Collaborative project casts
Princeton one of 17 universities and research centers studying upwelling events and how they influence biological production
by Sara Peters
Some really good science happens at Phil's Fish Market.
Okay not at Phil's, exactly, but in an office on the floor
above the fish market is the base of operations for the Autonomous
Ocean Sampling Network II (AOSN II). Plenty of good science
Naomi Leonard, professor in the Department of Mechanical
and Aerospace Engineering, leads the Princeton team
for the Autonomous Ocean Sampling Network II. Here,
Professor Leonard is checking the autonomous underwater
vehicles prior to their release in Monterey Bay this
by Sara Peters
This small office in Moss Landing, Calif., belongs to the
Monterey Bay Aquarium Research Institute (MBARI), whose main
office is right down the street. The AOSN project, sponsored
by the Office of Naval Research and coordinated by MBARI,
is a collaborative project involving pros from 17 universities
and research centers, including a group from Princeton.
The central objective is to gain the ability to predict marine upwelling events and see how they influence biological production. To do so, they conduct adaptive sampling, gathering relevant physical and biological data about marine and atmospheric conditions.
The sampling tools include remote sensors employed via aircraft and satellites and a network of autonomous underwater vehicles (AUVs) equipped with a regiment of scientific sampling instruments. Several models designed to predict ocean phenomena also guide the sampling plans.
Naomi Ehrich Leonard '85, professor in the Department
of Mechanical and Aerospace Engineering (MAE), leads the Princeton
team, which is in charge of the glider dynamics, control,
and adaptive sampling.
Other members of the Princeton team are--also from MAE--Assistant
Professor Clancy Rowley '95, Research Staff Member
Ralf Bachmayer, graduate students Pradeep Bhatta,
Edward Fiorelli, Joshua Graver, and Derek
Paley, and undergraduate Spring Berman '05.
This summer, the AOSN II team gathered in Moss Landing to conduct fieldwork. I was fortunate enough to tag along for a few days to watch and learn about dynamics--dynamics of gliders, the ocean, and large, collaborative research teams.
Long, winding road
base of operations for the Autonomous Ocean Sampling
Network II is in an office above Phil's Fish Market.
The road to the AOSN office smells like a mixture of the salty sea air to the west and the strawberry farms to the east. Off California Highway One, I take a right-hand turn onto a one-lane dusty road, pass a sign saying "Welcome to Moss Landing: the Heart of Monterey Bay," drive carefully over a questionably sturdy one-lane bridge, and I'm almost there.
Moss Landing claims to have a population of 500, but that sounds like a creative overestimation. The town consists of six elements, and six elements only: a post office, a cemetery, a marina, restaurants, antique shops, and marine science research facilities.
With the exception of the MBARI headquarters a few doors down from Phil's Fish Market/AOSN headquarters, the buildings are somewhat crumbly and weather-beaten. Nonetheless, there is an inviting, bustling feel to the place.
At the marina across the street, I watch a fisherman poke at a robustious sea lion until he slides off the catwalk. While I, the East Coaster, am both thrilled and mystified by the cacophony of sea lions barking, the Monterey Bay fishermen see them only as dangerous nuisances.
AOSN headquarters is a small office full of scientists in shorts and sandals, plenty of maps, several computers, a few projector screens, and a dry-erase board liberally daubed with red equations.
Here the AOSN Executive Team--13 people representing different aspects of the research--meet daily to discuss progress and plan for the upcoming days. On Saturday morning, Aug. 2, they met to brief one another on some relevant details of all branches of the project.
Yi Chao from the NASA Jet Propulsion Laboratory, Allan Robinson of Harvard, Steve Haddock from MBARI, Russ Davis from the Scripps Institution of Oceanography, and Professor Leonard presented important information about atmospheric modeling, ocean modeling, marine ecology, AUV dynamics, and adaptive sampling control theory.
The ringleader of this group is Jim Bellingham, the AOSN project leader from MBARI, a tall man with a boyish face who sits cross-legged on a table. He does an admirable job of summing up and pulling together the presentations of the engineers, oceanographers, and ecologists. Project details begin to become clear.
The project focuses on coastal upwelling in Monterey Bay. The surface of the ocean is heated by the sun, creating a stratified effect, with a warm surface and a cold bottom.
graduate students Derek Paley and Edward Fiorelli are
at the marina in Monterey Bay.
Upwelling occurs when winds blow the warm surface layer offshore, allowing the cold water to rise from the bottom.
This bottom water brings nutrient-rich sediment to the surface. Phytoplankton and zooplankton feed on the sediment, creating blooms, which draw other wildlife to the area to feed. Thus, these upwelling sites are particularly rich with information, seemingly begging for scientists to harvest discoveries from them.
Upwelling events have been difficult to predict, because they rely upon interactions between changeable atmospheric and marine climate systems. Understanding them more thoroughly would help ecologists better manage aquatic resources, estimate populations of endangered species, and predict the effects of climate change on ocean dynamics and marine life. Understanding upwelling also may have implications in military surveillance.
The propeller-driven AUVs and gliders collect data that will bring the ecologists closer to these goals. They gather information about temperature, salinity, bioluminescence, zooplankton levels, phytoplankton levels, etc.
The AUVs will be used in different ways to collect this data over a variety of depths. The propeller-driven AUVs are faster and can travel in a more direct fashion than the gliders, which are buoyancy-driven, relatively slow, and largely influenced by currents. The gliders, however, are low-power, and can thus function autonomously for far longer than the AUVs.
There are two different kinds of gliders that the AOSN team is using. Both are propelled by changes in buoyancy, using a ballast system to suck in or expel water to sink or rise. Both are 1.5 meters long (~4.9 feet), sort of torpedo-shaped, with stationary wings that help them steer and glide. In one case the gliders use a rudder to steer. In the other case, the gliders move an internal battery pack around to roll, bank, and turn.
Both gliders are powered by an electric battery. The glider that uses a rudder has an endurance of 10 to 30 days, depending in part on how much data is transmitted, and a maximum depth of 200 meters (~ 656 feet), and will be used mainly for coastal surveys. The glider that does not use a rudder has an endurance that exceeds 30 days and a maximum depth of 1,000 meters (~ 3,281 feet), and will be used for deeper surveys.
gliders run simulations in preparation for launch.
The ocean is a big place and there are lots of possible points at which to collect data, making it very difficult to get lucky enough to find an upwelling event or another data-rich site.
Thus, the AOSN team seeks to narrow it down a bit through a process called adaptive sampling. This is where Professor Leonard's group comes in.
At one level, adaptive sampling means that the data collected on Monday will influence the sampling plan for Tuesday, and so on. Professor Leonard's group proposes adaptive sampling at a second level. The concept is to enable the research tools to adapt in near real-time--every couple of hours--updating and changing plans on their own as a way to complement the the daily human decision making and to maximize the overall gain from adaptive sampling. In this summer's fieldwork, key developments for real-time adaptive sampling are being tested.
The sampling plan will be updated with three goals in mind:
* The sampling should provide the best data possible to be used in scientific hypotheses;
* the data should improve the predictive capabilities of the model; and
* the sampling should make the best use of the tools--in this case, the gliders--and their unique endurance and maneuverability.
The Princeton team will be working mainly with the gliders owned by the lab of Dave Fratantoni, an associate scientist at Woods Hole Oceanographic Institution. They are electric-powered, shallow-diving models.
The Princeton team, in collaboration with dynamic systems modelers at Caltech and the team from Woods Hole, are using the gliders in small, coordinated mini-fleets. They send out two or three gliders together, in a formation fixed around an imaginary axis. They give the gliders instructions for maintaining a certain (fixed or varying) distance from the axis and each other. They plot a course for the fleet and let their feedback control algorithms compute the detailed plans for the gliders piece by piece, directing them toward new way-points about every two hours.
"It's like putting imaginary springs in between each glider and the center of this formation," Professor Leonard said. "As we move the center around, it tugs the gliders around with it."
Long time gone
The John H. Martin awaits to transport scientists out
into Monterey Bay.
Once the gliders are launched in the bay, they'll be gone for several weeks, slowly cruising around on the mission that the scientists have laid out for them. Approximately every two hours, they'll complete a segment of their mission, surface, raise their tails in the air, connect to the team's computer, transmit a GPS reading and all their data, receive new waypoints, then continue on their way.
Using the gliders in this coordinated manner will provide data more useful than that gathered by a collection of independent gliders. For example, data from the coordinated gliders will be helpful in identifying gradients in temperature, salinity, or other physical or biological variables, and may thus point to the front of an upwelling event. Then the AOSN team, or ultimately the gliders themselves, can adapt the gliders' missions to move closer to this front.
Professor Leonard's team based their algorithms on the behaviors of fish schools (EQuad News, Winter 2002-2003). Schools exhibit extraordinary and highly adaptive collective behaviors. Foraging for food is an important example of an activity that is done much better by a school than by independent individuals. A fish in a school relies on tracking its neighbors as much as following whatever food signal it can sense individually. In this way, the individuals make use of each other, and the school as a whole finds the best available lunch.
"We're trying to mimic what fish do," Professor Leonard said. "Only, instead of looking for sites rich with food, our gliders are looking for sites rich with information."
Every day after gathering data the AOSN executive group, of which Professor Leonard is a member, go into "War Room" mode. They review all of the data gathered as well as the predictions of the models, then decide where, how, and what to sample the next day. These meetings go on for several hours and can be spiked with debates, which can sometimes only be quelled by lunch from Phil's.
dockside at Monterey Bay are, from left, front row,
MAE graduate students Pradeep Bhatta and Edward Fiorelli,
and MAE research staff member Ralf Bachmayer. In the
back is Christina Courcier from WHOI and MAE undergraduate
student Spring Berman.
While the executive team listened, discussed, planned, and
debated Saturday morning, graduate student Edward Fiorelli
ran practice simulations with the three gliders, which actually
sat in the MBARI parking lot down the street from Mr. Fiorelli
and his laptop.
Although at the moment the gliders aren't actually going anywhere, their ballast mechanisms are working, their acoustic depth detectors pinging, and as far as they're concerned, they're on a real mission.
The simulation program even manufactures false obstacles and current differences so that the unfettered gliders don't all "arrive" simultaneously.
Mr. Fiorelli gives the gliders their missions, supplying them with several way-points to hit. Gliders number 12, 13, and 7 go off on their mission (let's call that Mission Alpha). They surface and transmit their GPS point and data. Mr. Fiorelli gives them their next mission (Mission Beta), and off they go again. While they're off on Beta, Mr. Fiorelli will enter all their data into a complex algorithm, generate Mission Gamma for them, and have it ready when they surface after Mission Beta.
The catch, however, is that Mr. Fiorelli can't initiate the algorithms until all three gliders have completed Mission Alpha and resurfaced. This simulation shows Mr. Fiorelli just how tricky this can be.
Gliders 12 and 13 both surfaced, transmitted data, and dove down again to begin Mission Delta, while Mr. Fiorelli waited for glider 7...and waited...and waited. Sipping his Diet Vanilla CokeT through tense lips, he fixed both sore eyes on the computer screen, clearly trying to will glider 7 to return, and fast.
When he finally sees the message that glider 7 has returned, he sends it off on Mission Delta, and immediately gets to work on Mission Gamma, hoping to have it ready by the time gliders 12 and 13 have surfaced again.
During this hectic time, Mr. Fiorelli's understanding fiancée calls. He responds "Hi-I-can't-talk-I-can't-talk-now-I-really-can't-talk," and hangs up. He completes the mission plan in time, but learns an important lesson that he will remember on Monday when the gliders are actually in the water.
Leonard and John Lund plan a safe and efficient way
of launching the next glider.
The fog rolls in thick over Moss Landing, and I step aboard the John H. Martin, a 56-foot research vessel that can hit speeds of up to 23 knots. The John Martin is captained by Lee Bradford, a remarkably friendly, welcoming, and knowledgeable skipper.
After the captain is finished giving us the safety precautions, he proceeds to educate us about whales, Pacific weather patterns, and the migration habits of sea turtles. Also aboard are Professor Leonard, and John Lund and Christina Courcier, engineering assistants from Woods Hole.
The cabin of the Martin is Mr. Lund's and Ms. Courcier's base of operations. Their laptops can track and communicate with the gliders, reading information about their health and power. They can also communicate with Dr. Fratantoni's computer back on shore.
After launching the gliders and testing them for any flaws, Mr. Lund and Ms. Courcier will hand control of them back over to Dr. Fratantoni.
The Martin speeds to our first mark, about 4.5 miles off shore, as the fog burns away and reveals a startling blue sky. Mr. Lund and Ms. Courcier and Professor Leonard launch gliders 12 and 13. Each glider is then wheeled over to the stern on a little lorry, unbuckled, and given wings.
Then, as one person lifts the back end of the lorry up, the other person guides it down into the water. With little assistance, the glider slips into the water and resurfaces a few meters away, tail in the air.
"You really don't have to help these guys into the water, much," Professor Leonard said. "They're about a hundred pounds, so they just slide in pretty quickly."
After launching the gliders, Mr. Lund and Ms. Courcier run a series of tests to make sure that each one is operating satisfactorily.
They examine the gamut of details and send them on two trial runs, one of which is a 40-minute mission. During this mission, the John Martin idles and everyone aboard relaxes and enjoys the sun.
Graver, graduate student in the Department of Mechanical
and Aerospace Engineering, works at AOSN II headquarters
analyzing the performance of the gliders.
Suddenly Mr. Lund rushes out of the cabin, hastening to the bow, camera in hand saying, "We're surrounded by dolphins!"
A large number of Pacific White-Sided Dolphins, a pod of Risso's Dolphins, as well as a handful of Northern Right Whale Dolphins frolic nearby. They make a circle around the boat, while everyone aboard turns into six-year-olds--pointing, squealing, gasping in delight, stumbling from bow to stern, from port to starboard to get a closer look.
The Risso's Dolphins are about 3 meters (~10 feet) long, and light gray, covered with little white scratchy scars. I later learn that these scars are identifying characteristics of Risso's Dolphins, who scratch each other with their teeth while playing.
The Right Whale dolphins aren't immediately recognizable as dolphins, since they lack a dorsal fin. Yet they are incredibly vivacious, leaping entirely out of the water, flashing their sleek, black bodies through the air. The music of blowholes makes a calypso rhythm around us, each beat about two-thirds puff, one-third pop.
Unfortunately the dolphins are not as excited to see us as we are to see them, and they move on, jumping, fluking, and diving away.
Back to work
Thomas from MBARI and Joshua Graver prepare to launch
Eventually gliders 12 and 13 surface from their test mission, momentarily neglected during the dolphin craze. The tests show that both gliders are healthy, and are, therefore, sent off on their real mission.
As I watch them sink beneath the surface and dive to unseen depths, I suspect that I'm not the only one feeling a pang of worry--a protective feeling over the small machines that will be braving the briny depths alone for several weeks.
The captain spurs the John Martin toward our next launch point, about eight miles offshore. As the boat races along, before we plunge into a thick wall of fog, we catch a glimpse of a blue whale off the port-stern. Although we're only a few miles offshore, in places Monterey Bay is deep enough to be inviting to the whales, which can exceed 100 feet in length.
Glider 7 is relaunched, and once again the team awaits his completion of the test runs.
Mr. Lund rested against a stanchion, sighed expansively, and said to me, "You know, there really is a lot of stress involved in these launchings."
Coupling this comment with the relaxed manner of the crew and the ease of each launching, I'm not sure if I should take him seriously. But he assured me, "This is really an A-plus day."
He went on to explain that there are "hundreds of details" to remember to check carefully before launching each glider. They remain ever diligent on the precautions, because they are still haunted by events that happened in the Caribbean Sea back in January.
For unknown reasons, glider 3 was lost, and the team never retrieved him. They are now adamant that no other gliders share the fate of glider 3.
After another lengthy wait, the computer noted that glider 7 has surfaced after its test mission, but the fog cloaks its location from us.
"Can anyone see her out there?" Ms. Courcier asked, as everyone peers into the gray for a clue.
Captain Bradford slowly and cautiously poked around looking for a glimpse of the glider, making sure not to come upon it too quickly. If a million-dollar boat runs over a $60,000 glider, nobody wins.
"I got him!" Captain Bradford said, pointing to a misty yellow image off the port-bow.
The difficulty created by the fog brings up a good question. How well will the gliders be able to communicate? Will fog, rain, or waves disrupt the signal? The sea was flat this Monday morning, but the captain said he's seen 30-foot swells in these waters. Only time will tell what problems will arise.
All good things...
As the days and weeks passed on, the AOSN project encountered and attacked many obstacles. Gliders would be "rescued" by boaters. The gliders would prove unable to simultaneously receive way-points and upload data.
Yet when fieldwork officially ended Sept. 8, no one could deny that a great deal of extraordinary data had been collected. Now that data will be perused for discovery.
The Princeton team members were altogether pleased to be a part of the project.
"As engineers, we're always trying to solve a problem," Mr.
Fiorelli said. "It's great to be here, talking to and working
with the people who actually have the problem."
For more information about the Autonomous Ocean Sampling
Network II, see: www.princeton.edu/~dcsl/aosn
Sara Peters is communications assistant in the School of Engineering and Applied Science.
next story ]
of page ]