Machine Learning in Remote Sensing and Climate Research
Prof. Dr. Wouter Dorigo is head of the research group Climate and Environmental Remote Sensing at TU Wien GEO. His main research interest is remote sensing of soil moisture and vegetation. Within these themes he focuses on several aspects, including biogeophysical variable retrieval through physical and semi-empirical methods, calibration and validation, data merging, time series analysis, and using EO data and machine learning for climate and Earth system studies. The products generated within these projects have been adopted by thousands of users of the international water and climate community. Wouter has (co-)authored more than 100 peer-reviewed journal publications, including in high impact journals (e.g. Nature). According to Google Scholar his work has been cited >10000 times and has an h-index of 47. In 2015 Wouter Dorigo received TU Wien Science Award. In 2018 and 2019 he was ranked among world’s Highly Cited Researchers 2019 published by Clarivate Analytics.
The Earth’s climate is a highly complex system with innumerable processes, interactions, and feedbacks. Disentangling the impact of individual processes, e.g. the effect of increasing drought severity on vegetation development, may therefore be extremely challenging. The Climate and Environmental Remote Sensing group at TU Wien focuses on the production and use of satellite Earth observation data to study land surface processes. In our presentation, we will give an overview how such data are used in combination with machine learning methods to better understand the Earth and climate system, for example by:
- Enhancing Earth observation data (gap filling, downscaling, noise reduction, …).
- Disentangling multiple climate drivers (rainfall, temperature, …) of vegetation growth.
- Predicting agricultural yield from Earth observations.
- Predicting wildfire danger.
