Monitoring Moss and Lichen in Antarctica with Hyperspectral Drones

Antarctica’s mosses and lichens are among the most fragile and climate-sensitive ecosystems on Earth. As primary producers in an otherwise barren landscape, these organisms offer early signals of environmental change—yet they are notoriously difficult to monitor. Their low stature, patchy distribution, and subtle spectral differences make traditional RGB and multispectral imaging unreliable for accurate assessment.

A peer-reviewed study published in Nature Scientific Reports (Sandino et al., 2025) demonstrated how hyperspectral drone imaging combined with AI dramatically improves the detection and classification of cryptogamic vegetation. During a 2023 expedition to East Antarctica, researchers deployed a Headwall VNIR sensor mounted on a DJI Matrice 300 RTK drone to evaluate vegetation health and composition in extreme polar conditions.

This work highlights not only the potential of hyperspectral drone imaging in polar science, but also the broader value of integrating high-resolution spectral data with machine learning for fragile environmental monitoring anywhere in the world.

Why Traditional Spectral Tools Fall Short

Vegetation indices such as NDVI are widely used in plant monitoring. However, Antarctic mosses and lichens have spectral characteristics that fall outside what these indices are designed to capture. In the broad bands detected by multispectral and RGB sensors, mosses and lichens often exhibit overlapping spectral signatures.

Additionally:

• Their pigment differences are subtle and difficult to detect
• Healthy and stressed moss can look nearly identical in RGB
• Lichen often resemble bare rock
• High-albedo surfaces like snow and ice complicate segmentation

As a result, the study found that multispectral imaging and classical threshold-based classification approaches consistently struggled, leading to incomplete or inaccurate assessments.

Hyperspectral drones solve this by capturing hundreds of narrow, contiguous spectral bands—revealing differences in plant chemistry and physiology that are invisible to conventional tools.

Hyperspectral Drones Deliver New Scientific Insight

Collecting rich spectral data across the visible and near-infrared (VNIR) range, hyperspectral drones provided researchers with a level of detail impossible to achieve with traditional sensors. Key findings included:

• Machine learning models achieved >98% accuracy in classifying moss and lichen
• Simplified “light” models using only eight wavelengths still maintained high performance
• Deep learning segmentation produced fine-scale vegetation maps suitable for long-term monitoring
• Hyperspectral imagery successfully distinguished vegetation from rock and soil, even in low-contrast conditions

“These results show how focusing on the right spectral bands allows us to deliver actionable insight, even in extreme environments,” noted David Blair, Headwall’s Chief Product Officer. “It’s a powerful example of how hyperspectral imaging and AI together can transform ecosystem monitoring.”

AI + Hyperspectral Imaging: A New Model for Environmental Monitoring

One of the study’s major contributions is its demonstration of how AI strengthens the value of hyperspectral drone data. Rather than relying on hand-crafted vegetation indices, the research team used machine learning to:

• Identify which wavelengths carry the most diagnostic information
• Learn spectral–spatial patterns across diverse terrain
• Classify vegetation types with high confidence
• Reduce human subjectivity in interpretation

This synergy sets a new standard for remote sensing workflows—particularly in remote, ecologically sensitive environments where manual sampling is limited.

From Antarctica to Global Drone Remote Sensing Applications

While this research focused on Antarctica’s cryptogamic vegetation, the implications extend far beyond polar ecosystems. Hyperspectral drones bring transformative value to:

• Climate research: monitoring stress responses in polar, alpine, and arid ecosystems
• Agriculture: detecting crop stress, nutrient imbalance, or disease before visual symptoms appear
• Forestry: mapping disturbances, invasive species, or canopy health
• Soil and water analysis: identifying contamination or nutrient variation
• Environmental assessment: establishing repeatable baselines for long-term ecosystem change

If hyperspectral drones can reliably detect moss and lichen in one of the harshest environments on Earth, they can deliver exceptional value across landscapes worldwide.

Headwall Hyperspectral Technology in Action

The Headwall airborne hyperspectral payload used in the study is engineered specifically for drone deployment, offering:

• Lightweight design for extended flight time
• High signal-to-noise ratio for precise spectral measurements
• Robust performance in extreme temperatures and challenging light conditions
• Accurate, repeatable data that integrates seamlessly with machine learning workflows

This deployment reinforces how Headwall technology enables researchers to generate accurate, scalable, and non-destructive insights—especially in environments where traditional spectral tools fall short.

Learn More

Headwall supports researchers, agencies, and industry leaders working to better understand ecosystems, improve decision-making, and address critical environmental challenges.

Contact us to learn how hyperspectral drones can support your mission-critical applications.