The research is a pioneering effort to employ drones for methane (CH4) emission quantification from ruminants in sub-Saharan Africa. It is one of the first field studies to measure methane emissions from camels, a largely understudied source.
Methane emissions from livestock accounts for one-third of global anthropogenic methane emissions. Yet it remains poorly mapped in many regions—especially in Africa.
Using drones equipped with methane sensors, researchers flew over herds of cattle, goats, sheep, and camels—before and after grazing—to capture methane concentration data. A ground-based flux tower provided wind measurements.
The study was conducted at ILRI’s Kapiti Research Station in Kenya, in a collaboration between the University of Oslo, ILRI, NIBIO, and the University of Milan. Initial test flights were carried out at NIBIOs research station Tjøtta in the north of Norway. The results may contribute to the development of more accurate livestock methane emission estimates, facilitating their use in climate models and national greenhouse gas inventories.
“This study demonstrates that drones can effectively monitor emissions in remote or challenging environments, such as where traditional chamber methods are unavailable or impractical, for example with larger animals like camels”, says Alouette van Hove, PhD candidate at the University of Oslo, and first author of the study. Her research aims to develop new methods for measuring and calculating emissions of greenhouse gases such as methane and CO₂, including those from agriculture, using unmanned aerial vehicles (drones) equipped with sensors.
“It’s a flexible approach that allows researchers to travel to the animals’ location and take measurements over several days and at various times, all without disturbing the animals” she adds.
The article, named “Inferring methane emissions from African livestock by fusing drone, tower and satellite data” is a part of CircAgric-GHG – A research project that will unravel mechanisms by which farming systems can enhance circularity, reduce GHG emissions, and provide ecosystem services at multiple scales.
The researchers applied a Bayesian inference method that integrates drone-based methane measurements and flux tower wind data with an atmospheric dispersion model. This probabilistic approach accounts for uncertainty in the data and integrates prior knowledge to produce more reliable emission estimates.
Three following approaches were compared:
The researchers compared how consistent the different methods were, when compared to the IPCC estimates.
“We found that the Bayesian inference method consistently produced results that aligned with IPCC Tier 2 estimates, even for low-emission animals like goats and sheep, which indicates that the method is both reliable and robust under the conditions of the study, says van Hove.
In contrast, the mass balance method often overestimated emissions from the low-emission animals, suggesting potential limitations when applied to weaker sources.
Vibeke Lind, Research Scientist in NIBIO, and project leader of CircAgric-GHG, was responsible for calculating the IPCC Tier 2 methane emission estimates for the livestock monitored by the drones.
“The Tier 2 values incorporate detailed herd-specific or animal-specific data. In this case the data accounted for the local variations in Kenya and the livestock breed and feed qualities in the region. We used the method to estimate emissions from single animals, based on their weight, age, feed intake and feed quality”, says Lind.
Another innovation is the integration of hyperspectral satellite data. Researchers from the University of Milan used hyperspectral imagery from the Italian Space Agency satellite mission PRISMA acquired over the Kapiti farm at same time as the drone study was carried out. The goal was to see if hyperspectral satellite sensors could detect landscape features associated to herd location and potential methane emissions at Kapiti.
Usually, satellite imagery is applied to spot larger emission sources, like gas leaks from a factory. Therefore, the researchers where excited to investigate whether the satellites were able to detect spatial anomalies linked directly or indirectly to animal emissions. And it turned out; the satellite images did detect anomalies exactly at the locations of the herds. However, further research is needed to investigate whether these anomalies were caused by elevated methane levels or by other factors, such as changes in vegetation or soil moisture.
“While this was an exploratory assessment, the encouraging results suggest a potential for multi-scale assessment of point emissions sources at landscape scale, combining drones and last-generation spaceborne hyperspectral sensors” says Associate Professor, Francesco Pietro Fava, at the University of Milan.
The method presented in the study may be compared to what is happening when zooming in on a map from a far distance, to identify the detail in a single pixel.
“Satellites help identify emission sources, while drones provide an overview of emission concentrations and daily variations, such as before and after feeding”, says Lind.
To further zoom into the details, it’s possible to measure emissions from individual animals. Lind explains that this may be done using methods such as respiration chambers, the SF6 tracer gas technique, or face masks.
“Although drones are promising, they require certified operators and data experts, making them currently inaccessible for most farmers. Still, this technology shows potential for identifying emission sources and estimating volumes”,
Lind concludes.
Accurate, localized measurements are essential for developing effective mitigation strategies. This research could inform debates on feed subsidies, grazing practices, and emission reduction policies—especially in regions where data is scarce. The study also lays the groundwork for expanding methane mapping to other sources, such as wetlands, landfills, and thawing permafrost. Adaptations of the framework will be needed to handle multiple, diffuse sources and overlapping emission plumes, improving its relevance for complex landscapes.
As Alouette van Hove notes,
“It’s exciting to work on something that can actually be changed. Measuring methane from cows is not just a technical challenge—it’s a way to support better decisions for climate and agriculture.”
She is now working on optimizing drone flight paths using machine learning, enabling drones to autonomously detect and estimate methane sources in environments where their locations are unknown — “like smelling where the cows are,” as van Hove puts it. This could further improve efficiency and accuracy in future monitoring campaigns.
