Abstract:El Niño-Southern Oscillation (ENSO) is one of the significant climate phenomena, which appears periodically in the tropic Pacific. The intermediate coupled ocean-atmosphere Zebiak-Cane (ZC) model is the first and classical one designed to numerically forecast the ENSO events. Traditionally, the conditional nonlinear optimal perturbation (CNOP) approach has been used to capture optimal precursors in practice. In this paper, based on state-of-the-art statistical machine learning techniques, we investigate the sampling algorithm proposed in [Shi and Sun, 2023] to obtain optimal precursors via the CNOP approach in the ZC model. For the ZC model, or more generally, the numerical models with dimension $\mathrm{O}(10^4-10^5)$, the numerical performance, regardless of the statically spatial patterns and the dynamical nonlinear time evolution behaviors as well as the corresponding quantities and indices, shows the high efficiency of the sampling method by comparison with the traditional adjoint method. The sampling algorithm does not only reduce the gradient (first-order information) to the objective function value (zeroth-order information) but also avoids the use of the adjoint model, which is hard to develop in the coupled ocean-atmosphere models and the parameterization models. In addition, based on the key characteristic that the samples are independently and identically distributed, we can implement the sampling algorithm by parallel computation to shorten the computation time. Meanwhile, we also show in the numerical experiments that the important features of optimal precursors can be still captured even when the number of samples is reduced sharply.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: in the prediction of El Niño - Southern Oscillation (ENSO) events, how to efficiently obtain the optimal precursors. Specifically, the paper explores the application of sampling algorithms based on statistical machine - learning techniques in the Zebiak - Cane (ZC) model to replace the traditional adjoint method for capturing the conditional nonlinear optimal perturbation (CNOP). The ZC model is a classic coupled ocean - atmosphere model for numerically predicting ENSO events. The traditional method has the problem of high computational complexity when dealing with high - dimensional numerical models, while the sampling algorithm proposed in the paper can not only reduce the amount of computation but also further improve efficiency through parallel computing. ### Main research content 1. **Problem background**: - ENSO is one of the most significant inter - annual fluctuation phenomena in the global climate system, which has a huge impact on the earth's ecosystem and human society. - The ZC model is the first coupled ocean - atmosphere model for numerically predicting ENSO events. - Traditionally, the conditional nonlinear optimal perturbation (CNOP) method has been used to capture the optimal precursors, but this method is computationally expensive in high - dimensional models. 2. **Research methods**: - The paper proposes a sampling algorithm based on statistical machine - learning techniques for obtaining the optimal precursors in the ZC model. - This algorithm approximates the gradient by sample averaging, thus avoiding the use of the adjoint model, which is difficult to develop in the coupled ocean - atmosphere model. - The sampling algorithm also takes advantage of the independent and identically distributed characteristics of samples and can shorten the computing time through parallel computing. 3. **Experimental results**: - Numerical experiments show that even when the number of samples is significantly reduced, the sampling algorithm can still capture the important features of the optimal precursors. - Compared with the traditional adjoint method, the sampling algorithm shows a high degree of similarity in static spatial patterns and dynamic nonlinear time - evolution behaviors. - The computing time of the sampling algorithm is significantly reduced, especially in a parallel computing environment, and the computing efficiency is greatly improved. ### Formula display - **Standardized sea - surface temperature anomaly and thermocline depth anomaly**: \[ \mathbf{T}=\frac{\mathbf{T}}{\vert T\vert}=\frac{\mathbf{T}}{2}, \quad \mathbf{H}=\frac{\mathbf{H}}{\vert H\vert}=\frac{\mathbf{H}}{50} \] - **Euclidean norm of the initial error**: \[ \|\mathbf{u}_0\| = \left\| (\mathbf{T}(0), \mathbf{H}(0)) \right\|=\sqrt{\sum_{i,j}(T_{ij}^2 + H_{ij}^2)} \] - **Objective function**: \[ J(\mathbf{u}_0)=\|\mathbf{T}(\tau)\|^2 \] - **Constrained nonlinear optimization problem**: \[ \max_{\|\mathbf{u}_0\|\leq\delta}J(\mathbf{u}_0) \] - **Expected representation of the sampling algorithm**: \[ \hat{J}(\mathbf{u}_0)=\mathbb{E}_{\mathbf{v}_0\in B_d}[\nabla J(\mathbf{u}_0+\epsilon\mathbf{v}_0)] \] - **Expected representation of the gradient**: \[ \nabla\hat{J}(\mathbf{u}_0)=\mathbb{E}_{\mathbf{v}_0\in B_d}[\nabla J(\mathbf{u}_0+\epsilon\mathbf{v}_0)]=\frac{d}{\epsilon}\cdot\mathbb{E}_{\mathbf{v}_0\in S^{d - }

The Sampling Method for Optimal Precursors of ENSO Events

Echo State Network Based on Phase Space Reconstruction: El Niño 3 Index Forecasting

A Robust Strategy to Account for Data Sampling Variability in the Development of Hydrological Models

The first kind of predictability problem of El Niño predictions in a multivariate coupled data‐driven model

An adjoint-free algorithm for conditional nonlinear optimal perturbations (CNOPs) via sampling

A new subsurface precursor across the spring predictability barrier for the ENSO prediction

Mechanism and Prediction Studies of the ENSO

An Ensemble CNOP Method Based on a Pre-Screening Mechanism for Targeted Observations in the South China Sea

Rigorous Derivation of Stochastic Conceptual Models for the El Niño-Southern Oscillation from a Spatially-Extended Dynamical System

Complexity based approach for El Nino magnitude forecasting before the "spring predictability barrier"

Using an ensemble nonlinear forcing singular vector data assimilation approach to address the ENSO forecast uncertainties caused by the "spring predictability barrier" and El Niño diversity

Improved reliability of ENSO hindcasts with multi-ocean analyses ensemble initialization

Spatio-temporal data generation based on separated attention for ENSO prediction

Nonlinear Prediction of ENSO

A Hybrid Neural Network Model for ENSO Prediction in Combination with Principal Oscillation Pattern Analyses

Maximum Impacts of the Initial and Model Parametric Errors on El Niño Predictions

Impacts of Initial Zonal Current Errors on the Predictions of Two Types of El Niño Events

Temporal Convolutional Networks for the Advance Prediction of ENSO

Quantifying the Predictability of ENSO Complexity Using a Statistically Accurate Multiscale Stochastic Model and Information Theory

A Simple Multiscale Intermediate Coupled Stochastic Model for El Niño Diversity and Complexity

ENSONet: a data-driven ENSO forecasting model with concise spatial location learning parameters and temporal embedding