1 FUTURE DEVELOPMENTS IN EARTHQUAKE-INDUCED LANDSLIDES SCENARIOS Roberto W. Romeo University of Urbino, ITALY The Next Generation of Research on Earthquake-Induced Landslides

2 SCENARIO: from the Latin scǽna (a stage with a plot) A scenario is a sketch, outline or description of an imagined situation or sequence of events. (from the Oxford English Dictionary) Misuse and overuse of the term has attracted hostile attitudes by the people It doesn’t represent a prediction tool but rather a way to induce people to think about unthinkable even if unpredictable events

3 Task of a Scenario To provide planners and managers with a plot of a possible sequence of events so that they may test and plan strategies against uncertain future developments a story about what happened in the future (from The Forecasting Dictionary)

4 Characteristics of a SLS A SLS refers to a wide area, usually at basin scale for the purpose of investigating uniform climatic and geomorphological features… …thus requiring a large number of landslides of different type (i.e., falls, slumps and flows) to be analyzed through an appropriate … tool for a complete DAD process, namely Draw-Analyze-Display, usually accomplished by GIS-tools

5 The SLS in the scientific literature In the scientific literature:  213,000 articles refer directly or indirectly to: GROUND MOTION SCENARIOS  21,400 to LANDSLIDE SCENARIOS  9,700 to SEISMIC landslide SCENARIOS

6 Landslide inventories and triggering earthquakes In the last decade of the 20th century only 4 to 12 earthquakes have well documented inventories (Keefer, SG 2002 and Rodriguez et al., SDEE 1999) of landslides triggered by earthquakes versus about 1,500 earthquakes above magnitude of 6 (USGS Earthquake Hazard Program) according to the minimum reported magnitude in the inventories

7 Empirically-based SLSs: a mission impossible (?) Landslides triggered by earthquakes are RARE events in a RARE cause : P SL = P[L|Y]  P[Y] Being P[Y] = seismic hazard and P[L] = failure probability, that is f(type of movement, geotechnical properties, morphology, hydraulic conditions)

8 … and here’s why most of the methods to produce SLSs rely on theoretical models (mainly geotechnical) rather than on empirical ones (mainly statistical: look at the next talk by C.T. Lee)

9 The milestone of SLSs

10 Slope performance instead of slope stability They replaced the approach based on the pseudo-static safety factor with the Newmark’s sliding displ. analysis as implemented by Wilson and Keefer (BSSA, 1983) introducing the concept of slope performance rather than that of stability

11 Further developments Jibson-Harp-Michael (Eng.Geol., 2000) sketched the procedure for a GIS-based SLH map SL susceptibility SH

12 Probability assessment … and developed a method to estimate failure probabilities from Newmark displacements

13 Towards a Performance-based approach Romeo-Jibson-Pugliese (2007, 1 st NALC, Vail- CO ) They performed a landslide scenario based on the results of a formal seismic hazard analysis, and … they afforded the task of estimating the potential damage to exposed goods in the area AcAc I A = f (M,R,S) Yearly F exc (D>D c ) or mean T D as a function of the I A values Probablistic AND/OR Deterministic Landslide scenarios capacity demand

14 Common features of SLSs 1)Simplified Newmark’s sliding block analysis to asses slope performance on the basis of computed displacements 2)Shallow landslide model (i.e., infinite slope analysis) 3)A deterministic approach to the seismic hazard

15 Merits … They allow the integration of the whole analysis into a single computer program (GIS environment) They usually (but not necessarily always) provide conservative results They have the potential for a real-time upgrade of slope performance under a given (even observed) seismic shaking

16 … and shortcomings They provide approximate results usually affected by large uncertainties They are limited to shallow landslides (Keefer’s type I) that although being the predominant ones during eqks. nevertheless bring other important slides such as slumps and flows to be neglected They refer to a very limited seismic and boundary conditions thus reducing their capability to effectively mitigate the risk

17 1st key-Q: May simplified Newmark’s sliding block analysis be replaced by an improved one? A simplified analysis replaces a rigorous one (double integration of a time history) through a regression equation like this: Log(D n ) = k 1 + k 2 Log(gm) + k 3 Log(a c )   It is easy to be implemented into a GIS- environment for rapid and extensive analyses

18 Uncertainty of simplified methods

19 The use of a rigorous displacement analysis (e.g., Miles & Ho, SDEE 1999) is limited by: Computations become cumbersome for a wide area and necessarily require external calls (offline software) A substantial lack of improvement in absence of a formal local seismic response analysis that would allow amplification phenomena due to site effects being taken into account

20 1st key-A: simplified methods still represent the most suitable for SLSs, nevertheless they may be improved provided that: they are computed on a regional basis due to the current availability of a huge amount of strong motion records Amplified motion due to site effects can be taken into account in the regression models or through site amplification factors such as those used in the seismic codes (e.g., Eurocode 8)

21 Regionally-based simplified equations

22 Influence of site effects on the seismic displacements

23 2nd k-Q: may the infinite slope model be replaced by a less approximate one? This is the only one that can be implemented into a GIS without requiring external calls even if GIS is a 2-½ dimensional tool, it can analyze heights but not depths … and this rigorously would restrict the analysis to cohesionless soils in dry or wet conditions alone

24 2nd key-A: the infinite slope model is still the only one directly applicable in a GIS-environment, anyway: Failure depths for cohesive soils and water table depths for partially saturated soils should be assessed with their own reliability Multiple seasonal scenarios may substitute the probabilistic estimate of the hydraulic conditions Care shall be taken over the choice of the most appropriate geotechnical properties to differentiate ductile from brittle behaviors

25 3rd key-Q: may we refer to a seismic scenario which is not deterministic? A reference earthquake is usually selected whose gm-values may induce the largest ground failures, anyway: Over a large area usually many sources are active and more than one earthquake is required to encompass all the possible causes of failure A deterministic eqk. can be effectively useful mainly in view of a rigorous displacement analysis and a LSRA which are anyway not feasible for the reasons cited above

26 3rd key-A: a deterministic scenario may be well replaced by a probabilistic one, since: It is consistent with the simplified displacement analysis whose regression equations derive from a set of earthquakes It allows performing a full probabilistic analysis of slopes performance under seismic shaking It may be better implemented in a conventional risk assessment procedure

27 Displacement hazard curves figure from Rathje & Saygili, JGGE 2008

28 A PB approach Romeo (2008), technical report

29 GM from earthquake Far - Strong Near - Moderate All inclusive (PSHA) Hydraulic conditions Spring Sp/F-SSp/N-M Full Probabilistic* Summer Su/F-SSu/N-M Fall F/F-SF/N-M Winter W/F-SW/N-M SCENARIO The SLS matrix * with the possibility of hazard deaggregation

30 THANK YOU VERY MUCH FOR YOUR ATTENTION !