The recent development and miniaturization of hyperspectral sensors embedded in drones has allowed the acquisition of hyperspectral images with high spectral and spatial resolution. The characteristics of both the embedded sensors and drones (viewing angle, flying altitude, resolution) create opportunities to consider the use of hyperspectral imagery to map and monitor macroalgae communities. In general, the overflight of the areas to be mapped is conconmittently associated accompanied with measurements carried out in the field to acquire the spectra of previously identified objects. An alternative to these simultaneous acquisitions is to use a hyperspectral library made up of pure spectra of the different species in place, that would spare field acquisition of spectra during each flight. However, the use of such a technique requires developed appropriate procedure for testing the level of species classification that can be achieved, as well as the reproducibility of the classification over time. This study presents a novel classification approach based on the use of reflectance spectra of macroalgae acquired in controlled conditions. This overall approach developed is based on both the use of the spectral angle mapper (SAM) algorithm applied on first derivative hyperspectral data. The efficiency of this approach has been tested on a hyperspectral library composed of 16 macroalgae species, and its temporal reproducibility has been tested on a monthly survey of the spectral response of different macro-algae species. In addition, the classification results obtained with this new approach were also compared to the results obtained through the use of the most recent and robust procedure published. The classification obtained shows that the developed approach allows to perfectly discriminate the different phyla, whatever the period. At the species level, the classification approach is less effective when the individuals studied belong to phylogenetically close species (i.e., Fucus spiralis and Fucus serratus).
Geostatistical models with the use of hyperspectral data and seasonal variation – A new approach for evaluating the risk posed by invasive plants
