Imagine trying to understand the weather "by looking out your window once a week." That gives you an idea of what it's like to be a modern ocean scientist, says Richard Dewey, an ocean-modeling expert at the University of Victoria, in British Columbia, Canada.

To explore the deep, most oceanographers must jockey for a slice of time on a limited number of research ships, which generally leave port in the summer, when weather is most cooperative. After many days of chugging out to a particular spot on the ocean surface, they have just a few more days to launch tethered sensors, minisubs, or divers for a close-up look at what lies below. The result is choppy and piecemeal data, making it difficult to sort out cause and effect. And what data they get, of course, is skewed to the conditions that prevail during calm seas and summer months.

Hardly anyone doubts that such data is inadequate to understanding some of the most important questions about Earth's oceans—questions whose answers will be critical to our ability to feed a booming human population, power our cities, and fathom Earth's climate. Technology is responding to this challenge, but the efforts so far have been incremental and inadequate.

More and more, researchers are leaving instruments behind to continue sensing after research vessels depart, but severely constricted power and communications bandwidth limit their usefulness; they run on batteries, and many deliver data via painfully slow acoustic modems. Minisubs go deeper than ever but can't stay down long. A minisub such as the famous Alvin, operating out of Woods Hole, Mass., can take scientists down 4500 meters (a few go even deeper), but only for 5 to 10 hours. And the high cost of minisubs makes them a rare breed—by one recent count, only 40 minisubs operate today worldwide.

Dewey and his colleagues at a consortium of Canadian and U.S. universities and research labs think they've come up with a better way to get the job done. If they're right, ocean science will never be the same. They are in the vanguard of a growing movement to push ocean science toward the kind of 24/7 observation that scientists on dry land take for granted.

How do they intend to do it? By literally wiring the deep, seeding thousands of square kilometers of the ocean bottom with hundreds of sensors, cameras, and instruments powered from shore via an undersea 100-kilowatt grid. Linked together, the instruments are part of a multigigabit-per-second data network designed to continuously pour information onto the Internet, where scientists on land can access it. It is called the North-East Pacific Time-Series Undersea Networked Experiments, or NEPTUNE.

If NEPTUNE is completed, 200 000 square kilometers of ocean floor off the coasts of Oregon, Washington, and British Columbia will become one vast round-the-clock online undersea observatory pumping ashore a petabyte of invaluable deep-sea data every year.