Granger, C. W. J. (1969). Investigating causal relations by econometric models and cross-spectral method' Samarasinghe, S. M., Connolly, C., Barnes, E. A., Ebert-Uphoff, I., and Sun, L. (2021). Strengthened causal connections between the mjo and the north atlantic with climate warming. Geophysical Research Letters, 48:e2020GL091168. e2020GL091168 e2020GL091168 McGraw, Marie C., and Elizabeth A. Barnes. “Memory Matters: A Case for Granger Causality in Climate Variability Studies.” AMETSOC, American Meteorological Society, 15 Apr. 2018, https://doi.org/10.1175/JCLI-D-17-0334.1. Di Capua, G., Kretschmer, M., Donner, R., Hurk, B., Vellore, R., Raghavan, K., and Coumou, D. (2019). Tropical and mid-latitude teleconnections interacting with the indian summer monsoon rainfall: A theory-guided causal effect network approach. Earth System Dynamics Discussions, pages 1–27. Ramsey, J., Zhang, K., Glymour, M., Romero, R. S., Huang, B., Imm´e, Ebert-Uphoff, Samarasinghe, S. M., Barnes, E. A., and Glymour, C. (2018). Tetrad - a toolbox for causal discovery
Date of this Version
Machine Learning, Causal Inference, Time series, Climate Science
Machine learning has become a helpful tool for analyzing data, and causal Inference is a powerful method in machine learning that can be used to determine the causal relationship in data. In atmospheric and climate science, this technology can also be applied to predicting extreme weather events. One of the causal inference models is Granger causality, which is used in this project. Granger causality is a statistical test for identifying whether one time series is helpful in forecasting the other time series. In granger causality, if a variable X granger-causes Y: it means that by using all information without X, the variance in predicted Y is larger than the variance in predicted Y by using all information included X. In other words, the prediction of the value of Y based on its own past values and on the past values of X is better than the prediction of Y based only on Y's own past values. In the project, Granger Causality is applied to determine the causal relationship between the Nino-3.4 index and the surface temperature. Nino 3.4 index refers to the Sea Surface Temperature month running means in the region from the dateline to the South American coast, and surface temperature data is from NOAA Twentieth Century Reanalysis project. Nino 3.4 index and surface temperature are calculated in time series form to meet the application of granger causality. The results are shown in the form of figures to reflect the directed causal relationship clearly. Besides Granger causality, other causal methods, such as Bayesian Network, TETRAD, and PCMCI, that can be used in the relevant study are also introduced and discussed their possible use in future relevant research.