Johannes Breidenbach

Comparison of nearest neighbour approaches for small area estimation of tree species-specific forest inventory attributes in central Europe using airborne laser scanner data

The three nonparametric k nearest neighbour (kNN) approaches, most similar neighbour inference (MSN), random forests (RF) and random forests based on conditional inference trees (CF) were compared for spatial predictions of standing timber volume with respect to tree species compositions and for predictions of stem number distributions over diameter classes. Various metrics derived from airborne laser scanning (ALS) data and the characteristics of tree species composition obtained from coarse stand level ground surveys were applied as auxiliary variables. Due to the results of iterative variable selections, only the ALS data proved to be a relevant predictor variable set. The three applied NN approaches were tested in terms of bias and root mean squared difference (RMSD) at the plot level and standard errors at the stand level. Spatial correlations were considered in the statistical models. While CF and MSN performed almost similarly well, large biases were observed for RF. The obtained results suggest that biases in the RF predictions were caused by inherent problems of the RF approach. Maps for Norway spruce and European beech timber volume were exemplarily created. The RMSD values of CF at the plot level for total volume and the species-specific volumes for European beech, Norway spruce, European silver fir and Douglas fir were 32.8, 80.5, 99.0, 137.0 and 261.1%. These RMSD values were smaller than the standard deviation, although Douglas fir volume did not belong to the actual response variables. All three non-parametric approaches were also capable of predicting diameter distributions. The standard errors of the nearest neighbour predictions on the stand level were generally smaller than the standard error of the sample plot inventory. In addition, the employed model-based approach allowed kNN predictions of means and standard errors for stands without sample plots.

Use of models in large-area forest surveys: comparing model-assisted, model-based and hybrid estimation

This paper focuses on the use of models for increasing the precision of estimators in large-area forest surveys. It is motivated by the increasing availability of remotely sensed data, which facilitates the development of models predicting the variables of interest in forest surveys. We present, review and compare three different estimation frameworks where models play a core role: model-assisted, model-based, and hybrid estimation. The first two are well known, whereas the third has only recently been introduced in forest surveys. Hybrid inference mixes design-based and model-based inference, since it relies on a probability sample of auxiliary data and a model predicting the target variable from the auxiliary data..We review studies on large-area forest surveys based on model-assisted, model-based, and hybrid estimation, and discuss advantages and disadvantages of the approaches. We conclude that no general recommendations can be made about whether model-assisted, model-based, or hybrid estimation should be preferred. The choice depends on the objective of the survey and the possibilities to acquire appropriate field and remotely sensed data. We also conclude that modelling approaches can only be successfully applied for estimating target variables such as growing stock volume or biomass, which are adequately related to commonly available remotely sensed data, and thus purely field based surveys remain important for several important forest parameters.

Early Detection of Bark Beetle Green Attack Using TerraSAR-X and RapidEye Data

Bark beetles cause widespread damages in the coniferous-dominated forests of central Europe and North America. In the future, areas affected by bark beetles may further increase due to climate change. However, the early detection of the bark beetle green attack can guide management decisions to prevent larger damages. For this reason, a field-based bark beetle monitoring program is currently implemented in Germany. The combination of remote sensing and field data may help minimizing the reaction time and reducing costs of monitoring programs covering large forested areas. In this case study, RapidEye and TerraSAR-X data were analyzed separately and in combination to detect bark beetle green attack. The remote sensing data were acquired in May 2009 for a study site in south-west Germany. In order to distinguish healthy areas and areas affected by bark beetle green attack, three statistical approaches were compared: generalized linear models (GLM), maximum entropy (ME) and random forest (RF). The spatial scale (minimum mapping unit) was 78.5 m2. TerraSAR-X data resulted in fair classification accuracy with a cross-validated Cohen’s Kappa Coefficient (kappa) of 0.23. RapidEye data resulted in moderate classification accuracy with a kappa of 0.51. The highest classification accuracy was obtained by combining the TerraSAR-X and RapidEye data, resulting in a kappa of 0.74. The accuracy of ME models was considerably higher than the accuracy of GLM and RF models.

Empirical coverage of model-based variance estimators for remote sensing assisted estimation of stand-level timber volume

Due to the availability of good and reasonably priced auxiliary data, the use of model-based regression-synthetic estimators for small area estimation is popular in operational settings. Examples are forest management inventories, where a linking model is used in combination with airborne laser scanning data to estimate stand-level forest parameters where no or too few observations are collected within the stand. This paper focuses on different approaches to estimating the variances of those estimates. We compared a variance estimator which is based on the estimation of superpopulation parameters with variance estimators which are based on predictions of finite population values. One of the latter variance estimators considered the spatial autocorrelation of the residuals whereas the other one did not. The estimators were applied using timber volume on stand level as the variable of interest and photogrammetric image matching data as auxiliary information. Norwegian National Forest Inventory (NFI) data were used for model calibration and independent data clustered within stands were used for validation. The empirical coverage proportion (ECP) of confidence intervals (CIs) of the variance estimators which are based on predictions of finite population values was considerably higher than the ECP of the CI of the variance estimator which is based on the estimation of superpopulation parameters. The ECP further increased when considering the spatial autocorrelation of the residuals. The study also explores the link between confidence intervals that are based on variance estimates as well as the well-known confidence and prediction intervals of regression models.

The Influence of DEM Quality on Mapping Accuracy of Coniferous- and Deciduous-Dominated Forest Using TerraSAR‑X Images

Abstract: Climate change is a factor that largely contributes to the increase of forest areas affected by natural damages. Therefore, the development of methodologies for forest monitoring and rapid assessment of affected areas is required. Space-borne synthetic aperture radar (SAR) imagery with high resolution is now available for large-scale forest mapping and forest monitoring applications. However, a correct interpretation of SAR images requires an adequate preprocessing of the data consisting of orthorectification and radiometric calibration. The resolution and quality of the digital elevation model (DEM) used as reference is crucial for this purpose. Therefore, the primary aim of this study was to analyze the influence of the DEM quality used in the preprocessing of the SAR data on the mapping accuracy of forest types. In order to examine TerraSAR-X images to map forest dominated by deciduous and coniferous trees, High Resolution SpotLight images were acquired for two study sites in southern Germany. The SAR images were preprocessed with a Shuttle Radar Topography Mission (SRTM) DEM (resolution approximately 90 m), an airborne laser scanning (ALS) digital terrain model (DTM) (5 m resolution), and an ALS digital surface model (DSM) (5 m resolution). The orthorectification of the SAR images using high resolution ALS DEMs was found to be important for the reduction of errors in pixel location and to

Modeling Soil Carbon Dynamics in Northern Forests: Effects of Spatial and Temporal Aggregation of Climatic Input Data

Boreal forests contain 30% of the global forest carbon with the majority residing in soils. While challenging to quantify, soil carbon changes comprise a significant, and potentially increasing, part of the terrestrial carbon cycle. Thus, their estimation is important when designing forest-based climate change mitigation strategies and soil carbon change estimates are required for the reporting of greenhouse gas emissions. Organic matter decomposition varies with climate in complex nonlinear ways, rendering data aggregation nontrivial. Here, we explored the effects of temporal and spatial aggregation of climatic and litter input data on regional estimates of soil organic carbon stocks and changes for upland forests. We used the soil carbon and decomposition model Yasso07 with input from the Norwegian National Forest Inventory (11275 plots, 1960–2012). Estimates were produced at three spatial and three temporal scales. Results showed that a national level average soil carbon stock estimate varied by 10% depending on the applied spatial and temporal scale of aggregation. Higher stocks were found when applying plot-level input compared to country-level input and when long-term climate was used as compared to annual or 5-year mean values. A national level estimate for soil carbon change was similar across spatial scales, but was considerably (60–70%) lower when applying annual or 5-year mean climate compared to long-term mean climate reflecting the recent climatic changes in Norway. This was particularly evident for the forest-dominated districts in the southeastern and central parts of Norway and in the far north. We concluded that the sensitivity of model estimates to spatial aggregation will depend on the region of interest. Further, that using long-term climate averages during periods with strong climatic trends results in large differences in soil carbon estimates. The largest differences in this study were observed in central and northern regions with strongly increasing temperatures.

Interferometric SAR DEMs for Forest Change in Uganda 2000–2012

Monitoring changes in forest height, biomass and carbon stock is important for understanding the drivers of forest change, clarifying the geography and magnitude of the fluxes of the global carbon budget and for providing input data to REDD+. The objective of this study was to investigate the feasibility of covering these monitoring needs using InSAR DEM changes over time and associated estimates of forest biomass change and corresponding net CO2 emissions. A wall-to-wall map of net forest change for Uganda with its tropical forests was derived from two Digital Elevation Model (DEM) datasets, namely the SRTM acquired in 2000 and TanDEM-X acquired around 2012 based on Interferometric SAR (InSAR) and based on the height of the phase center. Errors in the form of bias, as well as parallel lines and belts having a certain height shift in the SRTM DEM were removed, and the penetration difference between X- and C-band SAR into the forest canopy was corrected. On average, we estimated X-band InSAR height to decrease by 7 cm during the period 2000–2012, corresponding to an estimated annual CO2 emission of 5 Mt for the entirety of Uganda. The uncertainty of this estimate given as a 95% confidence interval was 2.9–7.1 Mt. The presented method has a number of issues that require further research, including the particular SRTM biases and artifact errors; the penetration difference between the X- and C-band; the final height adjustment; and the validity of a linear conversion from InSAR height change to AGB change. However, the results corresponded well to other datasets on forest change and AGB stocks, concerning both their geographical variation and their aggregated values.

Remote sensing and forest inventories in Nordic countries – roadmap for the future

ABSTRACT The Nordic countries have long traditions in forest inventory and remote sensing (RS). In sample-based national forest inventories (NFIs), utilization of aerial photographs started during the 1960s, satellite images during the 1980s, laser scanning during the 2000s, and photogrammetric point clouds during the 2010s. In forest management inventories (FMI), utilization of aerial photos started during the 1940s and laser scanning during the 2000s. However, so far, RS has mostly been used for map production and research rather than for estimation of regional parameters or inference on their accuracy. In recent years, the RS technology has been developing very fast. At the same time, the needs for information are constantly increasing. New technologies have created possibilities for cost-efficient production of accurate, large area forest data sets, which also will change the way forest inventories are done in the future. In this study, we analyse the state-of-the-art both in the NFIs and FMIs in the Nordic countries. We identify the benefits and drawbacks of different RS materials and data acquisition approaches with different user perspectives. Based on the analysis, we identify the needs for further development and emerging research questions. We also discuss alternatives for ownership of the data and cost-sharing between different actors in the field.

Longitudinal height-diameter curves for Norway spruce, Scots pine and silver birch in Norway based on shape constraint additive regression models

BackgroundGeneralized height-diameter curves based on a re-parameterized version of the Korf function for Norway spruce (Picea abies (L.) Karst.), Scots pine (Pinus sylvestris L.) and silver birch (Betula pendula Roth) in Norway are presented. The Norwegian National Forest Inventory (NFI) is used as data base for estimating the model parameters. The derived models are developed to enable spatially explicit and site sensitive tree height imputation in forest inventories as well as future tree height predictions in growth and yield scenario simulations.MethodsGeneralized additive mixed models (gamm) are employed to detect and quantify potentially non-linear effects of predictor variables. In doing so the quadratic mean diameter serves as longitudinal covariate since stand age, as measured in the NFI, shows only a weak correlation with a stands developmental status in Norwegian forests. Additionally the models can be locally calibrated by predicting random effects if measured height-diameter pairs are available. Based on the model selection of non-constraint models, shape constraint additive models (scam) were fit to incorporate expert knowledge and intrinsic relationships by enforcing certain effect patterns like monotonicity.ResultsModel comparisons demonstrate that the shape constraints lead to only marginal differences in statistical characteristics but ensure reasonable model predictions. Under constant constraints the developed models predict increasing tree heights with decreasing altitude, increasing soil depth and increasing competition pressure of a tree. A two-dimensional spatially structured effect of UTM-coordinates accounts for the potential effects of large scale spatially correlated covariates, which were not at our disposal. The main result of modelling the spatially structured effect is lower tree height prediction for coastal sites and with increasing latitude. The quadratic mean diameter affects both the level and the slope of the height-diameter curve and both effects are positive.ConclusionsIn this investigation it is assumed that model effects in additive modelling of height-diameter curves which are unfeasible and too wiggly from an expert point of view are a result of quantitatively or qualitatively limited data bases. However, this problem can be regarded not to be specific to our investigation but more general since growth and yield data that are balanced over the whole data range with respect to all combinations of predictor variables are exceptional cases. Hence, scam may provide methodological improvements in several applications by combining the flexibility of additive models with expert knowledge.

Assessing uncertainty: Sample size trade-offs in the development and application of carbon stock models

Many parties to the United Nation’s Framework Convention on Climate Change (UNFCCC) base their reporting of change in Land Use, Land-Use Change and Forestry (LULUCF) sector carbon pools on national forest inventories. A strong feature of sample-based inventories is that very detailed measurements can be made at the level of plots. Uncertainty regarding the results stems primarily from the fact that only a sample, and not the entire population, is measured. However, tree biomass on sample plots is not directly measured but rather estimated using regression models based on allometric features such as tree diameter and height. Estimators of model parameters are random variables that exhibit different values depending on which sample is used for estimating model parameters. Although sampling error is strongly influenced by the sample size when the model is applied, modeling error is strongly influenced by the sample size when the model is under development. Thus, there is a trade-off between which sample sizes to use when applying and developing models. This trade-off has not been studied before and is of specific interest for countries developing new national forest inventories and biomass models in the REDD context. This study considers a specific sample design and population. This fact should be considered when extrapolating results to other locations and populations.