Abstract

<p>Recent advances in statistical learning have opened the possibility to forecast the behavior of chaotic systems using recurrent neural networks. In this letter we investigate the applicability of this framework to geophysical flows, known to be intermittent and turbulent.  We show that both turbulence and intermittency introduce severe limitations on the applicability of recurrent neural networks, both for short term forecasts as well as for the reconstruction of the underlying attractor. We test these ideas on global sea-level pressure data for the past 40 years, issued from the NCEP reanalysis datase, a proxy of the atmospheric circulation dynamics.  The performance of recurrent neural network in predicting both short and long term behaviors rapidly drops when the systems are perturbed with noise. However, we found that a good predictability is partially recovered when scale separation is performed via a moving average filter. We suggest that possible strategies to overcome limitations  should be based on separating the smooth large-scale dynamics, from the intermittent/turbulent features. </p>

Highlights

  • Several efforts have been made to apply machine learning to the prediction of geophysical data [8], to learn parameterizations of subgrid processes in climate models [9–11], to the forecasting [12–14] and nowcasting of weather variables [15–17], and to quantify the uncertainty of deterministic weather prediction [18].One of the greatest challenge is to replace equations of climate models with neural network capable to produce reliable long and short term forecast of meteorological variables

  • A first great step in this direction was the use of Echo State Networks (ESN, [19]) a particular case of Recurrent Neural Networks (RNN) to forecast the behavior of chaotic systems, such as the Lorenz 1963 [20] and the Kuramoto-Sivashinsky [21] dynamics

  • Good performance of regularized RNN in the short-term prediction of multidimensional chaotic time series was obtained, both from simulated and real data [24]. This success motivated several follow-up studies with a focus on meteorological and climate data. These are based on the idea of feeding various statistical learning algorithms with data issued from dynamical systems of different complexity, in order to study short-term predictability and long-term capabilities of RNN in producing a surrogate dynamics of the input data

Read more

Summary

Introduction

Several efforts have been made to apply machine learning to the prediction of geophysical data [8], to learn parameterizations of subgrid processes in climate models [9–11], to the forecasting [12–14] and nowcasting (i.e. extremely short-term forecasting) of weather variables [15–17], and to quantify the uncertainty of deterministic weather prediction [18].One of the greatest challenge is to replace equations of climate models with neural network capable to produce reliable long and short term forecast of meteorological variables. For all of those systems, as well as for the sea-level pressure data, we show how the performance of ESN in predicting the behavior of the system deteriorates rapidly when small-scale dynamics feedback to large scale is important.

Objectives
Results
Conclusion
Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call

Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.