💻 computer science

Towards Drone-based Mapping of Volcanic Gases using Gas Tomography

This paper demonstrates that model-based gas tomography using drone-mounted open-path sensors, combined with a Lagrangian wind advection model, successfully overcomes rotor downwash limitations to accurately map volcanic CO2 emissions where traditional in-situ sensors fail.

Original authors: Marius Schaab, Niklas Karbach, Antonia Rabe, Thomas Wiedemann, Patrick Hinsen, Dmitriy Shutin, Thorsten Hoffmann, Achim J. Lilienthal

Published 2026-05-27

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Marius Schaab, Niklas Karbach, Antonia Rabe, Thomas Wiedemann, Patrick Hinsen, Dmitriy Shutin, Thorsten Hoffmann, Achim J. Lilienthal

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine trying to smell a campfire while standing directly under a giant, roaring leaf blower. No matter how strong the smell is, the blast of air from the blower blows the smoke away before it can reach your nose. This is exactly the problem scientists faced when trying to measure volcanic gases using drones.

Here is a simple breakdown of what the researchers did and how they solved it, based on their paper:

The Problem: The "Leaf Blower" Effect

Volcanoes (and mud volcanoes) release gases like carbon dioxide (CO₂). To predict eruptions or understand climate impact, scientists need to map where these gases are coming from.

The Old Way: Scientists used to climb dangerous craters to measure gas directly.
The Drone Attempt: To stay safe, they tried using drones equipped with gas sensors. They hoped the drone could fly over the gas vents and "sniff" the air.
The Failure: Drones have propellers. When a drone hovers, its propellers push air down (called "downwash"). This air blast acts like that leaf blower, scattering the gas plume before the sensor can detect it. In the experiment, the drone flew right over the gas sources, but its sensors saw nothing because the gas was blown away.

The Solution: The "Laser Beam" Detective

To fix this, the team stopped trying to "touch" the gas with the drone. Instead, they used a technique called Gas Tomography, which is like taking an X-ray of the air.

The Setup: They used a special laser sensor (TDLAS) mounted on a tripod on the ground. This sensor shoots a laser beam to a reflector.
The Drone's New Job: Instead of carrying a sensor, the drone carried a mirror (reflector). The drone flew around, moving the mirror to different spots.
How it Works: The laser beam travels from the ground, hits the mirror on the drone, and bounces back. As the laser travels through the air, it measures the total amount of gas in that entire path.
- Analogy: Imagine shining a flashlight through a foggy room. You can't see the fog right next to the bulb, but you can see how much light is blocked by the fog in the middle of the room. The drone moves the mirror around so the laser scans the room from many different angles.

The "Wind" Glitch

There was one more hurdle: The Wind.
The laser measures gas along a straight line. But wind blows the gas sideways. If the wind is blowing the gas to the left, the laser might detect the gas on the right side of the map, making it look like the gas is coming from the wrong place.

The Fix: The scientists used a math model (a "Lagrangian model") to guess how the wind moved the gas. They measured the wind speed and calculated that it takes about 3 seconds for gas to drift from the ground up to the laser's height. They then "shifted" their data back by 3 seconds to correct the map, effectively undoing the wind's trickery.

The Results

The team tested this at the Salinelle dei Cappuccini mud volcanoes in Italy.

The Drone with the Sensor: Failed to find any gas because the propellers blew it away.
The Drone with the Mirror (Laser System): Successfully mapped the gas.
The Proof: They compared the laser map to a person walking around with a handheld sensor (which had no propellers to blow the gas away). The laser map matched the walking sensor's findings perfectly, showing the same high-concentration spots.

The Bottom Line

The paper concludes that if you want to map volcanic gases with a drone, don't let the drone touch the gas. Instead, use the drone to carry a mirror for a laser scanner on the ground. This avoids the "leaf blower" problem and, with a little math to correct for the wind, creates an accurate map of where the dangerous gases are hiding.

Technical Summary: Towards Drone-based Mapping of Volcanic Gases using Gas Tomography

Problem Statement
Volcanic gas monitoring is essential for eruption forecasting and understanding environmental impacts, yet traditional in-situ measurements pose significant safety risks to researchers. While drones offer a safer alternative for accessing hazardous terrain, they introduce a critical technical limitation: propeller downwash. This aerodynamic disturbance dilutes and disperses gas plumes directly beneath the drone, preventing in-situ sensors from detecting emissions from low-rate sources, such as the mud volcanoes at Salinelle dei Cappuccini. Consequently, standard drone-based sampling often fails to capture accurate gas concentrations or source locations in these specific environments.

Methodology
To address the downwash issue, the authors propose a remote sensing approach utilizing Gas Tomography with Tunable Diode Laser Absorption Spectroscopy (TDLAS). Unlike in-situ sensors that measure concentration at a single point, the TDLAS system measures the integral of gas concentration along a laser path between a sensor and a reflector (measured in ppm·m).

The study involved two primary experimental setups at the Salinelle dei Cappuccini mud volcanoes:

In-situ Validation: A drone equipped with a SmartGAS FlowEvo CO2 sensor was flown over the vents. Additionally, a manual "walking tour" was conducted with the same sensor detached from the drone to serve as a baseline for ground-truth detection without aerodynamic interference.
Remote Sensing & Tomography: A TDLAS sensor was mounted on a tripod, while a reflector was mounted on a drone (or held manually). The system collected line-integrated measurements from multiple angles to reconstruct a 2D gas distribution map.

A key methodological innovation is a model-based gas tomographic reconstruction that incorporates a Lagrangian wind compensation model. Since wind advection displaces gas molecules between emission and measurement, the authors compensate for this by offsetting the sensor and reflector locations based on the wind vector ( $w_t$ ) and a time delay ( $\Delta t$ ). The correction is applied as:
$x^{new}_t = x_t - w_t \Delta t$
where $\Delta t$ was empirically set to 3 seconds to account for the travel time of gas from the ground to the measurement plane.

Key Contributions

Demonstration of Downwash Limitations: The paper provides empirical evidence that drone-mounted in-situ sensors failed to detect CO2 emissions from the mud volcanoes due to propeller downwash, whereas manual in-situ measurements successfully identified elevated CO2 levels.
Remote Sensing Success: The study demonstrates that open-path TDLAS sensing, when combined with a drone-mounted reflector, can successfully map gas distributions without the sensor itself disturbing the plume.
Wind-Compensated Tomography: The authors present a novel reconstruction algorithm that integrates a Lagrangian model to correct for wind-induced advection, stabilizing the gas distribution map against varying wind conditions.

Results

Detection Capability: The drone-mounted in-situ sensor recorded only one CO2 detection during the flight, while the manual walking tour identified multiple high and medium gas detection zones.
Reconstruction Accuracy: The gas tomography reconstructions (both from handheld and drone-mounted reflectors) produced 2D maps where the CO2 distribution aligned with the locations of elevated gas detected by the manual in-situ sensor.
Comparison: While the reconstruction area for the drone-based dataset was smaller due to human precision limits and reflector size, the maximum CO2 values and spatial locations of the emissions were consistent between the drone-based tomography and the manual in-situ data. The authors note that the comparison is limited because tomography provides a time-averaged map, whereas in-situ sensors provide instantaneous snapshots.

Significance
The paper claims that model-based gas tomography effectively overcomes the aerodynamic limitations of drone downwash, enabling accurate mapping of volcanic emissions in environments where direct drone sampling fails. By combining remote sensing with wind compensation, this approach offers a safer and more reliable method for monitoring volcanic degassing, particularly for low-emission sources. The work validates the feasibility of using drone-mounted reflectors for rapid, large-area gas distribution mapping without the sensor physically entering the gas plume.