1. Australian Antarctic Data Centre
2. Discover and Manage Data
3. Records
4. AAS_4628_ASPA131_GeoAI

Description

This dataset contains a comprehensive multi-sensor ecological remote sensing dataset collected from the Canada Glacier Antarctic Specially Protected Area (ASPA 131), Taylor Valley, McMurdo Dry Valleys, Antarctica. The dataset integrates field observations, UAV multispectral and RGB imagery, Sentinel-2 satellite imagery, manually generated reference labels, deep learning predictions, fractional vegetation cover products, and supporting geospatial data collected between 2018 and 2025.

The dataset was developed to investigate ecological scale mismatch between field observations, drone imagery, and satellite observations by implementing a novel GeoAI framework that integrates biologically validated field data with UAV and satellite remote sensing. The workflow includes UAV photogrammetry, radiometric calibration, Sentinel-2 super-resolution processing, progressive weak-to-strong labelling, semantic segmentation using deep learning, cross-scale fractional cover generation, and validation products.

The dataset includes original imagery, processed orthomosaics, derived raster products, vector training labels, prediction maps, metadata, and supporting documentation required to reproduce the analyses. These data support ecological monitoring of Antarctic vegetation communities including cyanobacteria, mosses, lichens, soil, snow and water, and provide a reusable benchmark dataset for remote sensing, ecology, artificial intelligence and environmental monitoring.

The dataset comprises field observations, UAV surveys, satellite imagery, and derived products used to develop and evaluate a multi-sensor GeoAI framework for Antarctic vegetation mapping.

The dataset includes:

• UAV deep learning outputs
• Digital elevation models
• S2 fractional vegetation cover products
• Ground reference photographs
• UAV machine learning outputs
• Manual training polygons
• Metadata and documentation
• UAV orthomosaics
• Sentinel-2 native imagery
• Species identification records
• UAV spectral library
• Super-resolved sentinel-2 imagery
• Supporting GIS layers
• UAV vegetation indices

All spatial datasets are referenced to WGS84 / UTM Zone 58 South (EPSG:32758).

Purpose

The dataset was developed to:

- investigate ecological scale mismatch between field, UAV and satellite observations;
- develop and evaluate a multi-sensor GeoAI framework for Antarctic vegetation mapping;
- generate biologically validated training datasets for artificial intelligence;
- produce high-resolution vegetation maps and fractional cover products;
- support long-term monitoring of Antarctic terrestrial ecosystems; and
- provide an openly available benchmark dataset for ecological remote sensing and GeoAI research.

Quality

Extensive quality assurance procedures were implemented throughout data collection and processing.

Field observations were supported by GPS measurements and species identification. UAV imagery underwent radiometric calibration and photogrammetric quality assessment. Orthomosaics and satellite datasets were spatially aligned using a common coordinate reference system. Manual annotations were quality checked before expansion using machine learning and deep learning workflows. Independent validation datasets were used to evaluate model performance, and spectral preservation of super-resolved Sentinel-2 imagery was quantitatively assessed.

Access

These data are not yet publicly available for download.

Temporal Coverages

• Start date: 2018-01-01 - Stop date: 2025-12-31

Spatial Coverages

Science Keywords

• EARTH SCIENCE > BIOSPHERE > VEGETATION > VEGETATION SPECIES
• EARTH SCIENCE > BIOSPHERE > VEGETATION > VEGETATION COVER
• EARTH SCIENCE > CLIMATE INDICATORS > CRYOSPHERIC INDICATORS > SNOW COVER
• EARTH SCIENCE > LAND SURFACE > LAND USE/LAND COVER > LAND USE/LAND COVER CLASSIFICATION
• EARTH SCIENCE > BIOSPHERE > VEGETATION > VEGETATION INDEX
• EARTH SCIENCE > BIOLOGICAL CLASSIFICATION > FUNGI > LICHENS

Use Constraints

This data set conforms to the CCBY Attribution License (http://creativecommons.org/licenses/by/4.0/).

Please follow instructions listed in the citation reference provided at http://data.aad.gov.au/aadc/metadata/citation.cfm?entry_id=AAS_4628_ASPA131_GeoAI when using these data.