Improving Risk Estimates for Extreme Rain and Snow

Article Content

A new study led by scientists from UC San Diego’s Scripps Institution of Oceanography details an improved method for estimating the likelihood of extreme precipitation events in the western United States.

The traditional approach estimates the frequency of severe rain and snow by analyzing meteorological records looking only at precipitation intensity over arbitrarily defined time periods, such as one hour or 24 hours. The new model — called Trivariate Event Distribution or TED — analyzes storm events by considering consecutive days of rainfall, maximum intensity and total rainfall. By considering these two additional variables, TED can offer risk assessments for extreme weather events that would not be considered extreme on the basis of maximum intensity alone. One such event occurred in February 1980, when Southern California experienced damaging floods as a result of sustained, moderate rainfall that in some places lasted more than a week.

In a test of the model’s accuracy, the researchers found that TED provided a good statistical fit for the historical weather data for 87% of more than 4,000 weather stations across the western United States. The model's ability to provide a more holistic assessment of extreme rain and snow storm risk could inform infrastructure planning, insurance assessment and emergency preparedness.

The study, published Feb. 7, 2025 in the journal Scientific Reports, was led by Alexander Weyant and co-authored by Alexander Gershunov and Julie Kalansky of Scripps Oceanography. Other co-authors include mathematicians Anna Panorska and Tomasz Kozubowski at the University of Nevada, Reno. The research was funded by the California Department of Water Resources via the Atmospheric River Program.

Read the study: A holistic stochastic model for precipitation events