Artificial Intelligence (AI) can significantly improve our understanding of the climate and Earth science, says a study by German scientists. AI can be applied to data related to extreme events such as fire spreads or hurricanes, which are very complex processes influenced by local conditions.
It can also be applied to atmospheric and ocean transport, soil movement and vegetation dynamics data -- some of the classic topics of Earth system science. "From a plethora of sensors, a deluge of Earth system data has become available, but so far we've been lagging behind in analysis and interpretation," said Markus Reichstein of the Max Planck Institute for Biogeochemistry in Jena, Germany.
"This is where deep learning techniques become a promising tool, beyond classical machine learning applications such as image recognition, natural language processing or AlphaGo," added co-author Joachim Denzler, from the Friedrich Schiller University in Jena (FSU).
However, deep learning approaches are difficult. All data-driven and statistical approaches do not guarantee physical consistency per se, are highly dependent on data quality and may experience difficulties with extrapolations, according to the study published in the journal Nature. Besides, the requirement for data processing and storage capacity is very high.
If both techniques are brought together, so-called hybrid models are created. They can, for example, be used for modelling the motion of ocean water to predict sea surface temperature. While the temperatures are modelled physically, the ocean water movement is represented by a machine learning approach.
"The idea is to combine the best of two worlds, the consistency of physical models with the versatility of machine learning, to obtain greatly improved models," Reichstein explained.
It can also be applied to atmospheric and ocean transport, soil movement and vegetation dynamics data -- some of the classic topics of Earth system science. "From a plethora of sensors, a deluge of Earth system data has become available, but so far we've been lagging behind in analysis and interpretation," said Markus Reichstein of the Max Planck Institute for Biogeochemistry in Jena, Germany.
"This is where deep learning techniques become a promising tool, beyond classical machine learning applications such as image recognition, natural language processing or AlphaGo," added co-author Joachim Denzler, from the Friedrich Schiller University in Jena (FSU).
However, deep learning approaches are difficult. All data-driven and statistical approaches do not guarantee physical consistency per se, are highly dependent on data quality and may experience difficulties with extrapolations, according to the study published in the journal Nature. Besides, the requirement for data processing and storage capacity is very high.
If both techniques are brought together, so-called hybrid models are created. They can, for example, be used for modelling the motion of ocean water to predict sea surface temperature. While the temperatures are modelled physically, the ocean water movement is represented by a machine learning approach.
"The idea is to combine the best of two worlds, the consistency of physical models with the versatility of machine learning, to obtain greatly improved models," Reichstein explained.
No comments:
Post a Comment