The Woo-Wan-Chai landslide area is located between 42~45K on the 18th Provincial Road (also named Alishan Road). It is a large-scale deep-seated landslide. During the early operation period of Alishan Road, the road would be distorted and landslides would occur whenever there was heavy rainfall. Although the slope in that area had been rectified several times, sliding would still continue to occur. Between 2004 and 2006, the Directorate General of Highways (MOTC) and the Land Engineer Consultant Co. conducted detailed geological surveys and groundwater surveys to find out the sliding mechanism, and strategized the multi-stage remediation plan. According to the survey results, the sliding area encompassed nearly 50 hectares, and the sliding depth was at about 80 meters. The annual displacement was about 20 centimeter. Between 2010 and 2017, 8 large drainage well projects were completed in two phases in the area with deep and active sliding bodies. During the construction period and operation phase, the activity of the slope was closely monitored and the remediation project was adjusted if necessary. For the active N2 and N3 main sliding bodies in the Woo-Wan-Chai landslide area, their displacement rates decreased from about 18 mm/month to less than 2 mm/month after the construction of the large drainage wells. This paper will document how the relevant investigation surveys were integrated into the formulation of effective countermeasures to control the deep-seated sliding of the Woo-Wan-Chai landslide area.
This study aims to discuss the deformation characteristics of dip and anti-dip slopes under the influence of scale effect. The dip slope model consists of a bedding angle of 60 degrees and a slope angle of 30 degrees, while the anti-dip slope model has a bedding angle of 75 degrees. Centrifuge modeling tests and numerical simulations based on discrete element method (DEM) were employed as the major research tools. Results show that mild to severe folding would appear near the toe of the dip slope. With an increase in slope scale, the deformed area within the slope would become more massive. For the anti-dip slope, the major failure pattern is flexural toppling. The location with the most significant deformation for each rock layer can be pin-pointed. If these locations from all layers are connected together, the potential sliding mass of the anti-dip slope can be identified. The sliding mass of an anti-dip slope would also become more extensive as the scale increases.
After the 2009 Typhoon Morakot, rainfall-induced large-scale landslides have become a highly-concerned issue related to slope disasters. The identification of rainfall condition that can cause landslides is the key to establish a disaster alert system. In the past, it is difficult to analyze the triggering rainfall because the exact time of occurrence of large-scale landslide is difficult to obtain. By analyzing the landslide signal from broadband seismic records (BATS), it is possible to extract the time information of the landslides and to correlate this information with the rainfall conditions for historical landslide events. In this study, the ground signals triggered by 158 large-scale landslides occurred in 2001-2017 are analyzed. The landslide-triggered seismic signals can provide the accurate initiation time of landslide which can then be used to identify the corresponding rainfall condition. Among the 158 landslides, 89 landslides are used to evaluate the average rainfall intensity, duration, and cumulative amount of rainfall. The rainfall threshold for large-scale landslides is defined as the critical rainfall condition with cumulative probability of 5%. Based on the analysis results, large-scaled landslides tend to occur when the amount of effective cumulative rainfall is larger than 497 mm, and the rainfall duration exceeds 24 hours.
For the past few decades, protecting infrastructure from slope hazards has been an important issue. The slope instability can be triggered by earthquakes, heavy rainfalls, or human activities. Slope stability problems have drawn a lot of attention after Chi-Chi earthquake because rock mass disturbance has occurred more often since then. In recent years, global warming has imposed a new challenge for civil engineers because it could induce slope instability problems. Past research indicated that using Hoek-Brown failure criterion in conjunction with finite element limit analysis can yield a more accurate rock slope stability evaluation than using Mohr-Coulomb failure criterion. In this study, Hoek-Brown model is employed to assess the rock slope stability. In addition, artificial intelligence technique has been adopted in the numerical analyses, which is shown to provide a quick and reliable rock stability assessment.
The coseismic process of the Aso-Bridge large-scale landslide induced by the 2016 Kumamoto Earthquake was investigated using a combined finite-discrete element method (FEA-DEA). The pre-failure mechanism and post-failure kinematic process of the coseismic landslide as determined from the numerical simulations were validated against the field investigation, digital elevation model, and published results by other researchers. Furthermore, the influence of the seismic acceleration, especially in the vertical direction, was examined. Based on the comparison, the combined FEA-DEA was found to be capable of analyzing the coseismic process of the large-scale landslide, including source displacement, failure mechanism, kinematic process, and deposition. Vertical seismic acceleration was found to impose a trivial influence on the runout behavior in the post-failure regime, but it played a significant role in identifying the initiation time of the landslide. This study has demonstrated that the proposed finite-discrete element method has the potential to simulate geotechnical problems involving pre-failure mechanism and post-failure kinematic process.
Typhoon Morakot had triggered many deep-seated landslides in 2009. In addition to the direct damages and affected areas caused by these deep-seated landslides, the landslide volumes could cause dam-up lakes and secondary hazards in later years. Therefore, accurate identification and mapping of these deep-seated landslides were vital for mitigation and management of deep-seated landslide hazard. The objective of this paper is to establish and verify the methods for identification and mapping of deep seated landslides. The deep-seated landslides can be identified by two methods: (1) aerial photos with digital elevation model (DEM) and landslide geomorphology characteristics; (2) LiDAR digital model for identification of linear structures of the deep-seated landslide scarps. Two case studies were conducted in this paper using these two methods. For Case Study No.1, the deep-seated landslide in Nantou area was identified using aerial photos and DEM. For Case Study No.2, the landslide in Kaohsiung area was identified by LiDAR. With the preliminary identification results, the stability analyses along with different remote sensing data were adopted to verify the identified deep-seated landslides. Based on the analysis results, the identified mapped area of the deep-seated landslide in Case Study No. 1 was consistent with the verification result. However, a more critical main scarp was found up-slope of the preliminary identified area for the deep-seated landside in Case Study No. 2. Hence, the mapped area of the deep-seated landslide was modified accordingly. Thus, we propose that multi-scale remote sensing data should be used for identification and cross-validation of the deep-seated landslide. Moreover, stability analyses should be conducted for further verification. Such procedure could provide a reliable and rational method for identification and mapping of the deep-seated landslides.
Exploring the Possibility of Establishing Mass Movement Warning System for Taiwan Using the Soil Water Index
土壤雨量指數(soil water index; SWI)可表現目前受當下及前期降雨影響的概估土壤含水量，是日本氣象廳做為全國坡地災害警戒參考的重要指標。近年來更有許多研究指出，坡地災害的發生與否和當時的SWI是否為同一點位SWI歷史排序之前位有關。因此，前人研究提出一量化之方法，計算某一點位的SWI與過去十年的最大值之比值(即為normalized soil water index; NSWI)，可作為坡地災害預測的重要指標。本研究利用2006–2017年間發生的坡地災害(n = 344)，分析臺灣誘發坡地災害發生之降雨事件期間的SWI與NSWI變化。結果顯示，坡地災害發生時的SWI介於2.90至700.93之間，平均為308.42；而NSWI則介於0.01至1.87之間，平均為0.84。平均來說坡地災害發生在NSWI最大值出現後的3.42小時之內。我們進一步將NSWI區分成0–0.2、0.2–0.4、0.4–0.6、0.6–0.8、0.8–1.0、及>1.0等六個範圍，依此對全臺灣的坡地災害發生的可能性，定義不同的警戒等級。本研究並藉由臺灣氣候變遷推估資訊平臺計畫(Taiwan Climate Change Projection Information and Adaptation Knowledge Platform; TCCIP)所開發的多資料多模式整合系統(Multi-data and Multi-model Integrated System; MMIS)串接過去的時雨量資料，建置可查詢SWI與NSWI在歷史降雨事件中的變化情形，以及在某特定時間下的數值之介面。未來若能進一步串接即時雨量，甚至未來數小時的預估雨量，將可提供臺灣在坡地災害預警系統上的一重大突破。
The soil water index (SWI) represents the estimated soil moisture content affected by current and antecedent rainfall, and is an important indicator used by the Japan Meteorological Agency (JMA) as a reference to issue nationwide mass movement warnings. In recent years, many studies had pointed out that the occurrence of mass movements is highly related to whether the value of SWI is within the top historical ranking (recorded at the same location). According to this finding, some studies hadproposed a quantitative method, which is to calculate the ratio of the SWI at a particular location to the maximum value recorded in the past decade (i.e. normalized soil water index; NSWI). This NSWI can serve as an important indicator for predicting mass movement occurrence. In this study, 344 mass movements occurred during 2006–2017 were analyzed. Changes in SWI and NSWI during numerous rainfall events were calculated. The result shows that the SWI at the time of mass movements was between 2.90 and 700.93, with an average of 308.42; while the NSWI was between 0.01 and 1.87, with an average of 0.84. On average, mass movements occurred within 3.42 hours after the maximum NSWI had been reached. We also defined different warning levels based on six ranges of NSWI: 0–0.2, 0.2–0.4, 0.4–0.6, 0.6–0.8, 0.8–1.0, and > 1.0. These levels would represent different possibilities of mass movement occurrence. This study also utilized the Multi-data and Multi-model Integrated System (MMIS), which was developed by the Taiwan Climate Change Projection Information and Adaptation Knowledge Platform (TCCIP), to link together the historical rainfall data and to establish an interface. This allows us to easily explore the changes in SWI and NSWI during historical rainfall events or at a specific time of interest. In the future, if we could further integrate this system with real-time rainfall data and the predicted rainfall amount for the upcoming time, it would provide a major breakthrough in the mass movement warning system in Taiwan.
In 2009, Typhoon Morakot hit Taiwan and triggered catastrophic failure of deep-seated landslides. The enormous impact thereof shocked people of the country, driving them to pay even more attention to the landslide early warning and monitoring systems. How to effectively interpret and track potential sliding surface has also become an important subject in the field with respect to hazard mitigation.
Space borne Synthetic Aperture Radar (SAR) imagery is a remote sensing technology that has been highly developed in recent years. With advantages of being applicable in all weather conditions and to large areas, SAR that has a fixed revisit period has now been widely used in slope disaster investigation. The records thereof include intensity, polarization, phase and other information. Among them, phase information can estimate centimeter-level deformation using the DInSAR technique. Nevertheless, the deformation data are easily affected by atmospheric effect, orbital errors and terrain effects, resulting in a decrease in measurement precision. Furthermore, as temporal baseline changes over time and ground surface deformations are inconsistent, it was hard to set the landslide warning threshold in the past. This study has adopted fuzzy membership function to reclassify vertical ground surface deformation measured by DInSAR.
The results indicate that, after eliminating terrain effects, potential sliding surfaces within deep-seated landslides can be rapidly identified; and that time-series analysis results can also be used to evaluate the activity of potential sliding areas. In this study, we have collected six pairs of SAR images of Gaoping River Basin using ALOS satellite between August 23 of 2009 and February 26 of 2011. Not only were these six pairs used to determine potential sliding surfaces, but also we have preliminarily verified the results of Namasia, Nangnisalu and Putanpunas River Basin; and established procedures for analyzing potential sliding surfaces using DInSAR.
Due to the poor geological conditions in the hillsides of Taiwan and the drastic changes in the global climate, landslides, earth slides and debris flows are prone to occur after extreme rainfall. The resulting complex hazards often lead to the loss of people's lives and properties. Well-organized site characterization and monitoring program can greatly reduce the risk. Engineering geophysical methods, especially seismic travel-time tomography and Electrical Resistivity Tomography (ERT), are effective ways to investigate the landslide area. On the other hand, slope inclinometer is widely used in practice to identify the sliding depths of landslides. This paper aims to evaluate the possible pitfall and difficulties in applying these three methods in the characterization of large-scale deep-seated landslides. According to the evaluation, passive seismic method is the most effective way to obtain the 3D structure of the sliding surface which can then be used for proper borehole planning. Rather than surface ERT, Borehole-to-Surface ERT can better provide the 2D geological structure. Hence, it would be the preferred method to obtain the high-resolution image of the underground structure after the boreholes are drilled. Furthermore, Time Domain Reflectometry is more economical than inclinometer in terms of monitoring the sliding depth for deep-seated sliding area.
On the Assignment of Groundwater Table in Soils Liquefaction Analysis
本文探討土壤液化分析過程地下水位參數之錯置問題、以及可能造成影響。由於進行土壤液化分析過程中，估算地層抗液化強度或反覆阻抗比CRR之鑽探水位(GWT0)常被視為計算地震力作用力或反覆應力比CSR之分析水位(GWT)，導致液化分析結果產生誤差。事實上，若鑽探水位設置高於實際應有水位，則地層抗液化強度和安全係數將被高估；若鑽探水位設置低於實際應有水位，則地層抗液化強度和安全係數將被低估。此外，當鑽探水位誤設為分析水位時(GWT0=GWT)，水位上升將牽動CRR與CSR的同向變化。但因變化速率不同，對於淺地層與高水位情況下之土壤抗液化安全係數(FL=CRR/CSR)其計算結果則可能產生異常現象；亦即，在水位上升時安全係數不降反升的情形。關於鑽探水位如何設置，本文認為以鑽探作業與迴水循環停止後1日~1週之水位相較最為合適。最後，由於目前使用以Prof. H.B. Seed團隊發展SPT-N-based之液化評估版本眾多，本文建議可以Youd et al. 於2001年ASCE期刊所發表者為主要版本；因其內容彙整1996NCEER與1998NCEER/NSF Workshops 21位與會國際土壤液化專家學者之共識結論、普遍公認度較高。若能以相同版本進行評估，則土壤液化分析結果一致性較能確保。
This article addresses the issue and potential influences of misassignment of groundwater table in the analysis of soil liquefaction. The groundwater table(GWT0)during subsurface exploration for evaluation of cyclic resistance ratio (CRR) is often mistakenly assumed to be the same as the groundwater table (GWT) for computing cyclic stress ratio (CSR) due to seismic shaking. This wrong assumption would lead to erroneous liquefaction analysis results. If the assigned GWT0 is higher than the actual location, then the CRR and the associated factor of safety (FL) would be overpredicted. Alternatively, if the assigned GWT0 is lower than the actual location, then the CRR and FL would be underestimated. If the groundwater table during exploration for evaluation of CRR is assumed to be the same as the groundwater table for computing the cyclic stress ratio (i.e., GWT0=GWT), the variation in the groundwater tables would lead to changes of CRR and CSR in the same direction. However, as the amount of changes in CRR and CSR would be different, the computed factor of safety against liquefaction (FL=CRR/CSR) may increase or decrease. For example, an increase in the groundwater table would likely cause an increase in the computed factor of safety. The authors suggest the GWT0 to be taken as the groundwater table measured between 1 day and 1 week after the end of exploration or drilling fluid circulation. In addition, the authors recommend the use of Youd et al. (2001) liquefaction analysis procedure as it represents the consensus of 21 international experts in soil liquefaction who participated in the 1996 NCEER and 1998 NCEER/NSF Workshops.