RELATIVE DATING OF ALLUVIAL SURFACES USING REFLECTIVE HYPERSPECTRAL REMOTE SENSING: A NEW TOOL FOR GEOMORPHOLOGIC MAPPING

The conventional mapping method of alluvial surfaces in arid areas is based on estimation of relative time characteristics of the surfaces and their underlying soils. Most of the mapping is made manually on air-photos, which manifest the surface differences. Yet, this method is subjective and qualitative, and usually enables only rough discrimination between mapping units. In this study we examined the feasibility of mapping arid alluvial surfaces using hyperspectral remote sensing. This method identifies the surficial mineralogy based on specific spectral absorption features. Wadi Raham alluvial fan, in the southern Arava valley, Israel, was used as a case study.

The fan was divided by conventional field mapping into 5 alluvial surfaces, ranging in age from the late Pleistocene to present. The lithology of the surficial clasts is similar on all surfaces and consists mostly of carbonate, chert, and rhyolite. The surfaces differ mainly by the degree of desert pavement development and the degree of rock coating accretion. The results of the spectral field measurements showed that the surface reflectance was controlled mainly by the desert pavement development, and was only slightly effected by the rock coating accretion. All surfaces had a distinct carbonate absorption feature at 2.33mm due to the dominance of carbonate clasts. The fine particles in-between the clasts contributed hydroxyl and ferric absorption features (2.2mm, and 0.87mm, respectively) to the surface spectra. Our results revealed that as the percent of the surface covered by clasts increased, the absorption depth of the carbonate clast increased; whereas the absorption depth of the fine particle in-between the clasts, decreased.

The relationship that was established in the field between the degree of desert pavement development and the 3 spectral absorption depths was used to classify the airborne hyperspectral data. We used best-fit curves to transform the relevant spectral absorption depths of each pixel into percent of clast coverage. These classifications proved to be more accurate than the conventional map.

The methodology developed in this study for mapping alluvial surfaces can be used for other arid piedmonts throughout the world. The proposed approach is more objective, and potentially more accurate than the conventional mapping method.