Juho Pitkänen

An International Comparison of Individual Tree Detection and Extraction Using Airborne Laser Scanning

The objective of the “Tree Extraction” project organized by EuroSDR (European Spatial data Research) and ISPRS (International Society of Photogrammetry and Remote Sensing) was to evaluate the quality, accuracy, and feasibility of automatic tree extraction methods, mainly based on laser scanner data. In the final report of the project, Kaartinen and Hyyppa (2008) reported a high variation in the quality of the published methods under boreal forest conditions and with varying laser point densities. This paper summarizes the findings beyond the final report after analyzing the results obtained in different tree height classes. Omission/Commission statistics as well as neighborhood relations are taken into account. Additionally, four automatic tree detection and extraction techniques were added to the test. Several methods in this experiment were superior to manual processing in the dominant, co-dominant and suppressed tree storeys. In general, as expected, the taller the tree, the better the location accuracy. The accuracy of tree height, after removing gross errors, was better than 0.5 m in all tree height classes with the best methods investigated in this experiment. For forest inventory, minimum curvature-based tree detection accompanied by point cloud-based cluster detection for suppressed trees is a solution that deserves attention in the future.

Airborne laser scanning-based decision support for wood procurement planning

We present a decision support tool for guiding the selection of marked stands based on airborne laser scanning (ALS) data. We describe three stages, namely (1) wall-to-wall mapping of the stands matured for cutting using low-density ALS data; (2) tree-level inventory of these stands using high-density ALS data and (3) theoretical bucking of the imputed tree stems to produce detailed information on their characteristics. We tested them in a Scots pine dominated boreal forest area in Eastern Finland, where 79 sample plots were measured in the field. The detection of the stands matured for cutting had a success rate of 95% and our results demonstrated a further potential to limit the result towards stands dominated by certain species by means of intensity values derived from the low-density ALS data. The applied single-tree detection and estimation chain produced detailed tree-level information and realistic diameter distributions, yet the detection was highly emphasised on the dominant tree layer. The error levels in the estimates were generally less than standard deviations of the field attributes. Finally, plot-level accumulations of saw-log volumes were found rather similar, whether the input was based on the imputed tree data or trees measured in the field. The results are considered useful for ranking the stands based on their properties, whether the aim in the wood procurement is to focus on certain species or to select stands suitable for production needs.

Predicting the spatial pattern of trees by airborne laser scanning

The spatial pattern of trees can be defined as a property of their location in relation to each other. In this study, the spatial pattern was summarized into three categories, regular, random, and clustered, using Ripleys L-function. The study was carried out at 79 sample plots located in a managed forest in Finland. The goal was to study how well the spatial pattern of trees can be predicted by airborne laser scanning (ALS) data. ALS-derived predictions were based upon individual tree detection (ITD), semi-individual tree detection (semi-ITD), and plot-level metrics calculated from the canopy height model, AREA. The kappa value for ITD was almost zero, which indicates no agreement. The semi-ITD and AREA methods performed better, although kappa values were only 0.34 and 0.24, respectively. It appears difficult to detect a particularly clustered spatial pattern.

Edge-Tree Correction for Predicting Forest Inventory Attributes Using Area-Based Approach With Airborne Laser Scanning

We describe a novel method to improve the correspondence between field and airborne laser scanning (ALS) measurements in an area-based approach (ABA) forest inventory framework. An established practice in forest inventory is that trees with boles falling within a fixed border field measurement plot are considered “in” trees; yet their crowns may extend beyond the plot border. Likewise, a tree bole may fall outside of a plot, but its crown may extend into a plot. Typical ABA approaches do not recognize these discrepancies between the ALS data extracted for a given plot and the corresponding field measurements. In the proposed solution, enhanced ABA (EABA), predicted tree positions, and crown shapes are used to adjust plot and grid cell boundaries and how ALS metrics are computed. The idea is to append crowns of “in” trees to a plot and cut down “out” trees, then EABA continues in the traditional fashion as ABA. The EABA method requires higher density ALS data than ABA because improvement is obtained by means of detecting individual trees. When compared to typical ABA, the proposed EABA method decreased the error rate (RMSE) of stem volume prediction from 23.16% to 19.11% with 127 m2 plots and from 19.08% to 16.95% with 254 m2 plots. The greatest improvements were obtained for plots with the largest residuals.

Classification of forest land attributes using multi-source remotely sensed data

Abstract The aim of the study was to (1) examine the classification of forest land using airborne laser scanning (ALS) data, satellite images and sample plots of the Finnish National Forest Inventory (NFI) as training data and to (2) identify best performing metrics for classifying forest land attributes. Six different schemes of forest land classification were studied: land use/land cover (LU/LC) classification using both national classes and FAO (Food and Agricultural Organization of the United Nations) classes, main type, site type, peat land type and drainage status. Special interest was to test different ALS-based surface metrics in classification of forest land attributes. Field data consisted of 828 NFI plots collected in 2008–2012 in southern Finland and remotely sensed data was from summer 2010. Multinomial logistic regression was used as the classification method. Classification of LU/LC classes were highly accurate (kappa-values 0.90 and 0.91) but also the classification of site type, peat land type and drainage status succeeded moderately well (kappa-values 0.51, 0.69 and 0.52). ALS-based surface metrics were found to be the most important predictor variables in classification of LU/LC class, main type and drainage status. In best classification models of forest site types both spectral metrics from satellite data and point cloud metrics from ALS were used. In turn, in the classification of peat land types ALS point cloud metrics played the most important role. Results indicated that the prediction of site type and forest land category could be incorporated into stand level forest management inventory system in Finland.

Integrating multi-source data for a tropical forest inventory—a case study in the Kon Tum region, Vietnam

Summary This article presents a processing chain for forest volume mapping based on multi-source forest inventory methodology and the existing inventory data collected from the Kon Tum region, Vietnam. The modelling framework for imputing tally tree heights was built based on a mixed-effects height generalisation model. Mapping of the stem volume was based on nearest neighbour techniques (k-NN) and Landsat TM data after relative calibration with MODIS image material as underlying reference. The use of optical image materials, together with the demanding conditions set by a tropical forest structure, resulted in a moderate root mean square error value (76.6%) for the stem volume. The resulting volume maps, which were based on an objective estimation procedure, create the appropriate model dataset needed for testing the optimal large-scale inventory designs of forthcoming forest inventories that will be carried out in Vietnam.

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

ABSTRACT 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.