Fabian Ewald Fassnacht

Comparison of Feature Reduction Algorithms for Classifying Tree Species With Hyperspectral Data on Three Central European Test Sites

Tree species information is a basic variable for forest inventories. Knowledge on tree species is relevant for biomass estimation, habitat quality assessment, and biodiversity characterization. Hyperspectral data have been proven to have a high potential for the mapping of tree species composition. However, open questions remain concerning the robustness of existing classification approaches. Here, a number of classification approaches were compared to classify tree species from airborne hyperspectral data across three forest sites to identify a single approach which continuously delivers high classification performances over all test sites. Examined approaches included three feature selection methods [genetic algorithm (GA), support vector machines (SVM) wrapper, and sparse generalized partial least squares selection (PLS)] each combined with two nonparametric classifiers (SVM and Random Forest). Two further setups included classifications applied to the full hyperspectral dataset and to an image transformed with a minimum noise fraction (MNF) transformation. Results showed that SVM wrapper and the GA slightly outperformed the PLS-based algorithm. In most cases, the best classification runs involving a feature selection algorithm outperformed those incorporating the full hyperspectral dataset. However, the best overall results were obtained when using the first 10-20 components of the MNF-transformed image. Selected bands were frequently located in the visual region close to the green peak, at the chlorophyll absorption feature and the red edge rise as well as in three parts of the short-wave infrared region close to water absorption features. These findings are relevant for improving the robustness of tree species classifications from airborne hyperspectral data incorporating feature reduction techniques.

Stratified aboveground forest biomass estimation by remote sensing data

Abstract Remote sensing-assisted estimates of aboveground forest biomass are essential for modeling carbon budgets. It has been suggested that estimates can be improved by building species- or strata-specific biomass models. However, few studies have attempted a systematic analysis of the benefits of such stratification, especially in combination with other factors such as sensor type, statistical prediction method and sampling design of the reference inventory data. We addressed this topic by analyzing the impact of stratifying forest data into three classes (broadleaved, coniferous and mixed forest). We compare predictive accuracy (a) between the strata (b) to a case without stratification for a set of pre-selected predictors from airborne LiDAR and hyperspectral data obtained in a managed mixed forest site in southwestern Germany. We used 5 commonly applied algorithms for biomass predictions on bootstrapped subsamples of the data to obtain cross validated RMSE and r2 diagnostics. Those values were analyzed in a factorial design by an analysis of variance (ANOVA) to rank the relative importance of each factor. Selected models were used for wall-to-wall mapping of biomass estimates and their associated uncertainty. The results revealed marginal advantages for the strata-specific prediction models over the unstratified ones, which were more obvious on the wall-to-wall mapped area-based predictions. Yet further tests are necessary to establish the generality of these results. Input data type and statistical prediction method are concluded to remain the two most crucial factors for the quality of remote sensing-assisted biomass models.

Forest inventories by LiDAR data: A comparison of single tree segmentation and metric-based methods for inventories of a heterogeneous temperate forest

Abstract Inventories of temperate forests of Central Europe mainly rely on terrestrial measurements. Rapid alterations of forests by disturbances and multilayer silvicultural systems increasingly challenge the use of conventional plot based inventories, particularly in protected areas. Airborne LiDAR offers an alternative or supplement to conventional inventories, but despite the possibility of obtaining such remote sensing data, its operational use for broader areas in Central Europe remains experimental. We evaluated two methods of forest inventory that use LiDAR data at the landscape level: the single tree segment-based method and an area-based method. We compared a set of structural forest attributes modeled by these methods with a conventional forest inventory of the highly heterogeneous forest of the Bavarian Forest National Park (Germany), which partially includes stands affected by severe natural disturbances. Area-based models were accurate for all structural attributes, with cross-validated average root mean squared error ranging from ∼3.4 to ∼13.4 in the best modeling case. The coefficients of variation for the mapped area-based estimations were mostly minor. The area-based estimations were varied but highly correlated (Pearson’s correlations between ∼ 0.56 and 0.85) with single tree segmentation estimations; undetected trees in the single tree segmentat-based method were the main sources of inconsistency. The single tree segment-based method was highly correlated (∼ 0.54 to 0.90) with data from ground-based forest inventories. The single tree-based algorithm delivered highly reliable estimates for a set of forest structural attributes that are of interest in forest inventories at the landscape scale. We recommend LiDAR forest inventories at the landscape scale in both heterogeneous commercial forests and large protected areas in the central European temperate sites.

Non-destructive estimation of foliar carotenoid content of tree species using merged vegetation indices

Leaf pigment content is an important indicator of plant status and can serve to assess the vigor and photosynthetic activity of plants. The application of spectral information gathered from laboratory, field and remote sensing-based spectrometers to non-destructively assess total chlorophyll (Chl) content of higher plants has been demonstrated in earlier studies. However, the precise estimation of carotenoid (Car) content with non-destructive spectral measurements has so far not reached accuracies comparable to the results obtained for Chl content. Here, we examined the potential of a recently developed angular vegetation index (AVI) to estimate total foliar Car content of three tree species. Based on an iterative search of all possible band combinations, we identified a best candidate AVIcar. The identified index showed quite close but essentially not linear relation with Car contents of the examined species with increasing sensitivity to high Car content and a lack of sensitivity to low Car content for which earlier proposed vegetation indices (VI) performed better. To make use of the advantages of both VI types, we developed a simple merging procedure, which combined the AVIcar with two earlier proposed carotenoid indices. The merged indices had close linear relationship with total Car content and outperformed all other examined indices. The merged indices were able to accurately estimate total Car content with a percental root mean square error (%RMSE) of 8.12% and a coefficient of determination of 0.88. Our findings were confirmed by simulations using the radiative transfer model PROSPECT-5. For simulated data, the merged indices again showed a quasi linear relationship with Car content. This strengthens the assumption that the proposed merged indices have a general ability to accurately estimate foliar Car content. Further examination of the proposed merged indices to estimate foliar Car content of other plant species is desirable to prove the general applicability of the index for non-destructive estimation of Car from leaf reflectance data.

Modeling forest biomass using very-high-resolution data - combining textural, spectral and photogrammetric predictors derived from spaceborne stereo images

Abstract We used spectral, textural and photogrammetric information from very-high resolution (VHR) stereo satellite data (Pléiades and WorldView-2) to estimate forest biomass across two test sites located in Chile and Germany. We compared Random Forest model performances of different predictor sets (spectral, textural, and photogrammetric), forest inventory designs and filter sizes (texture information). Best model performances were obtained with photogrammetric combined with either textural or spectral information and smaller, but more field plots. Stereo-VHR images showed a great potential for canopy height model (CHM) generation and could be an adequate alternative to LiDAR and InSAR techniques.

Using a Multistructural Object-Based LiDAR Approach to Estimate Vascular Plant Richness in Mediterranean Forests With Complex Structure

A multistructural object-based LiDAR approach to predict plant richness in complex structure forests is presented. A normalized LiDAR point cloud was split into four height ranges: 1) high canopies (points above 16 m); 2) middle-high canopies (8-16 m); 3) middle-low canopies (2-8 m); and 4) low canopies (0-2 m). A digital canopy model (DCM) was obtained from the full normalized LiDAR point cloud, and four pseudo-DCMs (pDCMs) were obtained from the split point clouds. We applied a multiresolution segmentation algorithm to the DCM and the four pDCMs to obtain crown objects. A partial least squares path model (PLS-PM) algorithm was applied to predict total vascular plant richness using object-based image analysis (OBIA) variables, derived from the delineated crown objects, and topographic variables, derived from a digital terrain model. Results showed that the object-based model was able to predict the total richness with an r2 of 0.64 and a root-mean-square error of four species. Topographic variables showed to be more important than the OBIA variables to predict richness. Furthermore, high-medium canopies (8-16 m) showed the biggest correlation with the total plant richness within the structural segments of the forest.

Land-use regime shift triggered the recent degradation of alpine pastures in Nyanpo Yutse of the eastern Qinghai-Tibetan Plateau

ContextThe eastern Qinghai-Tibetan Plateau is a cultural landscape where traditional pastoralism substantially shaped the present mosaic structure of the alpine grasslands. During the past two decades, however, severe grassland degradations of this region has been considered as the major ecological concern.ObjectivesIn this study we took an interdisciplinary approach to investigate the impact of the historical land-use regimes to the observed degradation, by conducting an in-depth case study in a local pastoral village in the Nyanpo Yutse region.MethodsFirstly, we mapped historical land-use intensities (LUIs) of the study area at land-use transition time points of 1970s, 1984, 1994 and 2015 with oral history and participatory GIS (PGIS) approaches. Secondly, we conducted Landsat and Moderate Resolution Imaging Spectroradiometer (MODIS) time series analysis to detect the temporal and spatial patterns of the degradation. Thirdly, we discussed the causal relations between the land-use and land-cover changes.ResultsLivestock and pasture privatization from 1984 to 1994 created the land-use regime shift which resulted in a marked increase in LUIs and decreased pastoral mobility. The LUI increase in this transition period fostered the establishment of short-grass vegetation which facilitated the spreading of plateau pikas. The installment of iron fences as private pasture borders from 2004 to 2007 eventually started the onset of degradation.ConclusionsOur case study illustrates that the past land-use regime shift triggered the recent ecological regime shift in Nyanpo Yutse. Severe grassland degradation occurred with a time lag during which cumulative LUIs surpassed the vulnerability threshold of the biophysical system.

ISS observations offer insights into plant function

In 2018 technologies on the International Space Station will provide ~1 year of synchronous observations of ecosystem composition, structure and function. We discuss these instruments and how they can be used to constrain global models and improve our understanding of the current state of terrestrial ecosystems.

Linking plant strategies and plant traits derived by radiative transfer modelling

Question Do spatial gradients of plant strategies correspond to patterns of plant traits obtained from a physically based model and hyperspectral imagery? It has been shown before that reflectance can be used to map plant strategies according to the established CSR scheme. So far, these approaches were based on empirical links and lacked transferability. Therefore, we test if physically-based derivations of plant traits may help in finding gradients in traits that are linked to strategies. Location A raised bog and minerotrophic fen complex, Murnauer Moos, Germany. Methods Spatial distributions of plant traits were modelled by adopting an inversion of the PROSAIL radiative transfer model on airborne hyperspectral imagery. The traits are derived from reflectance without making use of field data but only of known links between reflectance and traits. We tested whether previously found patterns in CSR plant strategies were related to the modelled traits. Results The results confirm close relationships between modelled plant traits and C, S and R strategies that were previously found in the field. The modelled plant traits explained different dimensions of the CSR-space. Leaf Area Index (LAI) and the reciprocal of Specific Leaf Area appeared to be good candidates for reproducing CSR scores as community traits using remote sensing. LAI has not been used in previous studies to allocate plant strategies. Conclusions Combining RTMs and the CSR model is a promising approach for establishing a robust link between airborne or spaceborne imagery and plant functioning. The demonstrated potential to map traits with close relation to CSR gradients using only our understanding of the relation between traits and reflectance is a step forward towards an operational use of the CSR model in remote sensing. This article is protected by copyright. All rights reserved.

Object-based extraction of bark beetle (Ips typographus L.) infestations using multi-date LANDSAT and SPOT satellite imagery

As major agents of biological disturbances, bark beetle infestations have been reported to account for a large portion of damage that occur in European forest stands. As a result, accurate spatiotemporal characterization of the vulnerable areas is crucial for subsequent post-infestation management. Remote sensing-assisted mapping of bark beetle-induced forest mortality has been an important research focus during the last decade. Due to the occurrence of mostly small- to medium-scale infestation patches in European stands, high-resolution optical data is commonly applied for mapping mortality. Despite this, we hypothesize the widely available satellite products to be potentially advantageous due to their multitemporal availability and reasonable costs. Here, we combined multi-date LANDSAT and SPOT scenes across an 11-year time span in which various epidemic and non-epidemic infestations occurred within the Bavarian Forest National Park in Germany. The aim was to map temporally adjacent mortality classes. The spectral, geometric and textural metrics extracted from the segmented imagery were applied to perform a full object-based classification, for which a digital terrain model was additionally employed. A number of potentially influential factors were also explored, including the spatial aggregation of image segments and the spatial enhancement of the multispectral imagery. The analysis resulted in a nearly perfect separation of non-infested and dead trees, while different levels of confusion were observed when classifying the transitional mortality classes. While the pan-sharpening of selected image scenes contributed to the stability of mapping results for non-infested and dead trees, no explicit trend was observed when aggregating small image segments prior to classification. Furthermore, combining the metrics from image objects and the digital terrain model suggested an obviously improved classification compared to the previously achieved pixel-based results across the same study site. In this paper, we thoroughly discuss the practical aspects of applying object-based image processing for monitoring bark beetle-induced forest mortality.