Timezone: »
Variational data assimilation optimizes for an initial state of a dynamical system such that its evolution fits observational data. The physical model can subsequently be evolved into the future to make predictions. This principle is a cornerstone of large scale forecasting applications such as numerical weather prediction. As such, it is implemented in current operational systems of weather forecasting agencies across the globe. However, finding a good initial state poses a difficult optimization problem in part due to the non-invertible relationship between physical states and their corresponding observations. We learn a mapping from observational data to physical states and show how it can be used to improve optimizability. We employ this mapping in two ways: to better initialize the non-convex optimization problem, and to reformulate the objective function in better behaved physics space instead of observation space. Our experimental results for the Lorenz96 model and a two-dimensional turbulent fluid flow demonstrate that this procedure significantly improves forecast quality for chaotic systems.
Author Information
Thomas Frerix (Technical University of Munich)
Dmitrii Kochkov (Google)
Jamie Smith (Google)
Daniel Cremers (TU Munich)
Michael Brenner (Google/Harvard)
Stephan Hoyer (Google Research)
Related Events (a corresponding poster, oral, or spotlight)
-
2021 Spotlight: Variational Data Assimilation with a Learned Inverse Observation Operator »
Tue. Jul 20th 02:25 -- 02:30 PM Room
More from the Same Authors
-
2021 : Using a Cross-Task Grid of Linear Probes to Interpret CNN Model Predictions On Retinal Images »
Katy Blumer · Subhashini Venugopalan · Michael Brenner · Jon Kleinberg -
2023 Poster: Beyond In-Domain Scenarios: Robust Density-Aware Calibration »
Christian Tomani · Futa Waseda · Yuesong Shen · Daniel Cremers -
2023 Poster: Learning Expressive Priors for Generalization and Uncertainty Estimation in Neural Networks »
Dominik Schnaus · Jongseok Lee · Daniel Cremers · Rudolph Triebel -
2020 : Panel Discussion 1 »
Daniel Cremers · Nemanja Djuric · Ingmar Posner · Dariu Gavrila -
2020 : Q&A: Daniel Cremers »
Daniel Cremers -
2020 : Invited Talk: Deep Direct Visual SLAM (Daniel Cremers) »
Daniel Cremers -
2019 Poster: Approximating Orthogonal Matrices with Effective Givens Factorization »
Thomas Frerix · Joan Bruna -
2019 Oral: Approximating Orthogonal Matrices with Effective Givens Factorization »
Thomas Frerix · Joan Bruna -
2019 Poster: Flat Metric Minimization with Applications in Generative Modeling »
Thomas Möllenhoff · Daniel Cremers -
2019 Oral: Flat Metric Minimization with Applications in Generative Modeling »
Thomas Möllenhoff · Daniel Cremers -
2018 Poster: Learning Memory Access Patterns »
Milad Hashemi · Kevin Swersky · Jamie Smith · Grant Ayers · Heiner Litz · Jichuan Chang · Christos Kozyrakis · Parthasarathy Ranganathan -
2018 Oral: Learning Memory Access Patterns »
Milad Hashemi · Kevin Swersky · Jamie Smith · Grant Ayers · Heiner Litz · Jichuan Chang · Christos Kozyrakis · Parthasarathy Ranganathan