A multi-river contactless airborne river bathymetry dataset

To understand a river, one must look beneath the surface. For scientists who study floods, manage water supplies, or track how landscapes change, the shape of the riverbed is just as important as the water flowing above it. This hidden landscape, known as bathymetry, determines how fast water moves, where it pools, and how high it rises during a storm. Traditionally, mapping this underwater terrain has been a slow, difficult, and often dangerous job. It usually requires people to wade into the water, climb onto boats, or lower heavy equipment into the current. These methods are limited by how deep the water is, how fast it flows, and whether the banks are accessible. In many remote or hazardous locations, getting this crucial data has been nearly impossible, leaving gaps in our understanding of how rivers behave.

A new study offers a different way to see what lies beneath. Researchers from universities and institutes across Europe and Africa have created a large, open collection of river depth measurements taken without ever touching the water. Using drones equipped with special sensors, they flew over seven different rivers, capturing detailed maps of the riverbeds from the air. The team found that these contactless methods work remarkably well, matching the accuracy of traditional ground surveys while being able to reach places that boats and people cannot. By making this data freely available to the public, the researchers have provided a valuable tool for improving flood predictions, understanding river ecology, and refining the computer models that help us manage our water resources.

The project began with a simple but ambitious goal: to gather high-quality depth data from a wide variety of real-world rivers. Between late 2023 and mid-2025, the team surveyed 81 specific cross-sections across seven rivers located in Sweden, Germany, Italy, Benin, and Nigeria. These rivers vary greatly in size and character, from the cold, clear waters of the Torne River in the north to the warmer, sediment-rich channels of the Ogun River in West Africa. The objective was not just to take a few measurements, but to build a comprehensive dataset that could be used to test and improve how we map rivers from the sky.

To achieve this, the team relied on a fleet of drones, primarily the DJI Matrice 300, carrying two main types of sensors. The first was a water-penetrating radar system, which sends electromagnetic waves down through the air and water. When these waves hit the surface of the water, some bounce back immediately, while the rest travel through the water, hit the riverbed, and reflect back up. By measuring the tiny difference in time it takes for these two signals to return, the system can calculate the depth of the water. The second sensor was a sonar device, which uses sound waves instead of radio waves. Unlike the radar, the sonar needs to be in direct contact with the water, hanging just below the drone on a tether. The researchers used both methods to compare their results, treating the sonar and traditional ground measurements as a known standard to check the accuracy of the new radar technique.

The results showed that the drone-based radar method is highly effective. In the sections where the team could compare the drone data against ground measurements taken by people with high-precision GPS equipment, the radar measurements were very close. The difference between the drone's reading and the ground truth was typically around 12 centimeters, with most errors being even smaller. This level of accuracy is sufficient for many important applications, such as designing flood defenses or modeling how water moves through a channel. The study also confirmed that the radar method has distinct advantages over the sonar. Because the radar does not need to touch the water, it can fly over fast-moving currents and areas thick with underwater plants where a tethered sonar would struggle or get tangled.

However, the researchers also noted that the radar method is not perfect in every situation. Its ability to see the riverbed depends heavily on the electrical properties of the water. In rivers with very high levels of dissolved salts or minerals, the radar signals can be absorbed before they reach the bottom, making it difficult to get a reading. The team found that when the water's electrical conductivity gets too high, the radar's performance drops significantly. In contrast, the sonar and traditional ground surveys are not affected by this issue. This means that while the radar is a powerful tool for many rivers, it works best in water that is relatively fresh and clear of certain minerals.

The dataset released by the team is more than just a collection of numbers; it is a resource designed to help other scientists and engineers. The data includes not only the depth measurements but also the precise location of every point, allowing users to reconstruct the three-dimensional shape of the riverbed. For the first time, researchers have access to a large set of river profiles from different continents, collected using the same modern technology. This allows for a direct comparison of how different rivers behave and how different measurement techniques perform under various conditions. The data is already being used to test new ways of estimating river flow from space and to improve the computer models that predict flooding.

By proving that drones can map riverbeds accurately without entering the water, this study opens the door to more frequent and safer monitoring of our waterways. Rivers are constantly changing, shaped by erosion, sediment, and human activity. Being able to quickly and safely map these changes is essential for keeping communities safe and managing water resources effectively. The success of this project suggests that in the near future, we may be able to update our maps of the world's rivers as easily as we update our weather forecasts, using the sky to see what lies beneath.