Svein Solberg

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.

Single Tree Segmentation Using Airborne Laser Scanner Data in a Structurally Heterogeneous Spruce Forest

In this study, we present a new method for single tree segmentation and characterization from a canopy surface model (CSM), and its corresponding point cloud, based on airborne laser scanning. The method comprises new algorithms for controlling the shape of crown segments, and for residual adjustment of the canopy surface model (CSM). We present a new criterion that measures the success of locating trees, and demonstrate how this criterion can be used for optimizing the degree of CSM smoothing. From the adjusted CSM segments, we derived tree height and crown diameter, and based on all first laser pulse measurements within the segments we derived crown-base height. The method was applied and validated in a Norway spruce dominated forest reserve having a heterogeneous structure. The number of trees automatically detected varied with social status of the trees, from 93 percent of the dominant trees to 19 percent of the suppressed trees. The RMSE values for tree height, crown diameter, and crown-base height were around 1.2 m, 1.1 m, and 3.5 m, respectively. The method overestimated crown diameter (0.8 m) and crown base height (3.0 m).

Ecologically implausible carbon response

Arising from: F. Magnani et al. 447, 849–851 (2007)10.1038/nature05847; Magnani et al. replyMagnani et al. present a very strong correlation between mean lifetime net ecosystem production (NEP, defined as the net rate of carbon (C) accumulation in ecosystems) and wet nitrogen (N) deposition. For their data in the range 4.9–9.8 kg N ha-1 yr-1, on which the correlation largely depends, the response is approximately 725 kg C per kg N in wet deposition. According to the authors, the maximum N wet deposition level of 9.8 kg N ha-1 yr-1 is equivalent to a total deposition of 15 kg N ha-1 yr-1, implying a net sequestration near 470 kg C per kg N of total deposition. We question the ecological plausibility of the relationship and show, from a multi-factor analysis of European forest measurements, how interactions with site productivity and environment imply a much smaller NEP response to N deposition.

Mapping gap fraction, LAI and defoliation using various ALS penetration variables

Four alternative airborne laser scanning (ALS) canopy penetration variables were compared for their suitability for mapping of gap fraction, leaf area index and disturbances in a Scots pine forest. The variables were based on either echo counting or intensity, and on either first or first and last echoes. ALS data and field-measured gap fraction and effective leaf area index (LAIe) were gathered before and after a severe insect defoliation by pine sawflies. LAIe is a commonly used form of leaf area index that is mathematically derived from gap fraction, and includes the areas of foliage, branches and trunks, and which is not corrected for the clumping of foliage. The ALS penetration variables were almost equally strongly related to field-measured gap fraction and LAIe. The estimated slopes in the LAIe models varied from 0.94 to 2.71, and had coefficient of determination R 2 values of 0.92–0.94. They were strongly correlated to each other (R 2 values of 0.95–0.98) and agreed fairly well for temporal changes of LAIe during the summer and the insect defoliation (R 2 values of 0.82–0.95). Counting of first and last echoes produced penetration rates close to the gap fraction, and this penetration variable was able to penetrate tree crowns. Ground-only echoes represented mostly between-tree gaps, and canopy-first-ground-last pulses represented mostly within-canopy gaps. However, the penetration variables based on first and last echoes suffered from the problem that a second echo might be impaired both in low and in tall canopies. In low canopies, two adjacent echoes from the same pulse would be too close in time to be separated by the sensor, while in tall canopies the pulse might apparently be fragmented down through the canopy. The intensity-based penetration variables needed to be supplemented with reflectance values, or at least the ratio between reflectance of the canopy and the ground, and this ratio was estimated from the data. The study demonstrated that one might be able to distinguish between disturbance types, e.g. between defoliation and cutting, by comparing alternative ALS penetration variables. Insect defoliation was dominated by an increase in within-canopy gaps and, correspondingly, the fraction of partly penetrating canopy-first-ground-last pulses. Tree removals from cutting were dominated by increases in between-tree gaps and the corresponding fraction of ground-only pulses.

Temporal trends in throughfall and soil water chemistry at three Norwegian forests, 1986-1997

The Norwegian intensive monitoring programme of forest condition has recorded rainfall, throughfall and soil water data from 1986 at 16 forest plots. Using covariance analysis, this study has examined short-term and episodic influences on soil water ionic concentration at three of the plots, and identified both seasonal and long-term temporal trends. Acidity has decreased in bulk precipitation and throughfall, and the concentrations of base cations in the organic soil horizon have increased. Nevertheless, there is evidence of continued acidification in the organic and mineral soil horizons, though of a small scale. The influence of sea salt and drought effects on soil water chemistry are examined, but found to be unimportant in causing acidification effects such as increased soil aluminium concentration.

Crown density assessments, control surveys and reproducibility

This study aims to evaluate the quality of crown density data, based on independent, pairwise tree assessments. The data originates from monitoring of forest health (crown condition) in Norway; 250 plots, comprising 12 000 individual trees of Norway spruce, have been reassessed by a single observer during 1990–95. Of the trees, 2300 were controlled more than twice, providing the possibility of evaluating the quality of assessed temporal changes of crown density. True errors (standard deviation) are estimated to be about 10% for single trees and 5% for plot means, while the real standard deviation of the differences were slightly higher. The errors of the temporal changes of crown density were of similar magnitude. Systematic differences in crown density were found between sites and plot types, partly resulting from observer bias. However, the results suggest that observer bias is really the result of each observers personal style in assessment.

Deriving forest monitoring variables from X-band InSAR SRTM height.

The suitability of interferometric X-band radar for forest monitoring was investigated. Working in a spruce-dominated forest in southeast Norway, top height, mean height, stand density, stem volume, and biomass were related to space shuttle interferometric height above ground. A ground truth dataset was produced for each radar data pixel in the study area by combining a field inventory and automatic tree detection with airborne laser scanning data. Pixels were aggregated to forest stands. Interferometric height was strongly related to all of the five forest variables, and most strongly to top height with R2 = 0.71 and RMSE = 13% at the pixel level and R2 = 0.82 and RMSE = 5.6% at the stand level. Interferometric height was linearly related to stem volume and biomass up to 400 m3/ha and 200 t/ha, respectively, and RMSE was approximately 19% for both variables. These errors contain error components caused by the 3.5-year time lag between the radar acquisition and the laser scanning. It is concluded that interferometric X-band radar has potential for use in forest monitoring.

Time series study of concentrations of SO42− and H+ in precipitation and soil waters in Norway

Along with a steady reduction of acid inputs during 14 years of intensive forest monitoring in Norway, the influence of acid deposition upon soil water acidity is gradually reduced in favour of other and internal sources of H+ and sulphate, in particular from processes in the upper soil layer. We used statistical analyses in two steps for precipitation, throughfall and soil water at 5, 15 and 40 cm depths. Firstly, we employed time series analyses to model the temporal variation as a long-term linear trend and a monthly variation, and by this filtered out residual, weekly variation. Secondly, we used the parameter estimates and the residuals from this to show that the long term, the monthly and the weekly variation in one layer were correlated to similar temporal variation in the above, adjacent layer. This was strongly evident for throughfall correlated to precipitation, but much weaker for soil water. Continued acidification in soil water on many plots suggests that the combined effects of anthropogenic and natural acid inputs exceed in places the buffering capacity of the soil.

Changes of forest health in Norwegian boreal forests during 15 years

Abstract The Norwegian Monitoring Programme for Forest Damage has now been running since 1984. Its main objective has been monitoring the boreal forest conditions in relation to air pollution. Surveys of forests are performed on plots in a nation-wide representative grid network, in a network of local countywise plots, and in a network of intensively monitored plots within the framework of the internationally co-ordinated UN/ECE ICP Forests. At intensively monitored plots, a number of measurements were performed, for example tree crown-condition assessments, foliar chemistry, air pollution, precipitation chemistry (open area and stand throughfall), soil and soil-water chemistry. Vitality criteria have shown a declining trend, expressed as reduced crown density and more of discoloured trees, particularly in spruce forests. These results are similar to findings in other Scandinavian countries and coincide with the general trend in Europe. Forest yield has increased in later years in most parts of Norway. Tree mortality in excess of normal is not recorded. Considering these results, it is reasonable to presume that most Norwegian forest ecosystems are generally in a good condition. Areas in the south-eastern part of the country, however, may have shown indications of a slightly reduced crown condition. These areas are located in a region with a relatively high load of air pollution and low buffering capacity against acidification. Although forest conditions generally depend on soil, tree age, climate, pests and diseases, and other natural stressors; air pollution loads add to, or may interact with, these factors. Most likely inciting factors (e.g. summer drought) have occurred and produced visual symptoms. The actual effect of the air pollution component is, therefore, difficult to estimate; however, its importance is not discounted.

Forest biomass change estimated from height change in interferometric SAR height models.

BackgroundThere is a need for new satellite remote sensing methods for monitoring tropical forest carbon stocks. Advanced RADAR instruments on board satellites can contribute with novel methods. RADARs can see through clouds, and furthermore, by applying stereo RADAR imaging we can measure forest height and its changes. Such height changes are related to carbon stock changes in the biomass. We here apply data from the current Tandem-X satellite mission, where two RADAR equipped satellites go in close formation providing stereo imaging. We combine that with similar data acquired with one of the space shuttles in the year 2000, i.e. the so-called SRTM mission. We derive height information from a RADAR image pair using a method called interferometry.ResultsWe demonstrate an approach for REDD based on interferometry data from a boreal forest in Norway. We fitted a model to the data where above-ground biomass in the forest increases with 15 t/ha for every m increase of the height of the RADAR echo. When the RADAR echo is at the ground the estimated biomass is zero, and when it is 20 m above the ground the estimated above-ground biomass is 300 t/ha. Using this model we obtained fairly accurate estimates of biomass changes from 2000 to 2011. For 200 m2 plots we obtained an accuracy of 65 t/ha, which corresponds to 50% of the mean above-ground biomass value. We also demonstrate that this method can be applied without having accurate terrain heights and without having former in-situ biomass data, both of which are generally lacking in tropical countries. The gain in accuracy was marginal when we included such data in the estimation. Finally, we demonstrate that logging and other biomass changes can be accurately mapped. A biomass change map based on interferometry corresponded well to a very accurate map derived from repeated scanning with airborne laser.ConclusionsSatellite based, stereo imaging with advanced RADAR instruments appears to be a promising method for REDD. Interferometric processing of the RADAR data provides maps of forest height changes from which we can estimate temporal changes in biomass and carbon.