“Sandy Island” – image from Google Earth

By Evan

And now for the amusing story of the week. Various news outlets (like the ABC) have been reporting on an amazing find by University of Sydney geologist, Maria Seton. While cruising between Queensland and New Caledonia, they searched for an island know as “Sandy Island”. Much to their astonishment, the ships found a water depth of 1400 m, right where the island appeared on Google Earth.

Google Earth uses a model for determining the bathymetry of the oceans known as GEBCO. This map is largely developed using satellite altimetry data (i.e. gravity measurements from orbiting satellites). The surface of the ocean is not flat – it actually changes depending on features such as mountains and trenches that are on the bottom of the ocean. These features affect the local gravitational field, which can be measured by satellites. The resolution of these satellites is pretty low, and so it is groundtruthed using ship based soundings. The only problem is that these measurements are only dense in major shipping routes. Clearly there isn’t much shipping going on through the Great Barrier Reef.

GEBCO is calculated using a mathematical method known as spherical harmonics. In areas where the data density is low, there is always the possibility of artefacts being introduced into the model. If I were to guess, this is exactly what happened in the case of Sandy Island. It is pretty amazing (to me) that no one checked satellite imagery to confirm the existence of an island. I would attribute this to general apathy. There are a bunch of barrier islandsΒ  just 100 km away from “Sandy Island”, so I am a bit surprised that no one actually visually confirmed this from the air, considering GEBCO gives it a length of 25 km. I have used the GEBCO model in my own work, but I recognize that there are limitations to it.