HOWTO Predict Surf at Mavericks [Hint: Data]

mark sponsler.jpg

Surfers have been catching massive waves at Pillar Point in Half Moon Bay, CA since the 1960s. The first legitimate contest was held in 1999-2000, but what most of us know as Mavericks didn't launch until 2004 with the help of big wave rider Jeff Clark. Unlike the Kentucky Derby or March Madness, Mavericks doesn't happen every year. It was a no-go in 2007 and again this past season.


The problem isn't a lack of big swells, but whether the perfect one will rock Pillar Point during the contest waiting period (Jan. 1 - March 31 in 2009) -- and then predicting the if and when so everyone can get in place. The 24 surfers who compete at Mavericks are given 24-hours notice to show up. Extending the waiting period and tethering the surfers to Mavericks wouldn't be fair since they make their living traveling to contests all over the world, says Keir Beadling, who co-founded Mavericks Surf Ventures with Clark. Plus, the arrival of late spring welcomes gray whale migration and seal pups, which results in crowded waters.

So how do you forecast one of the most celebrated big wave contests in the world?

"It's no longer a secret where you have three puffs of smoke announcing who the next Pope is gonna be," says Beadling, "You definitely couldn't hold a contest of this magnitude -- 50,000 spectators in person and another 1+ million on webcast -- and get mobilized in time if it weren't for all the technology."

The short answer: datahead surfer Mark Sponsler (surfing Mavericks above).

Find out how a former quality engineer and product manager at Kennedy Space Center moved to the west coast, started dropping in on big waves, tuned into data, founded, and currently crunches more than 2 terabytes of swell data per year.

photo by Doug Acton, provided by Mark Sponsler

sponsler headshot.jpg

The Road to Mavericks

In the early 1980s, Florida native Sponsler was surfing and working at Kennedy Space Center, where he did a 15-year stint. A self-described "bithead," he started to get weather curious. Back then, though, hobbyists and surfers had a heck of a time obtaining any real, tangible data.

"When I first started trying to put stuff together, there was no Internet," he says, "There were no publicly-available [wave] models. You'd get in your car and drive to the National Weather Service office, and try to make sense of some charts they had posted on a wall."

Around 1984, NOAA began installing some of its first weather buoys. A big leap forward for generating potentially-solid swell data. Yet, the best Sponsler could do -- aside from driving back to that office -- was to listen to NOAA weather radio. And even then, he'd only get a rough idea of what was going on to then try to extrapolate the oceanic tidbits in surf terms.

By the time Sponsler's wife got a job opportunity in San Francisco, he had already caught the big wave bug paddling out in Hawaii. Looking for a change, they moved out west and never looked back. Soon after, he'd not only befriended Mavericks legend Jeff Clark and tackled Mavericks, but Sponslor became a "power user" of the early 90s Web. NOAA and CDIP continued to anchor more buoys updated with fancier sensors. New and improved weather satellites launched. Universities and institutions started to create and freely distribute waveand weather models. And with that great, big data dump, of course, began Sponsler's efforts to mine as much as possible.

In the 90s, Sponler started forecasting surf Mavericks via email blast (he still takes daily calls from any one of the 30 big wavers he calls the "traveling gypsies floating around the planet chasing surf"). As the list ballooned in size, a web site started to make more sense. Plus, he wasn't satisfied with the way much of the disparate data were being presented online. He started coding.

The Next Wave of DIY Forecasting

Launched in 1998, Stormsurf operates today with just three par-time employees, not including Sponsler, whose day job is as a tech product manager at a major health care provider.

His back-end, he admits, isn't state of the art, but it gets the job done. He's running four, 3.0 ghtz single core boxes each with 1 GB of RAM. Ancient by today's standards -- and he knows an upgrade is due -- but for the past five years his application servers have run just fine maxed out at 100% CPU, at probably 50% of every minute they're running.

The total throughput -- the raw GRIB files and various other data he downloads specific to wave/weather models and flash output -- is roughly 1.5 GB per model run four times a day. That's 6GB a day or 2.16 terabytes every year to forecast worldwide. Not jaw-droppingly-impressive if you're a Google or Microsoft, but a far cry from driving to the NWS office to stare at 2D charts tacked up to a wall.

What are Stormsurf's servers crunching?

The main components are wind speeds, wave models and the Jason-1 satellite. The wind speeds are provided by satellites that pass over twice a day, offering almost 100% oceanic coverage. The satellites, including NASA's QuikSCAT, can predict wind speed to within 2-3 knots.

wind sat.jpg

The next step is to input the raw figures into several of the aforementioned wave models, which extrapolate a predicted wave height. (I'm simplifying the models which take into account huge swaths of data like seafloor mapping)

The last main component is Jason-1, which helped generate its first image in 2002. As the satellite polar orbits the Earth multiple times a day, it beams down radar into the center of a storm to calculate the wave height -- down to an accuracy of 6 inches.

From there, as a swell approaches, Sponsler turns again to those buoys to provide secondary confirmation. NOAA has 991 buoys worldwide, but for Mavericks the most vital ones are those off the coast (duh). Armed with solar panels, computers, accelerometers, and directional sensors, the buoys measure the sea height, swell height, heave and throw of the ocean, and the energy distribution across entire frequency spectrum (from 3 seconds up to 40 seconds). If there are multiple swell trains hitting at the same time, the spectrum will help Sponsler identity it.

The buoys uplink continuously to satellites so the data can be downloaded to a central processing center in Maryland, which then chews everything in real-time so it can post to the Web. Those 1s and 0s are distributed in multiple formats, including the raw ASCII text flat files Stormsurf chops into bite-size chunks to run in parallel, and then construct its own buoy reports, which detail current conditions, provide historical surf height and chart out barometric, air and water temperature over time (for surfers or aspiring weatherheads who want to learn).

forecast buoy.jpg

Passing buoy after buoy, too, the swell continues to disseminate new figures. With buoys positioned miles out and then right off shore (#46012 for Mavericks), Sponsler can tune his forecasts up to 24 hours in advance (a short-term forecast can be delivered 2-3 weeks out; we'll get into long-term below).

With swell forecasting, one of the biggest unknowns is the wind, especially at Mavericks. Typically, the best surf at Mavericks occur with a dead calm. A light onshore or light northern breeze up to 15-20 knots is doable, but less ideal. Offshore, is another story:

"The waves moving so fast, jacking so fast that any offshore wind basically sends you airborne. You never make it to the bottom of the wave," says Sponsler, "Most people want 4' and offshore, but if it's offshore at Mavericks, you're just gonna die."

Long-Term Voodoo: Will Mavericks Happen in 2010?

In a word: probably. This past winter, Sponsler says he knew the contest was a longshot by early January. Some huge swells were hitting the break in October/November. By the time the contest window opened, though, the seas had calmed dramatically and the forecast was grim. "It was like a lake."

The big issue: El Niño & La Niña. 2009 saw what Sponsler and other Mavericks surfers are hoping is the end of a 3-year La Niña.

During an El Niño year, warm water builds up off the coast of South America and enhances the Jet Stream in the North Pacific, therefore instigating big winter storms. During La Niña, you get the opposite -- colder water off the coast, which suppresses storm development. For Mavericks, you also want big storms, strong low pressure developing on the International Date Line pushing in towards the west coast of Alaska, but not too close. El Niño is good for that, but once you hit the core of winter, the storms can get driven right into the coast, which isn't good for surf.

Needless to say, the list of atmospheric variables is vast and nature is fickle. For long-term forecasting (3-6 months out), Sponsler says the data only go so far, too. He examines everything obsessively, but for the time being, can arrive only at a probability for the number of storms, not some decisive moment for a certain storm hitting a certain beach. Thus, he calls the long-term forecast the "domain of artistry, voodoo and witch doctor sort of stuff."

Sponsler says we're right on the cusp of an El Niño year. Just another few weeks, and he should know for sure. Provided the current patterns hold, including the relative activity of the Madden Julian Oscillation (MJO), the ducks will be in line to potentially serve up another season of contest-worthy. Either way, he's never quick to pull the trigger either direction.

"This is not an exact science," Sponsler reminds me, "Whenever you go and declare it dead or alive, then something will always happen. It is nature; it does it's own thing."

image by ISIphotos, with permission from Mark Sponsler

This entry was posted in Uncategorized. Bookmark the permalink.

2 Responses to HOWTO Predict Surf at Mavericks [Hint: Data]

  1. ESQ says:

    Good article. I still keep my ancient Rat Shack weather radio cube on my desk as a reminder of those days when you’d get at best a 12 hour advance warning of any swell hitting the offshore buoys. The first time I surfed Mavs was during the massive Christmas 94 swell; the next morning it sadly claimed the life of Mark Foo. I gave up on the place shortly after due to the media circus intruding upon nature’s calm beauty. Getting buzzed by helicopters and jetskis ruined the place in my eyes, but no matter – there are dozens of spots along this stretch of coast that can easily accommodate 18’+ swells without the distractions of the sideshow.

    Nowadays, it is possible to predict within an hour or two the timing of swell arrival to a particular beach and using Stormsurf’s collection of data I’ve been able to successfully predict incoming swells a full two weeks prior to arrival. I’d like to offer my thanks to Mark.

    Too bad Aggroville isn’t still around… that would have made an interesting wrap-up to this series of pieces! >=)



  2. Anonymous says:

    I’ve been wondering about this for a while now, Steven. Though you genuinely seem interested in technology and gadgets, your posts are extremely sports-centric. Though better than most sports writing, a lot of us pasty-faced sedentary nerds find sports to be anathema.

    So why are you doing this on a gadget blog? Trying to get us off our fat butts? It won’t work, you know. 😉

Leave a Reply

Your email address will not be published. Required fields are marked *


More BB

Boing Boing Video

Flickr Pool




Displays ads via FM Tech

RSS and Email

This work is licensed under a Creative Commons License permitting non-commercial sharing with attribution. Boing Boing is a trademark of Happy Mutants LLC in the United States and other countries.

FM Tech