Satellite (Spot, Landsat, ...) remote sensing is based on the measurement by optoelectronic sensors of the quantity of sunlight reflected by unit areas of sunlight reflected by unit areas of terrestrial or planetary surfaces, for different bands of wavelengths (visible (VIS), near infrared (NIR), thermal infrared (TIR)). These radiometric quantities are proportional to the reflectivities of the surface components, that is, for the Earth, mainly water, snow, ice, vegetation, rocks, sands and soils. Thus, remote sensing from an heliosynchronous spatial orbit is a kind of long distance (800 km) optical spectrometry, which is applied to unit sampled surfaces (ground pixels) measuring 100 to 400 m 2 (Spot) or 900 m 2 (Landsat). Where the ground pixel is composed of rocks and soils, mineralogical studies are essential to calibrate and analyse the photometric data recorded by the satellite sensor, since these data are directly correlated to the crystal-chemical composition of the outcropping rocks. The recognition, or more often the differentiation, of the lithological facies present on a satellite numerical image are enhanced due to the utilization of «spectral signatures», i.e. any parameter, or combination of parameters, extracted from optical spectra recorded on the facies, and which are characteristics of a rock, a mineral, a weathering degree or a particular chemical composition. The optical reflectivity of a mineral surface is mainly controlled by the sum of the absorption coefficients of the minerals which compose this surface, weighted by their modal concentration. In the VIS and in the NIR ranges (400-2500 nm), absorption is essentially due to electronic transitions ; i.e. intra-atomic (crystal-field transitions), inter-atomic or inter-ionic (charge transfers) or inter-band (valence to conduction) transitions. Taking into account the energies involved in these spectral ranges, absorption depends on the concentrations, oxidation states and coordination numbers of the transition metals present in the crystal structure of the rock-forming minerals. Iron, with its different valencies, represents practically 90% of all the transition metals at the Earth's surface : Fe II and Fe III are mainly responsible for the colour, and more generally for the VIS-NIR spectral signatures of the mineral world on satellite images. In the part of the IR range (0.8-14.0 μm) utilized for remote sensing purposes, absorption is due to fundamental or harmonic vibrations of molecules (H 2 O, OH, CO 3 , SiOi4, etc.) forming the mineral lattice. The number and location of the absorption bands are a function of the number and type of atoms, the geometry of their co-ordination and the inter-atomic forces in the molecules. Controlled by crystal-chemical compositions, remotely sensed optical data are also very dependent on the morphological state of the pixels' surface. A trivial example illustrates this phenomenon : the reflectivity of H 2 O surfaces. With the surface of the sea, it behaves like a mirror to which is applied Fresnel's specular reflection laws ; since the specular reflectance is proportionale to the very low absorbance of water, the surface appears very dark from space. If the land is covered with snow―which has however practically the same intrinsic optical properties as water in the VIS spectral region―it behaves likes a powder to which can be applied the Kubelka-Munk diffuse reflectance theory ; such a surface appears very bright because the diffuse reflectance varies with the inverse of the absorption. The Rayleigh rule describes the transition conditions between the two types of surfaces and the two associated phenomena. In nature, mineral surfaces are mainly subject to diffuse reflection of the sunlight. It depends on morphological parameters (roughness, insolation angle, topography of the surface) and on chemical parameters (soil moisture, degree of weathering, geochemical and atmospheric composition). The aim of mineralogical studies is to provide experts―or expert-systems―with spectral recorded on actual or simulated petrological, mineralogical or pedological scenarios, eventually under extraterrestrial conditions (pressure, temperature, atmospheric composition). The utilization in the near future of hyperspectral sensors, which can sample the spectrum over 200 or more reflective channels about 10 nm wide, increases the fundamental need for such spectral signatures concerning minerals and their mixtures (i.e. rocks and soils) under various environmental conditions