Abstract's details
Predicting short and long-term sea level changes using Deep learning
CoAuthors
Event: 2022 Ocean Surface Topography Science Team Meeting
Session: Science I: Climate data records for understanding the causes of global and regional sea level variability and change
Presentation type: Type Poster
Contribution: not provided
Abstract:
Predicting short- and long-term sea-level changes is a critical task with deep implications for both the safety and job-security of a large part of the world's population.
The satellite altimetry data record is now nearly 30 years and we may begin to consider employing it in a deep learning (DL)---and, by definition, data-hungry---context, a somewhat unexplored territory until now.
Because DL is capable of capturing non-linear processes, it seems well-suited for climate- and weather-influenced data, although the requirement of large (or, rather, diverse) datasets has hampered its use in altimetry settings.
Furthermore, explainability of DL models has been an issue, as has the computing requirements in the past, and most machine learning models do not output uncertainties in their predictions.
Global Mean Sea Level (GMSL) largely changes linearly with time (3 mm/year) but this global average exhibits large geographical variations and covers a suite of regional non-linear signals changing in both space and time and improving the mapping and understanding of these regional signals will enhance our ability to project sea level changes into the future.
Today, though, datasets have approached a suitable size, model explainability is solved by permutation importance and SHAP values, computing is cheap enough and through several methods we are able to include information on uncertainties as well, handled by either appropriate loss functions, meta-learners or Bayesian methods.
Thus the time has come to employ 30 years of satellite altimetry data to improve our predictive power in sea-level changes. This project focuses on the above problems in both global and regional settings.
The satellite altimetry data record is now nearly 30 years and we may begin to consider employing it in a deep learning (DL)---and, by definition, data-hungry---context, a somewhat unexplored territory until now.
Because DL is capable of capturing non-linear processes, it seems well-suited for climate- and weather-influenced data, although the requirement of large (or, rather, diverse) datasets has hampered its use in altimetry settings.
Furthermore, explainability of DL models has been an issue, as has the computing requirements in the past, and most machine learning models do not output uncertainties in their predictions.
Global Mean Sea Level (GMSL) largely changes linearly with time (3 mm/year) but this global average exhibits large geographical variations and covers a suite of regional non-linear signals changing in both space and time and improving the mapping and understanding of these regional signals will enhance our ability to project sea level changes into the future.
Today, though, datasets have approached a suitable size, model explainability is solved by permutation importance and SHAP values, computing is cheap enough and through several methods we are able to include information on uncertainties as well, handled by either appropriate loss functions, meta-learners or Bayesian methods.
Thus the time has come to employ 30 years of satellite altimetry data to improve our predictive power in sea-level changes. This project focuses on the above problems in both global and regional settings.