Stories tagged models

Our very own JGordon drops some knowledge...

Jun
03
2010

Sophisticated forecast modeling tools developed at the Center for Coastal Margin Observation & Prediction (CMOP) were recently used to assist in the rescue of a disabled underwater glider.

CMOP researchers spent two days using a particle-tracking model to predict where and when their glider, nicknamed “Phoebe,” would drift ashore. This helped researchers understand how much time they had to stage a recovery operation.

Rescue planning: This is an example of the particle tracking model used to predict the direction the glider would drift. Red represents the glider GPS signals. Green represents the particle tracking forecast.
Rescue planning: This is an example of the particle tracking model used to predict the direction the glider would drift. Red represents the glider GPS signals. Green represents the particle tracking forecast.Courtesy CMOP

“Once Phoebe became a drifting glider, we treated her as a major piece of scientific instrumentation at risk and an opportunity to test our computer models in a sea emergency,” says Antonio Baptista, director of CMOP. “The forecasting system used for Phoebe is the same that we are currently transferring to the U.S. Coast Guard and NOAA (National Oceanic and Atmospheric Administration) for inclusion in their respective operational and emergency response systems.”

Phoebe is a bright yellow glider that moves through the water, gathering information, and sending satellite signals back to land each time she surfaces. She was sent out on her first mission of the year on April 16, 2010 to collect data in the waters off the Washington coast as a collaborative research effort with the Quinault Indian Nation.

Five days into her mission, Phoebe stopped communicating.

Katie Rathmell and Michael Wilkin, members of the CMOP field team in Astoria, Oregon, waited and hoped to receive a signal from her. Hours passed and still no signal. Then almost 24 hours later, Phoebe called home. She had surfaced and transmitted a GPS signal of her current location.

“We reviewed the files she sent and determined that she had gotten stuck at 8.4 meters below the surface and was unable to come up to the surface,” says Rathmell.

The team theorized that Phoebe got tangled in a kelp bed. After a pre-programmed period of time, she jettisoned her emergency ballast weight, which gave her enough buoyancy to escape the entanglement and surface. But having dropped the ballast weight meant she could no longer dive or maneuver. Phoebe was adrift in the ocean.

Rathmell and Wilkin started talking about how to stage a rescue. The challenge was the gale force winds offshore were making the seas too rough for ships to get out of the harbor. The team would have to wait until weather conditions improved.

Even though Phoebe was disabled, she was capable of transmitting a GPS signal every 30 minutes. This allowed the team to track her location. She was drifting south and getting closer to the Columbia River plume. They were concerned she might get caught in the incoming tide. This would pull her into the river and possibly crash her into the jetty. Currents and winds could also push her onto the beach and the surf could break up the glider. The problem was the team was unsure which direction she would drift.

That is when they made the decision to use CMOP’s modeling tools to help narrow down Phoebe’s potential drifting trajectories, possible threats, and windows of time for a recovery operation.

“The team hoped the weather would break in time for a successful recovery. The models helped predict how much time they had to recover Phoebe,” says Paul Turner, senior research programmer.

The data for the particle tracking comes from the forecast models that CMOP runs on a continuous basis. Turner ran simulations for two days using the winds, currents and tides to predict where Phoebe might end up. He generated graphs that predicted drifting directions in one, two, three and four hour intervals.

“Paul Turner did a very good job of getting the modeling and drifter prediction tools working in a fashion that allowed the data to be useful for us,” says Wilkin.

The forecast model showed that time was running out for Phoebe. The prevailing winds and currents were pushing her closer to shore. It was imperative to rescue her soon.

For several days, the conditions were too dangerous to cross the Columbia Bar and get the glider safely aboard a ship. Then around 10:30 on Sunday morning, the research team received word there was a break in the weather and Captain Dan Schenk from Sea Breeze Charters in Ilwaco, Washington would take them out.

Rathmell and Wilkin boarded the “Nauti-Lady” and took a rough ride over the Columbia Bar en route to Phoebe’s last known location.

Finding Phoebe was a challenge. This time of year there are crab traps set out in the ocean and many of their floats are the same color as Phoebe. The team would spot something on the surface of the water that might be Phoebe but it turned out to be something else.

Then they spotted her tangled up in crab lines and floats. “She was surrounded by kelp, plastic, beer bottles, and all sorts of trash,” says Rathmell. They were successful in getting hold of her, removing the crab lines, and pulling her aboard the ship. The team safely returned Phoebe to shore.

Found!: "Phoebe" was found tangled in kelp and crab lines off the Washington coast.
Found!: "Phoebe" was found tangled in kelp and crab lines off the Washington coast.Courtesy CMOP

“The successful rescue of Phoebe, under difficult sea conditions, is a credit to the team work among the Astoria field team, boat operators, modelers and programmers,” says Baptista. “CMOP’s oceanographic knowledge, field observations, computer models, and cyber infrastructure all came together to allow people to make the right decisions at the right time.”

CMOP will use the lessons learned from Phoebe’s rescue operation to further improve their scientific and engineering infrastructure.

May
26
2010

I enjoy working with our team to develop on-line interactive education activities. We are in the final testing of whose goal is to teach about the balance of global water, land coverage, atmosphere and cloudiness required to create a "liveable planet". If you want to play with it and give us feedback - here is the link:

http://profhorn.meteor.wisc.edu/wxwise/climate/makeplanet.html
The goal is to make a habitable planet by adding enough water, atmosphere and clouds to reach a global average temperature of about 15°C (59°F). You can mix and match, add or remove.

* Drag (and drop) an item from the right side to the left to add that element
* Drag (and drop) from the left are back to the right to remove that element
* HINT You must put at least 3 clouds by the planet!!

There is a timer to see how fast you can make the planet livable.

May
05
2009

So this is how it gets snotty: I thought it would be more subtle.
So this is how it gets snotty: I thought it would be more subtle.Courtesy The Rapscallion
Buzzketeers—quick, for your own safety, de-cash yourself now! Come on!

There’s a flu pandemic brewing, and y’all are just sitting there, lining your pockets with little green rags that carry as much disease as monetary value. So, please, for health’s sake, empty your wallets of cash, stuff those plague bills into manila envelopes, and send them to JGordon, The Science Museum of Minnesota, The Western Hemisphere (I don’t remember the exact address here, but I’m sure the postal service can figure out the details). I’m willing to sacrifice my health—for you—and disinfect your cash money. None of that money will be returned (please, I’m not made of postage), but I’m sure that the knowledge that you have done your part to slow the pandemic is compensation enough.

(This message goes doubly for the younger, or “lil,” Buzzketeers out there. I understand that you have less money, but your immature immune systems are particularly vulnerable to viral infection. Trust me on this one, and send those piggybanks my way.)

Do you not believe me? I think I’ve proven my scientific reliability time and time again… but here, a real link to a real story: cash is a pretty good way to transmit the influenza virus.

See, according to researchers at the Central Laboratory of Virology in Switzerland, a lonely lil’ flu virus on a fresh and clean piece of paper money can only live for about an hour. Unfortunately, viruses are rarely lonely, and our cash money is not very clean. So the researchers observed how long a virus could live on cash when it was mixed with a little nasal mucus (we’ll call it “snot”).

Under a cozy little film of mucus, the flu viruses were much hardier. Some strains of influenza lived as long as 17 days on the bill. And while the scientists didn’t test the exact strain of swine flu that we’re dealing with now, they did see how long other varieties of the H1N1 virus would last. H1N1 influenza remained viable (it could still infect someone) on the cash for up to 10 days.

It turns out that about 94 percent of dollar bills may carry pathogens (germs, viruses, etc). So let me shoulder this burden of worry, and let’s see that cash.

On to part 2 of this post…

Researchers at Northwestern University and Indiana University are also using money to study the spread of disease, but in a totally different way. It’s a little more complicated, and a little cooler.

Even if cash is totally clean, and doesn’t act as a vector for passing the flu, a cash transaction represents a face-to-face exchange between multiple people, the sort of encounter that could result in the flu virus being passed on.

And, hey, look: a project designed to follow the journey of individual dollar bills across the country.

The Northwestern and Indiana scientists took data from this bill-tracing project (called Where’s George?, and combined it with information on air traffic and commuter traffic patterns for the country to make a mathematical model of how people move and interact in the US. They then added information about the H1N1 swine flu into the system—the locations of confirmed cases, rates of infection, the time it takes to become contagious… that sort of thing. With all the variables taken into consideration, the model becomes incredibly complex—so complex that it takes a supercomputer about ten hours to make all the calculations, and come up with a forecast of where future infections will be, and how many of them we might expect.

But the model seems to work. Both universities, working independently, came up with strikingly similar models, and when predictions from the models were compared to real-life figures they matched up pretty well.

So far the models’ estimates have been slightly lower than actual infections, but they predict that there will be about 2,000 cases of the swine flu in the United States by the end of May, with most of those occurring in New York, Miami, Los Angeles and Houston. The researchers didn’t run any predictions beyond about a month, however. The flu, they say, as well as public response to it, are so unpredictable that using the models to look too far ahead doesn’t work. (The flu could mutate into something more virulent, or the government could do something drastic to control its spread, or, you know, we could get invaded by space aliens.)

(Liza, by the way, talked about these models a little last week.)

How about that? Money follows us around, viruses follow us around, viruses follow money around, and we trade all of it.

Here’s the link to Northwestern’s flu model

Here’s the link to Indiana’s model.