Applying new spectral unmixing technique to Martian hyperspectral data

Hyperspectral datasets are now a common feature in planetary data, as recent missions have mapped planetary surfaces thanks to spectro-imagers with improved spatial and spectral resolutions. For those instruments the spatial extent of a pixel is usually large enough to have a mixture of various constituents contributing to the spectrum of a single pixel, involving both linear and non-linear mixing effects (which in the case of atmospheric contributions, for example, have to be corrected). The aim of spectral unmixing is to identify the materials present in each pixel of a hyperspectral image, and to evaluate the proportion of each of those materials in the pixels. This can yield important clues about the surface composition of planetary bodies and their geological history.