Hawaii’s water situation is unique. In terms of salt water, Hawaii is literally surrounded by the stuff. If you’re looking for fresh water to drink, you may have to search a little harder. Much of Hawaii’s drinking water is pumped out of the ground at points drilled into the slope of volcanic mountains.

However, a new discovery that was recently published in Science Advances may one day change how volcanic islands like Hawaii access fresh water.

Eric Attias, a research affiliate faculty at the University of Hawai’i (UH), used controlled-source electromagnetic (CSEM) imaging technique and high-performance computing (HPC) to discover freshwater reserves hiding under the ocean subsurface.

If his assumption that most volcanic islands have similar reservoirs is right, this revelation could help islands ensure their water security in the years to come.

A shocking discovery

Like most scientific discoveries, Attias and his colleagues were building on previous work. He states that the United States Geological Survey created a hydrological model that many scientists rely on.

“Conventional hydrogeologic models for onshore aquifers assume thinning of the basal freshwater lens as the coastline is approached, where freshwater floating on denser seawater discharges to the ocean through coastal springs,” says Attias. “These conventional models neglect the complexity of the offshore submarine environment, which is what we aimed to find out – is fresh water flowing and accumulating under the ocean?”

Attias describes a scenario where a volcano erupts and the lava cools into basalt, a relatively porous and fractured volcanic rock. The holes in the basalt can allow fresh water through, whereas regional-scale layers of ash and soil on top of the basalt keep the fresh water within this aquifer separated from top and bottom basaltic rock layers that are saturated by seawater.

Of course, basalt allows salt water in just as easily as fresh water, and the existence of a basalt formation doesn’t necessarily mean it’s filled with drinkable water. For this, Attias needed electromagnetic imaging. Since the CSEM method is sensitive to contrasts in electrical resistivity, it can distinguish between basaltic rocks saturated with seawater (conductive) and those saturated with freshwater (resistive).

Imagine a big boat towing a one-kilometer-long array of devices that is dragged on the surface of the water. Then, Attias and his colleagues would initiate a 100-amp electromagnetic wave from a 40-meter dipole. This wave diffuses through the water and the subsurface rock. The salt ions in seawater conduct electricity better than fresh water, so changes in electrical resistivity are noteworthy.

“Then, using geophysical inversion code (based on second partial derivatives), we can produce a data-driven resistivity model of the subsurface that shows you the areas where you have high electrical resistivity versus areas with low electrical resistivity,” says Attias. “Although it’s based on different physics, it’s kind of like mapping the brain using an MRI.”

After the imaging was complete—which included a boat accidentally running into their array on the first day—Attias and his colleagues concluded that underwater basalt systems extending four kilometers west of Hawaii’s coastline hold overall about 4.75 cubic kilometers of fresh water.

What’s more, Attias believes that this reservoir could be tapped with current technologies. He mentions that gas and oil companies sometimes drill in mid-ocean, where the water is about four kilometers deep, and some of the places Attias and his colleagues observed were as close as 200 meters from the shore, less than 80 meters deep.

Similar drilling techniques could be used, and a leakage of fresh water into the ocean won’t cause widespread damage – unlike an oil spill.

The right tool for the job

As Attias points out, this work would have been impossible without access to the HPC resources provided by UH. Attias used the Mana HPC cluster to perform the heavy lifting. Attias also states that UH consulted the Texas Advanced Computing Center (TACC) to create the cluster Attias relied on.

“I designed the starting models to be inverted on my laptop and then submitted the inversion job to the HPC cluster,” says Attias. “For each job I used between five to 10 nodes and each node has about 20 processors. One version would take me two or three days. You have about 150 processors working to produce a solution in a few days. That’s a lot of computational power that was required. Overall, it all took me many months.”

Attias continues: “I acquired CSEM data from 22 individual survey lines, which required a 2D inversion algorithm to produce an electrical resistivity model of the subsurface. 2D inversions are computationally expansive, mostly since I ran at least 30-40 inversions per each line data to seek a geologically plausible final model, by parameterizing each inversion slightly different in terms of errors, finite-element meshing, frequencies, signal-to-noise ratio, etc. In this way, the IT cluster really helped me speed up the work.”

Like a good scientist, Attias is looking for the next problem to solve. An upcoming paper will study fresh water escaping their basalt confines and rising as plumes into the ocean. What’s more, he’s hopeful his research could help give decision-makers the information they need to make future changes.

“I'm pretty certain that this is how fresh water is transported from onshore-to-offshore in most volcanic islands, because there is onshore evidence that the formation of basalts with alternating seawater/freshwater layers exist at several other different islands,” says Attias.

Attias continues: "But the thing is, I think some other articles misunderstood me, thinking that I support pumping offshore freshwater – but no. I'm a scientist; I just made a discovery, what the decision-makers want to do with it – it's their choice. I advocate for better management of current freshwater resources over exploiting new sources.”