RECONSTRUCTING SANDSTONE PROVENANCE: WHAT IS THE WITHIN-CATEGORY VARIANCE OF INDICATORS?

Petrographic and geochemical discrimination diagrams have long been popular for interpreting the provenance of ancient sandstones. This approach is limited, however, by lack of knowledge from modern sediments of the degree of success in making accurate assignments, a lack of quantitative assessments of which parameters are most successful and a lack of data on the within-category variance of parameters. We have addressed these issues by measuring chemical and petrographic variables for sands from a continental-margin volcanic arc using streams draining three of the principal Cascade volcanoes: Hood, Rainier, and St Helens.

Petrographic variables (QFL and subdivisions) for these sands were distinct from reported values from other the other major tectonic settings, but showed extensive overlap for the three volcanoes. Values also spanned the whole range for various subdivisions of the volcanic arc category. The best discriminator was the log-transformed ratio PAB/L (pyroxene+amphibole+biotite/total lithics), but even here the three volcanoes overlapped significantly.

Various proposed chemical discrimination schemes, both for major and trace elements, similarly showed a failure to distinguish the three volcanoes. The best discriminator was a cross plot of Ba v Na+K, but, again, the three volcanoes overlapped.

We found that samples from this single tectonic setting, a transitional magmatic arc (terminology of Dickinson) or the fore-arc to a continental-margin arc (terminology of Valloni & Maynard) plotted over the whole range of the generally applied subdivisions of arc-derived sandstones. That is, it appears that the subcategories of volcanic arcs cannot be differentiated. Volcanic and sedimentary processes, operating on the scale of a few hundred miles, homogenize provenance information from point sources such as volcanoes.