Geological mapping strategy using visible near‐infrared–shortwave infrared hyperspectral remote sensing: Application to the Oman ophiolite (Sumail Massif)

An airborne hyperspectral survey of the Oman ophiolite (Sumail Massif) has been conducted using the HyMap airborne imaging spectrometer with associated field measurements (GER 3700). An ASD FieldSpec3 spectrometer was also used in order to constrain the spectral signatures of the principal lithologies cropping out in the surveyed area. Our objective was to identify and map the various igneous lithologies by a direct comparison at high spectral resolution between field and airborne spectra despite strong variations in outcropping conditions such as (1) lighting, (2) surface roughness geometry, (3) blocks coated with red/brown patina and exfoliation products, or (4) deep hydrothermal weathering. On the basis of spectral signatures, we are able to distinguish three end‐members of olivine‐orthopyroxene bearing assemblages in the mantle sequence: (1) harzburgites, (2) dunites, and (3) a harzburgite with interstitial carbonate. Because plagioclase is spectrally featureless in the wavelength range studied it cannot be detected. In the crustal sequence, we therefore identified four end‐members with variable abundance of clinopyroxene: (1) massive gabbros, (2) amphibolized (upper) gabbros associated with intrusive dykes, (3) wehrlite with high serpentine content, and (4) gabbronorite (a lithology not previously recognized in the studied area). With the exception of wehrlite, spectra of olivine‐rich end‐members display characteristic Mg‐OH narrow absorption features caused by their high serpentine content. We take advantage of this observation to split the data into two subsets, corresponding to the mantle and crustal sequences, respectively. Pixels of an image often correspond to heterogeneous areas in the field and a direct comparison between airborne and in situ spectra is not straightforward. However, comparing spectra of pixels associated with the most homogeneous areas in the field with the spectra acquired in situ at the same location, we found a systematic change both in mean intensity and overall spectral shape. Dividing each spectrum by its low‐pass trend removes the effects caused by surface light scattering associated with each scale of analysis and results in an exceptional match between field and airborne spectra. However, the albedo information is lost and as a consequence, rock types only characterized by albedo change cannot be discriminated. A spectrum of a mixture of powdered minerals is usually seen as a linear combination of mineral spectra proportional to their abundance. However, this is no longer the case when minerals occur in complex arrangements in rock types. We thus develop a synthetic spectral library of all possible combinations of rock types covering the surface area of a pixel and use a simple distance calculation to identify the best match between each pixel and modeled spectra. This procedure allows the determination of the fractional cover of each rock type in a given pixel and to establish maps for each spectral end‐member. The final product is a geological map, derived from the combination of end‐member fractional cover maps, and is broadly consistent with the existing geological maps. Beyond this general agreement which demonstrates the potential of this new approach for geological mapping, imaging spectrometry allows (1) to map in detail the outline of the Moho north of Maqsad and (2) to identify a new crustal sequence enriched in silica south of Muqzah, revealing the presence of orthopyroxene, the nature and distribution of which are of relevance to the petrological and tectonic understanding of the Oman ophiolite evolution.