Data assimilation

Last updated Sep 19, 2022 Edit Source

• Data assimilation or machine learning?

• #COURSE ECMWF An introduction to data assimilation
• #COURSE ECMWF Training course: Data assimilation

• #TALK Ensemble Data Assimilation
• #TALK Data Learning: Integrating Data Assimilation and Machine Learning (CMCC)

• #PAPER A review of operational methods of variational and ensemble‐variational data assimilation (Bannister 2016)
• #PAPER Learning earth system models from observations: machine learning or data assimilation? (Geer 2020)
• #PAPER Towards an unbiased stratospheric analysis (Laloyaux 2020)
  • Weak-constraint 4D-Var
• #PAPER Attention-based Convolutional Autoencoders for 3D-Variational Data Assimilation (Mack 2020)
  • Proposed a new ‘Bi-Reduced Space’ approach to solving 3D Variational Data Assimilation.
  • Used Convolutional Autoencoders to create the reduced space for solving 3D Var
  • Proved that our approach has the same solution as previous methods for reducing 3D Var space
  • Lower computational complexity of previous methods
  • Tested the new method with data from a real-world application: a pollution model in London
• #PAPER Deep Data Assimilation: Integrating Deep Learning with Data Assimilation (Arcucci 2021)
• #PAPER Data Learning: Integrating Data Assimilation and Machine Learning (Buizza 2022)
• #PAPER Deep Learning Augmented Data Assimilation: Reconstructing Missing Information with Convolutional Autoencoders (Wang 2022)

• #POSTER A Neural Network-Based Observation Operator for Coupled Ocean-Acoustic Variational Data Assimilation
• #POSTER DAN - An optimal Data Assimilation framework based on machine learning recurrent Networks
• #POSTER Toward an integrated NWP-DA-AI system for precipitation prediction
  • Phased-Array Weather Radar (PAWR) scans the whole sky in the 60-km range every 30 seconds at 110 elevation angles
  • 3D extension of the Convolutional Long Short-Term Memory (Conv-LSTM; Shi et al., 2015) is applied to PAWR nowcasting
  • In addition to the Conv-LSTM with past observations, we also develop a Conv-LSTM that accepts forecast data
• #POSTER Using machine learning to correct model error and application to data assimilation with a quasi-geostrophic model
• #TALK Data Assimilation and Machine Learning Science at ECMWF (Bonavita)
• #TALK Artificial Neural Network at the service of Data Assimilation (and vice versa) (Arcucci)
  • See “Attention-based Convolutional Autoencoders for 3D-Variational Data Assimilation” paper
  • #POSTER ICLR - Urban air pollution forecasts generated from latent space representation
  • Combination of simulation (process driven) data and observations (data-driven)
  • The Ensemble Kalman filter, at least 1000 members to better capture underlying PDF
  • DA methods assume linearity, ML could capture non-linearities (LSTMs)
  • Computational fluid dynamics software could be replaced by ML
• #PAPER Combining data assimilation and machine learning to estimate parameters of a convective-scale model (Legler 2021)