Wetland ecosystems on both regional and global scales are valued for their irreplaceable hydrological, biological, and ecological functions. However, wetland area has been diminishing in recent decades due to human activities and ongoing climate change. In this context, conservation of wetlands ties into goals 13 (Climate action) and 15 (Life on land) of the UN Sustainable Development Goals. On the global scale, climatically significant effects of wetlands include methane emissions and a role in the global carbon cycle. On regional or local scales, wetlands represent dynamic landscapes with annual and interannual changes of both water level and vegetation cover, which promotes biodiversity in the area. For these reasons, monitoring wetlands and their various characteristics (e.g., soil moisture, water level, land cover and vegetation structure, biomass estimation) is important. While being reliable, in situ observations in wetlands are time consuming and hard to accomplish due to the extent, accessibility, and complexity of studied areas. In this regard, remote sensing has proved to be a suitable technology for wetland monitoring, especially in the last decade, with modern systems enabling frequent observations in different modalities, spatial scales, and spectral domains.
To address the challenge of wetland mapping/monitoring, this Ph.D. project focuses on mapping wetlands and wetland change by remote sensing. Emphasis is given to synthetic aperture radar (SAR), which offers unique potential for wetland monitoring by acquiring information on roughness and moisture content of the observed surface. As such, this Ph.D. project is defined with three objectives: The first objective is to investigate the response of a SAR signal and derived features to selected wetland characteristics and their changes in connection with different sensor parameters (wavelength, polarization, look angle). The second objective is to explore how the mapping/derivation of desired characteristics can be improved by fusion with high resolution optical (e.g., Sentinel-2, Planet) or LiDAR (e.g., GEDI) data. The third objective is to develop a methodology enabling processing of time series of remote sensing data suited for monitoring of selected wetland characteristics utilizing suitable machine learning methods, such as recurrent neural networks.
The proposed Ph.D. project requires solid knowledge of remote sensing (MSc. level), experience with SAR data processing, programming skills (Python). Experience with machine & deep learning techniques is of advantage.Deadline is closed