Multi-Method Provenance Analysis of Namibian Desert Sand

Abstract:

Mineralogical, geochemical and geochronological provenance proxies each have their own strengths and weaknesses:

a. Bulk geochemistry, framework petrography and heavy mineral compositions can differentiate between source areas characterised by different lithologies, but are sensitive to hydraulic sorting and chemical alteration.

b. Detrital zircon U-Pb geochronology is insensitive to winnowing effects, but is 'blind' to lithologies devoid of zircon and cannot differentiate between first cycle and recycled sediments.

c. Cosmogenic neon isotopes can be used to identify different generations of surface exposure while simultaneously tracking different magmatic sources.

The challenge is then to combine these different proxies into a self consistent story, and do so in as objective a manner as possible. We here present a case study of Namibia's Namib Sand Sea and Skeleton Coast ergs, in which all the aforementioned methods have been combined using a three-way multidimensional scaling (aka INDividual Differences SCALing or INDSCAL) analysis:

1. Each of the datasets was represented by a 'dissimilarity matrix' of pairwise distances between samples.

2. The set of these matrices was fed into the INDSCAL algorithm, which produces two pieces of graphical output: the 'group configuration', which is a scatter plot or 'map' in which similar samples plot close together and dissimilar samples plot far apart, and the 'proxy weights', in which not the samples but the proxies are plotted according to the weight they attached to the 'group configuration' axes.

The INDSCAL map of the Namibia dataset indicates that (a) long-shore drift of Orange River sediments dominates the coastal sediment compositions all along the Namibian coast until Angola, and (b) that light and heavy minerals tell complementary parts of the provenance story.