ASSESSING THE RELIABILITY OF QUATERNARY GEOCHRONOMETERS BY COMPARING INDEPENDENT DATES FROM ALLUVIAL AND FLUVIAL SURFACES

To assess whether deformation rates can be reliability estimated with only a single Quaternary geochronometer, we compiled ¹⁰Be exposure, U-series, luminescence, and radiocarbon (¹⁴C) dates from >300 alluvial and fluvial surfaces. Our results support the view that reliable rates should be based on multiple dates. We assess scatter in ¹⁰Be clast dates, depth profile modeling, methods for estimating inherited signal, and agreement between different geochronometers. Key conclusions of this review are:

(1) Scatter in ¹⁰Be clast dates is common: Of 284 datasets, the date range is a significant proportion (94% on average) of the median.

(2) A third of studies attempt to improve precision by removing outliers. However, half of the datasets contain around the same number of dates (≤5) as were considered outliers by a third of studies, raising questions regarding minimum sample sizes. Analysis of 47 datasets with >10 dates shows that samples of ≤5 dates have less than a 68% chance of replicating the median date within 1σ, casting doubt on smaller datasets.

(3) Even when contrary to theory, ¹⁰Be dates tend to be systematically older than others. Only 44 of 73 cases where surfaces were dated with two methods do the dates overlap at 1σ. 24 of 73 median dates differ by a factor of >2. Agreement improves somewhat when excluding outliers or all but the minimum ¹⁰Be date, but in general ¹⁰Be dates are safely interpreted as maximum age constraints.

(4) ¹⁰Be concentrations in active wash samples rarely match depth profile inheritance suggesting that such samples do not accurately approximate inherited nuclide components.

(5) Depth profiles constrain inheritance for smaller grain sizes, but in 14 of 61 cases were too scattered to be modeled. Only 22 of 61 depth profiles yielded reliable dates, of which only 8 adhered exactly according to theory, likely the result of incremental deposition or variable catchment residence times.

Implications: Slip rates for ~120 fault segments in the current Uniform California Earthquake Rupture Forecast were constrained with quantitative geochronometers. Of these only 16 used two dating methods, half of which do not agree within 1σ and individually would predict very different slip rates. Rates based on dates from a single geochronometer should only be accepted in the absence of datable material, not funding.