EVALUATING VARIABILITY IN 10Be BASIN-AVERAGED EROSION RATES FROM DRIFT CREEK, AN OREGON COAST RANGE WATERSHED

The use of terrestrial cosmogenic nuclide (TCN) concentrations in fluvial sediment as a measure of basin-averaged erosion rates relies on the assumption that sediments are well-mixed, such that nuclide concentrations accurately reflect the spatial contribution of source areas upstream. Models suggest that sediment delivery via landslides can significantly perturb the local inventory of TCN in channels, depending on the depth, size, and frequency of landslides (Niemi et al., 2005; Yanites et al., 2009). However, there are few empirical data that provide insight into the scale of drainage area that yields representative cosmogenic erosion rates in landscapes prone to landslides. Here we investigate variability in erosion rates derived from both cosmogenic ¹⁰Be and ²⁶Al within the Drift Creek watershed in the central Oregon Coast Range. Drift Creek is an ~176 km² watershed almost entirely underlain by Eocene sandstones of the Tyee Formation and has only been minimally perturbed by logging. The Oregon Coast Range is a highly dissected landscape prone to landslides which may bias TCN inventories at different spatial scales. Individual sample locations (n=24) in nested sub-watersheds span a range of drainage areas from 0.9 – 176 km², allowing an assessment of TCN inventories and mixing throughout the watershed. We find that ¹⁰Be erosion rates vary from ~60 to ~145 m/Myr across Drift Creek; we observe high variance in sub-catchments with drainage areas less than ~11 km² (61 ± 5 to 139 ± 13 m/Myr) with an average erosion rate of 99 ± 43 m/Myr. At larger drainage area, erosion rates rapidly converge toward the outlet average of ~127 ± 13 m/Myr. The average erosion rate for 12 individual trunk stream samples is 119 ± 45 m/Myr. Results from ²⁶Al track consistently with those from ¹⁰Be. The apparent spatial scale (~11 km²) required for adequate mixing in Drift Creek is consistent with model simulations suggesting that sufficient spatial averaging occurs over ~10¹ to 10² km² when landslides are shallow and/or infrequent (e.g., Yanites et al., 2009). Our results provide empirical support for the commonly employed strategy of sampling moderately sized tributary watersheds to estimate long-term erosion rates in active mountain belts.