This study examined the relationship between slope failure and precipitation using long-term data for the Chitou tract, which is in the Experimental Forest of National Taiwan University, located 20 km south of the Chi-chi Earthquake epicenter. The University Forest maintains a variety of records that are a useful source of long-term data. Repair records for forestry roads, which assure long-term uniformity of temporal data owing to the high priority given to repairing forestry roads, were used to discuss the characteristics of slope failure. The relationship before and after the earthquake was compared to clarify the indirect influence of the Chi-chi Earthquake on subsequent slope failures. As a result, we showed that the earthquake affected subsequent slope failures, although the earthquake itself caused no slope failures directly when it happened on 21 September 1999. After the earthquake, the first slope failure was triggered by a precipitation event involving 91.5 mm of rain over 5 days in February 2000. By contrast, the smallest precipitation event that triggered slope failure before the earthquake was 210.5 mm over 5 days. In July 2001, the second slope failure event following the earthquake set a record for the most slope failures in the history of the Chitou tract. By the time of the slope failure event in May 2002, the earthquake no longer had any notable influence.
Numerical simulation model that can reproduce the riverbed variation caused by a debris flow in a mountainous river is proposed. In the model, the sediment material is composed of two grain-size classes and the one-dimensional governing equations for a stony debris flow are employed. The governing equations are discretized according to the staggard scheme. The influence of a sabo dam upon the debris flow is built in the model by changing the outflow condition at the concerned point where the flow velocity, flow discharge and sediment discharge are calculated. The rapid deposition phenomena at the upstream of the sabo dam can not be represented by former models in which the deposition velocity equation is used. The calculated deposition velocity by such equation is too small in the upstream area of the dam, because it is proportional to the flow velocity. Usual deposition velocity equation was developed under the condition where the riverbed gradient changes gradually, therefore it is not applicable for the area where the flow depth and velocity suddenly change. In the present model, the rapid separation of sediment from the flow at the upstream of the sabo dam is taken into account, consequently the larger deposition velocity can be estimated. We try to calculate a long distance runoff process of a debris flow by the present model, and investigate the influence of the characteristics of sediment on the hydrograph.
A pipeflow model was developed to examine preferential flow through soil pipes within slopes and its influences on slope stability. To assess the new pipeflow model, flume experiments were conducted and its results were simulated by the model. Soils were pilled into a flume to prepare a comparatively large sloping soil domain (length = 5.0 m, width = 0.5 m, depth = 0.35 m, and gradient = 20°) and used for the experiments under 3 different conditions, i.e.; no - pipe (Run 1), a pipe from upslope end through downslope end (Run 2), and a pipe whose outlet located within the soil domain (Run 3). Output from the matrix at downslope end, pipe flow from pipe outlet, and distribution of pore water pressure at bottom of sloping soil domain responding to artificial rainfall on soil surface (82 mmlhr) were continuously measured. Artificial soil pipes (O.D. = 2.6 cm) were constructed of stainless steal mesh and felt so as to reduce flow resistance from soil matrix to the soil pipe, and placed 2.5 cm above the bottom and the center of the soil domain. Experimental results demonstrated that pipeflow (Run 2) decreased soil pore water pressure and made the slope more stable comparing to the no-pipe condition (Run 1), however backflow from the pipe outlet to the soil matrix (Run 3) increased the soil pore water pressure at the point and generated surface erosion. Through the experiments, pore water pressure distributions not only along the soil pipe direction but also along a direction across the soil pipe were detected ; this finding was usually neglected in similar conventional experimental studies. The pipeflow model presented in this paper simulated the experimental results of pore water pressure and flow rates, and its availability was confirmed. Furthermore, distribution of pore water pressure within the soil domain obtained from the model calculations were compared to surface morphology after the surface displacement obtained in the flume experiments. Results demonstrated that the model can calculate the pore water pressure distribution with high accuracy and may be applicable to the stability analysis for the actual slope in which preferential pathways such as soil pipes exist.
An unusually large number of typhoons visited Southeast Asia in 2004; 27 to the beginning of December, and 10 hit just in Japan. Heavy rains caused serious debris - related damage. Rainfall totaling over 1, 000 mm caused many shallow and some deep slope failures. In Taiwan, mountainous regions were also damaged severely by flash-floods and landslides because of the heavy rains accompanying Typhoon Aere. And deep slope failures causing debris flows measuring a million m3 killed 15 people in the east mountainous region of Hsinchu Prefecture, Taiwan. If, as seems likely, the sea temperature east of the Philippines remains high in the coming year, we can expect many more and large typhoons to be generated. If so, there is a high possibility of deep slope failures likely in Taiwan up to 10 times large as those that occurred in Izumi-shi, Kagoshima Prefecture, and Minamata-shi, Kumamoto Prefecture, Japan. The rainfall with largely amount and highly intensity could result in increased the number and area size of slope failures, however, we have little knowledge of enlarged deep slope failures. This paper describes debris disasters caused by deep slope failures and debris flows triggered by the heavy rainfall during Typhoon Aere from 23 to 25 August 2004 in the east mountainous region of Hsinchu Prefecture, Taiwan.
The purpose of this paper is to compare the characteristics of heavy rainfall disaster information in Taiwan with those of such information in Japan based on field research, interview survey, and Internet survey between 2003 and 2004. The Taiwan Central Weather Bureau (CWB) has 387 raingauge observatories. The density of observatories is 1 station per 93km2, which is higher than that established by the Japan Meteorological Agency (AMeDAS, 1 station per 287km2). The Taiwan Soil and Water Conservation Bureau publishes any real-time danger-level of sediment disaster in various places on their website. This information is calculated based on precipitation data of the CWB. In Japan, raingauge networks are administrated by the Meteorological Agency, River Bureau, and individual prefectures, and these data is using independently in general. That is, the method of using precipitation data in Taiwan is more efficient than that in Japan. On the other hand, we can safely say that the standard of heavy rainfall warning in Taiwan is rough than that in Japan ; only one value (1-hour precipitation over 15 mm and cumulative precipitation over 130 mm) is provided for all Taiwan. When this warning is announced, it is sent from CWB to all municipality offices directly. It is possible that the method of announcing warnings in Taiwan is achieves more rapid delivery than that in Japan.
A planted Japanese black pine (Pinus Thunbergii) shelterbelt, which had grown upward satisfactorily in the back of an artificial embankment, ceased growing after reaching a certain height. The apparent reason for this was that the tree tops of the shelterbelt had come out above the area where the wind speed had been mitigated by the embankment and were exposed directly to strong sea winds. A wind break fence was thus constructed on the embankment in order to protect the trees from the wind, and the trees have begun growing again. The embankment is 2.66 m in height, 2.0 m in width at the top and the slope is 1: 1.5. The fence is 2.35 m in height. The position where tree height had stagnated in the past was 1.1-1.9 m above the ground. The present tree height range is 4.2-7.5 m, and the trees have grown 1.6-1.9 m during the last five years (in the five years since the fence was constructed). The wind tunnel experiment was conducted in order to consider how effectively the combination of the embankment and the fence has mitigated the wind speed, both in absolute terms and in comparison with a large embankment which has the same height as the embankment plus the fence. Three models were made, i.e., of the embankment, of the combination of embankment and fence, and of the large embankment. The model scale was 1/25. Wind speed in the wind tunnel was 4.8 m/s and 6.8 m/s, assuming local wind speed of 14 m/s and 20 m/s. The addition of the fence on the embankment remarkably expanded the area where the wind decreased, and there was even an area where the wind speed ratio was under 0.2, which had not been seen before the fence was constructed. The combination of the smaller embankment and the fence was much more effective than the large embankment and it appears to have enabled the trees to begin growing upward again.