Machine Learning Techniques to Construct Patched Analog Ensembles for Data Assimilation

  title={Machine Learning Techniques to Construct Patched Analog Ensembles for Data Assimilation},
  author={Lucia Minah Yang and Ian G. Grooms},
  journal={J. Comput. Phys.},
  • L. Yang, I. Grooms
  • Published 27 February 2021
  • Computer Science, Environmental Science
  • J. Comput. Phys.

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

Data assimilation (DA) is a key component of many forecasting models in science and engineering. DA allows one to estimate better initial conditions using an imperfect dynamical model of the system

Extending the Capabilities of Data-Driven Reduced-Order Models to Make Predictions for Unseen Scenarios: Applied to Flow Around Buildings

We present a data-driven or non-intrusive reduced-order model (NIROM) which is capable of making predictions for a significantly larger domain than the one used to generate the snapshots or training

Evaluation of Machine Learning Techniques for Forecast Uncertainty Quantification

This research presents a probabilistic evaluation of the phytochemical properties of carbon dioxide, which has the potential to have an impact on temperature and carbon dioxide levels in the atmosphere.

An AI-based Domain-Decomposition Non-Intrusive Reduced-Order Model for Extended Domains applied to Multiphase Flow in Pipes

This paper presents a new AI-based non-intrusive reduced-order model within a domain decomposition framework (AI-DDNIROM), which is capable of making predictions for domains significantly larger than the domain used in training.



Analog ensemble data assimilation and a method for constructing analogs with variational autoencoders

  • I. Grooms
  • Environmental Science, Computer Science
    Quarterly Journal of the Royal Meteorological Society
  • 2020
The use of analogs from a modestly‐sized catalog is shown to improve the performance of EnOI, with limited marginal improvements resulting from increases in the catalog size, and a new method of constructing analogs using variational autoencoders is proposed.

Analog Forecasting with Dynamics-Adapted Kernels

A suite of forecasting methods which improve traditional analog forecasting by combining ideas from kernel methods developed in harmonic analysis and machine learning and state-space reconstruction for dynamical systems are introduced.

A framework for variational data assimilation with superparameterization

A 3D-Var variational data assimilation framework for use with Superparameterization, which approximates the large scale dynamics of the true system accurately and is a natural fit for relatively expensive multiscale SP models.

Comparing Adaptive Prior and Posterior Inflation for Ensemble Filters Using an Atmospheric General Circulation Model

This work has shown that sample errors and model errors are major drawbacks from which ensemble Kalman filters suffer and can be addressed by redesigning the ensemble size and model parameters.

Data Assimilation Using an Ensemble Kalman Filter Technique

The possibility of performing data assimilation using the flow-dependent statistics calculated from an ensemble of short-range forecasts (a technique referred to as ensemble Kalman filtering) is

A Hybrid NWP–Analog Ensemble

A hybrid ensemble (HyEn) is constructed by finding m analogs for each member of a small n-member NWP ensemble, to produce a total of m × n members, to combine the strengths of the AnEn andNWP ensemble approaches.

Ocean data assimilation: a case for ensemble optimal interpolation

Drawing on Bluelink outcomes and a series of experiments with simple models, a case for using EnOI for ocean data assimilation is made and a discussion of the benefits of EnOi as well as its known limitations are discussed.

Auto-Encoding Variational Bayes

A stochastic variational inference and learning algorithm that scales to large datasets and, under some mild differentiability conditions, even works in the intractable case is introduced.

Probabilistic Weather Prediction with an Analog Ensemble

The skill and value of AnEn predictions are compared with forecasts from a state-of-the-science NWP ensemble and a 12–15-month-long training period of forecasts and observations.

Sequential data assimilation with a nonlinear quasi‐geostrophic model using Monte Carlo methods to forecast error statistics

A new sequential data assimilation method is discussed. It is based on forecasting the error statistics using Monte Carlo methods, a better alternative than solving the traditional and