Watershed erosion estimated from a high-resolution sediment core reveals a non-stationary frequency-magnitude relationship and importance of seasonal climate drivers

Abstract:

The inclusion of paleo-flood events greatly affects estimates of peak magnitudes (e.g., Q₁₀₀) in flood-frequency analysis. Likewise, peak events also are associated with certain synoptic climatic patterns that vary on all time scales. Geologic records preserved in lake sediments have the potential to capture the non-stationarity in frequency-magnitude relationships, but few such records preserve a continuous history of event magnitudes. We present a 10-meter 2000-yr record from Upper Squaw Lake, Oregon, that contains finely laminated silt layers that reflect landscape erosion events from the 40 km² watershed. CT-scans of the core (<1 mm resolution) and a ¹⁴C-dated chronology yielded a pseudo-annual time series of erosion magnitudes. The most recent 80 years of the record correlates strongly with annual peak stream discharge and road construction. We examined the frequency-magnitude relationship for the entire pre-road period and show that the seven largest events fall above a strongly linear relationship, suggesting a distinct process (e.g., severe fires or earthquakes) operating at low-frequency to generate large-magnitude events. Expressing the record as cumulative sediment accumulation anomalies showed the importance of the large events in “returning the system” to the long-term mean rate. Applying frequency-magnitude analysis in a moving window showed that the Q₁₀₀ and Q₁₀ of watershed erosion varied by 1.7 and 1.0 orders of magnitude, respectively. The variations in watershed erosion are weakly correlated with temperature and precipitation reconstructions at the decadal to centennial scale. This suggests that dynamics both internal (i.e., sediment production) and external (i.e., earthquakes) to the system, as well as more stochastic events (i.e., single severe wildfires) can at least partially over-ride external climate forcing of watershed erosion at decadal to centennial time scales.