National Technical University of Athens (NTUA)
Τεχνικές τηλεπισκόπησης για την ανάδειξη της δυναμικής και των χαρακτηριστικών της μεσογειακής βλάστησης
Abstract
dc:descriptionThis thesis intends to highlight the characteristics and dynamics of the Mediterranean vegetation using Remote Sensing methods and techniques. Vegetation characteristics are inextricably linked to the pigments and biochemical properties of the vegetation. These parameters cannot be precisely captured even by the most modern satellite remote sensing instruments, as these have certain limitations in terms of spatial and spectral resolution. Hyperspectral sensors mounted on aircraft or other monitoring platforms is a source of high-resolution data with a good balance of the above discriminative capabilities. Remote Sensing instruments present errors and noise which are more noticeable for images with very high spatial and spectral resolution such as those of airborne hyperspectral sensors, with a consequent degradation of the quality of the images and a relative impact on the accuracy of the image processing and the analysis results. Vegetation dynamics are inextricably linked to factors such as climate, abiotic environment and biotic interactions. Understanding how these factors interrelate and influence species coexistence and productivity is an important subject of contemporary research. Atmosphere, soil, water and vegetation are key, dynamic abiotic and biotic factors whose evolution over time can now be recorded with modern satellite remote sensing instruments.The purpose of this thesis is to portray the Mediterranean vegetation, its characteristics and dynamics, through state-of-the-art Remote Sensing techniques, such as:1. the detection, quantification and restoration of the spectral curvature effect (spectral smile), a common type of radiometric distortion encountered in remote sensing data;2. the creation of a field-measured spectral library for the most common types of Mediterranean forest vegetation, that is compatible with all known spectral libraries of materials, as well as its integration in a graphical interface that facilitates its use.3. the use of appropriate vegetation indicators and pigment indicators (chlorophyll, xanthophyll, carotenoids, anthocyanins and other nitrogen-rich compounds) from spectral regions (channels) which reveal characteristic biophysical properties through the reflectance variations of the various vegetation species,4. the implementation of spectral unmixing, through techniques for reducing the dimension of hyperspectral data, exploiting the information of the spectral absorption zones of the vegetation and investigating the effects of band selection on endmember extraction. 5. the estimation of the abundance of materials (the last step of the linear spectral unmixing) with Bayesian inference and their penalization, for dealing with multicollinearity and producing biodiversity measurement maps6. the analysis of time values evolution, variability and autoregressions of Land Surface Temperature (LST) and Normalized Difference Vegetation Index (NDVI), in order to study how vegetation is affected by climate change and to identify potential disturbances (e.g. phenological changes) due to the occurrence of changes in their decomposition components, which can be interpreted by the effects of biotic and abiotic factors, in the species of Mediterranean vegetation.
Degree
thesis:*- Grantor dc:publisher
- National Technical University of Athens (NTUA)
- Year dc:date
- 2022
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Andronis, Vasileios
- Ανδρώνης, Βασίλειος
Subjects
dc:subject × 18- Φασματικό χαμόγελο
- Μπεϋζιανός Φασματικός Διαχωρισμός
- Ποινικοποίηση Elastic Net
- Χρονοσειρές
- Κλιματική αλλαγή
- Υπερφασματική τηλεπισκόπηση
- Spectral smile
- Bayesian spectral unmixing
- Elastic Net penalization
- Time series
- Climate change
- Hyperspactral Remote Sensing
- Περιβαλλοντική Μηχανική
- Επιστήμες Μηχανικού και Τεχνολογία
- Τηλεπισκόπηση
- Environmental Engineering
- Engineering and Technology
- Remote Sensing
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/52922