On the treatment of discordant detrital zircon U–Pb data

Abstract. Zircon U–Pb geochronology is a staple of crustal evolution studies and sedimentary provenance analysis. Constructing (detrital) age spectra straightforward for concordant 206Pb/238U 207Pb/206Pb compositions. But unfortunately, many datasets contain significant proportion discordant analyses. This paper investigates two decisions that must be made when analysing such data. First, the analyst choose whether to use or date. The method more precise young samples, whereas better suited old samples. However there no agreement which “cutoff” should used switch between two. subjective decision can avoided by using single-grain concordia ages. These represent kind weighted mean methods, offers precision than either latter methods. A second how define discordance cutoff “good” “bad” Discordance usually defined as (1) relative difference dates. However, this shows several other definitions are possible well, including (2) absolute difference; (3) common-Pb fraction according Stacey–Kramers mantle model; (4) p value concordance; (5) perpendicular log ratio (or “Aitchison”) distance line; (6) maximum likelihood composition on line. Applying these six filters 70 869-grain dataset zircon compositions reveals (i) filter tends suppress components in spectra, whilst inflating older components; (ii) likely reject grains less ones; (iii) p-value-based has undesirable effect biasing results towards least measurements; (iv) log-ratio-based strictest Proterozoic lenient Phanerozoic Archaean (v) all produces best results, sense it most closely match those unfiltered data: sharpens but does not change their shape. popular definition fares worst criterion. All methods presented have been implemented IsoplotR toolbox geochronology.