A comparison of two causal methods in the context of climate analyses
International audience Correlation does not necessarily imply causation, and this is why causal methods have been developedto try to disentangle true causal links from spurious relationships. In our study, we use two causal methods,namely, the Liang–Kleeman information flow (LKIF) and the Peter and...
Published in: | Nonlinear Processes in Geophysics |
---|---|
Main Authors: | , , , , , |
Other Authors: | , , , , , , , , , , , , |
Format: | Article in Journal/Newspaper |
Language: | English |
Published: |
HAL CCSD
2024
|
Subjects: | |
Online Access: | https://imt-atlantique.hal.science/hal-04591649 https://imt-atlantique.hal.science/hal-04591649/document https://imt-atlantique.hal.science/hal-04591649/file/npg-31-115-2024.pdf https://doi.org/10.5194/npg-31-115-2024 |
Summary: | International audience Correlation does not necessarily imply causation, and this is why causal methods have been developedto try to disentangle true causal links from spurious relationships. In our study, we use two causal methods,namely, the Liang–Kleeman information flow (LKIF) and the Peter and Clark momentary conditional independence(PCMCI) algorithm, and we apply them to four different artificial models of increasing complexity and onereal-world case study based on climate indices in the Atlantic and Pacific regions. We show that both methodsare superior to the classical correlation analysis, especially in removing spurious links. LKIF and PCMCI displaysome strengths and weaknesses for the three simplest models, with LKIF performing better with a smallernumber of variables and with PCMCI being best with a larger number of variables. Detecting causal links fromthe fourth model is more challenging as the system is nonlinear and chaotic. For the real-world case study withclimate indices, both methods present some similarities and differences at monthly timescale. One of the keydifferences is that LKIF identifies the Arctic Oscillation (AO) as the largest driver, while the El Niño–SouthernOscillation (ENSO) is the main influencing variable for PCMCI. More research is needed to confirm these links,in particular including nonlinear causal methods. |
---|