Test of thermoluminescence dating of loess from New Zealand and Alaska

Abstract

The accuracy of thermoluminescence (TL) ages for loess (and sediments in general) greater than ∼100 ka is disputed. We tested the accuracy of three common TL sediment-dating techniques applied to 4–11 μm sized polymineral grains from known-age loess, using 16 samples from North Island and South Island, New Zealand, and 2 samples from central Alaska. Estimated sample ages range from 20–26 to ∼800 ka. We varied the optical bleaching spectrum, the selected window of the emission spectra (ultraviolet or uv, blue and green wavelengths), the pre-readout heat treatment, and the TL equivalent-dose measurement technique. Most of the 20–26 ka samples gave TL age underestimates of 3–6 ka that may be attributed to post-burial open system behavior. For the older samples, the partial-bleach TL method gave expected ages up to ∼300 ka, and the total-bleach TL method produced expected ages above ∼100 ka. The partial-bleach regeneration TL method gave significant age underestimates for samples older than ∼100 ka, with a maximum TL age of ∼250 ka for samples having expected ages up to ∼350 ka. Two 300–360 ka,samples which gave total-bleach age underestimates with use of uv TL and green TL, gave expected ages with use of blue TL. These results demonstrate that reliable TL ages for loess from New Zealand and Alaska up to ∼800 ka can be obtained if uv (and green?) emissions and the various regeneration methods are avoided. This age range is well above the former putative 80–100 ka upper age limit for TL dating of loess from other regions, which was thought to be a global limit. Application of our successful procedures to such loess is encouraged.