Every river in the world
SCIENCE | Online game trains AI to track river data
Full access isn’t far.
We can’t release more of our sound journalism without a subscription, but we can make it easy for you to come aboard.
Get started for as low as $3.99 per month.
Current WORLD subscribers can log in to access content. Just go to "SIGN IN" at the top right.
LET'S GOAlready a member? Sign in.
With an online game and a deep learning tool, environmental scientist Shuo Zong at the ETH University in Zürich, Switzerland, studies river ecosystems on a global scale. A “game” at lab.citizenscience.ch shows two satellite photos of rivers. The player is asked to click on which river is “wilder.” The site gives pointers, but Zong says they’re hardly needed. “We trust the average perception of participants. It’s actually quite good,” Zong says.
The project aims to compile data about ecosystem health for every river in the world. It’s not just about wildness, Zong explains, but caring for resources that underpin human society. “Everything we rely on—society, business—comes directly or indirectly from nature.”
Rivers are critical for human development. Many power centers and AI data centers use river water as a cooling mechanism, and in many parts of the world, rivers remain the primary method of transportation.
But for these reasons, river ecosystems often suffer negative impact from human activity. Zong wants to create a tool that monitors river conditions worldwide over time. “Globalization and urbanization are everywhere, now we need data to match,” he says. Better data can guide policymaking and development strategies. Using satellite photos provides data about remote rivers that are difficult for ecologists to study.
Even with thousands of participants, analyzing every river is a colossal task. That’s where AI comes in. Zong uses the game responses to train a deep learning AI model. Deep learning AI is modeled after the human brain: It learns from examples, not instructions. With the game’s responses, the AI model creates a ranking that can be applied to every river.
Zong is currently publishing a paper about his work, but the online game stays open. More participation can refine the AI tool further. In the meantime, Zong has launched a similar game for mountain ecosystems.
Recovery is possible
For five years, volunteers in New Orleans have loaded cypress and tupelo saplings into boats and paddled to remote wetlands, aiming to restore the protective barrier that historically shielded New Orleans from storms. In 2005, Hurricane Katrina ripped through the wetlands, destroying forests already weakened by the Mississippi River-Gulf Outlet Canal. The 1960s-era canal brought saltwater into freshwater ecosystems, poisoning the trees. It was blamed for funneling Katrina’s surge into the city.
In 2009, the canal closed, making the first reforestation efforts possible. Several groups joined to form the Central Wetlands Reforestation Collective with a goal of replanting 30,000 trees.
“In our wildest dreams we never thought we’d be able to plant some of the areas that we are now planting,” Meraux Foundation’s Blaise Pezold said. “It was thought to be too low, too salty, Katrina messed it up too much.” But salinity levels in the water continued to decrease, and trees planted eight years ago stand 30 feet tall today.
Most of the collective’s organizations will meet their planting quotas this year.
“We’re one part of a larger movement to resist this sort of ‘doomerism’ mindset, and to show that recovery is possible,” said Common Ground Relief director Christina Lehew. —E.S.
Please wait while we load the latest comments...
Comments
Please register, subscribe, or log in to comment on this article.