Transcript 播放檔下載
Coupling TRIGRS and TOPMODEL in shallow landslide Prediction 報告者:李浩瑋 指導教授:李錫堤 2011.5.19 Out line Introduction Literature review Objective Methodology Result and discussion Conclusion 2 Landslide Susceptibility Analysis Qualitative analysis: Empirical method Quantitative analysis: Statistic method -based upon long-period landslide inventories Deterministic analysis -strength parameters, -failure depth -groundwater conditions 3 Literature review 1) By combining an infinite-slope stability calculation with a transient, one-dimensional analytic solution for pore pressure response to transient rainfall infiltration.[Iverson, 2000; Baum et al., 2002; Savage et al., 2003; Godt, 2004] 2) TRIGRS models were used for slope stability analysis. [吳佳郡, 2006;王姵兮,2007;鐘欣翰,2008] 1) Concept of topographic index, ln(a/tanβ).[Beven and Kirby, 1979] 2) A hydrological simulation based on a modified version of TOPMODEL was developed to estimate the temporal groundwater level for conducting the slope-instability analysis. [李光敦, 2009] 4 Objective rainfall-triggered shallow landslide 1) Consider the lateral flow 2) Compare the result of trigrs in shallow landslide prediction 5 Research Process Start Precipitation T=0, Input initial groundwater table Hydrological model KZ、D0、IZ TRIGRS (1-D infiltration) Simulating groundwater table Infinite slope model Prediction of shallow landslide T=t+dt TOPMODEL (Modification of water table) T=final t End 6 TRIGRS TRIGRS (Transient Rainfall Infiltration stability Model (Baum et al., 2002) Grid-based Regional Slope- Infiltration model (Iverson, 2000) D0 2 ( 2 ) 2 t cos Z D0 = Ksat/C0 D0 =飽和水力擴散率 Ksat = 飽和水力傳導係數 C0 =土壤最小之含水量 Z= z/cos α Conceptual sketch of the hydrological model Z =鉛直方向 in TRIGRS(after Godt 2004) z =垂直坡面方向 7 TOPMODEL TOPMODEL(TOPgraphy based hydrological MODEL)(Beven. et al., 1979) a: Specific area tan β :slope a TI ln tan a Z j Z m[ ln( )j] tan Zj Zj :Depth to groundwater table Z: Average depth of the groundwater table m : Recession constant λ : mean value of the Topographic index 8 Infinite slope 文獻回顧 resistance force C ( s D w hw ) cos 2 tan FS driving force s D sin cos FS FS>1 Stable FS<1 Unstable h tan C [1 w ( w )] s D sin cos s D tan 未來工作 TOPMODEL Calculate Zj w D Z j tan C FS [1 ( )] s D sin cos s D tan Schematic diagrams of the coupled hydrologicalslope instability model(after Lee 2009) 9 Study area 文獻回顧 10 Event Analysis Event-based Triggered Landslide Inventory Typhoon Aere Rainfall Records 2004/8/23~8/25 GAOYI Station 11 hydrological parameters 鍾(2008) 地質區 Kz(m/s) D0(m2/s) Iz(m/s) zone1 9×10-4 1.8×10-3 10-8 zone2 10-3 2×10-3 10-8 zone3 2×10-3 4×10-3 10-8 12 Event Analysis Antecedent rainfall 2004/8/20 Initial water table 2004/8/22 Typhoon Aere 13 Groundwater level simulation Rainfall period 14 Geologic parameters 鍾(2008) 地質區 γs(kN/m2) Cmax(kPa) ψ(°) zone1 18.2 8 36 zone2 19.6 9 28 zone3 18.9 9 27 15 Model result Rainfall period 16 Model validation Model validate • Error matrix (Stehman, 1997) success curve and prediction curve (Chung and Fabbri, 1999) 全區資料 網格數 Unstable (FS < 1) Stable (FS ≧1) Unstable N1 N2 stable N3 N4 總體正確率=(N1+N2)/(N1+N2+N3+N4) Success rate 分類結果網格數 Area in percentage Different rainfall situations for validation ─Using Typhoon Masta Rainfall(2005) Result Success Rate Curve Predict 121809 Actual Unstable (FS < 1) Stable (FS ≧1) Unstable 625 446 stable 11530 108488 山崩組正確率58.35%、非山崩組正確率89.59%,總體正確率89.10% 18 Model validation Typhoon Aere Typhoon Masta Prediction Success Rate Curve Typhoon Typhoon Aere Mastalandslide landslidemap map 19 Discussion Success Rate Curve TRIGRS Success Rate Curve Coupling TRIGRS and TOPMODEL TRIGRS 結合TRIGRS 與TOPMODEL 山崩組正確率 56.7% 58.4% 非山崩組正確率 87.4% 89.5% 總正確率 87.2% 89.1% 20 Discussion (C) (12) (11) (10) (A) (1) (9) (2) (8) (7) (3) (B) (6) (4) (5) Coupling TRIGRS and TOPMODEL與TRIGRS分析結果套疊圖 21 Discussion 集水區邊界 水系 艾利颱風崩塌地 Coupling TOPMODEL < 1 TRIGRS ≧ 1 Coupling TOPMODEL ≧ 1 TRIGRS < 1 Coupling TOPMODEL < 1 TRIGRS < 1 Coupling TOPMODEL ≧ 1 TRIGRS ≧ 1 (1) (3) (2) (4) 集水區西側(A區)之實際崩塌地分析成果 22 Discussion 集水區邊界 水系 艾利颱風崩塌地 Coupling TOPMODEL < 1 TRIGRS ≧ 1 Coupling TOPMODEL ≧ 1 TRIGRS < 1 Coupling TOPMODEL < 1 TRIGRS < 1 Coupling TOPMODEL ≧ 1 TRIGRS ≧ 1 (5) (6) (7) 集水區東側(B區)之實際崩塌地分析成果 23 Discussion (8) (9) (11) (10) (12) 集水區邊界 水系 艾利颱風崩塌地 Coupling TOPMODEL < 1 TRIGRS ≧ 1 Coupling TOPMODEL ≧ 1 TRIGRS < 1 Coupling TOPMODEL < 1 TRIGRS < 1 Coupling TOPMODEL ≧ 1 TRIGRS ≧ 1 集水區北側(C區)之實際崩塌地分析成果 24 Conclusion 本研究嘗試結合TOPMODEL將每個網格地下水位作適度 的修正,結果顯示結合TRIGRS與TOPMODEL能反應長 期暴雨期間側向補 注勢能對於整體地下水位分佈可能的 影響,分析結果顯示不穩定區多半遠離山脊而鄰近河岸, 與TOPMODEL之預期相符。 研究中以馬莎颱風作為驗證分析模型的事件,馬莎颱風 誘發山崩雖較為不足,得到的預測率曲線下之AUC為 0.767,預測結尚屬合理。 結合TRIGRS與TOPMODEL的總體正確率為89.4%,成 功率曲線下面積為0.822,TRIGRS的總體正確率為87.4%, 成功率曲線下面積AUC為0.787,從此兩種評估方法結果 可以看出,結合TRIGRS與TOPMODEL能有效地解釋崩 塌地分佈而可獲得較佳的預測效果。 25 Thank you for your attention