MEDIDA for QG

[project website]

Table of contents

• Introduction
• Requirements
• Experiments
  • Case 1
• Citation
• References

Introduction

Key points

• Model error discovery with interpretability and data assimilation (MEDIDA)[1]* is scaled up to geostrophic turbulence and sparse observations
• Naive use of neural nets (NNs) as interpolator does not capture small scales due to spectral bias, failing discoveries of closed-form errors
• Reducing this bias using random Fourier features enables NNs to represent the full range of scales, leading to successful error discoveries

Requirements

• python 3.6
  • scipy
  • numpy
• Pytroch
• RFF in Pytroch

Experiments

Case 1

Case 1 is disscused here Case 1 Location

Python code

will be updated

Citation

• Mojgani, R., Chattopadhyay, A., and Hassanzadeh, P. , Interpretable structural model error discovery from sparse assimilation increments using spectral bias-reduced neural networks: A quasi-geostrophic turbulence test case, (2023).(url)
  BibTeX

  @misc{mojgani2023interpretable,
  title={Interpretable structural model error discovery from sparse assimilation increments using spectral bias-reduced neural networks: {A} quasi-geostrophic turbulence test case},
  author={Rambod Mojgani and Ashesh Chattopadhyay and Pedram Hassanzadeh},
  year={2023},
  eprint={2309.13211},
  archivePrefix={arXiv},
  primaryClass={physics.comp-ph}
  }

References

• [1] Mojgani, R., Chattopadhyay, A., and Hassanzadeh, P. , Closed-form discovery of structural errors in models of chaotic systems by integrating Bayesian sparse regression and data assimilation., Chaos 32, 061105 (2022) arXiv:2110.00546. (Download)
  BibTeX

  @article{Mojgani_Chaos_2022,
  author = {Mojgani,Rambod  and Chattopadhyay,Ashesh  and Hassanzadeh,Pedram },
  title = {Discovery of interpretable structural model errors by combining {B}ayesian sparse regression and data assimilation: {A} chaotic {K}uramoto–{S}ivashinsky test case},
  journal = {Chaos: {A}n Interdisciplinary Journal of Nonlinear Science},
  volume = {32},
  number = {6},
  pages = {061105},
  year = {2022},
  doi = {10.1063/5.0091282},
  URL = {https://doi.org/10.1063/5.0091282},
  eprint = {arXiv:2110.00546}
  }