The Big Thompson Disaster: Reverberations of a Flash Flood, 40 Years Later

What began as a celebratory Saturday in the mountains ended in tragedy 40 years ago this weekend, when a catastrophic flash flood ripped through the narrow Big Thompson Canyon of Colorado’s Front Range. A total of 144 people were killed on that Saturday evening, July 31, 1976--the eve of the 100th anniversary of Colorado’s statehood. On just about any summer weekend, the canyons northwest of Denver are packed with vacationers and day-trippers. With the state’s centennial falling on this particular weekend, the mood was especially festive, and the weather seemed no more threatening than on many other summer days. Forecasts through the day called for a 40% to 50% chance of showers and thunderstorms, but there was no particular concern about flood risk. Only a few hours later, critical gaps in weather data, communication, and public awareness had teamed up with a slow-moving deluge to create a true disaster--one that’s had a noteworthy influence on how we deal with flash floods today.


Figure 1. The irony of tragedy: a structure labeled “DREAMLAND” torn apart by the Big Thompson Canyon flood on July 31, 1976. Image credit: USGS.

Working in a data vacuum
The mountains of the U.S. West are just as prone to flash flooding as they were in 1976. However, there has been phenomenal progress in the ability to foresee, monitor, and warn people about the risks. Looking back from the vantage point of our current data-saturated world, it is astonishing how little information was available to forecasters on that fateful day. Routine satellite observations were in their infancy, and the black-and-white images that came in every 30 minutes were crude by today’s standards (see Figure 2). On the ground, there were no official weather stations or stream gauges reporting from anywhere in the Big Thompson Canyon, which starts near Rocky Mountain National Park and rolls about 25 miles eastward and downward, ending near the city of Loveland.


Figure 2. Left: A 7:30 PM CDT radar image of the flood-producing thunderstorm complex on July 31, 1976, in the Big Thompson Canyon (upper left of image), which was located near the outer edge of the radar range. Right: Satellite imagery from 7:00 PM CDT provided little detail on the Big Thompson storm, part of an arc of intense thunderstorms stretching from north-central Colorado (point A) to southeast Kansas (point C). Image credit: NOAA.

Most crucially, there was only a smattering of radar data for local forecasters to draw on. Just six years earlier, in 1970, northeast Colorado had obtained its first National Weather Service weather radar. This “conventional” unit preceded the NEXRAD Doppler network of the 1990s. The radar was located at Limon, more than 100 miles southeast of the Big Thompson Canyon, which meant the canyon fell toward the outer edge of the radar’s useful range. There were other problems as well. During this pre-Internet era, forecasters based in Denver had no way to directly view the radar observations being gathered at Limon. Instead, standard practice was for black-and-white images to be transmitted from Limon to Denver via facsimile, or “fax”. On July 30, the fax transmitter at Limon broke, meaning that no images could be transmitted to Denver until repairs could be made. This didn’t happen until August 1. As a result, the radar technician working on July 31 had to make phone calls to an NWS forecaster in Denver and summarize what he was seeing on radar--a cumbersome process at best.


Figure 3. A massive thunderstorm complex builds over the upper reaches of the Big Thompson Canyon on the late afternoon of July 31, 1976. Image credit: NOAA


The cold truth about warm rain
The gaping observational holes of 1976 were accompanied by a limited understanding of the meteorology that drives flash floods in the mountainous West. We now know that the heaviest rainfall comes from “warm rain” processes, when the atmosphere is so unusually warm and moist that much of a storm lies below the freezing level (a hard thing to achieve in this high-altitude region). The warm-rain process can yield radar returns that are misleadingly low for the amount of rain actually being produced. The official NOAA report on the event notes that NWS staff were puzzled by a seeming contradiction: the storms extended upward to an impressive 62,000 feet, but the strongest radar returns were surprisingly weak (only about 30 dbZ). We don’t know exactly how much warm-rain processes may have boosted the rain totals, but the available data suggests there was at least some impact. If a higher-resolution radar had been available and located closer to the canyon, and if the warm-rain process had been recognized at the time, forecasters might have picked up on the gravity of the threat. As it happened, the storms were addressed with a fairly routine severe thunderstorm warning and a cursory reference to potential flooding.

Unbeknownst to virtually everyone outside the canyon, torrential rain was falling at the time, with amounts topping 12” in less than five hours toward the western (higher) end of the canyon. Before long, an enormous pulse of high water cascaded down the canyon, pushing 10-foot-wide boulders ahead of it. The flood wave demolished more than 570 structures and hundreds of vehicles--as well as much of U.S. Highway 34, the primary route into and out of the canyon for some 1800 full- and part-time residents and hundreds of visitors that night. Many tried in vain to escape in their vehicles, and a highway patrolman who drove into the canyon to investigate was among those killed. Some of the worst damage occurred near the downstream end of the canyon, where relatively little rain fell. Based on the 139 bodies recovered from the flood (several others were never found), the vast majority of victims were killed by traumatic injury, not by drowning. With communication tools so limited by today’s standards, it took many hours for the full scope of the tragedy to become evident. Not until the next day did most Coloradans find out anything about the Big Thompson disaster.

While it was shocking in its own right, the Big Thompson flood came amid a decade of major flash flood disasters. More than 200 people died in Rapid City, South Dakota, on June 9-10, 1972. A valley in West Virginia was devastated by the breach of a mine-tailings dam at Buffalo Creek on February 26, 1972, which took 125 lives (see my related post from this year). And the failure of six dams in heavy rain led to 84 deaths in Johnstown, Pennsylvania, on July 19-20, 1977. These disasters--on top of more than 300 deaths in the 1974 Super Outbreak of tornadoes--helped galvanize the world of weather research and policy, serving as motivation for the massive NWS modernization efforts of the 1990s.


Figure 4. Signs like these became a fixture in Colorado’s steep canyons following the Big Thompson Flood. Image credit: Courtesy Colorado State University Water Center.


The Big Thompson flood also played a major role in bringing social science into meteorology. Eve Gruntfest, then a graduate student at the University of Colorado Boulder, ended up studying individual and institutional responses to the flood for her master’s thesis. Gruntfest concluded that many victims in the Big Thompson might have survived had they scrambled up the hillside just a few feet rather than fruitlessly staying in their vehicles. “Before that research, people didn’t realize that you can’t ‘out-drive’ a flash flood,” Gruntfest said. Her findings led to “CLIMB TO SAFETY” signs that were deployed across the state’s canyons (see Figure 4). Gruntfest later founded the landmark WAS*IS program (Weather and Society Integrated Studies). Based at the National Center for Atmospheric Research, WAS*IS brought together hundreds of meteorologists and social scientists for mutual learning and brainstorming and to help integrate social science into meteorological research and practice.

Big Thompson, take two: What went better this time
Recurrence intervals--the amount of time one would expect to elapse between floods of a given magnitude--are a risky thing to calculate. One estimate, based on geological evidence, concluded that the streamflow in the Big Thompson Canyon in 1976 was on the order of a 10,000-year event. Incredibly, another catastrophic flood struck just 37 years later, in early September 2013, when another hydrologic disaster struck a much larger swath of the Colorado Front Range. Stretching over several days, and including both flash-flood and river-flood elements, the waters of 2013 inflicted more than $1 billion in damage, with more than 350 homes demolished and almost 500 miles of roadway damaged or destroyed—including much of U.S. Highway 34 through the Big Thompson. The peak water volume flowing through the canyon was somewhat lower than in 1976, but the flood crest was sustained over a much longer period, with multiple peaks. “I never thought I would ever see cars wrapped around trees again, and certainly not in the Big Thompson Canyon,” Gruntfest told me.


Figure 5. Vehicles lie upended in a swollen Coal Creek in Lafayette, CO, during the 2013 floods that struck large parts of Colorado. Credit: Will Von Dauster, NOAA.


Figure 6. A county road in Berthoud, CO, washed away by the September 2013 Colorado Front Range floods. Credit: Lornay Hansen/Cooperative Institute for Research in Environmental Science (CIRES).

For all the psychic and physical damage it inflicted, Colorado’s flooding of 2013 was far less deadly than the Big Thompson event of 1976, with just eight deaths directly attributable to the floods. We can chalk up the much-reduced death toll to a range of factors. Despite inconsistencies in timing and location, a number of high-resolution model runs suggested more than a day in advance that torrential rains of up to 8 inches could strike parts of northeast Colorado. The NWS stressed the potential for heavy rain in local outlooks issued a day in advance. Years of coordination among federal, state, and local agencies had already led to greatly improved awareness of the flash flood risk in Colorado’s canyons. The event itself was tracked far more closely than the 1976 storms, thanks to modern radar and satellites: a total of 78 flash flood warnings were issued by NWS offices in Denver and Pueblo.

Once the event was under way, much-improved lines of communication allowed for quick road closures and other rapid responses that helped keep many thousands of residents out of harm’s way. Cellphones, websites, and social media also helped people to obtain a wealth of information on the emerging floods (although cellphone coverage remains limited in many canyons). As with tornado warnings, flash flood warnings are now transmitted directly to newer cellphones as audio and text alarms through the Wireless Emergency Alert system, provided that the phone’s owner has not opted out of this feature. More progress is on the way: for example, NOAA’s experimental Flooded Locations And Simulated Hydrography Project (FLASH), based at the National Severe Storms Laboratory, is using high-resolution rainfall data to produce flood forecasts every five minutes with a resolution of just one kilometer (0.6 mile). FLASH is one example of a new generation of tools that are likely to revolutionize severe weather warnings in the 2020s and beyond. And starting late this year, the long-awaited GOES-R satellite debut will deliver a huge advance in the tempo and resolution of U.S. imagery available for forecasters.

These advances and many others should make us grateful, yet never complacent. In 2004, it was easy to think that the United States might never again see a hurricane kill many hundreds of citizens. Just a year later, Hurricane Katrina proved that assumption horribly wrong. In 2010, one might have thought that the days of a single tornado killing more than 100 people, or an outbreak killing several hundred, were long gone--and then we saw the Super Outbreak of April 2011 and the horrific Joplin tornado of May 2011. Likewise, a flash flood striking in the wrong place at the wrong time could still wreak a catastrophic toll. On an individual level, anyone who drives around a barrier and into high water during a flash flood is taking what could be a deadly personal risk. More than 100 of the deaths in the Big Thompson canyon were vehicle-related. Doing what we can to spread awareness of water’s deadly power would be a fitting tribute to those whose lives were lost in the Big Thompson Canyon.

We'll have an update on Saturday afternoon on the two systems in the tropical Atlantic that have some potential for development over the next few days, Invest 96L and Invest 97L. See also this morning's post from Jeff Masters on the twin systems.

Bob Henson

More background:
Big Thompson Canyon Flash Flood of July 31-August 1, 1976 (NOAA Natural Disaster Survey Report, October 1976)
The Record Front Range and Eastern Colorado Floods of September 11-17, 2013 (NWS Service Assessment, June 2014)
A Deadly Flood That Helped Improve Weather Forecasting [NOAA web feature, July 29, 2016]


Figure 7. Nezette Rydell, meteorologist in charge at the NWS Denver/Boulder office, addresses a gathering held near Loveland, CO, on Friday, July 29, 2016, in honor of the 40th anniversary of the Big Thompson Flood of 1976. Organized by NOAA and the U.S. Geological Service, the event was hosted by the Sylvan Dale Guest Ranch, located at the base of the canyon. Ranch co-owner Susan Jessup, who was an eyewitness to both the 1976 and 2013 floods, also spoke, along with experts and policymakers from the State of Colorado, UCAR, USGS, and Larimer County, CO. I delivered opening remarks and introduced speakers. Image credit: Bob Henson.