Nonprofit, independent journalism. Supported by readers.


Rising sea levels threaten to drown domestic internet

Within the next 15 years, persistent coastal flooding will submerge more than 4,000 miles of the fiber-optic conduits that carry internet traffic in the United States.

Waves hitting the rocks at Rockaway Beach in Queens, New York City.
REUTERS/Mark Kauzlarich

I had hoped I was done with the depressing subject of sea-level rise for a while, but a study by researchers at the University of Wisconsin–Madison and the University of Oregon reveals a new dimension that’s too interesting and important to pass up:

  • Within the next 15 years, persistent coastal flooding will submerge more than 4,000 miles of the fiber-optic conduits that carry internet traffic in the United States — with New York, Miami and Seattle topping the list of trouble spots.
  • More than 1,000 connection nodes will be above but surrounded by water, making them difficult to reach and repair.
  • These projections are based on a widely anticipated rise of just one foot from today’s levels; if the pace is slower, the inundations won’t be lessened — just postponed. (And because the study doesn’t factor in the additional impacts of storm surge atop the high-tide trend, these projections could be taken as conservative.)

It’s impossible to quantify the likely disruptions suggested by these findings, laid out in a peer-reviewed analysis presented in Montreal on Monday to a joint conference of the Internet Society, the Association for Computing Machinery, and the Institute of Electrical and Electronics Engineers. But based on the large though relatively short-term service interruptions associated with recent big storms — think Hurricanes Sandy and Irma — I think it’s fair to consider them staggering. 

And we’re not talking only about email, Facebook and online banking. Besides enabling the critical long-distance communications vital to both modern commerce and virtual socializing, the internet allows the control, via “short-haul links” or “metro links,” of things like traffic signals and railroad routing; the sharing of medical records among doctors and hospitals, and an increasingly vast “internet of things” that includes household appliances at one end of the scale, and regional grids of electric power production and transmission at the other. These short links are at greater risk than the long-distance lines, the new analysis finds.

An unexamined threat, until now

While increasing attention is being given to a rising sea’s risks for roads, homes, hospitals, military bases and other aspects of the built landscape, it appears that nobody has published an examination of threats to the digital infrastructure until now.

Article continues after advertisement

The study’s authors include Paul Barford, a computer scientist at the University of Wisconsin-Madison; his wife Carol Barford, who directs UW-Madison’s Center for Sustainability and the Global Environment; and Ramakrishnan Durairajan, a Ph.D. candidate mentored by Paul Barford who is now on the computer science faculty at the University of Oregon. And as a result of thier conclusions, the notion of an underwater internet has set off considerable buzz, not only at this week’s international conference of internet engineers in Montreal, but in media outlets both professional and popular.

Rae Zimmerman, a planning expert at New York University who focuses on climate adaptation, told National Geographic’s Alejandra Borunda  (with audible understatement) that “this will be a big problem.” 

In the same piece, Mikhail Chester, the director of the Resilient Infrastructure Laboratory at the University of Arizona, concurs: 

“Considering how interconnected everything is these days, protecting the internet is crucial,” says Mikhail Chester, the director of the Resilient Infrastructure Laboratory at the University of Arizona. Even minor hits, like when storms knock out internet connectivity for a few days, can affect things we take for granted, from traffic lights to flight patterns.

Because high-speed data lines are still laid in pretty much the same way as the copper snakes that powered America’s first electric streetlights — that is to say, by private companies following their own agendas — this system has grown in an uncoordinated and largely regulated way, with plenty of gaps and redundancies. A fair portion of it is also beginning to show its age after 20-some years of service.

Some of its structure also remains proprietary. Over the years, though, Barford has gathered data from a wide range of sources to produce the Internet Atlas, which maps the known lines. His maps, in turn, make visible the large corridors formed by these lines’ proliferation. 

Because companies lay individual lines in areas of high population and service demand — and try to piggyback on other infrastructure elements like roadways, rail routes and power lines — the cables that form the physical internet reach their greatest densities along the nation’s coastlines.

The significance of this geographic pattern was plain to Carol Barford, since the corridors lie in zones that have been seeing a steady increase in tidal flooding over the last century of record-keeping. It was also of keen interest to Durairajan, the study’s principal author, who took Barford’s maps of the internet and laid them over  (or, I guess, under)  maps of projected tidal flooding prepared by the National Oceanic and Atmospheric Administration.

What water does to fiber optic

The researchers also made a special effort to focus on places where transoceanic cables, which are actually waterproof, join coastal networks of fiber cables, which are merely water resistant — like a cheap watch, perhaps. Key excerpt, lightly compressed:

Water, humidity and ice have long been recognized as threats to fiber optic strands and conduit. Water-related threats include signal attenuation due to water molecules embedding in fiber micro-cracks, corrosion damage to connectors, signal loss in optical-electrical-optical connections, and fiber breakage due to freezing. Cable construction techniques (e.g., cladding and hydrophobic gels) along with careful deployments enable fiber to function for decades under normal/expected environmental conditions.

Article continues after advertisement

The starting point for our work is that while standard Internet infrastructure deployments are designed to be weather and water resistant, they are not designed to be surrounded by or under water. Thus, we posit the following risks due to sea level rise. The first is physical damage at certain nodes (e.g., submarine cable landing stations) and at termination points (i.e., colocation facilities). A majority of the cable landing stations are near a tidally active region and terminate at the nearest colocation facility.

Second, buried conduits will become submerged, which will expose them continuously to all of the threats mentioned above, and the possibility of physical damage due to exposure caused by tides and storms. The fact that a great deal of conduit infrastructure was deployed over the past twenty years and is aging means that all seals and cladding are likely to be more vulnerable to damage, especially if they are under water.

What do the ISPs have to say about this? A couple of them spoke to National Public Radio’s Rebecca Hersher:

AT&T uses fiber optic cable “designed for use in coastal areas as well as being submerged in either salt- or fresh-water conditions,” spokesman Jeff Kobs says. “In certain locations where cabling will be submerged for long periods of time or consistently exposed, such as beaches or in subways, we use submarine underwater cabling.”

“After Sandy, we started upgrading our network in earnest, and replacing our copper assets with fiber assets,” says Verizon spokeswoman Karen Schulz. “Copper is impacted by water, whereas fiber is not. We’ve switched significant amounts of our network from copper to fiber in the Northeast.”

I guess I put less stock in Schulz’s assertion of fiber-optic invulnerability than Durairajan’s listing of all the things that can go wrong, especially since AT&T appears to be replacing some of its fiber lines proactively. Time will clarify who’s being square with the public on this point.

Speaking of time and the tides, it’s another intriguing finding of this analysis that while NOAA expects six feet of sea-level rise by the year 2100, nearly all the Internet inundation suggested by its maps occurs with the first 12 inches — another way of highlighting the low-lying locations of the network as it stands today. (Also, the scale of effort required to move it to permanently higher, drier ground.)

“That surprised us,” Paul Barford said in a UW announcement of the findings. “The expectation was that we’d have 50 years to plan for it. We don’t have 50 years.

“When it was built 20-25 years ago,” following established rights of way, “no thought was given to climate change. This is a wake-up call. We need to be thinking about how to address this issue.”

Article continues after advertisement

And, he suggested, the thinking must move quickly beyond higher seawalls and such:

The first instinct will be to harden the infrastructure. But keeping the sea at bay is hard. We can probably buy a little time, but in the long run it’s just not going to be effective.

* * *

The full paper, “Lights Out: Climate Change Risk to Internet Infrastructure,” can be read here without charge.