How a megaflood could swamp California’s Central Valley

Posted on 30 January 2023 by Guest Author

This is a re-post from Yale Climate Connections by Jeff Masters

When early settlers came to the confluence of the Sacramento and American Rivers before the California Gold Rush, Indigenous people warned them that the Sacramento Valley could become an inland sea when great winter rains came. The storytellers described water filling the valley from the Coast Range to the Sierra during these rare events.

And their warnings became realized when Great Flood of 1861-62 hit. A six-week onslaught of at least 10 powerful Pacific storms in December and January carried mighty “atmospheric rivers” of subtropical moisture into California, dumping torrential rains in the valleys and prodigious snows in the mountains. When an unusually warm storm struck in January, heavy rains fell on the enormous Sierra snowpack, melting it.

A cataclysmic flood ensued, inundating the Central Valley and transforming it into a lake 300 miles long and over 20 miles wide; much of the now densely populated coastal plain in present-day Los Angeles and Orange counties was also inundated. As summarized in a 2013 Scientific American overview, the flood killed thousands of people, drowned one-quarter of the state’s estimated 800,000 cattle, and submerged downtown Sacramento under more than 10 feet of brown water laden with debris from countless mudslides. With the state’s capital city paralyzed, the California legislature was forced to move to San Francisco until the summer of 1862. By that point, the state was bankrupt, as one-third of its taxable properties had been destroyed.

California’s long history of megastorms

Sediment research has found that six storms similar to or even more severe than the 1861-62 storm hit California in the past 2,000 years, arriving about every 200 to 400 years. One study estimated the arrival dates as 212, 440, 603, 1029, 1418, and 1605 AD; the dates vary by a century or more from study to study, but the data makes clear that such megastorms have recurred regularly. The storm that occurred around 1605 appears to have been the mightiest of them all — and far stronger than storm that brought the Great Flood of 1861-62.

Given this history, it is inevitable that another great flood will hit the state someday, and climate change is thought to boost the odds of such an event. And when the next great flood comes, the damages could well dwarf those of any previous global weather disaster, adding up to more than $1 trillion — an extraordinary catastrophe with triple the cost of the feared great quake on the San Andreas fault.

According to a 2011 government scenario, waters, winds, and landslides from such a megastorm would likely overtop dozens of levees, flood nearly a quarter of the square footage of the state’s buildings, wipe out key roads for weeks to months, and leave some communities without power for months.

In this first of a three-part series on California’s vulnerability to a megaflood, we examine the results of this 2011 study, called the “ARkStorm” scenario, which simulated what a repeat of the Great Flood of 1861-62 might do. Part Two looks at the poor state of the U.S. dam infrastructure in general, and more specifically at the California dams at the highest risk of failure in a megaflood. And since the ARkStorm research is more than a decade old, a new study that presents an “ARkStorm 2.0” scenario will be covered in Part Three, which will discuss the future of California megafloods and how climate change likely increases their odds.

California is highly vulnerable to a great flood

Many of the prosperous cities and fertile farmlands of California are built on the flood plains of the rivers that once inundated the valleys every few hundred years. Now the rivers are dammed and lined with levees and drainage channels, protecting the critically important development on California’s flood plains.

A megaflood would be a catastrophe for two main reasons:

What is the ARkStorm scenario?

Were the Great Flood of 1861-62 to recur, it might resemble the ARkStorm scenario — a plausible hypothetical storm conjured up in a 2011 study by 117 experts led by the United States Geological Survey, or USGS. In the acronym, the “AR” stands for atmospheric river, and the “k” stands for the number 1,000 because the storm could be expected to bring 1-in-1000-year rains to some locations.

The storm modeled by these experts could flood up to 25% of all buildings in the state, breach approximately 50 levees, and force the evacuation of 1.5 million people. The hypothetical ARkStorm would flood about 4,000 square miles, much of it agricultural, with a population of about 6.5 million, including much of Sacramento, Silicon Valley, and Stockton. Along with the mammoth evacuation required in the inland region and delta counties, over 220,000 people would need short-term shelter.

Since there is little data on the meteorological conditions during 1861-62, the USGS ARkStorm scenario used a computer-modeled hybrid storm that combined two actual storms that hit California: a Southern California storm from Jan. 19-27, 1969, which killed 115 people and caused inflation-adjusted damages of $3.3 billion, followed by a northern California storm from Feb. 8-20, 1986, which killed 13. An additional tweak was applied to produce a sufficient amount of precipitation to approximately match the limited observations of 1861-62.

The simulated USGS ARkStorm did $725 billion (in 2007 dollars) in damage. Approximately 55% of the damage was to buildings, infrastructure, and agriculture (of which just 5-8% would be covered by insurance), while 45% of the damage resulted from business interruption. Adjusted for inflation, the ARkStorm would cost about $1.1 trillion in 2023 dollars, and additional costs would occur because of increases in wealth and population. In recent decades, wealth on the West Coast has increased by about 3.3% per year (using GDP growth stats); the population of California has increased by about 6% since 2007. This suggests that the ARkStorm would cost approximately $1.7 trillion today (about 7% of U.S. GDP). Complicating this estimate is the fact that many billions of dollars in levee improvements have occurred in California since the ARkStorm report was issued, which would likely lead to a modest decrease in damages.

The authors cautioned that the impacts modeled were not exhaustive, since they did not consider tourism and recreation or loss of cultural value as a result of damage to historic artifacts and sites. And a repeat of one of the stronger storms documented in the past 2,000 years might cause damages much higher.

Much of the rest of this post covers the specific sectors considered in the ARkStorm scenario, demonstrating the extraordinary danger such a flood would likely pose to people, infrastructure, agricultural lands, and property. All costs are in 2007 dollars; adjusting for inflation would make these costs about 50% higher in 2023 dollars.

A megastorm would breach levees and flood buildings

The ARkStorm scenario generated flooding with an estimated return period of 100 to 500 years over many critical California watersheds, in addition to some 1-in-1000-year flooding. A 100-year return period means that a flood has a 1% probability in any given year, or a 26% probability over a 30-year time span. Similarly, flooding with a 500-year return period has a 0.2% probability of occurrence in any given year, or 6% over a 30-year period.

Nearly one-quarter of the total building square footage in California was affected by flooding in the ARkStorm scenario. Most flooded buildings were not a total loss, but rather experienced damage requiring repair costs between 10-50% of replacement cost. (Just one inch of water in a 2,500-square-foot home can cause $27,000 in damage, and 12 inches can cause $72,000 in damage, according to FEMA.) Residential buildings dominated the flood-related building repair costs. Total flood damage to buildings was estimated at $195 billion, with another $103 billion in damage to building contents. Business interruption from building downtimes of one to three years would cost an additional hundreds of billions of dollars.

The scenario hypothesized that urban levees might be threatened or overtopped at 60 to 75 critical sites and that 15-20 breaches might realistically occur. In addition, 30 breaches of levees protecting the islands of the Sacramento-San Joaquin Delta were considered realistic, with two to three breaches occurring per island. The Delta is about 73% farmland, with much of the land lying 10-25 feet below sea level; these fields are some of the most productive agricultural land in the nation. The Delta’s levees have experienced one major breach in the past 30 years, a June 2004 event that flooded the entire island of Jones Tract.

Urban levees in California are generally constructed to withstand a flood with a 1-in-200-year recurrence interval. There are, however, a few levees protecting urban areas near Sacramento and Stockton that offer protection at lower flood thresholds. For example, the levee protecting the community of Woodland (population of 9,000, property value $1.6 billion) offers protection against only a 1-in-20-year flood — one that has an 80% chance of occurring over a 30-year period (Figure 4). Fortunately, the state and federal governments are partway through the construction of a $1.85-billion flood protection project that will help shore up levees in the Sacramento area. About 361 miles of urban and 120 miles of non-urban levees have been repaired or improved since 2007, according to the state’s Central Valley Flood Protection Plan.

The city of Stockton, a disadvantaged community whose metro-area population of nearly 800,000 is almost 80% nonwhite, has a particularly high flood risk because of inadequate levees. The city lies on the San Joaquin River, and a 2018 study by the Army Corps found that much of the city could flood to depths of 10-20 feet if the city’s levees were breached. According to the National Levee Database, the main levee protecting the city offers only 1-in-100-year flood protection. The levee protects over 129,000 people and property worth over $16 billion.

Levees protecting rural areas in California have much lower failure thresholds, with some privately-owned ones offering only 1-in-10-year protection. Multiple levees of this nature failed in the Stockton region and on the Salinas River near Monterey during the January 2023 floods (see Tweet above). And until they are tested against a 1-in-100-year level flood, we should not be overconfident that levees rated to that level of protection will perform as designed. Extreme floods are notorious for exposing the unforeseen flaws in levees. In an interview with the Los Angeles Times, Jeffrey Mount, a geomorphologist and senior fellow at the Public Policy Institute of California, said, “There are two kinds of levees: Those that have failed, and those that will fail.”

Figure 4. Sacramento area levee system (red lines). The figures within each levee area show the value of property protected, the level of protection offered by the levee, and the number of people living behind the levee. For example, the levee surrounding downtown Sacramento protects 384,000 people and $6.8 billion in property and is rated to withstand a 1-in-500-year flood. (Image credit: National Levee Database, U.S. Army Corps of Engineers)

Would a California megaflood cause dams to fail?

The ARkStorm study did not foresee any of California’s approximately 1,500 dams failing, though minor spillway damage and downstream erosion were plausible, it found. As will be discussed in Part Two of this series, there is reason to question this finding in light of the serious damage caused to the spillways of America’s tallest dam, the Oroville Dam in California, during flooding in 2017.

Important roads closed for weeks to months

The ARkStorm could cause tens of thousands of landslides, the vast majority of them being debris flows. These would cost on the order of $3 billion, including $1 billion in damage to private property, $1 billion to state highways, and $1 billion to other infrastructure. Indirect costs because of disruption of infrastructure were not considered in the estimate.

The damage to roads would be long-lasting in many cases, particularly along mountain roads in regions subject to landslides, such as coastal Highway 1 in Big Sur and near Santa Cruz (see Tweet above). Important roads would be closed for weeks or months, significantly slowing recovery efforts. The ARkStorm largely cut off traffic from Los Angeles to the north and east for one to two weeks, with gradual recovery. For Sacramento, traffic to the north, south, and west was largely cut off for one week or so, with gradual recovery thereafter.

Road damage from landslides could hinder repairs of critical water pipelines and limit access to repair wastewater treatment plants like the one in Long Beach. Water supply systems and wastewater treatment plants rely on supplies of chemicals that are carried on railways and trucks every few days, which might be cut off. One possible preparation effort: stockpiling Bailey bridges, which are portable, prefabricated trusses, primarily used by military engineering units to bridge gaps up to 200 feet. They do not require heavy equipment for construction, can be brought to the job site in trucks, and are strong enough for heavy traffic.

Figure 5. Road restoration percentages in northern California 30 days after the ARkStorm and 17 days after the ARkStorm in southern California. (Image credit: USGS)

Power outages for months in some areas

Damage to the power infrastructure by wind and flooding was estimated at $1 billion, with long restoration times of months predicted in mountainous areas experiencing high winds. Business interruption costs from lack of power were estimated at $18 billion. Flood damage to power infrastructure was also cited as a major concern. While most power plants are located out of the areas predicted to flood, some are inside, especially in Santa Clara County and Los Angeles.

High-voltage substations and generating plants have high-voltage transformers that can be damaged by flooding – for example, by flood-borne debris impacting the transformer. These transformers are custom-made, designed to operate only at that location, and serve a large population, often in excess of 200,000 people. The transformers are not interchangeable and are too expensive to stockpile beyond those available for normal operational redundancy. If one of these large transformers were damaged, it could take six months or more to replace.

Figure 6. Power restoration curves at a few key locations in California, showing the percentage of customers capable of receiving power at selected times, assuming the onset of the storm on Jan. 27, 2011. Mountainous areas experiencing higher winds, like Mono County in the Sierras (light blue curve), were modeled to have significant power outages lasting months. (Image credit: USGS)

Contaminated and interrupted water supplies

An ARkStorm would wreak serious damage on the water supply systems for California, causing an estimated $3 billion in damage. Restoration of service would take days, weeks, or months; the report suggested that the water treatment plant serving the entire city of Sacramento would be inoperative for up to three to six months. The costs of business interruption because of water system failures were put at a massive $54 billion.

About half of southern California’s water comes from the Sacramento-San Joaquin Delta, where multiple levee failures were anticipated, leading to a lengthy interruption in water transport. The levee repairs necessary to restore water transport to southern California might take three months, but that region would have other sources of water. In fact, water quality might be a far more significant problem than quantity, because of runoff carrying sediments into reservoirs and from erosion of the banks of reservoirs. Contaminants from runoff could potentially require extended boil-water orders.

Raw sewage in the water

The ARkStorm scenario had 21 of the 113 wastewater treatment plants in California experiencing flooding, at a cost of $300 million, taking days or weeks to repair. The resulting business interruption would cost an additional $28 billion. In Los Angeles, the scenario imagines that the Donald C. Tillman and Terminal Island wastewater treatment plants are flooded, putting raw sewage into the central and western San Fernando Valley and into the Los Angeles River. This causes a hazardous-material condition that could trigger evacuation of homes and businesses that were not otherwise flooded and could shut down roads through the affected area.

Billions in damage to agriculture

Damages to crops, livestock, and fields were estimated to range between $4 and $7 billion, and damages to farm structures and equipment were estimated separately at $13 billion. The scenario hypothesized that 31 of the approximately 60 islands in the Sacramento-San Joaquin Delta would flood and that it could take up to 1.5 years to remove that water because many of these islands lie 10-25 feet below sea level.

Wind damage in mountainous areas

Wind-related building repair costs are estimated to be $6 billion – small in comparison with the nearly $300 billion loss from flood damage. Wind damage contributes such a small fraction of the overall loss because the areas of highest winds (higher elevation mountainous areas) are relatively sparsely populated.

Coastal flooding would cause damage, too

While the vast majority of the flood damage California would experience from an ARkStorm would be from inland flooding from heavy rains, coastal flooding from large storm surges riding in on elevated sea levels would also cause significant damage. Coastal storms that hit California during strong El Niños are particularly problematic; ocean warming, wind patterns, and runoff from heavy rains inland have brought sea levels up to one foot above average during the past three strong El Niños (Figure 7).

Figure 7. Monthly mean water levels in San Francisco during the past three strong El Niño events (1982-83, 1997-98, and 2015-16), compared to average. (Image credit: Abe Doherty, updated with data from Reinhard Flick)

Even a 1-in-100-year flood would be devastating in California

It wouldn’t take a California flood event as severe as an ARkStorm to cause one of the most expensive disasters in U.S. history. A 2022 study estimated a lesser flood with a 1-in-100-year return period affecting only the Los Angeles area would likely inundate property — worth $56 billion with a population of about 425,000 — to a depth of a foot or more, with the risks disproportionately higher for non-Hispanic Black and disadvantaged populations.

The other two parts of this three-part series:
Part Two: If a megaflood strikes California, these dams might be at risk
Part Three: Climate change is increasing the risk of a California megaflood