Mohometry gone astray: realistic assignment of uncertainties and reevaluation of calculated paleothicknesses

Prof. Matthew J. Kohn, Boise State University (Humboldt Professor at Mainz)

Abstract

Calculation of crustal thickness based on the geochemistry of arc magmas (chemical “mohometry”) relies on empirical correlations between seismically determined thickness and either elemental concentrations or elemental ratios in modern arcs. Such calculations ignore the effects on magma chemistry of other intensive parameters, including oxygen fugacity (fO2), water fugacity (fH2O), and temperature, as well as crystallization history. 

Review of the global arc database shows that data scatter is intrinsic to arc and volcano compositions. Minimum uncertainties are uniform and large, ranging from ±17-19 km (2) for arc segments to ±21-23 km for volcanic complexes. Systematic compositional differences among volcanoes and arc segments, and strong correlations among proposed mohometers, prevent reduction of this uncertainty through either increased number of analyses or application of multiple proxies. Other factors besides crustal thickness – especially fO2, fH2O, temperature, and crystallization history of major and accessory minerals – have major effects on trace-element chemistry of magmas and increase uncertainties by tens of km. 

New and published trace element compositions of detrital zircon and whole-rocks from the east-central (Lhasa) sector of the Gangdese arc, southern Tibet, show changes in chemistry only after initiation of collision: no changes in crustal thickness are resolvable (within ±20 km) throughout nearly 200 Myr of oceanic subduction. Post-collision trends in trace elements correlate with both temperature and fO2, rendering past estimates of crustal thickness and trends through time unreliable. Application to the Paleozoic “Acadian Altiplano” (New England Applachians) shows that, in contrast to prior interpretations, no variations in crustal thickness through time and space are resolvable. In the western interior United States, crustal thinning is resolvable only because modern thicknesses are known precisely. Even where thinning is most extreme, the magnitude is poorly determined – between ~10 and ~50 km.

Quantitative mohometry is currently no better than ~±20 km, irrespective of number and quality of analyses, and may be much worse depending on effects of intensive parameters – an uncertainty that is similar to or greater than random variation in arc crustal thickness (±25 km). Quantitative mohometry may be improved through analysis of detrital zircon, simultaneous characterization of fO2 and T, development of a geochemical proxy for fH2O, and further modeling.