Abstract:Because atmosphere itself is a nonlinear system and there exist some problems using the linearized equations to study the initial error growth, in this paper we try to use the error nonlinear growth theory to discuss its evolution, based on which we first put forward a new concept: nonlinear local Lyapunov exponent. It is quite different from the classic Lyapunov exponent because it may characterize the finite time error local average growth and its value depends on the initial condition, initial error, variables, evolution time, temporal and spatial scales. Based on its definition and the atmospheric features, we provide a reasonable algorithm to the exponent for the experimental data, obtain the atmospheric initial error growth in finite time and gain the maximal prediction time. Lastly, taking 500 hPa height field as example, we discuss the application of the nonlinear local Lyapunov exponent in the study of atmospheric predictability and get some reliable results: atmospheric predictability has a distinct spatial structure. Overall, predictability shows a zonal distribution. Prediction time achieves the maximum over tropics, the second near the regions of Antarctic, it is also longer next to the Arctic and in subtropics and the mid-latitude the predictability is lowest. Particularly speaking, the average prediction time near the equation is 12 days and the maximum is located in the tropical Indian, Indonesia and the neighborhood, tropical eastern Pacific Ocean, on these regions the prediction time is about two weeks. Antarctic has a higher predictability than the neighboring latitudes and the prediction time is about 9 days. This feature is more obvious on Southern Hemispheric summer. In Arctic, the predictability is also higher than the one over mid-high latitudes but it is not pronounced as in Antarctic. Mid-high latitude of both Hemispheres (30°S–60°S, 30°–60°N) have the lowest predictability and the mean prediction time is just 3–4 d. In addition, predictability varies with the seasons. Most regions in the Northern Hemisphere, the predictability in winter is higher than that in summer, especially in the mid-high latitude: North Atlantic, North Pacific and Greenland Island. However in the Southern Hemisphere, near the Antarctic regions (60°S–90°S), the corresponding summer has higher predictability than its winter, while in other areas especially in the latitudes of 30°S–60°S, the prediction does not change obviously with the seasons and the average time is 3–5 d. Both the theoretical and data computation results show that nonlinear local Lyapunov exponent and the nonlinear local error growth really may measure the predictability of the atmospheric variables in different temporal and spatial scales.

Attractor radius and global attractor radius and their application to the quantification of predictability limits

The Relationship Between Deterministic and Ensemble Mean Forecast Errors Revealed by Global and Local Attractor Radii

LOCALIZED FEATURES OF CHAOTIC SYSTEM AND ATMOSPHERIC PREDICTABILITY

A New Technique to Quantify the Local Predictability of Extreme Events: The Backward Nonlinear Local Lyapunov Exponent Method

Temporal-Spatial Distribution of Atmospheric Predictability Limit by Local Dynamical Analogs

Predictability of Ensemble Forecasting Estimated Using the Kullback-Leibler Divergence in the Lorenz Model

Information theory in nonlinear error growth dynamics and its application to predictability: Taking the Lorenz system as an example

Information Theory in Nonlinear Error Growth Dynamics and Its Application to Predictability:Taking the Lorenz System As an Example

Quantitative Study of the Relative Effects of Initial Condition and Model Uncertainties on Local Predictability in a Nonlinear Dynamical System

Determination of the Backward Predictability Limit and Its Relationship with the Forward Predictability Limit

Application of Backward Nonlinear Local Lyapunov Exponent Method to Assessing the Relative Impacts of Initial Condition and Model Errors on Local Backward Predictability

Comparison Of The Influences Of Initial Error And Model Parameter Error On The Predictability Of Numerical Forecast

Heterogeneity of the Attractor of the Lorenz '96 Model: Lyapunov Analysis, Unstable Periodic Orbits, and Shadowing Properties

Nontrivial Global Attractors in 2-D Multistable Attractor Neural Networks

Quantifying Local Predictability of the Lorenz System Using the Nonlinear Local Lyapunov Exponent

Nonlinear Local Lyapunov Exponent and Quantification of Local Predictability

Relationships Between the Limit of Predictability and Initial Error in the Uncoupled and Coupled Lorenz Models

A study of predictability of coupled ocean–atmosphere system using attractor radius and global attractor radius

Estimation of Atmospheric Predictability for Multivariable System Using Information Theory in Nonlinear Error Growth Dynamics

Study on the Regularity of Predictability Limit of Chaotic Systems with Different Initial Errors

Nonlinear Local Lyapunov Exponent and Atmospheric Predictability Research