Error analysis and modelling of double saturating exponential dose response curves from SAR OSL dating

Abstract

Increasingly observed in single-aliquot regenerative dose (SAR) optically stimulated luminescence (OSL, also termed photon stimulated luminescence or PSL) dating studies of sedimentary quartz are dose response curves that at high doses are not satisfied by a single saturating exponential (SSE) regression model. Commonly these can appear to be satisfied by a SSE+Linear (E+L) regression model, but some authors have proposed that a double saturating exponential (DSE) model would more closely fit the observed dose response curves (DRCs), especially in the region of highest applied doses. As error analysis for SAR equivalent dose (De) values derived from a DSE model is not yet available through the widely available Risø supplied software (Analyst), we present here a regression and error analysis scheme for DSE SAR data, and also a simple charge traffic model that generates approximate DSE dose responses. To illustrate results from our error analysis, we employ two SAR high dose data sets for fine-grain quartz, and compare graphically the SSE, E+L and DSE fits for each data set. These comparisons show clearly that such data are more closely fitted by a DSE regression than by the other two models. This result, and the charge traffic model, lend validity to the physical reality of DSE regression models, and have implications for quartz SAR dating of older sediments.