Deep learning-enhanced ensemble-based data assimilation for high-dimensional nonlinear dynamical systems

no code implementations • 9 Jun 2022 • Ashesh Chattopadhyay, Ebrahim Nabizadeh, Eviatar Bach, Pedram Hassanzadeh

With small ensembles, the estimated background error covariance matrix in the EnKF algorithm suffers from sampling error, leading to an erroneous estimate of the analysis state (initial condition for the next forecast cycle).

Long-term stability and generalization of observationally-constrained stochastic data-driven models for geophysical turbulence

1 code implementation • 9 May 2022 • Ashesh Chattopadhyay, Jaideep Pathak, Ebrahim Nabizadeh, Wahid Bhimji, Pedram Hassanzadeh

In this paper, we propose a convolutional variational autoencoder-based stochastic data-driven model that is pre-trained on an imperfect climate model simulation from a 2-layer quasi-geostrophic flow and re-trained, using transfer learning, on a small number of noisy observations from a perfect simulation.

Transfer Learning Weather Forecasting

Analog forecasting of extreme-causing weather patterns using deep learning

1 code implementation • 26 Jul 2019 • Ashesh Chattopadhyay, Ebrahim Nabizadeh, Pedram Hassanzadeh

The trained networks predict the occurrence/region of cold or heat waves, only using Z500, with accuracies (recalls) of $69\%-45\%$ $(77\%-48\%)$ or $62\%-41\%$ $(73\%-47\%)$ $1-5$ days ahead.