Reliability of palaeoclimatic reconstructions from synchrotron-radiation X-ray fluorescence (SR-XRF) geochemical data: a fitness-for-purpose assessment of measurement uncertainty

Abstract

Commercial XRF core scanners (ITRAX, Avaatech) cannot reliably quantify bromine (Br) in oligotrophic lake sediments at the background concentrations typical of these environments (1–10 mg kg−1), where their effective limit of detection (LOD) is 20–25 mg kg−1. Synchrotron-radiation X-ray fluorescence (SR-XRF) achieves sub-mg kg−1 detection limits but the certified analytical procedure assigns a target relative measurement uncertainty of 30% (k = 1) for Br ≤ 10 mg kg−1. Whether this uncertainty compromises palaeoclimatic reconstructions has not been evaluated systematically. We apply a fitness-for-purpose framework to SR-XRF datasets from three Siberian lake-sediment cores (n = 1652 measurement points total). For the log-ratio palaeoproductivity proxy ln(Br/Rb), the natural variation in the records exceeds the propagated standard measurement uncertainty by a factor of 10 or more in all examined cores. For Kucherlinskoye Lake, 78% of measurement points fall below the LOD of both commercial scanners, establishing SR-XRF as the only viable measurement method for this system. Palaeotemperature reconstructions calibrated against instrumental records reproduce known climatic events and yield statistically significant correlations (r = 0.58–0.86; p < 0.01). We conclude that the 30% target uncertainty is fit for purpose for log-ratio-based palaeoclimatic reconstruction, and we propose concentration-dependent applicability ranges to guide the complementary use of both measurement approaches.