Filtern
Erscheinungsjahr
Dokumenttyp
Volltext vorhanden
- ja (25) (entfernen)
Schlagworte
- Fernerkundung (25) (entfernen)
A sustainable development of forests and their ecosystem services requires the monitoring of the forests" state and changes as well as the prediction of their future development. To achieve the latter, eco-physiological forest growth models are usually applied. These models require calibration and validation with forestry reference data. This data includes forest structural parameters such as tree height or stem diameter which are easy to measure and can be used to estimate the core model parameters, i.e. the tree- biomass pools. The methods traditionally applied to derive the structural parameters are mainly manual and time-consuming. Hence, the in situ data acquisition is inefficient and limited in its ability to capture the vertical and horizontal variability in stand structure. Ground-based remote sensing bears the potential to overcome the limitations of the traditional methods. As they can be automated, ground-based remote sensing methods allow a much more efficient data acquisition and a larger spatial coverage. They are also able to capture forest structure in its three dimensions. Nevertheless, at present further research is required, in particular with respect to the practical integration of ground-based remote sensing data into forest growth models as well as regarding factors influencing the structural parameter retrieval from this data. Therefore, the goal of this PhD thesis was to investigate the influencing factors of two ground-based remote sensing methods (terrestrial laser scanning and hemispherical photography), which have not or only scarcely been studied to date. In addition, the use of forest structural parameters derived from these methods for the calibration of a forest growth model was assessed. Both goals were achieved. The results of this thesis could contribute significantly to a comprehensive assessment of ground-based remote sensing and its potential to derive the forest structural parameters. However, the use of these methods to calibrate forest growth models proved to be limited. An optimized data sampling design is expected to eliminate the major limitations, though. Furthermore, the combination of ground-based, airborne, and satellite remote sensing sensors was suggested to provide an optimized framework for the general integration of remotely sensed data into forest growth models. This combination of remote sensing observations at different scales will contribute greatly to a modern forest management with the purpose of warranting a sustainable forest development even under growing economic and ecological pressures.
Soil and water conservation are cross-sectional assignments. The respective objectives of the individual interest groups cause conflicts of use and lead to different assessments of the soil's potential. Necessary decisions and the practical implementation of soil and water conservation measures require the use of data. These data, which are both spatial and temporal, characterise past, present and, in the case of predictions, also future environmental conditions. The multitude of relevant data necessitates the use of geographic information systems as an instrument for successful resource management. With the use of problem-oriented case studies, it was possible to show that an improved understanding of the system is necessary for both optimisation of the site-specific resource management within the framework of Precision Farming and for the assessment of local to regional conflicts of use with regard to land usage and soil and water conservation. By changing the method, sufficient respective measures regarding documentation, prevention and risk assessment were able to be introduced and implemented. With the objective of practical implementation of a sustainable resource management, the possibilities of short- to long-term initiation of self-organised systems through the networking of available (geo-)information as well as the respective interest groups involved in the conflict of use formed the focal point of this investigation. The creation of networks linking agriculture, water extractors and nature conservation promotes necessary synergies and emergences, due to increased communication. Not the conveyance of knowledge alone, but rather new forms of understanding cause the interest groups involved to change their behaviour, thus facilitating efficient resource management for the interests of soil and water conservation.
Time series archives of remotely sensed data offer many possibilities to observe and analyse dynamic environmental processes at the Earth- surface. Based on these hypertemporal archives, which offer continuous observations of vegetation indices, typically at repetition rates from one to two weeks, sets of phenological parameters or metrics can be derived. Examples of such parameters are the beginning and end of the annual growing period, as well as its length. Even though these parameters do not correspond exactly to conventional observations of phenological events, they nevertheless provide indications of the dynamic processes occurring in the biosphere. The development of robust algorithms for the derivation of phenological metrics can be challenging. Currently, such algorithms are most commonly based on digital filters or the Fourier analysis of time series. Polynomial spline models offer a useful alternative to existing methods. The possibilities of using spline models in the analytical description of time series are numerous, and their specific mathematical properties may help to avoid known problems occurring with the more common methods for deriving phenological metrics. Based on a selection of different polynomial spline models suitable for the analysis of remotely sensed time series of vegetation indices, a method to derive various phenological parameters from such time series was developed and implemented in this work. Using an example data set from an intensively used agricultural area showing highly dynamic variations in vegetation phenology, the newly developed method was verified by a comparison of the results of the spline based approach to the results of two alternative, well established methods.
Die organische Bodensubstanz (OBS) ist eine fundamentale Steuergröße aller biogeochemischen Prozesse und steht in engem Zusammenhang zu Kohlenstoffkreisläufen und globalem Klima. Die derzeitige Herausforderung der Ökosystemforschung ist die Identifizierung der für die Bodenqualität relevanten Bioindikatoren und deren Erfassung mit Methoden, die eine nachhaltige Nutzung der OBS in großem Maßstab überwachen und damit zu globalen Erderkundungsprogrammen beitragen können. Die fernerkundliche Technik der Vis-NIR Spektroskopie ist eine bewährte Methode für die Beurteilung und das Monitoring von Böden, wobei ihr Potential bezüglich der Erfassung biologischer und mikrobieller Bodenparameter bisher umstritten ist. Das Ziel der vorgestellten Arbeit war die quantitative und qualitative Untersuchung der OBS von Ackeroberböden mit unterschiedlichen Methoden und variierender raumzeitlicher Auflösung sowie die anschließende Bewertung des Potentials non-invasiver, spektroskopischer Methoden zur Erfassung ausgewählter Parameter dieser OBS. Dafür wurde zunächst eine umfassende lokale Datenbank aus chemischen, physikalischen und biologischen Bodenparametern und dazugehörigen Bodenspektren einer sehr heterogenen geologischen Region mit gemäßigten Klima im Südwesten Deutschlands erstellt. Auf dieser Grundlage wurde dann das Potential der Bodenspektroskopie zur Erfassung und Schätzung von Feld- und Geländedaten ausgewählter OBS Parameter untersucht. Zusätzlich wurde das Optimierungspotential der Vorhersagemodelle durch statistische Vorverarbeitung der spektralen Daten getestet. Die Güte der Vorhersagewahrscheinlichkeit gebräuchlicher fernerkundlicher Bodenparameter (OC, N) konnte für im Labor erhobene Hyperspektralmessungen durch statistische Optimierungstechniken wie Variablenselektion und Wavelet-Transformation verbessert werden. Ein zusätzliches Datenset mit mikrobiellen/labilen OBS Parametern und Felddaten wurde untersucht um zu beurteilen, ob Bodenspektren zur Vorhersage genutzt werden können. Hierzu wurden mikrobieller Kohlenstoff (MBC), gelöster organischer Kohlenstoff (DOC), heißwasserlöslicher Kohlenstoff (HWEC), Chlorophyll α (Chl α) und Phospholipid-Fettsäuren (PLFAs) herangezogen. Für MBC und DOC konnte abhängig von Tiefe und Jahreszeit eine mittlere Güte der Vorhersagewahrscheinlichkeit erreicht werden, wobei zwischen hohen und niedrigen Konzentration unterschieden werden konnte. Vorhersagen für OC und PLFAs (Gesamt-PLFA-Gehalt sowie die mikrobiellen Gruppen der Bakterien, Pilze und Algen) waren nicht möglich. Die beste Prognosewahrscheinlichkeit konnte für das Chlorophyll der Grünalgen an der Bodenoberfläche (0-1cm Bodentiefe) erzielt werden, welches durch Korrelation mit MBC vermutlich auch für dessen gute Vorhersagewahrscheinlichkeit verantwortlich war. Schätzungen des Gesamtgehaltes der OBS, abgeleitet durch OC, waren hingegen nicht möglich, was der hohen Dynamik der mikrobiellen OBS Parameter an der Bodenoberfläche zuzuschreiben ist. Das schränkt die Repräsentativität der spektralen Messung der Bodenoberfläche zeitlich ein. Die statistische Optimierungstechnik der Variablenselektion konnte für die Felddaten nur zu einer geringen Verbesserung der Vorhersagemodelle führen. Die Untersuchung zur Herkunft der organischen Bestandteile und ihrer Auswirkungen auf die Quantität und Qualität der OBS konnte die mikrobielle Nekromasse und die Gruppe der Bodenalgen als zwei mögliche weitere signifikante Quellen für die Entstehung und Beständigkeit der OBS identifizieren. Insgesamt wird der mikrobielle Beitrag zur OBS höher als gemeinhin angenommen eingestuft. Der Einfluss mikrobieller Bestandteile konnte für die OBS Menge, speziell in der mineralassoziierten Fraktion der OBS in Ackeroberböden, sowie für die OBS Qualität hinsichtlich der Korrelation von mikrobiellen Kohlenhydraten und OBS Stabilität gezeigt werden. Die genaue Quantifizierung dieser OBS Parameter und ihre Bedeutung für die OBS Dynamik sowie ihre Prognostizierbarkeit mittels spektroskopischer Methoden ist noch nicht vollständig geklärt. Für eine abschließende Beurteilung sind deshalb weitere Studien notwendig.
Although gravitropism forces trees to grow vertically, stems have shown to prefer specific orientations. Apart from wind deforming the tree shape, lateral light can result in prevailing inclination directions. In recent years a species dependent interaction between gravitropism and phototropism, resulting in trunks leaning down-slope, has been confirmed, but a terrestrial investigation of such factors is limited to small scale surveys. ALS offers the opportunity to investigate trees remotely. This study shall clarify whether ALS detected tree trunks can be used to identify prevailing trunk inclinations. In particular, the effect of topography, wind, soil properties and scan direction are investigated empirically using linear regression models. 299.000 significantly inclined stems were investigated. Species-specific prevailing trunk orientations could be observed. About 58% of the inclination and 19% of the orientation could be explained by the linear models, while the tree species, tree height, aspect and slope could be identified as significant factors. The models indicate that deciduous trees tend to lean down-slope, while conifers tend to lean leeward. This study has shown that ALS is suitable to investigate the trunk orientation on larger scales. It provides empirical evidence for the effect of phototropism and wind on the trunk orientation.
Determining the exact position of a forest inventory plot—and hence the position of the sampled trees—is often hampered by a poor Global Navigation Satellite System (GNSS) signal quality beneath the forest canopy. Inaccurate geo-references hamper the performance of models that aim to retrieve useful information from spatially high remote sensing data (e.g., species classification or timber volume estimation). This restriction is even more severe on the level of individual trees. The objective of this study was to develop a post-processing strategy to improve the positional accuracy of GNSS-measured sample-plot centers and to develop a method to automatically match trees within a terrestrial sample plot to aerial detected trees. We propose a new method which uses a random forest classifier to estimate the matching probability of each terrestrial-reference and aerial detected tree pair, which gives the opportunity to assess the reliability of the results. We investigated 133 sample plots of the Third German National Forest Inventory (BWI, 2011"2012) within the German federal state of Rhineland-Palatinate. For training and objective validation, synthetic forest stands have been modeled using the Waldplaner 2.0 software. Our method has achieved an overall accuracy of 82.7% for co-registration and 89.1% for tree matching. With our method, 60% of the investigated plots could be successfully relocated. The probabilities provided by the algorithm are an objective indicator of the reliability of a specific result which could be incorporated into quantitative models to increase the performance of forest attribute estimations.
For grape canopy pixels captured by an unmanned aerial vehicle (UAV) tilt-mounted RedEdge-M multispectral sensor in a sloped vineyard, an in situ Walthall model can be established with purely image-based methods. This was derived from RedEdge-M directional reflectance and a vineyard 3D surface model generated from the same imagery. The model was used to correct the angular effects in the reflectance images to form normalized difference vegetation index (NDVI)orthomosaics of different view angles. The results showed that the effect could be corrected to a certain scope, but not completely. There are three drawbacks that might restrict a successful angular model construction and correction: (1) the observable micro shadow variation on the canopy enabled by the high resolution; (2) the complexity of vine canopies that causes an inconsistency between reflectance and canopy geometry, including effects such as micro shadows and near-infrared (NIR) additive effects; and (3) the resolution limit of a 3D model to represent the accurate real-world optical geometry. The conclusion is that grape canopies might be too inhomogeneous for the tested method to perform the angular correction in high quality.
With the ongoing trend towards deep learning in the remote sensing community, classical pixel based algorithms are often outperformed by convolution based image segmentation algorithms. This performance was mostly validated spatially, by splitting training and validation pixels for a given year. Though generalizing models temporally is potentially more difficult, it has been a recent trend to transfer models from one year to another, and therefore to validate temporally. The study argues that it is always important to check both, in order to generate models that are useful beyond the scope of the training data. It shows that convolutional neural networks have potential to generalize better than pixel based models, since they do not rely on phenological development alone, but can also consider object geometry and texture. The UNET classifier was able to achieve the highest F1 scores, averaging 0.61 in temporal validation samples, and 0.77 in spatial validation samples. The theoretical potential for overfitting geometry and just memorizing the shape of fields that are maize has been shown to be insignificant in practical applications. In conclusion, kernel based convolutions can offer a large contribution in making agricultural classification models more transferable, both to other regions and to other years.
In order to discuss potential sustainability issues of expanding silage maize cultivation in Rhineland-Palatinate, spatially explicit monitoring is necessary. Publicly available statistical records are often not a sufficient basis for extensive research, especially on soil health, where risk factors like erosion and compaction depend on variables that are specific to every site, and hard to generalize for larger administrative aggregates. The focus of this study is to apply established classification algorithms to estimate maize abundance for each independent pixel, while at the same time accounting for their spatial relationship. Therefore, two ways to incorporate spatial autocorrelation of neighboring pixels are combined with three different classification models. The performance of each of these modeling approaches is analyzed and discussed. Finally, one prediction approach is applied to the imagery, and the overall predicted acreage is compared to publicly available data. We were able to show that Support Vector Machine (SVM) classification and Random Forests (RF) were able to distinguish maize pixels reliably, with kappa values well above 0.9 in most cases. The Generalized Linear Model (GLM) performed substantially worse. Furthermore, Regression Kriging (RK) as an approach to integrate spatial autocorrelation into the prediction model is not suitable in use cases with millions of sparsely clustered training pixels. Gaussian Blur is able to improve predictions slightly in these cases, but it is possible that this is only because it smoothes out impurities of the reference data. The overall prediction with RF classification combined with Gaussian Blur performed well, with out of bag error rates of 0.5% in 2009 and 1.3% in 2016. Despite the low error rates, there is a discrepancy between the predicted acreage and the official records, which is 20% in 2009 and 27% in 2016.
Abstract: Thermal infrared (TIR) multi-/hyperspectral and sun-induced fluorescence (SIF) approaches together with classic solar-reflective (visible, near-, and shortwave infrared reflectance (VNIR)/SWIR) hyperspectral remote sensing form the latest state-of-the-art techniques for the detection of crop water stress. Each of these three domains requires dedicated sensor technology currently in place for ground and airborne applications and either have satellite concepts under development (e.g., HySPIRI/SBG (Surface Biology and Geology), Sentinel-8, HiTeSEM in the TIR) or are subject to satellite missions recently launched or scheduled within the next years (i.e., EnMAP and PRISMA (PRecursore IperSpettrale della Missione Applicativa, launched on March 2019) in the VNIR/SWIR, Fluorescence Explorer (FLEX) in the SIF). Identification of plant water stress or drought is of utmost importance to guarantee global water and food supply. Therefore, knowledge of crop water status over large farmland areas bears large potential for optimizing agricultural water use. As plant responses to water stress are numerous and complex, their physiological consequences affect the electromagnetic signal in different spectral domains. This review paper summarizes the importance of water stress-related applications and the plant responses to water stress, followed by a concise review of water-stress detection through remote sensing, focusing on TIR without neglecting the comparison to other spectral domains (i.e., VNIR/SWIR and SIF) and multi-sensor approaches. Current and planned sensors at ground, airborne, and satellite level for the TIR as well as a selection of commonly used indices and approaches for water-stress detection using the main multi-/hyperspectral remote sensing imaging techniques are reviewed. Several important challenges are discussed that occur when using spectral emissivity, temperature-based indices, and physically-based approaches for water-stress detection in the TIR spectral domain. Furthermore, challenges with data processing and the perspectives for future satellite missions in the TIR are critically examined. In conclusion, information from multi-/hyperspectral TIR together with those from VNIR/SWIR and SIF sensors within a multi-sensor approach can provide profound insights to actual plant (water) status and the rationale of physiological and biochemical changes. Synergistic sensor use will open new avenues for scientists to study plant functioning and the response to environmental stress in a wide range of ecosystems.