Image

Taking Pictures of the Seafloor with Sound

 / 

June 22, 2026
 / 

 / 

How Project Recover uses sonar technology to find U.S. aircraft and the service members still missing aboard them

By Mark Moline, Co-Founder Project Recover and University of Delaware Professor

One of Project Recover’s biggest challenges is searching the seafloor for aircraft that went down during World War II and other conflicts. We’re not looking in small areas, either. Some search zones are as large as Walt Disney World.

To survey an area that big on land, you’d fly over it and take photographs. Cameras capture light reflected off the surface. But underwater, that doesn’t work.

Here’s why. Sunlight gets absorbed by water the deeper you go. And ocean water is full of particles that cloud visibility, even in relatively shallow areas. Sometimes a diver can’t see their hand in front of their face. Trying to photograph the seafloor from a distance would be like trying to take aerial photos through a thick cloud of smoke. You’d get nothing useful.

So if light doesn’t work underwater, what does?

The answer is sound.

Sound travels exceptionally well through water, far better than it does through air. That’s why submarines, whales, and sonar all rely on it. For Project Recover, the specific tool we use is called side scan sonar.

How Side Scan Sonar Works

Think of a bat navigating in the dark using echolocation. It sends out pulses of sound, listens for what bounces back, and builds a picture of its surroundings. Side scan sonar works the same way.

The device sends out pulses of sound in wide, fan-shaped beams aimed out to both sides (that’s where the “side scan” name comes from). When those pulses hit something on the seafloor, like a rock, a reef, or an aircraft, the sound bounces back. A computer measures how long the echo took to return and how strong it was. Dense, hard objects reflect stronger signals. Soft sand or mud reflects weaker ones. And if an object is sticking up off the seafloor, it casts a sound shadow behind it, which tells us roughly how tall the object is.

An Autonomous Underwater Vehicle (AUV) over the crash site of a Marine Corsair in the South Pacific. AUVs are used extensively to assist in locating underwater crash sites. Photo by Eric Terrill

From Towfish to Underwater Drones

Traditionally, sonar devices were towed behind a boat inside a torpedo-shaped housing called a towfish. As the boat moved forward, the towfish would scan strips of the seafloor on both sides, building up a detailed image the same way a lawnmower covers a yard in neat rows.

The problem? The seafloor isn’t flat. There are hills, valleys, coral reefs, and slopes. Keeping a towfish at a consistent height above all of that is nearly impossible. The sonar angles constantly shift, which degrades the data. Occasionally the seafloor rises so fast that the towfish slams into the bottom and tears loose from the cable. A very expensive mistake.

To solve this, Project Recover mounts side scan sonar onto Autonomous Underwater Vehicles, or AUVs. These underwater drones are pre-programmed to fly a precise grid pattern over the seafloor. They use onboard sensors to maintain a constant altitude above the bottom, no matter how the terrain changes. They aren’t affected by surface waves the way towed systems are. And because they have highly accurate navigation systems, we always know exactly where the vehicle is, and by extension, exactly where any detected object sits on the seafloor.

Two Passes, Two Frequencies

Not all sonar frequencies work the same way. Low-frequency sound travels farther but produces lower-resolution images. High-frequency sound gives us sharper detail but covers less distance.

So Project Recover uses a two-pass approach. We first survey large areas with low-frequency sonar to identify anything that looks like it could be an aircraft. Then we go back over those specific areas with high-frequency sonar for a clearer look.

AI That Does the Heavy Lifting

Side scan sonar generates an enormous amount of data. After a full day on the water, it’s not realistic for a small team to manually review everything before doing it all again the next morning. So Project Recover is developing AI machine learning models to help.

We train the model on data from previous missions, essentially teaching it what aircraft wreckage looks like on the seafloor. The more data it sees, the better it gets at flagging objects worth a closer look. What used to take four people four hours every evening now takes about one to two minutes. The team still reviews all the data after the mission is complete, but in the field, the AI buys back hours that used to be lost to late nights, and makes sure no one is too exhausted to do their best work the next day.

Confirming What We Find

Once sonar gives us a strong lead, we confirm it visually. If the water is shallow enough, our team dives on the site with SCUBA gear. If it’s too deep, we send down a Remotely Operated Vehicle, or ROV, equipped with HD or 4K cameras. When we confirm that we’ve found the aircraft we’ve been searching for, we spend many hours documenting the site in detail. That documentation becomes the foundation for planning the recovery of our MIAs, which is the ultimate goal of everything we do.

A Win for the Host Countries, Too

Project Recover operates in many countries that have little to no existing data about their own seafloor. Because these governments have generously allowed us to work in their waters, we share our sonar maps with them at the end of every mission. Those maps often reveal coral reefs and seagrass habitats that have never been charted before, giving local governments new tools for protecting their marine ecosystems for generations to come.

It’s a good deal for everyone.

How You Can Help Bring Them Home

Behind every sonar image, every AUV dive, and every late night reviewing data is a family somewhere still waiting for an answer. There are more than 72,000 Americans still unaccounted for from World War II alone. The technology to find them exists. What makes the difference is the resources to deploy it.

Project Recover is a 501(c)(3) nonprofit. Every mission is funded by people who believe no one should be left behind.

Here are a few ways to get involved:

Make a gift. A donation of any size directly funds mission operations, from AUV deployments to ROV dives to the AI development that helps our team work faster and smarter in the field. Every dollar goes toward bringing someone’s loved one home.

Give in honor of a MIA. A memorial gift is a meaningful way to pay tribute to someone who served. We’ll acknowledge your gift and the person you’re honoring.

Become a recurring donor. Monthly giving allows Project Recover to plan missions further in advance and respond quickly when new leads emerge. It’s one of the most impactful things a supporter can do.

Spread the word. Share this post. Tell someone about Project Recover. The more people who know this work is happening, the more families we can reach and the more missions we can fund.

These service members kept their promise. Help us keep ours.

Leave a Reply