Predicting the Big One
Not an earthquake but a catastrophic series of storms
Growing up in Southern California, the "Big One" always referred to earthquakes. There were still at that time some with living memories of the devastation of the 1906 San Francisco quake, and my father had strong memories of a smaller but still tragic quake centered off Long Beach in 1933.
Our preparation for The Big One in school was to practice earthquake drills that called for everyone to dive under their desk. That drill was extended during the cold war era to include the same response to nuclear attacks. Maybe the drill was reasonable for earthquakes.
That Big One has yet to happen, but the concept has been reinvented in recent, somewhat sensationalized stories in the press. The new incarnation is not an earthquake, but a devastating series of storms predicted to cause unprecedented rainfall and flood damage in California.
My first encounter with this Big One was in an article in the New York Times that presented the concept and its potential for destruction in graphic terms. The descriptors accompanying the opening images (deluge, pummel, ravaged) make a scientist cringe, but it is fascinating to see how a professional science paper or report gets turned into a media presentation. Maybe like the description of Zombie Ice, it takes sensationalized names and vivid descriptions to get the public's attention.
So let's dig into the story of this Big One.
Two threads are joined in the telling. The first is the relatively recent discovery of a previously invisible weather phenomenon: Atmospheric Rivers. The second is a disaster preparedness exercise led by the U.S. Geological Survey using the name ARKStorm. ARK in this case is an acronym for Atmospheric River 1,000 (K for 1,000), or an extreme series of storms with an expected recurrence frequency of once every 1,000 years. In a minute, we will see that 1,000 years may be overly optimistic.
We tend to think that now-familiar weather features have been understood for a long time, but that is often not the case. Atmospheric Rivers were only first described in the 1990s. They are very long, narrow ribbons of moisture originating in the western tropical Pacific that run unimpeded all the way to the west coast of North America, and can remain in place for hours to weeks. They tend to form between extended low and high pressure areas that throw these long streams of moisture west to east.
Discovery of these rivers followed the development of improved satellite sensors for measuring the amount of moisture in the air. This image captures some of the dynamics of an Atmospheric River, but the full NOAA video is truly worth many thousands of words in terms of visualizing how this works. A more detailed explanation of these Rivers is in an earlier essay.
North America is not the only region where Atmospheric Rivers reach landfall. One source identifies five areas where they can originate, including the north and south Pacific and Atlantic Oceans, and the Caribbean.
Atmospheric Rivers are now recognized as a common feature of the global weather system whose key characteristics are the rate at which moisture is carried towards the coast, and how long a river stays in place. A rating system (1-5) has been developed based on these two parameters that expresses potential impacts from beneficial to damaging.
Mild flows of short duration can be key to relieving drought and providing both rain for immediate use and snow in coastal and inland mountains that will melt to provide water through the dry summer months. Long-lasting and intense rivers can lead to major flooding events.
So Atmospheric Rivers are something like the Jekyll and Hyde of California hydrology. Most of the rainfall delivered to this semi-arid region comes from these rivers, and so do most of the damaging, flood-generating megastorms.
Which brings us to the source of the redefinition of the "Big One."
This part of the story begins with a project called the Multi Hazards Demonstration Project organized by the U.S. Geological Survey. This research and outreach project:
…uses hazards science to improve resiliency of communities to natural disasters including earthquakes, tsunamis, wildfires, landslides, floods and coastal erosion. The project engages emergency planners, businesses, universities, government agencies, and others in preparing for major natural disasters.
An early product of this project was a description of the probable impact of a magnitude 7.8 earthquake along the San Andreas fault. The goal of the project, and this kind of scenario generation, is to make planning agencies aware of possible worst-case scenarios for natural disasters - those Big Ones that may not happen in your lifetime, but will happen eventually.
Having examined the traditional earthquake Big One, the next study addressed another worst-case scenario, the ARKStorm, or a series of category 5 Atmospheric River events.
Scenario testing is interesting and intriguing, but to gain credibility, it helps to have a historical example to provide context. If the earthquakes of 1906 and 1933 set the stage for planning for the 7.8 quake to come, the context for the Atmospheric River Big One was established in the winter of 1861/62.
An excellent description of that tragic winter can be found here. The authors summarize the devastation this way:
The rivers and rains poured into the state’s vast Central Valley, turning it into an inland sea 300 miles long and 20 miles wide. Thousands of people died, and one quarter of the state’s estimated 800,000 cattle drowned. Downtown Sacramento was submerged under 10 feet of brown water filled with debris from countless mudslides on the region’s steep slopes. California’s legislature, unable to function, moved to San Francisco until Sacramento dried out—six months later. By then, the state was bankrupt.
In that same paper, the authors summarize geological evidence showing that this was not a one-time occurrence. A history of major floods of this dimension, recreated by analysis of sedimentation events across 4 western locations, suggests 4 events over the last 1,000 years, and perhaps 7 over the last 2,000 years. Another source describes as many as 7 events in the last 1,000 years, and 10 over the last 2,000, some more potent than our benchmark of 1861/2.
So the historical and geological records say that a megastorm season like 1861/2 has occurred every 150-300 years. And the last one happened 160 years ago.
How to generate a scenario for such an event?
The detailed ARKStorm report issued by the U.S. Geological Survey notes that weather data collection in 1861/2 was not complete enough to drive the kind of detailed predictions that were the goal of the project. Instead, the team combined three more recent events for which the data were available: Extended events in January 1969 and February 1986, and a single 24-hour event also in January 1969. This synthetic storm was thought to be similar to the 1861-1862 sequence.
Fine-scaled weather models were then used to map the distribution of precipitation across California. Combining these with spatial maps of elevation, topography and coastal conditions yielded predictions for flooding, landslides, and coastal inundation, among other impacts. This map of predicted maximum flood water depths, such as across the central valley, matches well with descriptions of the actual 1861/2 event.
Statements about emergency preparedness and response in the final ARKStorm report echo those presented in essays posted on this Substack site about sea level rise and the impacts of hurricane Ian on the choice between paying up front for disaster prevention versus paying for recovery.
An ARkStorm raises serious questions about the ability of existing federal, state, and local disaster planning to handle a disaster of this magnitude. A core policy issue raised is whether to pay now to mitigate, or pay a lot more later for recovery. Innovative financing solutions are likely to be needed to avoid fiscal crisis and adequately fund response and recovery costs from a similar, real, disaster.
And then what of climate change?
We hear repeatedly that a warming climate system will increase the amount of water vapor in the atmosphere and produce more energetic storm systems. How might this play out for an ARKStorm?
Predicting climate futures in the kind of detail required to include Atmospheric Rivers means accessing the complex computer models developed by the climate science community for that purpose. Unlike simpler models that can capture and communicate our climate future at the global scale, these models, when nested with finer scale weather models, can be used to envision something like an ARKStorm.
But there is a limitation in using existing models of climate change for this application: while models vary in structure and predicted outcomes, they are intended to describe average conditions, rather than rare extreme events. How to use them in the ARKStorm context?
Instead of looking at the average outcome of a set of different models, as most climate change projections do, a recent paper makes use of that variation among models to envision how an ARKStorm might look near the end of the 21st century. The authors essentially selected the most extreme 30 day period across a set of 40 models to characterize a single ARKStorm for each of two periods (1986-2005 and 2071-2080) representing roughly current climatic conditions and a warmer climate near the end of the century.
The paper suggests that the simulation selected by this process for the current period is very similar to the 1861/2 event. Comparing this with the selected simulation for 2071-2080 suggests that a future ARKStorm would deliver more total precipitation at higher rates over more days across the simulated 30 day period. The future map of flooding in the Central Valley of California would be more extreme than the one generated above by the U.S. Geological Survey ARKStorm process.
The paper also uses a broader set of model projections to suggest that the frequency of ARKStorms might increase dramatically as the planet warms. With as many as 3 ARKStorms predicted to occur before 2100, the frequency would increase from the historical average of one every 150-300 years to one every 30 years. While there are uncertainties in this analysis, if the actual increase in the frequency of ARKStorms is even close to this prediction, these extreme events will become more of a pressing issue for current generations, rather than future ones.
And that was a major point of the New York Times article that led to this essay. The dramatic opening graphics set the stage for a series of interviews with scientists engaged in the ARKStorm project, and for detailed presentations of particular locations where impacts, such as a failing dam, could prove catastrophic. The impression presented, and stated by some, is that Californians have been "lucky" with respect to ARKStorms for the last 160 years.
Preparing for an ARKStorm in the absence of climate change is challenge enough. It appears that the challenge may be even greater in a warmer world.
Sources
Background on the San Francisco and Long Beach Earthquakes can be found here:
https://en.wikipedia.org/wiki/1906_San_Francisco_earthquake
https://en.wikipedia.org/wiki/1933_Long_Beach_earthquake
The New York Times article on the Big One is here:
https://www.nytimes.com/interactive/2022/08/12/climate/california-rain-storm.html
Information on Atmospheric Rivers is here:
https://psl.noaa.gov/arportal/
https://en.wikipedia.org/wiki/Atmospheric_river
https://cw3e.ucsd.edu/wp-content/uploads/2013/10/Dettinger_Ingram_sciam13.pdf
This one has graphic of return times and map with arrows of AR paths
The NOAA video embedded in the essay is from:
Information on the rating scale for ARs is here:
https://water.ca.gov/News/Blog/2019/Feb-19/New-Scale-for-Atmospheric-Rivers-Impact
https://www.scientificamerican.com/article/warning-scale-unveiled-for-dangerous-rivers-in-the-sky/
The quotation on the impact of the California flood of 1861/2, and an excellent graphic on the frequency of such major events derived from geologic evidence is here:
https://cw3e.ucsd.edu/wp-content/uploads/2013/10/Dettinger_Ingram_sciam13.pdf
The list of more frequent ARKStorm-like events is from:
https://en.wikipedia.org/wiki/Great_Flood_of_1862
The ARKStorm process is described here:
https://pubs.usgs.gov/of/2010/1312/of2010-1312_text.pdf
The flooding figure is from
http://geography.wr.usgs.gov/science/mhdp/arkstorm.html
This source has predicted economic impacts
https://www.usgs.gov/centers/western-geographic-science-center/science/arkstorm
The article predicting ARKStorms in California's future is here:
https://www.science.org/doi/10.1126/sciadv.abq0995
Another paper discussing probable increases in precipitation rates in a warming world is here:
https://agupubs.onlinelibrary.wiley.com/doi/full/10.1002/grl.50334