物質點法輔助崩塌監測預警系統-以光華崩塌地為例

Abstract

邊坡穩定性評估是山坡地治理與防災規劃的核心工作，其分析結果往往直接影響後續工程設計與現場監測策略。傳統上，學界與工程師多利用極限平衡法（Limit Equilibrium Method, LEM）或有限元素法（Finite Element Method, FEM）來計算安全係數，並據此判斷邊坡是否穩定。然而，這兩種方法皆無法全面再現邊坡從初始變形到破壞的整個運動過程。 為了克服此問題，本研究採用物質點法（Material Point Method, MPM）進行二維數值模擬，基於實地地質調查、既有文獻地層剖面與高解析度 DEM，建立物質點法模型，並提出「崩塌臨界位移」概念，作為坡面失效的判定門檻。接著透過 MPM 數值模擬分析主滑動面位移分佈，並據此制定伸縮計、傾斜管及 GPS 等監測儀器的最佳佈設策略，以確保關鍵區域能在變形初期即時掌握異動。 除此之外，也提出動態調整警戒門檻（20 mm／日至 40 mm／日）之建議，以兼顧實用性與應變效率。比較二維與三維模型在坡度（30.96°–34.44°）與長寬比條件下的相似度分析，則指出在特定 L/D 範圍內，二維模擬可在 5% 誤差內替代三維計算，從而在資源受限時提供快速模擬；當條件不適合時，則需採用三維模型以捕捉更全面的滑動機制。 本研究所提出的 MPM 模型構建流程、監測佈設策略、地下水位情境分析及二維/三維模型適用性比較，不僅為光華崩塌地的預警與應急決策提供了全面的技術框架，也將大幅提高現場監測與防災效率，為類似山坡地區的災害風險管理提供重要參考。

Slope stability assessment is fundamental to hillside management and disaster mitigation, directly informing engineering design and field monitoring strategies. Traditionally, Limit Equilibrium Method (LEM) and Finite Element Method (FEM) have been used to compute safety factors, but neither captures the full kinematic evolution from initial deformation to failure. To address this limitation, this study employs the Material Point Method (MPM) for two-dimensional numerical simulations. We integrate field geological surveys, published stratigraphic profiles, and high-resolution digital elevation models to build a detailed MPM representation of the Guanghua landslide site, introducing the concept of a “critical collapse displacement” as a failure threshold. Using MPM, we analyze displacement distributions along the primary sliding surface and derive an optimized layout for extensometers, inclinometers, and GPS stations to ensure early detection of anomalous movements in high-risk zones. We also recommend dynamically adjusting the warning threshold from 20 mm/day to 40 mm/day to balance practical constraints with timely response. A comparative study of two- and three-dimensional models—across slopes of 30.96°–34.44° and varying length-to-width ratios—indicates that, within a specific “L/D range”, two-dimensional simulations can reproduce three-dimensional results within 5% error, enabling rapid screening when computational resources are limited; outside this range, three-dimensional modeling is necessary to capture the full sliding mechanism. Overall, the methodology presented—encompassing MPM model development, instrument deployment strategies, groundwater-level scenario analysis, and two- versus three-dimensional model applicability—provides a comprehensive technical framework for real-time warning and emergency decision-making at Guanghua and serves as a valuable reference for landslide risk management in analogous mountainous terrains.