Exploring the Uncertainties of Detrital Zircon Provenance Analysis With Statistical Provenance Modelling

Abstract

Traditional provenance analysis often relies on simplifying assumptions, such as uniform erosion rates within source areas or homogeneous zircon fertility across source units, which can introduce significant uncertainty into interpretations. While prior work has explored implications of isolated examples of one or more of these assumptions being violated, a comprehensive sensitivity analysis of the potential influence of these factors is lacking. Here we address this gap with a set of statistical simulations where erosion rates and zircon fertility are systematically varied. Zircon fertility is modulated by zircon grain size and zirconium concentration, to simulate sediment provenance and quantify their effects on resultant detrital zircon U–Pb age distributions. Using PDP cross-correlation and K–S tests, we compare a model based on area-weighted contributions against two alternative models: one weighted by zircon mass and another by grain abundance. By exploring a wide parameter space, we evaluate the magnitude and sources of uncertainty that influence provenance interpretations. To investigate these effects, we conduct three suites of numerical experiments including: (1) varying synthetic source population complexity; (2) varying erosion rate, zircon fertility, and zircon grain size; and (3) using large sample size empirical datasets as sources to assess realism. In each of these experiments we assessed the divergence between an area-only mixture model and ones weighted by zircon mass or number of zircon grains. Results show that increased complexity and overlap among source populations lead to higher variance and reduced distinctiveness between the different models. For the empirical datasets, even substantial variation in parameters yields only subtle differences between the different weighting schemes. While these simplified scenarios may not capture the full complexity of real-world landscapes, they offer valuable insights into the underlying controls and potential biases that affect interpretations of detrital zircon datasets.