Sprecher
Beschreibung
The Baltic Sea is under pressure. Eutrophication and global warming are threatening the ecosystem. Among other things, coastal hypoxia continues to increase despite many efforts to reduce nutrient inputs. New, dynamic evaluation tools are needed to decipher the complex drivers behind these problems and take more effective measures. Applying machine learning to complex environmental data could be a key to better understanding the drivers. However, this method is not trivial due to temporal and spatial irregularities in the data. Based on a reanalysis, a probalistic neural network is trained and results are applied to data to bridge irregularities. This approach will enable accurate and timely assessments of the state of the environment in the future. We here show our current approach and status of the INSYST project.
