PNNL

The Science

Atlantic tropical cyclones (TCs) cause enormous damage. Scientists have sought to understand TC risk at local-to-regional scales. The challenge in understanding that risk lies in the limited historical data and rarity of these storms making landfall, combined with the high cost of simulating storms using advanced climate models. To overcome these hurdles, scientists have crafted a new approach that uses a mix of physics, simple math, and advanced computing (deep neural networks) to create a synthetic record of tropical cyclones. This breakthrough allows them to simulate tens of thousands of synthetic storms with realistic paths, strengths, and rainfall. This innovative model provides a richer, more detailed picture of Atlantic tropical cyclone behavior.

The Impact

This study pioneers the use of advanced artificial intelligence to create a detailed synthetic record of TCs, offering a groundbreaking tool for enhancing our understanding and preparedness for these natural disasters. This innovative method addresses the critical issue of limited observational data and the prohibitive computational demands of traditional models. Its contributions promise significant advancements in disaster readiness and climate risk analysis, potentially benefiting multiple sectors such as urban development, infrastructure planning, and insurance by refining our capacity to assess and manage the risks posed by TCs.

Summary

TCs pose a significant threat to the socio-economic stability of the U.S. and Caribbean coastal regions, making the precise assessment of TC risks at local and regional levels crucial. Conventional methods are limited by the brief historical record of observations and the substantial computational demands of high-resolution climate simulations, leading to challenges in accurately gauging these risks. To bridge this gap, researchers developed a groundbreaking database that includes 40,000 synthetic TCs, crafted using the Risk Analysis Framework for Tropical Cyclones (RAFT) and pioneering the application of advanced artificial intelligence for this purpose. This comprehensive database not only mirrors the historical patterns of TC tracks and intensities with high fidelity but also incorporates data on storm-induced rainfall, thus providing an all-encompassing resource for the analysis of wind and rainfall hazards posed by North Atlantic TCs. Demonstrated by its strong alignment with actual observed data, researchers methodology marks a pivotal advancement in the meticulous evaluation of TC risks, setting the stage for enhanced disaster readiness and more strategic risk management approaches. Further research is ongoing to improve the precipitation simulation and evaluate TC impacts such as urban flooding, power outages, and damage to infrastructure.

PNNL Contact

David Judi, Pacific Northwest National Laboratory, david.judi@pnnl.gov

L. Ruby Leung, Pacific Northwest National Laboratory, Ruby.Leung@pnnl.gov

Funding

This work was supported by the Multisector Dynamics and Regional and Global Model Analysis program areas of the Department of Energy, Office of Science, Office of Biological and Environmental Research as part of the multi-program, collaborative Integrated Coastal Modeling project.