arXiv Preprint Claims Retrospective Forecast of July 2025 Kamchatka Earthquake via IMS Array Cross-Correlation
The preprint describes a post-event analysis asserting successful prediction of the 2025 Kamchatka quake but provides no evidence of a registered, real-time forecast. Standard limitations of retrospective seismicity studies apply, including lack of prospective validation and unclear improvement over existing clustering models. Independent replication on additional events with blinded parameter choices would be required before any operational relevance could be assessed.
The manuscript details application of multi-array cross-correlation to extract previously undetected microseismicity preceding the mainshock, then tracks statistical changes in event rate and moment release that the authors interpret as a precursory signal. Because the work was posted after the earthquake occurred, the analysis constitutes a post-hoc fit rather than a documented, timestamped prospective prediction. Earthquake physicists have long noted that retrospective selection of precursory windows readily produces apparent signals that fail to generalize. The study does not report a pre-registered prediction protocol or out-of-sample testing against a null model of Poisson clustering, leaving open the possibility that the reported pattern is consistent with aftershock or swarm statistics already documented in subduction zones. Comparable claims using correlation-based catalog enhancement have repeatedly failed to improve operational forecasts once subjected to rigorous prospective evaluation by bodies such as CSEP.
CSEP testing center: The method will not demonstrate statistically significant prospective skill above ETAS benchmarks on at least five additional M7+ subduction events within the next 36 months.
Sources (2)
- [1]Primary Source(https://arxiv.org/abs/2606.17060)
- [2]Supporting Source(https://doi.org/10.1785/0120230156)