Spectral Mountains – Enabling oblique hyperspectral mapping for steep targets

Light reflected or emitted from a natural surface contains material-characteristic, spectral signatures like fingerprints. Hyperspectral imaging sensors can capture this information in image rasters with hundreds of discrete spectral channels. The resulting spectral data cube can be analysed to create detailed maps of the surfaces’ material composition. Hyperspectral imaging is experiencing rapid transformation, mostly due to the ongoing miniaturization of sensors, the boost in computer processing power and the need for fast and non-invasive characterization technologies. The technology is versatile in regards to application type and scale, and can be applied using passive illumination, e.g., sun- or skylight. Hyperspectral imaging already supports a variety of application fields in earth observation, such as agriculture, geoscience, urban planning and environmental monitoring, ranging from global (satellite-borne) down to sample scale (lab).

Mountainous environments pose a specific challenge for hyperspectral imaging, as topographic complexity often requires oblique (non-nadir) acquisition, while illumination conditions strongly vary in time and space, and entrenched 2-D data analysis techniques (e.g. using 2-D gridded data such as DEMs and orthomosaics) are of limited applicability. It is important to move beyond the current usage of hyperspectral data as 2D rasters and go towards a more complex, but also more realistic 3D representation. This avoids occlusion and false-neighbourhood effects and allows us to accurately correct illumination effects induced by the geometry of the target with respect to the illumination source and the sensor positions. It enables the deployment of hyperspectral sensors from innovative, yet challenging platforms and non-nadir observation angles, occurring with tripod- and drone-based acquisitions (Fig. 1). The required transfer of hyperspectral data to a 3D “hypercloud”, i.e., a geometrically and spectrally accurate combination of a photogrammetric point cloud and the hyperspectral datacube (Fig. 2), ultimately allows the fusion of multi-scale and multi-platform scenes as well as the integration of sample data or subsurface information. With careful correction, the resulting dataset can provide tremendous value in mountain research, e.g., for the estimation of variation in mineralogical composition for better understanding of rock-forming processes, the detection of plant species and lichen coverage or monitoring of environmental changes.

With this contribution, we give an overview of the challenges and opportunities of spectral imaging in the context of mountain research. We showcase best practices and trends in data acquisition, platforms and data correction workflows, and highlight the advantages of the hypercloud approach for the mapping of steep and complex targets. We give examples from geoscience and mineral exploration perspective, covering natural alpine outcrops and cliffs of different scale and geological setting, as well as artificial outcrops in mining and exploration context.