on a combination of photogrammetric 3D data and orthoimagery from UAV-borne aerial imaging was at a similar level to those based on airborne laser scanning data and aerial imagery acquired using purpose-built aerial camera from the same study area.K

In Finland, there are currently two, parallel sample-plot-based forest inventory systems, which differ in their methodologies, sampling designs, and objectives. One is the National Forest Inventory (NFI), aimed at unbiased inventory results at national and regional level. The other is the Forest Centreâs management-oriented forest inventory based on interpretation of airborne laser scanning and aerial images, with the aim of locally accurate stand-level forest estimates. The National Forest Inventory utilises relascope sample plots with systematic cluster sampling. This inventory method is optimised for accuracy of regional volume estimates. In contrast, the management-oriented forest inventory utilises circular sample plots with an allocation system covering certain pre-defined forest classes in the inventory area. This method is optimised to produce reference data for interpretation of the remote-sensing materials in use. In this study, we tested the feasibility of the National Forest Inventory sample plots in provision of additional reference data for the management-oriented inventory. Various combinations of NFI plots and management inventory plots were tested in the interpretation of the laser and aerial-image data. Adding NFI plots in the reference data generally improved the accuracy of the volume estimates by tree species but not the estimates of total volume or stand mean height and diameter. The difference between the plot types in the NFI and management inventories causes difficulties in combination of the two datasets.

Information on forest biomass is required for several purposes, including estimation of forest bioenergy resources and forest carbon stocks. Airborne laser scanning is today considered the most accurate remote sensing method for forest inventory. The three-dimensional nature of laser scanning data enables estimation of the volumes of the tree canopies. The dimensions of the tree canopies show high correlation with the amount of forest biomass. Optical aerial photographs are often used to complement laser data, for improved distinction between tree species. The paper reports on a study testing the estimation of forest biomass variables in two study areas in Southern Finland. The biomass variables were derived on the basis of tree-level field measurements, with biomass models used for pine, spruce, and birch. The sample-plot-level biomass components were derived on the basis of tree-level data and used as reference data for airborne-laser- and aerialâphotograph-based estimation. Results were slightly better for total biomass (RMSE 22.5% and 23.6% for the two study areas) than total volume (RMSE: 23.4% and 26.1%). Species-specific estimation errors were large in general but varied between the study areas, because of differences in their forest structures.

Map form information on forest biomass is required for estimating bioenergy potentials and monitoring carbon stocks. In Finland, the growing stock of forests is monitored using multi-source forest inventory, where variables are estimated in the form of thematic maps and area statistics by combining information of field measurements, satellite images and other digital map data. In this study, we used the multi-source forest inventory methodology for estimating forest biomass characteristics. The biomass variables were estimated for national forest inventory field plots on the basis of measured tree variables. The plot-level biomass estimates were used as reference data for satellite image interpretation. The estimates produced by satellite image interpretation were tested by cross-validation. The results indicate that the method for producing biomass maps on the basis of biomass models and satellite image interpretation is operationally feasible. Furthermore, the accuracy of the estimates of biomass variables is similar or even higher than that of traditional growing stock volume estimates. The technique presented here can be applied, for example, in estimating biomass resources or in the inventory of greenhouse gases.

Satellite sensor data have traditionally been used in multi-source forest inventory for estimating forest characteristics. Their advantages generally are large geographic coverage and large spectral range. Another remote sensing data source for forest inventories offering a large geographic coverage is high altitude aerial photography. In high altitude aerial photographs the spectral range is very narrow but the spatial resolution is high. This allows the extraction of texture features for forest inventory purposes. In this study we utilized a Landsat 7 ETM satellite image, a photo mosaic composed of high altitude panchromatic aerial photographs, and a combination of the aforementioned in estimating forest attributes for an area covering approximately 281 000 ha in Forestry Centre Häme-Uusimaa in Southern Finland. Sample plots of 9th National Forest Inventory (NFI9) were used as field data. In the estimation, 6 Landsat 7 ETM image channels were used. For aerial photographs, 4 image channels were composed from the spectral averages and texture features. In combining both data sources, 6 Landsat channels and 3 aerial image texture channels were selected for the analysis. The accuracy of forest estimates based on the Landsat image was better than that of estimates based on high altitude aerial photographs. On the other hand, using the combination of Landsat ETM spectral features and textural features on high altitude aerial photographs improved the estimation accuracy of most forest attributes.

A two-phase sampling design has been applied to forest inventory. First, a large number of first phase sample plots were defined with a square grid in a geographic coordinate system for two study areas of about 1800 and 4500 ha. The first phase sample plots were supplied by auxiliary data of Landsat TM and IRS-1C with principal component transformation for stratification and drawing the second phase sample (field sample). Proportional allocation was used to draw the second phase sample. The number of field sample plots in the two study areas was 300 and 380. The local estimates of five continuous forest stand variables, mean diameter, mean height, age, basal area, and stem volume, were calculated for each of the first phase sample plots. This was done separately by using one auxiliary data source at a time together with the field sample information. However, if the first phase sample plot for which the stand variables were to be estimated was also a field sample plot, the information of that field sample plot was eliminated according to the cross validation principle. This was because it was then possible to calculate mean square errors of estimates related to a specific auxiliary data source. The procedure produced as many estimates for each first phase sample plot and forest stand variable as was the number of auxiliary data sources, i.e. seven estimates: These were based on Landsat TM, IRS-1C, digitized aerial photos, ocular stereoscopic interpretation from aerial photographs, data from old forest inventory made by compartments, Landsat TM95âTM89 difference image and IRS96âTM95 difference image. The final estimates were calculated as weighted averages where the weights were inversely proportional to mean square errors. The alternative estimates were calculated by applying simple rules based on knowledge and the outliers were defined. The study shows that this kind of system for finding outliers for elimination and a weighting procedure improves the accuracy of stand variable estimation.

Five auxiliary data sources (Landsat TM, IRS-IC, digitized aerial photographs, visual photo-interpretation and old forest compartment information) applying three study areas and three estimators, two-phase sampling with stratification, the k nearest neighbors and regression estimator, were examined. Auxiliary data were given for a high number of sample plots, which are here called first phase sample plots. The plots were distributed using a systematic grid over the study areas. Some of the plots were then measured in the field for the necessary ground truth. Each auxiliary data source in combination with field sample information was applied to produce a specific estimator for five forest stand characteristics: mean diameter, mean height, age, basal area, and volume of the growing stock. When five auxiliary data sources were used, each stand characteristic and each first phase sample plot were supplied with five alternative estimates with three alternative estimators. Mean square errors were then calculated for each alternative estimator using the cross validation method. The final estimates were produced by weighting alternative estimates inversely according to the mean square errors related to the corresponding estimator. The result was better than the final estimate of any of the single estimators. The improvement over the best single estimate, as measured in mean square error, was 16.9% on average for all five forest stand characteristics. The improvement was fairly equal for all five forest stand characteristics. Only minor differences among the accuracies of the three alternative estimators were recorded.

There is considerable interest in identifying and locating natural forests as accurately as possible, because they are deemed essential in preventing biodiversity loss. In the boreal region, natural forests contain a substantial amount of dead wood and exhibit considerable variation in tree age, size, and species composition. However, it is difficult to define natural forests in a quantitative manner. This is an issue, for example, in the Finnish national forest inventory. If naturalness could be related to the metrics derived from tree measurements, it would be easier to locate natural forests based on the inventory data. In this study, we investigated the value of metrics computed from tree locations and tree sizes for the characterization of a key aspect of naturalness, namely, structural naturalness as defined in the Finnish national forest inventory. We used L-moments, Gini coefficient, Lorenz asymmetry, and interquartile range to quantify the variations in tree size at the plot level. We summarized the spatial pattern of trees with a spatial aggregation index. We compared the structural metrics, species proportions, and stand age using the classes of structural naturalness described in the Finnish national forest inventory, which have been determined in the field without strict numerical rules. These categories are 'natural', 'near-natural', and 'non-natural'. We found that the forests evaluated as structurally natural had larger variations in tree size and species composition and showed a more clustered spatial pattern of trees on average, although the variation in the structural metrics was considerable in all three classes. In addition, we used the structural metrics to predict naturalness by employing a random forest algorithm. Based on the structural metrics, it was possible to obtain high precision in the classification only if we simultaneously accepted low recall, and vice versa; the link between the inspected metrics and naturalness evaluated in the field was weak. The stand age separated the three classes more clearly and it also improved the classification. [ABSTRACT FROM AUTHOR]

In forest management planning, errors in predicted stand attributes might lead to suboptimal decisions that result in decreased net present value (NPV). Forest inventory data will have higher value if the amount of suboptimal decisions can be decreased. Therefore, the value of information can be measured through the decrease in inoptimality losses, which are the NPV differences between the optimal and suboptimal decisions. In this study, four alternative sample plot selection strategies with different numbers of sample plots were compared in terms of expected mean inoptimality losses. Stand-level mean inoptimality losses varied between 41.1 x [ha.sup.-1] [euro] and 80.7 x [ha.sup.-1] [euro], depending on the sample plot selection strategy and the number of sample plots used as training data in the k-nearest neighbors imputation method. Mean inoptimality losses decreased substantially when the number of sample plots increased from 25 to 100, and the decreasing trend continued until 500 sample plots. Total inoptimality losses can decrease by approximately 1 million [euro] in an inventory area of 100 000 ha when the number of sample plots is increased from 100 to 500. The measurement of more sample plots can be justified as long as the field measurement costs do not exceed the decrease in inoptimality losses. Key words: forest inventory, forest management planning, inoptimality loss, value of information, cost-plus-loss analysis. Lors de la planification de l'amenagement forestier, des erreurs dans les predictions des attributs de peuplement pourraient conduire a des decisions sous-optimales, entrainant une diminution de la valeur actuelle nette (VAN). Les donnees d'inventaire forestier auront une plus grande valeur si elles servent a diminuer le nombre de decisions sous-optimales. Par consequent, la valeur de l'information peut etre mesuree par la diminution des pertes dues a la sous-optimalite, qui correspondent a la difference de VAN entre les decisions optimales et sous-optimales. Dans cette etude, quatre strategies differentes de selection des placettes d'echantillonnage, avec differents nombres de placettes d'echantillonnage, ont ete comparees en termes de moyennes attendues des pertes dues a la sous-optimalite. Les pertes moyennes dues a la sous-optimalite a l'echelle du peuplement ont varie entre 41,1 [euro] et 80,7 [euro] par hectare, en fonction de la strategie de selection des placettes d'echantillonnage et du nombre de celles-ci utilisees comme donnees d'apprentissage dans la methode d'imputation des k-plus proches voisins. Les pertes moyennes dues a la sous-optimalite ont sensiblement diminue lorsque le nombre de placettes d'echantillonnage est passe de 25 a 100, et la tendance a la baisse s'est poursuivie jusqu'a 500 placettes d'echantillonnage. Les pertes totales dues a la sous-optimalite peuvent diminuer d'environ 1 million [euro] dans une zone d'inventaire de 100 000 ha lorsque le nombre de placettes d'echantillonnage augmente de 100 a 500. La mesure d'un plus grand nombre de placettes d'echantillonnage peut etre justifiee a condition que les couts de mesure sur le terrain ne depassent pas la diminution des pertes par sous-optimalite. [Traduit par la Redaction] Mots-cles: inventaire forestier, planification de l'amenagement forestier, perte due a la sous-optimalite, valeur de l'information, analyse combinee de cout plus perte., 1. Introduction Forest inventories based on airborne laser scanning (ALS) usually consist of data acquisition, model-building, and accuracy assessment with training data, the prediction of stand attributes in a wall-to-wall [...]

In this study, we investigate the use of model-based inference in forest surveys in which auxiliary data are available as a probability sample. We evaluate the effects of model form and sample size on estimators of growing stock volume, based on different types of remotely sensed auxiliary data. The study was performed through Monte Carlo sampling simulation using a two-phase sampling design within a simulated study area resembling the conditions in mid-western Finland. We show that the choice of model has a minor to moderate effect on the precision of model-based estimators. Similarly, the choice of estimator of the variance-covariance matrix of model parameter estimates, which is at the core of uncertainty assessment in model- based inference, was also found to have a minor to moderate effect on the precision of model-based estimators. Regarding sample sizes, the model error contribution to the total variance remains the same regardless of the sample size of the first phase (i.e., the size of the sample of auxiliary data); to reduce the model-error contribution, there is a need to increase the sample size of the second phase (i.e., the size of the sample of field plots for developing regression models). As a baseline for comparisons, model-assisted estimators were applied and found to be about equally precise as the model-based estimators, in accordance with the theory for the case when models are estimated from the sample data. Key words: airborne LiDAR, heteroscedasticity, Landsat, model-based estimator, sampling simulation. Dans cette etude, nous examinons l'utilisation de l'inference fondee sur un modele dans le cas des inventaires forestiers pour lesquels des donnees auxiliaires sont disponibles comme echantillon aleatoire. Nous evaluons les effets de la forme du modele et de la taille de l'echantillon sur les estimateurs du volume sur pied a partir de differents types de donnees auxiliaires de teledetection. L'etude a ete realisee a l'aide d'une simulation d'echantillonnage de Monte Carlo en utilisant un plan d'echantillonnage a deux phases dans une aire d'etude simulee avec des conditions semblables a celles du centre-ouest de la Finlande. Nous montrons que le choix du modele a un effet variant de mineur a modere sur la precision des estimateurs provenant des modeles. De meme, le choix de l'estimateur de la matrice des variances-covariances des estimations des parametres des modeles, qui est au centre de l'estimation de l'incertitude de l'inference fondee sur un modele, a aussi eu un effet variant de mineur a modere sur la precision. En ce qui concerne la taille de l'echantillon, la contribution de l'erreur du modele a la variance totale est demeuree la meme peu importe la taille de l'echantillon de la premiere phase (c.-a-d. la taille de l'echantillon des donnees auxiliaires). Pour reduire la contribution de l'erreur du modele, il faut augmenter la taille de l'echantillon de la deuxieme phase c.-a-d. la taille de l'echantillon des placettes de terrain visant a mettre au point des modeles de regression). A titre de base de reference pour des comparaisons, des estimateurs assistes par des modeles ont ete appliques et ont produit une precision a peu pres egale a celle des estimateurs fondes sur les modeles, ce qui est conforme a la theorie dans le cas des modeles estimes a partir de donnees echantillonnees. [Traduit par la Redaction] Mots-cles: lidar aeroporte, heterocedasticite, Landsat, estimateur fonde sur un modele, simulation d'echantillonnage., 1. Introduction The global interest in forests is increasing due to the multiple services provided by this ecosystem (e.g., Mery et al. 2005). As a consequence, several international agreements such [...]