Abstract:Abstract. Despite the great success of machine learning, its applications in climate dynamics have not been well developed. One concern might be how well the trained neural networks could learn a dynamical system and what can be the potential applications of this kind of learning. Detailed studies show that the coupling relations or dynamics among variables in linear or nonlinear systems can be well learnt by reservoir computer (RC) and long short-term memory (LSTM) machine learning, and these learnt coupling relations can be further applied to reconstruct one series from the other dominated by common coupling dynamics. In order to validate the above conclusions, toy models are applied to address the following three questions: (i) what can be learnt from different dynamical time series by machine learning; (ii) what factors significantly influence machine learning reconstruction; and (iii) how to select suitable explanatory or input variables for the reconstructed variable for machine learning. The results from these toy models show that both of RC and LSTM can indeed learn coupling relations among variables, and the learnt implicit coupling relation can be applied to accurately reconstruct one series from the other. Both of linear and nonlinear coupling relations between variables can influence the quality of the reconstructed series. If there is a strong linear coupling between variables, all of variables can be taken as explanatory variables for the reconstructed variable, and the reconstruction can be bi-directional. However, when the linear coupling among variables is much weaker, but with stronger nonlinear causality among variables, the reconstruction quality is direction-dependent and it may be only uni-directional. We propose using convergent cross mapping causality (CCM) index ρa→b to determine which variable can be taken as the reconstructed one and which can be taken as the explanatory variable. For example, the Pearson correlation between the average Tropical Surface Air Temperature (TSAT) and the average Northern Hemispheric SAT (NHSAT) is as weak as 0.08, but the CCM index of that NHSAT cross maps TSAT is ρN→T = 0.70, it means that NHSAT could be taken as the explanatory variable. Then we find that TSAT can be well reconstructed from NHSAT by means of RC. However, the reconstruction quality in the opposite direction is poor, because the CCM index of that TSAT cross maps NHSAT is only ρT→N = 0.24. These results also provide insights on machine learning approaches for paleoclimate reconstruction, parameterization scheme, and prediction in related climate studies.

Reconstructing Coupled Time Series in Climate Systems Using Three Kinds of Machine-Learning Methods

Reconstructing coupled time series in climate systems by machine learning

A Hybrid Atmospheric Model Incorporating Machine Learning Can Capture Dynamical Processes Not Captured by Its Physics‐Based Component

Evaluation of statistical climate reconstruction methods based on pseudoproxy experiments using linear and machine learning methods

Predicting global patterns of long-term climate change from short-term simulations using machine learning

Current Progress in Subseasonal-to-decadal Prediction Based on Machine Learning

Revealing recurrent regimes of mid-latitude atmospheric variability using novel machine learning method

The effectiveness of machine learning methods in the nonlinear coupled data assimilation

Deep learning tool: reconstruction of long missing climate data based on spatio-temporal multilayer perceptron

Machine learning for climate physics and simulations

Climate-invariant machine learning

Reconstructing Historical Climate Fields With Deep Learning

Machine Learning Climate Model Dynamics: Offline versus Online Performance

Combining Neural Networks and CMIP6 Simulations to Learn Windows of Opportunity for Skillful Prediction of Multiyear Sea Surface Temperature Variability

Super-resolution of subsurface temperature field from remote sensing observations based on machine learning

Principle of Machine Learning and Its Potential Application in Cli-mate Prediction

Towards Learned Emulation of Interannual Water Isotopologue Variations in General Circulation Models

Efficient Climate Simulation via Machine Learning Method

Stress-testing the coupled behavior of hybrid physics-machine learning climate simulations on an unseen, warmer climate

Predicting climate change using an autoregressive long short-term memory model