TRACERS DERIVED IN SITU REPLICATE AND MIX BETTER THAN FALLOUT TRACERS INCORPORATED INTO GRAIN COATINGS (Invited Presentation)

Cosmogenic isotopes, short-lived radionuclides, and thermochronometric indicators are measured in river sand to quantify erosion rates, trace sediment sources, and/or infer erosional processes. Interpretations are often based on the rarely-tested assumptions that (1) tracer concentration is time-invariant and (2) sediment is well-mixed and representative of the upstream area. Better understanding of when and where these assumptions break down and what sampling strategies minimize variance will improve science done using detrital river sediment. We tested these assumptions in SW China by measuring mixing at 13 river junctions in 3 watersheds and by collecting 113 replicate pairs at two different time periods or from two different nearby sites. All samples were analyzed for in situ ¹⁰Be (¹⁰Be_i), meteoric ¹⁰Be (¹⁰Be_m), ²¹⁰Pb_ex, and ¹³⁷Cs. Due to low ¹³⁷Cs and ²¹⁰Pb_ex activity in Chinese samples, we also tested replicability of ¹³⁷Cs and ²¹⁰Pb_ex at 3 sites in northern Ohio. We find that 2 of the 3 China watersheds mix well for ¹⁰Be_i and only 1 mixes well for ¹⁰Be_m when considering the concentration of ¹⁰Be at the outlet compared to the area-weighted mean of headwater samples. No watersheds mix well for ²¹⁰Pb_ex. Individual junctions mix poorly for all three systems. Faster eroding basins, larger basins, and junctions where one tributary is much larger than the other tend to mix better. The most important factor in setting mixing efficacy is the fraction of the basin area contributing to the downstream sample but not the upstream samples. Individual pairs of samples often replicate poorly (-150-100% different), but the population of all replicate samples replicates within 2σ for ¹⁰Be_i but not ¹⁰Be_i. ²¹⁰Pb_ex and ¹³⁷Cs replicate poorly. ¹⁰Be_i replicability is worse in steeper basins, suggesting that stochastic processes affect reproducibility. ¹⁰Be_m­­ replicability is worse in larger basins suggesting that it is a non-conservative tracer_­­. Poor mixing and replicability of short-lived fallout radionuclides suggest that these isotopes may be best used as tracers over short distances and that sampling should happen within as short a period of time in a study area. For ¹⁰Be, our analyses suggest that samples should be amalgamated from multiple sites or multiple time periods to increase representativeness.