Sea floor image in Google Earth before (left) and after update. Grid-like markings in original were rumored to be evidence of the lost city of Atlantis. Google/Scripps Institution of Oceanography

When Google Ocean was added to the vast image database of Google Earth in 2009, the Internet lit up. Here was a real treasure trove - the floor of the world's seas in remarkable detail, free for use by scientists, teachers, shipping companies, and anyone else with a spirit of curiosity.

There was, as well, a spot about 600 miles off the coast of northern Africa with a curious matrix of lines (see picture, left). A British tabloid ran a boldfaced headline: "Is This Atlantis?" and quoted an aeronautical engineer as saying, "It must be man-made."

At Google - as well as NASA, NOAA, the Scripps Institution of Oceanography, and other agencies that collected the sea floor data - one could almost hear a collective gnashing of teeth. They report Google Ocean has now been updated after three years of painstaking work, and "Atlantis" (see upper right) is all but gone.

The grid in the original image was very man-made, said Google in 2009 - it was essentially a computing error. "Bathymetric [or sea floor terrain] data is often collected from boats using sonar to take measurements of the sea floor. The lines reflect the path of the boat as it gathers the data."

Now, new data have been added to the layers that went into creating the ocean images. About 15 percent is now from sonar readings by ships; much is from satellites measuring minute variations in Earth's gravity as they go over mountains or ocean trenches. The different data sources (Scripps alone sent ships on 287 sonar expeditions) had to be pieced together by computer, and much of the work was done at the University of California, San Diego, where the Scripps Institution is headquartered. The scientists say the 2012 version is more precise than the 2009 original, and the known errors have been corrected.

"The problem is very simple," said David Sandwell, a Scripps geophysicist who worked on the project, in an email to ABC News. "When you measure seafloor depth you actually measure the two-way travel time of a sonar echo. The ship sends it, it bounces off the bottom, and returns to the ship. To convert the time to depth you need a velocity. If the velocity used in the conversion is a bit too high, the depth from that cruise will be greater than all the surrounding cruises. The particular cruise that created the 'Atlantis' problem was a systematic mow-the-lawn type survey in a regular grid pattern, so the resulting depth errors looked like a regular grid."

Atlantis was mentioned by Plato around 360 B.C., a city that "in a single day and night of misfortune … disappeared into the depths of the sea." It has been the stuff of legend ever since. Did it ever really exist? Will Scripps' updated data do anything to settle the matter?

"I see this as like the 'Face on Mars,'" said Sandwell. "Of course it is nonsense, but in a way brings public attention to the deep ocean."