Climate Change: Linear and Nonlinear Causality Analysis

The goal of this study is to detect linear and nonlinear causal pathways toward climate change as measured by changes in global mean surface temperature sea level over time using a data-based approach contrast the traditional physics-based models. Monthly data on potential factors, including greenhouse gas concentrations, sunspot numbers, humidity, ice sheets mass, coverage, from January 2003 December 2021, have been utilized analysis. We first applied vector autoregressive model (VAR) Granger causality test gauge relationships among factors. then adopted error correction (VECM) well distributed lag (ARDL) quantify long-run equilibrium short-term dynamics. Cointegration analysis has also examine dual directional causalities. Furthermore, work, we presented novel pipeline based artificial neural network (ANN) VAR ARDL models embedded data. results indicate that rise affected sheet mass (both linearly nonlinearly), (nonlinearly), extent coverage (nonlinearly weakly); whereas specific humidity concentration warming nonlinearly) number (only nonlinearly weakly). tend fit closer than expected due increased parameter dimension Given information criteria are not generally applicable comparison statistical series models, our next step robustness compare forecast accuracy these two soon-available 2022 monthly