[H-GM04] Geomorphology

[HGM04-P01] Relationship between hillslope angles and denudation rates deduced from sediment yields in catchments: A case study considering bedrock exposure and landslide

*Shigeru Sueoka¹, Yuki Hiura², Tetsuya Komatsu¹ (1.Tono Geoscience Center, Japan Atomic Energy Agency, 2.JGI, Inc.)

Keywords:sediment yield in catchment, denudation rate, hillslope angle, bedrock exposure, landslide

In general, the denudation rate in mountainous areas has a good correlation with some topographic parameters, e.g., the hillslope angle and altitude dispersion (e.g., Ahnert, 1970; Ohmori, 1978; Portenga & Bierman, 2011). Using this property, an attempt is being made to reveal the relationship between denudation rates and topographic parameters based on sediment yields in catchments across Japan, and to compute the denudation rate distribution in the Japanese Islands (Fujiwara et al., 1999; Hasegawa et al., 2005). Recently, however, it has been pointed out that it is difficult to predict the denudation rate based on such a single topographic parameter (e.g., Montgomery & Brandon, 2002; Binnie et al., 2007; Matsushi et al., 2014; Korup et al., 2014). This is because the denudation rate tends to behave nonlinearly in steep mountainous terrain, diverging to infinity near a threshold (critical angle), and shows extremely large variation near the critical angle. One of the reasons for this is that the mechanism of denudation changes because bedrock is exposed on slopes that exceed the angle of repose of the soil. In other words, soil creep is dominant on soil-mantled hillslopes, whereas rockfall and landslides are dominant in bare-bedrock hillslopes. Recently, DiBiase et al. (2023) proposed a model applicable to bare-bedrock hillslopes based on the nonlinear sediment transport model for soil-mantled hillslopes proposed by Roering et al. (2007). The validity of this model was tested in the San Gabriel Mountains of the western U.S., where denudation rates predicted by their model were more consistent with those estimated by the ¹⁰Be dating method than the model of Roering et al. (2007). In this study, we attempted to apply the model of DiBiase et al. (2023) to the denudation rate data obtained from sediment yields in catchments across the Japanese Islands for better prediction of the denudation rates. In addition, we also examined the relationship between the percentage of landslide areas and the dispersion in denudation rates because DiBiase et al. (2023) pointed out the influence of landslides as a major cause of the discrepancy between model predictions and actual denudation rates. The 100 dams were selected based on the total water storage, the period for which sediment volume data were obtained, the initial sediment volume during the period, and the linearity of the sediment volume increase (JAEA & CRIEPI, 2019, 2020). The bare-bedrock areas and landslides were identified based on the Fundamental Land Classification Map at 1:200,000 (GIS data) and the Landslide Distribution Maps of NIED, respectively. Based on the relationship between denudation rate and mean slope in the 100 dam catchments, the critical angle was determined to be ca. 35° from the fitting of the equation of Roering et al. (2007). In addition, based on the relationship between the mean slope of the catchment area and the percentage of bare-bedrock area, the slope at which bedrock begins to be systematically exposed (S*) was estimated to be ca. 29°, and based on the relationship between the mean slope of the entire catchment area and the mean slope of the bare-bedrock area, the maximum and minimum slopes of the bare-bedrock hillslopes (S_{max rock}, S_{min rock}) were estimated to be ca. 40° and 35°, respectively. These values were used to approximate the relationship between denudation rate and mean slope in the catchment of 100 dams using the equation of DiBiase et al. (2023). The approximation by Roering et al. (2007) tended to underestimate the denudation rate at gentle slopes and overestimate it at steep slopes, and the misfit values between the model and measured values increased near the critical angle. In contrast, the DiBiase et al. (2023) equation kept both misfit values within a certain range, including angles beyond the critical angle. In contrast, by the DiBiase et al. (2023) equation, the misfit values were within a certain range, including angles beyond the critical angle. Therefore, it can be expected that their model is valid for predicting denudation rates in the last few years to few decades estimated from the dam sedimentation rate. On the other hand, unlike the case of the San Gabriel Mountains, there was no clear trend of increase in the misfit values between the modeled and measured denudation rates as the landslide area increased. The reason for this might be that in the San Gabriel Mountains, the ¹⁰Be method calculates an average denudation rate of ca. 10² to 10⁴ years, so the probability of including a landslide event during the period of interest is higher than with the dammed sediment approach, which covers a period of ca. 10⁰ to 10¹ years.