With climate change dramatically accelerating, scientists continue to struggle to predict the shape of a substantially warmer world. This is particularly true with regard to weather and storms, which – due to the granular, mercurial processes at play – elude climate scientists more than, say, ice melt projections. Recently, a climate study commissioned by the City and County of San Francisco (and powered by two leading supercomputers) worked to understand how tomorrow’s climate would affect the storms that pass through the Bay Area.
The researchers were specifically focused on extreme precipitation – which, they wrote in Weather and Climate Extremes, “poses a major challenge for local governments … as flooding can damage and destroy infrastructure and property.”
The team started with five “historically impactful” storms ranging from a few days to over a week in duration. These storms were selected with input from stakeholders, who advised the researchers on which storms had proved most disruptive and damaging to city infrastructure and services.
The five storms (so few in number as to not strain the available supercomputing resources) were simulated at a 3km resolution over the Bay Area via convection-permitting regional climate model simulations. Further, the researchers ran three ten-member sets of simulations for each storm, representing the historical climate (aiming to reproduce the storm accurately) and two future time periods (2040-2060 and 2080-2100) under an aggressive global warming scenario (Representative Concentration Pathway 8.5).
This type of study is known as “hindcasting,” which aims to use previous weather and climate events as a point of comparison and benchmark to understand how altered variables might influence similar events in the future. “If a hurricane like Katrina happened at the end of the 21st century, what could it be like?” said Christina Patricola, an assistant professor of geological and atmospheric sciences at Iowa State University and lead author of the paper, in an interview with TACC’s Aaron Dubrow. “More rainfall, higher winds? Our method can be used for any type of weather system that can be hindcasted.”
To run these simulations, the researchers used a pair of top-ranked supercomputers: Cori, at the National Energy Research Scientific Computing Center (NERSC), an HPE system that delivers 14.02 Linpack petaflops and ranks 37th on the Top500; and Stampede2, at the Texas Advanced Computing Center (TACC), a Dell system that delivers 10.68 Linpack petaflops and ranks 44th.
“It was a very big help to have the resources from TACC and NERSC for these simulations,” Patricola said. “We’re interested in extreme precipitation totals and hourly rainfall rates. We had to go to a high resolution of 3km to make these predictions. And as we increase resolution, the computational expense goes up.”
Some increase in rainfall with warmer temperatures is to be expected, as warmer air can carry more water. But the effect shown by the simulations outpaced this predictable increase: the researchers found that the types of storms assessed in the simulations might carry 26 percent to 37 percent more precipitation by the end of the century. “We found something very interesting,” Patricola said. “Precipitation increased substantially for events with an atmospheric river and a cyclone together, whereas precipitation changes were weak or negative when there was only an atmospheric river.”
The new level of resolution provided by these simulations is also, in itself, transformative for simulating local climates.
“Having this level of detail is a game-changer,” said Dennis Herrera, general manager of the San Francisco Public Utilities Commission (SFPUC). “The SFPUC has been at the forefront when it comes to studying climate change and using that data to guide our decisions. This study will help us stay out front on this issue. This groundbreaking data will help us develop tools to allow our port, airport, utilities, and the City as a whole to adapt to our changing climate and increasingly extreme storms.”
“This project is relatively unique and one of the initial projects like this, working in very close collaboration between city agencies and climate scientists,” added Patricola. “It can serve as a good example of what climate science can do to provide the best possible information to cities as they prepare for the future.”
About the research
The research discussed in this article was published as “Future changes in extreme precipitation over the San Francisco Bay Area: Dependence on atmospheric river and extratropical cyclone events” in the June 2022 issue of Weather and Climate Extremes. The paper was written by Christina M. Patricola, Michael F. Wehner, Emily Bercos-Hickey, Flor Vanessa Maciel, Christine May, Michael Mak, Olivia Yip, Anna M. Roche and Susan Leal.