Abstract
Worldwide, biodiversity is threatened by drivers like habitat loss, over-exploitation, and pollution. As a consequence, large numbers of plant and animal species are endangered or even face extinction. Due to the manifold ecosystem services provided by biodiversity, including food and resource provision, the consequences of declining biodiversity are substantial. Conservation actions like protecting areas aim to foster biodiversity and stop the decline. For the detection of biodiversity loss hotspots and the assessment of conservation actions, reliable and frequent monitoring of biodiversity is needed. Recently, optical remote sensing has emerged as a valuable tool for monitoring.
Optical remote sensing is based on the measurement of sunlight, reflected on Earth's surface, using special instruments - so-called imaging spectrometers. These instruments measure the reflected light in many wavelengths, reaching from the visible to the short wavelength-infrared region. The amount of light reflected per wavelength, measured in radiance units, is determined by the surface reflectance, a distinctive material property driven by its structural and chemical composition. Thus, materials can be identified, characterized, and quantitatively described by analyzing their reflectance. In order to retrieve the surface reflectance, the measured radiance is brought into relation to concurrent irradiance, which has to be estimated.
However, atmospheric constituents and surface topography induce absorption and scattering processes, modifying radiance and irradiance. Therefore, to obtain reliable surface reflectance data, several processing steps are applied to compensate for these modifications and to reveal the actual target surface reflectance. The processing steps include an atmospheric correction to compensate for atmospheric absorption and scattering processes, a topographic correction to compensate for terrain-induced irradiance variations, and an anisotropy correction to compensate for the directionally uneven scattering of incoming radiation by many natural surfaces.
Much knowledge about these effects in the atmosphere and on the surface, but also many available correction approaches, are based on observations in coarse spatial resolution satellite data with pixel sizes of several tens of meters. However, airborne and newer satellite systems dispose of a much higher spatial resolution with pixel sizes of a few meters down to the sub-meter region. With the higher spatial resolution, observed surfaces are much more heterogeneous. This larger heterogeneity increases the complexity of the influence of atmosphere, topography, and anisotropy on reflectance.
At its core, this Ph.D. thesis systematically investigates how atmosphere, topography, and anisotropy influence state-of-the-art high-resolution imaging spectrometer data and derived downstream products applicable for biodiversity monitoring and how currently available correction methods can compensate for these influences. In the first part, topographic correction methods, initially developed for coarse-resolution space-based data but also used for high-resolution airborne data, are investigated. By simultaneously correcting airborne and space-based data of the same study site with the same correction methods, the role of topographic effects in both datasets and the transferability from space-based to airborne data is examined. In the second part, the relative influence of atmospheric, topographic, and anisotropy effects on reflectance is studied systematically. For this purpose, the accordance of reflectance in the overlapping area of neighboring airborne flight lines is examined after each applied correction step. In the third part, finally, the role of such effects and their correction, as well as the role of the acquisition geometry, is investigated on the estimation of functional richness, a widely used biodiversity metric. To achieve this, functional richness estimates of a study site, which was observed in a short timeframe in three flight lines with varying acquisition geometries, are derived with and without corrections for atmosphere, topography, and anisotropy.
The obtained results show that retrieved surface reflectance and derived downstream products show significant variations in values due to atmospheric, topographic, and anisotropy effects. These observations emphasize that a comprehensive correction is essential to get reliable data for biodiversity monitoring from remote sensing. The results also show that currently available correction approaches, although often developed for coarse-resolution data, can effectively reduce the influence of atmosphere, topography, and anisotropy on retrieved reflectance and derived products in high-resolution data. However, residual effects remain, which could necessitate new correction approaches specifically designed for the increasing complexity of data.
The thesis concludes with an outlook, pointing to the potential of recent and upcoming imaging spectrometers on space-based, airborne, and drone-based platforms. These advancements promise a more accurate and comprehensive data collection. Furthermore, new correction approaches, specifically designed for the increasing complexity of high-resolution data, are introduced. These developments could enable more effective and reliable biodiversity monitoring in the future.
Vögtli, Marius