The analysis of slope stability over large areas is a demanding task for several reasons, such as the need for extensive datasets, the uncertainty of collected data, the difficulty of accounting for site-specific factors, and the considerable computation time required due to the size of investigated areas, which can pose major barriers, particularly in civil protection contexts where rapid analysis and forecasts are essential. However, as the identification of zones of higher failure probability is very useful for stakeholders and decision-makers, the scientific community has attempted to improve capabilities to provide physically based assessments. This study combined a transient seepage analysis of an unsaturated-saturated condition with an infinite slope stability model and probabilistic analysis through the use of a high-computing capacity parallelized platform. Both short- and long-term analyses were performed for a study area, and roles of evapotranspiration, vegetation interception, and the root increment of soil strength were considered. A model was first calibrated based on hourly rainfall data recorded over a 4-day event (December 14–17, 1999) causing destructive landslides to compare the results of model simulations to actual landslide events. Then, the calibrated model was applied for a long-term simulation where daily rainfall data recorded over a 4-year period (January 1, 2005–December 31, 2008) were considered to study the behavior of the area in response to a long period of rainfall. The calibration shows that the model can correctly identify higher failure probability within the time range of the observed landslides as well as the extents and locations of zones computed as the most prone ones. The long-term analysis allowed for the identification of a number of days (9) when the slope factor of safety was lower than 1.2 over a significant number of cells. In all of these cases, zones approaching slope instability were concentrated in specific sectors and catchments of the study area. In addition, some subbasins were found to be the most recurrently prone to possible slope instability. Interestingly, the application of the adopted methodology provided clear indications of both weekly and seasonal fluctuations of overall slope stability conditions. Limitations of the present study and future developments are finally discussed.

中文翻译：

---

大面积非饱和火山碎屑土多季节概率边坡稳定性分析

由于多种原因，大面积边坡稳定性分析是一项艰巨的任务，例如需要大量数据集、收集数据的不确定性、难以解释场地特定因素以及由于规模而需要大量计算时间可能构成主要障碍的调查区域，特别是在快速分析和预测必不可少的民防环境中。然而，由于识别故障概率较高的区域对利益相关者和决策者非常有用，科学界一直试图提高提供基于物理的评估的能力。本研究通过使用高计算能力并行平台，将非饱和饱和条件的瞬态渗流分析与无限边坡稳定性模型和概率分析相结合。对研究区进行了短期和长期分析，并考虑了蒸散、植被截留和土壤强度根系增量的作用。首先根据在 4 天事件（1999 年 12 月 14-17 日）中记录的每小时降雨数据校准模型，该事件导致破坏性滑坡，以将模型模拟结果与实际滑坡事件进行比较。然后，将校准模型应用于长期模拟，其中记录了 4 年期间（2005 年 1 月 1 日至 12 月 31 日，2008) 被考虑研究该地区对长期降雨的反应。校准表明，该模型可以正确识别观测到的滑坡时间范围内较高的失效概率，以及计算为最易发生滑坡的区域的范围和位置。长期分析允许确定安全斜率系数低于 1.2 的天数 (9)。在所有这些情况下，接近斜坡不稳定的区域集中在研究区域的特定部门和集水区。此外，发现一些子流域最容易出现可能的斜坡失稳。有趣的是，所采用方法的应用清楚地表明了整体斜坡稳定性条件的每周和季节性波动。最后讨论了本研究的局限性和未来的发展。