Case study by SPH Engineering · UgCS · Polar Research
Terrain-following hyperspectral flights map Antarctic red snow algae
Researchers flew a hyperspectral sensor over a glacier on Livingston Island, Antarctica, at a constant height above the ice using UgCS terrain following. The drone data became the spectral reference that let Sentinel-2 satellite images map red snow algae across the South Shetland Islands from 2018 to 2024.
The challenge
Red snow algae, mainly Sanguina nivaloides, darken Antarctic snowfields. The blooms can cut surface albedo by up to 20%, which speeds up melting. Scientists had no map of where they occur at archipelago scale.
Satellites can cover that area, but Sentinel-2 pixels are 10 to 20 m across, so algae, dust, rock and clean snow blend together. Mineral dust can look like algae and cause false positives. Only 45 of 109 downloaded scenes were cloud-free. The classifier needed a clean spectral fingerprint of real algae, and that had to come from much finer data on the ground.
The solution
In March 2022 the team flew a DJI Matrice 600 Pro with a Headwall Photonics pushbroom sensor over Charrúa Glacier and Shelter Point. With a pushbroom sensor, pixel size depends directly on height above the surface. Over a sloping glacier, a fixed altitude above sea level would let the resolution drift and break the overlap between passes.
- Missions were planned in the UgCS desktop flight planner with terrain following, holding a constant 80 m above the ice.
- This kept the GSD at 6.5 cm/px across the survey, with a planned 40% lateral overlap.
- An APX-15 GNSS-inertial unit, used with the drone's GPS, brought georeferencing to ±0.02 m vertical and ±0.05 m horizontal.
With a remote site reached by ship, mostly cloudy weather and a short flying window, there was no chance for a second attempt.
The results
- Two clean hyperspectral cubes gave 1,215 spectrally pure pixels, nine endmembers and 214 pixels of red snow algae for the spectral library.
- The library, resampled to Sentinel-2 bands, trained a classifier (Spectral Angle Mapper followed by a Support Vector Machine) applied to 45 scenes from 2018 to 2024.
- Validation against 20 field spectra and six known bloom sites gave an F1 score of 0.757, an AUC of 0.935, precision of 0.868 and recall of 0.672.
- Blooms covered up to 12% of a single island (about 176 km²), up to 53 km² on King George Island (about 4.2%) and up to 10 km² on Robert Island (about 7.23%).
Where this applies in India
Holding a fixed height over uneven ground matters for any sensor whose resolution depends on altitude:
- hyperspectral and multispectral surveys in agriculture and forestry research;
- Himalayan glacier and snow studies by research institutes;
- mapping of hill slopes, coastlines and mining areas;
- ground reference data collection for satellite remote sensing projects.
SPH Engineering's UgCS training courses are also available through HobiTech.
The equipment, available from HobiTech
| Flight planning | UgCS - Drone Flight Planning Software: see UgCS PRO, UgCS EXPERT and UgCS ENTERPRISE |
| Drone and sensor | DJI Matrice 600 Pro with a Headwall Photonics VNIR-SWIR sensor was used in the study. Ask us for current compatible drones. |
| Positioning | APX-15 GNSS-inertial unit (Trimble Applanix), used in the study |
| Satellite data | Sentinel-2 Level 2A, 45 scenes, 2018–2024 |
Case study by SPH Engineering, published July 2026, based on Román et al. (2026), Communications Earth & Environment, 7, 53. This page summarises it in our own words; the research was carried out by the study team of Román et al., not by HobiTech. Read the original case study on SPH Engineering ↗
Need steady GSD over uneven terrain?
Tell us your drones, sensors and survey sites. We will recommend the right UgCS edition and licence, quoted in rupees with a GST invoice.