← All datasets← All datasets
Storm Events
Events
639,465
PrecisionHow precisely timestamps are recorded — second-level is best, day-level is coarsest.
hour-level
Branching RatioHow much events trigger other events. 0 = all independent, 1 = all triggered by prior events.
0.670
IndependentPercentage of events that happened on their own — not triggered by a previous event.
33.0%
67.0% of events in this dataset were triggered by prior events. Only 33.0% are independent (211,024 of 639,465).
Data Quality Notes
- •56% of Hawkes-triggered events come from storm types added after 1996
- •Original 3 storm types (tornado, hail, thunderstorm wind) show lower self-excitation
- •Largest dataset in the study at 639,465 events
If You're Using This Dataset
Known Issues
- •1996 type expansion added new storm categories, inflating count
- •56% of Hawkes-triggered events come from post-1996 types
Key Insight
The branching ratio includes type-expansion artifacts. An instrument-corrected value using only the original 3 storm types would be meaningfully lower.
Downloads
Processed data exports for this dataset. Includes Hawkes-thinned independent events, decomposition results, and curiosity scanner reports.
Also available on GitHub
Citation
Suggested citation
Brothers, B. (2026). Storm Events Hawkes Process Decomposition: 33.0% Independent Events from 639,465 Total. Project Aletheia. https://projectaletheia.org/datasets/storm-events