VARIABILITY IN EROSION RATES DUE TO DISCRETE SUPPLY PROCESSES IN THE EASTERN CENTRAL RANGE, TAIWAN

Suspended sediment records are commonly used as a direct measurement of erosion rates in mountain basins where the majority of eroded material is transported as suspended sediment. In order to use these erosion rates in understanding larger scale processes we must first understand the accuracy of such erosion rates. The main limitations in such analyses are due to the inadequacy of the data and in the discrete nature of sediment supply, typically in the form of bedrock-involved landslides. While many authors have looked at the role of inadequate data in determining erosion rates, we have investigated the inherent variations in such erosion rates that arise due to discrete sediment supply.

We calculate erosion rates for the 12 rivers of the eastern Central Range (ECR) in Taiwan directly from suspended sediment data using mean sediment loads, traditional rating curves and a new corrected rating curve. Our resulting erosion rates vary between .7 to 21.3 mm/yr. In order to address the uncertainty in these erosion rates due to the short records, infrequency of observations and, in particular, the discrete nature of the sediment supply, we have developed a model of sediment supply and transport in a landslide dominated basin. The model consists of stochastic water discharge generation, stochastic landslide generation and a deterministic sediment transport algorithm. We calibrate the model to the Chih-Pen river of the ECR using the observed water and sediment time series as well as known landslide magnitude-frequency statistics within the ECR. Through Monte Carlo simulations, we are able to estimate the expected variation in erosion rates for a range of parameters. Our simulations show that a 27 year record for the Chih-Pen, with an observed erosion rate of 9.5 mm/yr, has an expected standard deviation of 6 mm/yr. Similar variances are expected for other rivers within the ECR.