We investigated natural disasters such as floods and sediment movements in the Nakatsugawa watershed over a period of 406 years, using literatures including historical documents. We found that natural disasters have occurred irregularly, and classified the 406-years period into three categories, High Frequency (HF), Middle Frequency (MF), and Low Frequency (LF), based on disaster frequency. The HF category is divided into three periods, (A. C. 1687-1705, A. C. 1789-1794, and A. C. 1842-1870), MF into two periods (A. C. 1795-1841 and A. C. 1928-2005), and LF into three periods (A. C. 1600-1686, A. C. 1706-1788, and A. C. 1871-1927). Historical activities using the riverfront or using its water resource were recorded mainly during LF, indicating that LF periods would have been considered safe because few disasters occurred. We compared time series of the categories in the watershed with other areas and found that the time series were not dominated by external events (affecting areas outside the watershed) such as climate changes, or large earthquakes. Watershed-specific events leading to classification of periods as HF., MF,. or LF depended on the existence of unstable remanent sediment in the watershed. We verified these findings with historical documents and photographs.
The sediment-related disaster warning system as one of the non-structural measures for the prevention of disasters has been operated since 2005. But the critical line of the sediment-related disasters to predict the occurrence of the disasters is not well understood. Also the information to judge the issuance of the evacuation orders is limited. As a result it is difficult to warn in a timely manner, residents to evacuate, at present. To solve these problems, the utilization of the logistic regression model is investigated in this paper. It is experimented to express the indices which are utilized to judge the issuance of the sediment-related warning information or the evacuation orders, by the easily understood probabilistic values using the logistic regression model. It is confirmed that the critical line of the sediment-related disasters can be decided straightly from the calculated disaster occurrence probability and the hourly probability of disaster occurrence can also be obtained. It becomes clear that the logistic regression model can be applied to areas where the data of disasters are scarce and that the calculated exceeding probability can be utilized as the warning line.
Stemflow serving as point input to the forest floor could have great implications for soil water dynamics and slope stability around a tree. To clarify the effect of stemflow on soil water dynamics and slope stability, we conducted field observations and numerical simulations of rainwater infiltration processes on a forested hillslope. The results of field observations indicated that locally concentrated rainwater input attributable to stemflow on the downslope side of the tree trunk caused the large and rapid increases in water content and pore water pressure in the region downslope of the tree stem, resulting in the development of an asymmetric saturated zone around the tree. Rainwater infiltration simulations were conducted with the proposed model (Liang et al., 2009 a), in which stemflow was parameterized as a source flux spring in soil layers. The simulation results were then used for the slope stability analysis. The proposed model showed more rapid decreases in the minimum safety factor attributable to greater generation rates of saturated zones downslope of trees. This model also resulted in the timing of minimum safety factor < 1.0 at an earlier stage than in the conventional model. Furthermore, the proposed model showed evidence of the risk for slope failure throughout all parts of the slope, though the conventional model showed only the risk of slope failure in the lower part of the slope.
To estimate debris flow peak discharge and hydrograph are important for effective disaster prevention. Ikeda et al. (2007) has presented occurrence criteria of debris flow must be evaluated not only rainfall condition but also supply process of materials. At the Kitamata Valley of Name River, three debris flows occurred in 1999 as the result of landslide dam that blocked the riverbed near an elevation 1800 m in 1998. The largest debris flow of these was occurred on June 27, forming a conspicuous peak with relatively high velocity. Analyzing the occurrence situation and flow property of this debris flow by field research, aerial photograph interpretation, and CCTV camera figures, we estimate the outbreak process and hydrograph of this debris flow. Using the two layer sediment transport model Takahama et al. (2000) have developed, shape of hydrograph and peak discharge was well explained with the exception of total volume and a duration of debris flow. Also the hydrograph shows that the supply process of materials will be an outburst of a landslide dam. Moreover, comparison the peak discharge and hydrograph between the case with landslide dam and without it, the peak discharge is larger than without landslide dam, also the total volume of debris flow.
Heavy rainfall and snowmelt cause sediment disasters. The wetter the soil moisture is, the higher the sediment disaster frequency becomes. In addition, antecedent precipitation and snowmelt influence sediment disaster generation. We investigated the moisture environment around the epicenter before and after the Iwate-Miyagi Nairiku earthquake of 2008 based on an antecedent soil moisture index (ASI30) using AMeDAS data (precipitation, air temperature, snow depth). In order to calculate ASI30, amount of snowmelt during snow cover period was estimated using advanced degree-day method. The ASI30 was highest during the snowmelt season and after heavy rainfall more than 100 mm. When the earthquake occurred, the ASI30 values at the Matsurube and Komanoyu sites were 18.0 and 20.4 mm, respectively. These values did not appear to affect the damage caused by the earthquakes. Previously recorded ASI30 values after heavy rainfall events or during the snowmelt seasons of heavy snowfall years were larger than were those recorded after the earthquakes. It was important to advance the remediation activity of the disaster, and to continue the monitor around the epicenter in the future.