Quantifying landslide recurrence intervals and tracing landslide-derived sediment with in-situ 10Be and 14C

Abstract:

Landslides are a primary erosional process in many orogens, and one of the most sensitive surface process to tectonic and climatic perturbations. However, it remains extremely difficult to constrain long-term rates of landslide activity, and hence landslide-derived sediment generation, because the physical records of landsliding are often removed in <10² yrs. Here, we use the concentrations of in-situ ¹⁰Be and ¹⁴C from recent landslide deposits from Fiordland and the Southern Alps of New Zealand to estimate landslide frequencies over 10³-10⁴ yr timescales and to test whether paired ¹⁴C-¹⁰Be measurements can be used to trace erosional depth-provenance.

We show that ¹⁰Be concentrations in landslide deposits can be used to estimate landslide recurrence intervals and frequency, when combined with detailed DEMs of the landslide scars built from photogrammetry. Our ¹⁰Be-based, long-term landslide frequency estimates from samples collected at lower elevations are an order of magnitude lower than the frequencies inferred from short-term published landslide inventories based on aerial photography observations up to the tree line. We derive sediment flux estimates from both data sets, and compare them to a new data set of ¹⁰Be catchment-averaged erosion rates, to examine whether these differences reflect spatial or temporal gradients in landslide activity. Finally, we show how paired in-situ ¹⁴C-¹⁰Be measurements offer some promise as a tool to trace erosional depth-provenance, and hence potentially fingerprint landslide-derived sediment. These new approaches open up possibilities to quantify sediment generation from landslides on >10² yr timescales.