An Ocean of Data

Alaska has 33,900 miles of shoreline, including this scenic section of Glacier Bay. PolArctic has designed an engine to measure and chart nearshore depths on the changing Alaska coastline.

Lauren Decker, M.S. ’09, a Yup’ik Alaska Native, is chief scientific officer for PolArctic, an ocean and data science company that uses artificial intelligence and machine learning—informed by Indigenous knowledge—to help businesses in the Arctic operate safely, profitably, and sustainably.

By Alexander Castro

Lauren Decker loved bugs. Butterflies, beetles, ants, moths, cockroaches and other grubs. She loved them so much that she thought they’d be her life’s work. She told her first-grade teacher she’d be an entomologist one day.

Childhood fantasies tend to lack in specifics, however, and Decker’s dreams of entomology were slowly squelched. By the time she arrived at the University of Washington as an undergraduate, Decker realized the sheer variety of sciences she could study. Decker remembers thinking, “I don’t know what kind of scientist I want to be now!”

You might say she still works with bugs, albeit not the squishy kind. Debugging is only part of what Decker does as chief scientific officer at PolArctic LLC. Working alongside the company’s CEO (who is also her sister), Decker is in charge of programming sophisticated frameworks that map, model, and predict the ever-changing Arctic.

Coding can be demanding, solitary work, but it’s a good fit for Decker: “I am very much a ‘sit-in-my-office-and-program’ kind of person,” she says. “It’s so ingrained. My dad’s a programmer. I’ve been programming since I had to sit on his lap to reach the keyboard.”

Fishing boats in Seward, Alaska. Seafood and fishing are vital to Alaska’s economy. PolArctic recently secured a $1 million NSF grant to forecast climate change impacts on fisheries stocks.

PolArctic Helps Keep the Arctic Working

So, what does Decker do in her office all day? “I’m doing exactly what I wanted to do in terms of AI (artificial intelligence) and machine learning,” she says. She’s coding new technologies for businesses, governments, and Indigenous communities to help decipher a capricious (and climate-ravaged) Arctic.

One hot product from PolArctic is the Ice3 Model, a neural network algorithm that can render high-resolution predictions about when and where sea ice will be present. Using historical data and trends, the algorithm helps determine where ice may break or freeze. It’s a valuable tool for trades like fishing. Crabbers, for instance, tend to congregate at the ice edge, where frost poses a tangible and present threat.

“You don’t want to be overwhelmed by it and have your crab pots frozen in,” Decker explains. Using Ice3, fishers can make smarter moves about where to set up camp. So, too, can ships or barges use Ice3 to help steer through narrow rivers to deliver heating oil and other supplies to Alaskan communities.

“Alaska is really on the edge of nowhere. We have to transport our seafood harvest and all resources tremendous distances,” says Garrett Evridge, director of AFDF Startup Accelerator, an Anchorage-based program that seeks to supercharge Alaskan businesses in a sustainable way.

PolArctic is part of the AFDF Startup Accelerator portfolio of businesses, a group of 12 that includes similarly eco-oriented companies like Net Your Problem, a company that recycles fishing gear, to more traditional commodity producers like AlaSkins, a brand of dog treats made from dried Alaskan fish.

Evridge says the organization helps its cohort “engage with potential customers,” a role he’s happy to play when it comes to PolArctic: “I’m a fan and a supporter of PolArctic.”

John Farrell, the executive director of the U.S. Arctic Research Commission—and former associate dean for research and administration at URI’s Graduate School of Oceanography, Decker’s alma mater—is impressed with PolArctic, too.

He says, “I don’t often cross paths with two young female entrepreneurs who decide to strike out on their own and start up a company focusing on oceanographic and data science products for the Arctic region. That’s an unusual and rare combination.”

Evridge agrees: “As an Alaskan, I’m glad to know we have scientists like Lauren working on these challenges.”

PolArctic ICE3 High Resolution Regional Forecast for Ice Extent

1 October 2020

15 October 2020

31 October 2020

PolArctic’s Ice3 algorithm forecasts when and where sea ice will be present. They incorporate historical data along with climate change variables and trends to determine where ice may break or freeze to help vessels determine sea ice risk along their routes. This example shows the predicted range of ice (in white) extending into the water in Baffin Bay—between the Canadian Archipelago and Greenland—in October 2020.

Navigating the New Arctic

And there are challenges aplenty in the 49th state. Aside from the overarching damage inflicted by climate change, the economic ripples from more recent crises like the COVID pandemic and the Ukraine-Russia war have necessitated some heavy lifting when it comes to collecting and creating actionable data. Add to that the fact that the biggest state boasts 33,900 miles of shoreline, and it becomes apparent why Evridge says Alaska is “uniquely exposed to supply chain disruptions,” including everything from storms and seasonal ice to “geopolitical disruptions.”

“PolArctic is able to inject higher quality information into their system, so operators are able to make better decisions,” Evridge says.

“So much of the diminished ice impacts are current. It’s not something we’re going to prevent, because it’s already happened. Learning to live with that, navigating the new Arctic, is just an entirely new field.”
Lauren Decker, M.S. ’09, chief scientific officer, PolArctic

Decker sees one big question hovering over Alaska, as well as parts of Canada: “How do we build our economy? We can’t just do resource extraction,” she says. “We have to start thinking about the green energy transition. So much of the diminished ice impacts are current. It’s not something we’re going to prevent, because it’s already happened. Learning to live with that, navigating the new Arctic, is just an entirely new field.”

Helping Decker navigate this brave, broken new world are AI-fueled processes of maritime modeling, database architecture, and a wide assortment of visualization and forecasting tools.

Robots may take over the world one day, but for now they’re the allies of data scientists like Decker, who notes that AI was hardly on people’s radars when she started attending GSO in 2006: “There wasn’t a data science degree when I was going through school. It seems like a lot of tools and a lot more people are moving in that direction now.”

Her graduate work in physical oceanography still involved the same methods she uses today: data visualization, modeling, and, of course, programming. This skill set proved almost immediately useful: Decker defended her master’s thesis and interviewed for her first postgraduate job on the same day in 2009. Both went smoothly. Decker scored an oceanographer position at Applied Science Associates, a scientific data and consulting firm in Wakefield, R.I., only a mile from GSO’s Narragansett Bay Campus.

PolArctic leaders (and sisters) Lauren Decker (left) and Leslie Canavera were honorees at
the 2022 Women in AI Awards. Here, they’re enjoying some downtime together in their
home state, Alaska.

Lauren Decker gave a PechaKucha talk (a concise presentation style for telling a story using images) at the 2022 Arctic Frontiers Conference in Tromsø, Norway. Her PechaKucha—about herself, PolArctic, and AI—included this photo of her great-great-grandmother, great-grandmother, and grandmother (the baby in the photo), along with her great-uncles, in Southwest Alaska.

Decker was cozy in this role until she started getting calls—calls from her sister, Leslie Canavera. An Air Force veteran with experience in satellite technology, Canavera had a idea: You love what you’re currently doing, but let’s apply it to the ocean, in the Arctic.

“She started calling me every day. I would go on a walk and talk with her about it,” Decker says. “It took like a year for her to convince me.”

PolArctic is a neatly split venture, with Canavera handling the business end and Decker the science. Like many siblings, Decker and Canavera were close as kids, or, in Decker’s words, they were “make-snowmen-together kinda people.” Born and raised in Anchorage, Alaska, Decker and Canavera are both Indigenous; specifically, Yup’ik.

Science Informed by Indigenous Knowledge

Decker recently incorporated Indigenous knowledge of the best fishing spots into a training set for an AI model. This combination of ethnography and AI isn’t a novelty, but it is a rarity in the AI sector, which remains largely white and male. PolArctic hasn’t gone unnoticed in this regard. One example: the Women in AI Awards, where both members of PolArctic scored some honors in 2022.

Debates around AI have only intensified in the past year. Concerns over AI’s sometimes-theft of intellectual property has migrated from art and design into higher education, where colleges are scrambling to reckon with the likes of ChatGPT and its ability to summon entire term papers on demand. If text generation provokes unease, one can only imagine the potential thicket awaiting researchers who want to feed Indigenous secrets to the ghosts in the machine.

“Doing it correctly is hard,” Decker acknowledges. “You can’t go and ask someone you don’t necessarily know where their favorite fishing spot is. It’s sensitive information in terms of how you build on it and incorporate modeling for it.”

Her own Yup’ik heritage is one compass guiding Decker’s sensitivity on these issues, but she also takes a cue from the Interagency Arctic Research Policy Committee (IARPC), where she was recently appointed “data management nonfederal co-lead.” Chaired by the National Science Foundation, IARPC brings together Arctic communities, researchers, and federal agencies with business in the chilliest reaches of the northernmost state. Their common goal? Ensure Arctic research is not just high-quality, but equitable, too.

IARPC touts the data management principles of “FAIR” (findability, accessibility, interoperability, and reusability) and “CARE” (collective benefit, authority to control, responsibility, ethics) with both acronyms advocating “nuanced approaches that protect private and sensitive data, [and] respect Indigenous data sovereignty and governance,” per the organization’s website. “Traditional knowledge is in that zone of having to balance the two data management principles: of having it be open and of having the authority to control it,” Decker says.

In PolArctic’s case, Decker sees this as a positive: “We might have Indigenous communities reach out to us and then we support them in developing something, like commercial fisheries.”

In other words: PolArctic’s data can help Indigenous people not only protect their knowledge but put it to work in their own business ventures.

A barge navigating the waters in Southeastern Alaska. In 2020, PolArctic was awarded an NSF Small Business Innovation Research grant for coastline modeling and bathymetry. PolArctic is helping to chart the shifting Arctic coastline so big ships can navigate coastal waters safely.

Using Data to Map the Arctic Coastline

The data PolArctic provides is dynamic and responsive enough for commerce and, in some cases, it’s remapping areas that have never been friendly or easily accessible to humans.

Consider PolArctic’s Coastline Evolution & Nearshore Approximation (CENA), an AI engine designed for nearshore bathymetry. It identified a previously unknown subsurface reef in Hudson Bay, Decker says, and it’s also charted parts of Prince William Sound that hadn’t been mapped since 1939.

“The Arctic coastline is poorly charted,” Decker says, explaining that big ships can’t navigate it effectively. “With CENA, you can look at that really shallow range.”

The utility of CENA surely helped convince the National Science Foundation to award Decker and Canavera a Small Business Innovation Research grant (SBIR) in early 2020 for coastline modeling and bathymetry. Grant money in hand, Decker and Canavera went networking for PolArctic right as the pandemic emerged. Landing back in their West Coast headquarters of Seattle, the sisters found an airport full of people wearing masks. No need to recap the quarantine era, but for Decker it was “an excuse to program for six months.”

They initially thought PolArctic would work mostly with governments, but this hasn’t been the case: “We’ve worked a lot more with business-to-business contracts,” Decker says, citing “an enormous gap” in data services for Arctic businesses.

Forecasting sea ice is a valuable form of prophecy for companies in the Arctic—and the federal government is interested in PolArctic’s capacity for this uncommon skill. As Farrell explains, “By securing support from the National Science Foundation, first through a Small Business Innovation Research grant, and more recently with a $1 million award to forecast climate change impacts on fisheries stocks, PolArctic boosted its credibility and professional standing.”

The million dollar prize was awarded in January and comes attached to a Phase II continuation of PolArctic’s original SBIR grant. The Phase I research, Decker says, used “satellite imagery to estimate remote shore depths,” which “NOAA (National Oceanic and Atmospheric Administration) has a hard time doing, especially in the Arctic,” where the landscape itself tends to rebuff cartographic efforts.

An Ocean of Data to Help Fisheries

Phase II, meanwhile, applies the research to habitats just off the shore. It involves “agent-based modeling,” a method that follows multiple individuals of a particular species to better project a holistic understanding of entire ecosystems.

Decker says the target species to examine are crabs, cod, and possibly kelp: “Hopefully, Phase II will get us down to regular ocean depths,” where more of these creatures take up residence. The resulting data could help address and predict their presence (or absence) near and beyond the shoreline. Decker says this knowledge is desirable to fisheries, which benefit from knowing whether certain areas contain lucrative crops like scallops.

“PolArctic is uniquely positioned to play a role in our fisheries and in modernizing them by responding more effectively to climate change.”
Garrett Evridge, director, AFDF Startup Accelerator

The ultimate idea is to manifest a “digital twin,” Decker says, or in other words, an ocean of data, with depths of information that parallel the ocean’s own nuance and complexity. Decker’s not exactly sold on the name “digital twin,” though, and she opens the floor to suggestions.

Regardless of the sobriquet, it’s likely Evridge and PolArctic’s other true believers will rave about the startup’s latest venture. PolArctic, Evridge says, is “uniquely positioned to play a significant role in our fisheries and in modernizing them by responding more effectively to climate change.”

Indeed, it’s mostly shipping and fishing companies that have come to rely on PolArctic—two powerhouse industries in such a remote, wet, and cold part of the world. Tools like Ice3 help relieve some of the anxieties that come with Arctic commerce.

There are other Arctic anxieties, too, most of them related to the fragility of the Arctic itself. By generating top-shelf, actionable data, PolArctic seeks ways for businesses to work sensibly and with minimal ecological impact.

“That’s really our big thing: How do we do this responsibly?” Decker says.

In the Arctic proper, laws prohibit commercial fisheries, but fishing remains a plentiful, profitable trade in the subarctic. Places like Dillingham and Bristol Bay in Alaska have seen a recent crash in crabbing. But other fish are doing really well.

“Salmon has skyrocketed,” Decker says. “The ocean itself is becoming a really great place for salmon to thrive.”

The salmon are thriving. Lauren Decker is, too.

Photos: Courtesy PolArctic and Lauren Decker; iStock