Land-use changes, climate change, and other stressors are causing drastic changes in the state and composition of vegetation. To understand these processes and anticipate future developments, there is an urgent need for: 1) documenting the current state of vegetation, 2) tracking its changes over long periods, and 3) obtaining near-real-time information on acute developments. Earth observation methods offer an efficient and comprehensive way to achieve this. One approach is to identify vegetation properties that are both robust and easily detectable through Earth observation data, while also serving as indicators for more difficult-to-measure biodiversity parameters. There is still a need for research, particularly in the use of artificial intelligence (AI) to develop a versatile and accurate detection approach. This project aims to use AI and the fusion of hyperspectral and high-resolution spatial imagery to develop innovative remote sensing products. These products are designed to describe various plant and vegetation characteristics typical of Central Europe, providing high indicative value. Compared to conventional remote sensing data, they will offer improved accuracy, interpretability, and high spatial resolution for ecological and conservation-focused vegetation mapping.

Open master's thesis topics:

1. EnMap scenes of Baden-Württemberg are analysed to determine whether the conservation status of FFH grassland (classification into status classes A, B and C) can be identified. For this purpose, maps of plant characteristics are derived and compared with observed conservation statuses from FFH mapping. In this way, features are to be deductively developed that allow the conservation status to be classified by remote sensing. The classification obtained by remote sensing is then compared with the observations in order to identify patterns in the deviations. Some of the differences between the observation and the model are likely to arise from observer bias in the observations, which can be detected in this way. What cannot be explained by this can serve as a basis for further refinement of the models. Supervision: M. Ewald, S. Schmidtlein, express interest

2. On the island of Sylt, the spread of the invasive species Rosa rugosa and Campylopus introflexus was investigated in earlier projects using airborne hyperspectral remote sensing. Now, the approach taken at that time is to be expanded by creating maps of plant traits, which are then ‘translated’ into maps of invasive species in the next step. On the one hand, this should generalise previous approaches to recording invasive species using a trait-based approach. On the other hand, it opens a window into the black box that the empirical models based on reflection and field data on species occurrence represent. Supervision: M. Ewald, S. Schmidtlein. express interest