CORRELATION OF TEPHRA: INSIGHTS FROM THE LATE CENOZOIC TEPHRA RECORD FOR THE WESTERN USA

An extensive record of explosive silicic eruptions in the late Cenozoic of the western USA is now encapsulated in databases assembled by the U.S. Geological Survey and several North American universities. These databases have been assembled using: 1) chemical analyses of tephra glass shards (chemical “fingerprinting”); 2) stratigraphic context of tephra relative to other tephra, magnetopolarity events, fossils, etc.; and 3) dating either directly by isotopic techniques or indirectly by correlation to dated units or by interpolation relative to other dated events. With these tephra databases it is now feasible for workers in the region to submit a newly sampled tephra to a tephrochronology lab with the reasonable expectation that the sample can be correlated to a dated tephra. Nonetheless, there are important issues of tephra correlation that remain poorly understood, unresolved, or even unrecognized by tephrochronologists. These issues need to be resolved and a standardized approach to correlation developed before individual tephra databases are fully utilized and before the more challenging goal of integrating all western U.S. tephra databases into a single database is realized.

Issues of tephra correlation and possible approaches to their resolution are explored using the examples from the University of Utah western USA tephra database and from computer simulations designed to elucidate the issues. Results to date indicate that correlation is best done via a stepped approach as follows. First a database search based on mean glass shard composition is used to get candidate correlations. Next, when individual glass shard analyses are available, a pattern-matching search winnows the candidate list down to those tephra with compatible patterns of compositional variation, be it a unimodal or polymodal compositional pattern or a broad compositional range. Using microprobe analyses, or even higher precision trace element analyses, a unique compositional correlation may not be found. Further refinement is possible if information on stratigraphic context is available, particularly information on stratigraphic position relative to other tephra or other event horizons. Several of these steps are now automated at the Utah lab, and, in principle, all can be automated and integrated as a single application.