WHITS (Wind-focused Hurricane Interactive Track Simulator) has been used to create a dataset representing 10,000 years of global tropical cyclone activity, including tropical cyclone tracks and associated wind speeds. Such large unbiased synthetic databases are essential for insurance and risk modeling, as they allow for robust estimation of rare, high-impact events and more reliable assessment of return periods for catastrophic losses. The download package includes the full stochastic catalog, and all supporting Python code used for data generation and figures.
Unmet Need: Accurate Global cyclone risk assessment tools for insurance and resilience planning
Accurate simulation of tropical cyclone risk is critical for insurance pricing, disaster response, and infrastructure resilience. Historical records, especially for rare events, are limited, and simple bias correction methods often fail to capture true variability across regions and storm intensities. Many existing ML-based models lack the track realism and accuracy required for insurance applications.
The Technology: Global tropical cyclone simulator with memory-based track modeling
WHITS uses machine learning to sample, adapt, and reassemble historical storm track segments, preserving memory across segments to generate realistic storm paths and wind speeds. This approach improves track density, landfall statistics, and wind field consistency. Unlike many academic models, WHITS produces statistically robust and unbiased simulations applicable for global insurance risk analysis.
Applications:
- Catastrophe risk modeling for reinsurance
- Insurance analytics platforms
- Parametric insurance design and climate risk disclosure
- Infrastructure and evacuation planning for operational average risk over the next 5 years
- Coastal resilience and disaster preparedness
Advantages:
- Insurance-relevant accuracy: Track density and landfall distributions match historical behavior
- Global coverage: Every tropical cyclone ocean basin
- No bias correction needed: Realistic, statistically grounded track and wind outputs
- Preserves storm memory: Enhances realism through segment-based stochastic modeling
- Highly scalable: Large storm database for modeling and analytics
Lead Inventors:
Jennifer Nakamura, Ph.D.
Upmanu Lall, Ph.D.
WHITS Web:
https://rainbow.ldeo.columbia.edu/~jennie/WHITS/
Related Publications:
Nakamura, J., Lall U., Kushnir Y., and Rajagopalan B., “HITS: Hurricane Intensity and Track Simulator with North Atlantic Ocean Applications for Risk Assessment.” J. Appl. Meteor. Climatol., 2015 Jul 1; 54(7): 1620–1636.
Nakamura, J., Lall U., Kushnir Y., Harr PA., McCreery K., “Early Season Hurricane Risk Assessment: Climate-Conditioned HITS Simulation of North Atlantic Tropical Storm Tracks.” J. Appl. Meteor. Climatol. 2021 Apr 15; 60(4): 559–575.
Nakamura, J., Lall, U., Kushnir Y., Harr, PA. “A saturated stochastic simulator: synthetic US Gulf coast tropical cyclone precipitation fields.” Nat Hazards. 2023 Oct 24; 120(2024): 1295-1318.
Tech Ventures Reference:
