They saw the heat coming for Northern California several weeks before it arrived. By mid-August forecasters and weather followers predicted that a hot blanket would smother the Western United States the first week in September. Every long and then medium-range forecast model painted the West Coast red. A week before Labor Day weekend, the National Weather Service (NWS) forecast office in Monterey — which rarely deploys adjectives — warned of “oppressive heat” over the upcoming weekend.
But the Weather Service, and everyone else, placed the epicenter of the heat inland. Forecasts focused on the potential for breaking all-time temperature records in the interior. There was talk of Livermore reaching 116 degrees, an all-time high only a degree short of the all-time record high in Las Vegas. There were also jokes on social media as people around the country saw the NWS-issued “excessive heat watch” for what weather app forecasts suggested would be temperatures in the high 70s in San Francisco. Even in the early morning on the Friday the heat wave began, the official NWS forecast discussion called for “upper 80s to mid 90s near the coast.”
Downtown San Francisco hit 106 by Friday afternoon. That’s the hottest it’s ever been at that station in 143 years of temperature records. Three people died and 50 people were hospitalized for heat-related illness in the city. The number of 911 calls overwhelmed ambulances and forced San Francisco to request mutual aid from neighboring counties. There were 1,342 emergency calls on Friday, San Francisco Department of Public Health Director of Communications Rachael Kagan said, more than double the number on the previous Friday.
Coastal San Francisco didn’t suffer alone: Salinas hit 107, the Farallon Islands hit 89, Eureka hit 89, and San Luis Obispo hit 115, all all-time records. The heat didn’t relent on Saturday, when San Francisco tied a 103-year-old record for warmest minimum temperature, cooling to a Vegas-like 75 degrees overnight before temperatures started climbing again. With no nighttime relief, Kagan said the city logged 1,413 emergency calls on Saturday Sept. 2, the highest since New Year’s Eve 2012.
“We knew the air mass was going to be really warm above us, and we knew the hot temperatures were going to be everywhere,” said Roger Gass, a meteorologist at the National Weather Service Bay Area office in Monterey. “I guess the surprise was that the marine influence would really shut down.”
How did dangerous heat in San Francisco come as a surprise? The regional forecast was accurate, and accurate early. High pressure building over the Pacific Coast caused a statewide heat wave, as just about everyone saw would happen. The intensity of the heat in the atmosphere was high, as forecast. But that it got so extremely hot near the coast startled even experts.
It’s similar in some ways to how Hurricane Irma underperformed its worst-case forecast for Florida, or how the 2015 El Niño didn’t bring a wet winter to Southern California. Big-picture, the forecasts for both those events were extremely accurate well in advance. The National Hurricane Center spotted Irma a full week early, recognized the potential for danger, and effectively spread the word to evacuate southeast Florida. But in its final hours before landfall in Florida the hurricane edged west by a matter of a few dozen miles, resulting in widespread damage but not the annihilation of Miami. In the El Niño, one of the strongest ever recorded, the global ocean behaved just as forecasters predicted it would almost a year in advance, and the downstream effects that followed went reasonably according to expectation — except in Southern California, where an unusual local pattern meant it didn’t rain like most forecasters thought it would.
“There’s been a flurry of recent weather events and weather extremes in California and elsewhere that from a large-scale perspective have been a predictive success,” said Daniel Swain, a climate scientist at UCLA who writes about California weather at WeatherWest.com. “But on the ground the perception is it was a miss. If you’re in exactly location X the forecast was different than you expected a day or two before. Despite the fact that by any objective metric we saw it coming from a mile away.”
So this unprecedented Labor Day heat wave: Was it a predictive success, or a failure? Why is it so hard to accurately predict temperatures in the immediate vicinity of the California coast, anyway?
There are two general patterns of heat in Northern California, Swain says. The first, which is common in the earlier months of summer, is for high pressure to bulge out from the desert southwest, like a giant dome bubbling out of Phoenix to share its insulated warmth. Generally, though, the ocean is still cold from spring upwelling, and so the dome stalls as it reaches the coast, driven back inland by marine air. Those are the familiar June or July conditions that make it 105 in Livermore, 75 in downtown San Francisco, and 65 at Ocean Beach.
The second heat pattern is what happened in early September. Later in the summer high pressure often builds over the Eastern Pacific, moving the center of the heat directly over Northern California and allowing it to envelop even the coast. That’s the pattern that leads to the Bay Area’s notorious Indian summers, and why September and October are the warmest months of the year here. That’s the pattern weather models predicted, and that ultimately materialized, over Labor Day weekend. High pressure paired with warmer coastal waters often leads to it being 110 in Livermore and 85 in downtown San Francisco, and this high pressure also featured such a warm air mass that it seemed it might push everything even a few degrees above typical heat wave highs and into record territory.
What forecasters couldn’t have foreseen, Gass said, was that the marine influence would disappear entirely. In the event, the marine air mass flattened down to zero, overwhelmed and pushed far offshore by winds blowing out of the east. Essentially, Gass said, “your air conditioning unit died.”
Those east winds were just a little stronger than forecast. Such a subtle shift, one all but imperceptible to a global weather model, explains the 20 degree difference between forecast and reality in San Francisco. And the precision underlying global weather models just isn’t sufficient to determine heat wave temperatures in a seven-mile-square coastal strip like San Francisco and the upper Peninsula.
“Meteorology is a difficult science,” Gass said. “We knew it was going to be hot, we just didn’t know the coastline wouldn’t give us any relief. There was almost nothing to indicate it would be record-breaking at the coast.”
“Meteorology is a difficult science. We knew it was going to be hot, we just didn’t know the coastline wouldn’t give us any relief. There was almost nothing to indicate it would be record-breaking at the coast.”
It’s not necessarily a problem of needing better forecasting tools, Swain said. Part of what makes meteorology difficult is that there’s a lot of randomness in the atmosphere, and ultimately forecasts depend on those uncertainties sorting themselves out. Climate science can sidestep those problems by taking the big picture and the long view. But in the chaotic, messy short-term of predicting next week’s weather to within a few miles and a few degrees, it’s hard to think we’ll ever perfectly know the future. It’s hard to even evaluate the quality of a forecast for something so extreme it has only happened one or two times before — as pointless as filling 24 hours of ESPN airtime with judgments based on the first game of a new NFL season.
“Clearly, any time it’s not perfect there’s room for improvement, but at a certain level you wonder how you best spend your time,” Swain said. “Do you really try and hone it down, or is it good enough to say there’s a record warm air mass and someone’s going to hit their record high? Is it good enough to say there’s a 50/50 chance Miami’s going to get a catastrophic strike from a hurricane? It’s not clear right now it’s possible to do any better than that. Those are really good forecasts. And yet as we’re seeing on the ground it’s not necessarily good enough for people to inform their decisions. And that’s a problem. If we can make forecasts that are objectively good but not great for people, that’s almost more a science communication problem than a science problem.”
Beth Rubenstein, a legislative aide to San Francisco Supervisor Aaron Peskin, said the forecast might have undersold the heat somewhat, but the city should have geared up anyway. The Board of Supervisors Government Oversight Audit Committee will hold a hearing Wednesday Sept. 20 to analyze preparation and communication during the heat wave. “We knew there were going to be high temperatures starting on Friday,” Rubenstein said. “And yet it seemed like there was not a clear emergency plan. There was only certain information coming out about cooling centers on end of day Friday. There was communication with the Board of Supervisors at the end of the day on Friday. Things were happening a day late. So what was the emergency plan? Why wasn’t it activated two days earlier?”
Rubenstein said the response brings up concerns for emergency services in an earthquake “for which we have no warning.”
Department of Public Health spokesperson Kagan, though, said it wasn’t a lack of planning so much as the sheer volume of demands on city services. “Overall the city’s response was very coordinated,” she said. “It was just a lot to respond to.”
78 in downtown San Francisco where it has only cooled to 76. If it doesn’t cool further, would be all-time record high minimum temperature. pic.twitter.com/6Gwe28IwGp— NWS Bay Area (@NWSBayArea) September 2, 2017
Heat waves are worth getting used to, even in San Francisco. One of the signatures of climate change, Swain said, is that in general our weather patterns remain mostly the same but with either greater magnitude or greater endurance. Right now we see a September heat wave push San Francisco to 106 instead of 96, or we see triple digit temperatures last two days instead of one. Events like that aren’t going to be common in the future, either. But maybe 30 years from now we’ll see a three-day Labor Day weekend heat wave in which downtown San Francisco hits 110, or 115. Maybe we’ll see Bakersfield, where it already sometimes reaches 115, reach 120. “Those are numbers,” Swain said, “that are hard to handle.”
In an odd footnote, San Francisco broiled over Labor Day following a summer in which it had largely been spared having to notice our altered climate. This has been the hottest summer in California history, but it hasn’t necessarily felt that way at the coast. It’s like one of those “Clinton archipelago” maps of the 2016 presidential election: while most of the area of California has baked red under intense and unprecedented heat, the reasonably small real estate where almost all Californians live — the coastal Bay Area and Los Angeles — has stayed within normal ranges. While Death Valley burned through the hottest month ever recorded on Earth in July, San Francisco huddled under blankets in a blue bubble.
“We had some of the hottest temperatures ever in June and July, and I kept getting emails saying what are you talking about? It’s cold and foggy in San Francisco,” Swain said. “But I’ve got none of those emails from people who live in Bakersfield or Redding. This has been completely bonkers, persistently hot all summer, and I hope this isn’t an indication of what the future is going to be like. But what I hate to tell them, it kind of is. That’s sobering.”