Publication Details |
| Category | Text Publication |
| Reference Category | Journals |
| DOI | 10.1175/AIES-D-24-0114.1 |
Licence  |
|
| Title (Primary) | Evaluating the robustness of PCMCI+ for causal discovery of flood drivers |
| Author | Miersch, P.
; Günther, W.; Runge, J.; Zscheischler, J.
|
| Source Titel | Artificial Intelligence for the Earth Systems (AIES) |
| Year | 2025 |
| Department | CER |
| Volume | 4 |
| Issue | 4 |
| Page From | e240114 |
| Language | englisch |
| Topic | T5 Future Landscapes |
| Data and Software links | https://doi.org/10.5281/zenodo.14765911 |
| Keywords | Extreme events; Statistical techniques; Time series; Flood events; Machine learning |
| Abstract | Estimating causal drivers of high-impact extreme events such as
floods from data is an aspiring pursuit. Time series causal discovery
methods, such as the conditional-independence-based PC Momentary
Conditional Independence (PCMCI) framework, are designed to identify
causal relationships from complex multivariate observational time
series. However, the application to extreme event data remains a
challenge due to, by the nature of extremes, data length limitations,
conditional independence testing for nonlinear relationships, and
potential violations of the methods’ assumptions. So far, these
challenges have mostly been explored on synthetic data with limited
transferability to real-world applications. In this study, we evaluate
causal discovery on real and pseudoreal data generated with a
hydrological model across 45 catchments with varying flood-generating
processes. Because no detailed causal ground truth exists, we focus on
the robustness of output graphs. To this end, we simulate a large
sample, identify discharge peaks, and investigate the robustness of the
causal discovery algorithm PCMCI+ when applied to different realizations
of the same setting for various sample sizes. We find that the
robustness generally increases with sample size, yet a significant
proportion of inferred causal edges remain inconsistent even for large
datasets. Notably, while some flood drivers are reliably identified,
other key hydrological mechanisms are systematically missed even for
very large sample sizes, highlighting methodological limitations. Our
study provides a blueprint for investigating the real-world performance
of causal discovery methods and illustrates their current limitations
for identifying causal drivers of floods. Significance Statement Revealing the causes of extremes in the Earth system, like floods, is important for climate risk assessments. Causal discovery is a modern machine learning approach aiming to find these causal drivers in ever more abundant observational data. However, the reliability of causal discovery algorithms depends on many sources of uncertainty. Here, we evaluate the robustness of causal discovery on real and pseudoreal data generated with a state-of-the-art hydrological model and find that current observational sample sizes may not be enough to reliably estimate causal drivers in such challenging settings. |
| Persistent UFZ Identifier | https://www.ufz.de/index.php?en=20939&ufzPublicationIdentifier=31540 |
| Miersch, P., Günther, W., Runge, J., Zscheischler, J. (2025): Evaluating the robustness of PCMCI+ for causal discovery of flood drivers Artificial Intelligence for the Earth Systems (AIES) 4 (4), e240114 10.1175/AIES-D-24-0114.1 |
|