At 4:42 p.m. on Feb. 12, 2017, one year ago today, the California Department of Water Resources issued a tweet: “EMERGENCY EVACUATION: Auxiliary spillway at Oroville Dam predicted to fail within the next hour. Oroville residents evacuate northward.”
An estimated 188,000 residents were ordered to evacuate. In the end, the emergency spillway did not fail. Months later, forensic investigators couldn’t say for sure how close the structure came to breaking apart.
But what if had?
Here’s a conceptual chronology of the near-catastrophe, based on a sophisticated UC Irvine computer analysis that wasn’t available to state officials during the emergency. The state’s inundation maps at the time only provided modeling of a complete failure of the main dam itself.
Never miss a local story.
5:42 p.m.: Undermined by erosion less than 36 hours after it was put into use for the first time ever, the emergency spillway begins to slide from its position next to the dam’s main spillway, which fractured days earlier.
More water begins to pour around the emergency spillway’s edges, loosening it further, and eventually a 1,000-foot-long block of concrete tumbles completely away. The top 30 feet of Lake Oroville, filled to the brim, pours down the unpaved mountainside and into the Feather River channel below.
The peak flow is an astonishing 620,000 cubic feet per second, or more than twice as much as the dam is ever supposed to release. More than 450,000 acre-feet of water – about half the volume of Folsom Lake – is gushing into the Sacramento Valley. Tens of thousands of area residents are sitting in their cars, stuck in traffic as they try to flee.
6:12 p.m.: The leading edge of water escapes the Feather River canyon and enters the city of Oroville. The water is still about 30 feet high.
The river’s levees can handle up to 170,000 cubic feet per second of water, but they are completely overwhelmed by a flow that is more than three times capacity.
Water spills into the city, running fast enough to rip wood-framed houses from their foundations. Near the river, its scouring power is strong enough to tear up roads and highways next to the river channel. Further away from the river, plants are stripped from the soil.
6:42 p.m.: Much of central Oroville is flooded. The water has spread out, but is more than 10 feet deep in many blocks near the Feather River. The city’s bridges over the river are impassible. Parts of highways 70 and 162 are covered in water – much of it deep enough to cover cars entirely. The leading edge of the water is just passing the city limits on the Feather River, near Georgia Pacific Road. Hundreds of homes are swept off their foundations.
7:42 p.m.: For the past hour, the flood waters have raced across the floodplain southwest of the city. The area has few structures or roads. Water now reaches the edge of the city airport, but it will not advance further in that direction.
9:42 p.m.: The front of the flood has traveled 18 miles down the Feather River from the spillway. It has spread across orchards to the west, inundating rural homes along Larkin Road and Vance Avenue, and it now reaches Highway 99. There it is waist-high, and still strong enough to push cars off the road.
11:42 p.m.: The town of Biggs, population 1,700, is next to flood. It sits on a bit of a hill, so parts of downtown are under less than a foot of water. But the water is 3 feet deep or more near Biggs High School – and still flowing fast enough to knock a person over in that part of the city.
1:42 a.m., Feb. 13: The flooding reaches Gridley, population 6,584, early the next morning. Flooding on the north side reaches 3 feet deep. Parts of the city will eventually be an island in the flood – but the police department, hospital and high school will be in the wet areas.
4:30 a.m.: Live Oak is next to flood. Unlike Gridley and Biggs, it’s not on a rise – so flooding is deepest in the center of town. But by this point, that’s only about 18 inches of water and people can walk through the flow without falling over.
5 a.m.: The main course of water rushing down the Feather River arrives in Marysville and Yuba City. Even though these cities are about 10 miles downstream from Live Oak, they sit on lower ground and actually might be more vulnerable. The release from the dam has raised the river, which runs between the two cities, by more than 5 feet.
Because the river has been running fairly low up until now, it’s not certain the 5-foot increase would overtop the levees at Yuba City or Marysville. But water that’s escaped the river channel from points upstream could enter the cities and flood them with 3 to 5 feet of water.
60 hours after the breach: Water reaches the Sacramento River via the Sutter Bypass and the Feather River. The flood has mostly run its course, but it still manages to raise water levels on the Sacramento by about 3 feet.
Behind this scenario
This chronology of a hypothetical Oroville spillway failure is based on an analysis by FloodRise, a research project led by UC Irvine that uses computer models to map flood risks.
The team did an analysis of the auxiliary spillway failure after noting that the dam inundation flood maps used in the crisis were based on a failure of the main dam and not the auxiliary spillway alone, according to Adam Luke, one of the program’s engineers.
“We thought we could run our models and have a more relevant flood map to the situation that was happening at the time,” Luke said.
A detailed version of the the flood maps they generated is available here.
The model takes into account reservoir conditions, the size of the breach, the downstream geography and topography, and the friction of the floodplain.
The scenario makes assumptions about several aspects of the situation that are difficult to predict. It assumes the spillway is completely dislodged from its position, and the remaining gap doesn’t erode further.
The model also lacks precise measurements for the levee system, so flood hazards are only valid if the levees fail following the spillway collapse. This assumption becomes more tenuous farther downstream from Oroville. The team estimates the model is 60 to 80 percent accurate.
UC Irvine engineers said the impacts they predicted from the floodwaters – people knocked over, cars pushed off roads – were based on earlier floods, such as the 1963 failure of the Baldwin Hills dam in Los Angeles. That failure killed 5 people and damaged more than 250 homes.
The FloodRise model – a simulation of the four days following a failure – took three days for the computers to run in real time.
“If you don’t have those maps available prior to the situation, then you’re not going to be able to have data describing the exact scenario in real time,” Luke said.
But Luke said they are working to change that.
“If we could get these models to run fast enough you could really provide a valuable resource for protecting property and lives,” he said.
Where were you?
If were you evacuated or otherwise impacted, tweet your experience with the hashtag #orovilleevac and we’ll consider including it here.
Nathaniel Levine: 916-321-1026, @NathanielLevine