SPATIAL ANALYSIS AND DISASTER RISK MANAGEMENT: RETROSPECTIVE AND FUTURE DIRECTIONS

Transcription

1 SPATIAL ANALYSIS AND DISASTER RISK MANAGEMENT: RETROSPECTIVE AND FUTURE DIRECTIONS CEES VAN WESTEN International workshop on Multi-Hazard and Risk March 2015, Universiti Teknologi Malaysia (UTM), Kuala Lumpur, Malaysia

2 CONTENTS Introduction Case study: Fella river, Italy Hazard Vulnerability Risk Web-based system for changing risk analysis: Concept & objectives Input data requirements System architecture Risk analysis Cost-benefit analysis Conclusions

3 RISK ASSESSMENT A: Input data B: Susceptibility assessment C: Hazard assessment D: Exposure analysis E: Vulnerability assessment F: Risk assessment G: Quantitative risk Economic risk Direct Indirect Population risk Societal risk Individual risk H: Qualitative risk 3

4 Two EU projects The analysis was carried out in two EU FP7 projects CHANGES: 4 PhDs and 5 ESRs for 18 months each (programming) INCREO: Data acquisition, design and coordination

5 Quantitative multi-hazard risk assessment in the Eastern Italian Alps using GIS based flood and landslide modelling Medium scale landslide risk assessment is useful for decision making, spatial planning and allocating budgets and resources to multiple municipalities, provinces and local governments in larger areas. This use case shows an example of the use of LiDAR data for multi-hazard risk assessment: Debrisflows Flashfloods River floods

6 Study area Austria Fella River Basin (Eastern Italian Alps) Italy Slovenia Legend Legend Legend 5mdem_int 2753m Elevation 5mdem_int MainBasin_To Polygan1 Elevation 5mdem_int 2753m Elevation 426m 2753m 426m MainBasin_To Polygan1 2753m 426m 426m Hillshade_5mdem_int Elevation 5 Low : 0 ± 2753m High : 255 km Multi-hazard environment Value!! River Low High : 255 Hillshade_5mdem_int and : 0 flash flooding, Rock falls, deep- seated 0 and 2.5 shallow 5 km 426m landslides, debris flows Value High : 255 Low : 0 Legend MainBasin_To Polygan1 MainBasin_To Polygan1 5mdem_int Elevation 426m Hillshade_5mdem_int Value Hillshade_5mdem_int High : 255 Value Low : 0 Legend 5mdem_int Elevation 5 ± 2753m km km Legend 5mdem_int

7 Hydro-meteorological event: 29 August 2003 Convective storm system Rainfall peaks ~ 390 mm in 12h Liquid peak discharge ~ 20 m³ s -1 km -2 Flash floods, floods, debris flows, erosion Damages ~ 389 mil. Euros, 2 casualties ( Civil Protection, FVG Region, Italy)

8 METHODOLOGY

9 Cumulative Probability Precipitation (mm) Number of days Rainfall frequency analysis Total precipitation, top 1% rainy days # days with rain totals > 10mm 1 GEV CDF for Malborghetto, GEV fit GEV low GEV hi Empirical CDF 10 yr 25 yr 50 yr 100 yr Middle mm Low limit (95%) Higher limit (95%) Precipitation (mm)

10 Probability of not being exceeded F(x) Changing return periods? Analyzing possible effects of climate change Return Period Change in intensity depends on where you are in the precipitation distribution (left) Wide range of possible future scenarios (right) First results from the new CMIP5 RCM climate projections (CORDEX: wcrp-cordex.ipsl.jussieu.fr/) 1 Fella, annual daily maximum rainfall 100 Change in return period (base = 20 year RP ) Increase or decrease in intensity Both show increase obs low high Precipitation (mm) Median RP decreasing over time

11 Generating a multi-annual landslide inventory

12 Debris flow inventory in the Fella River Basin Multi-temporal debris flow inventory consisting of point sources and run-out polygons Year Number of debris flows 137 MainBasin_To Polygan1 gend dem_int vation m 426m Landslide AVI (CNR IRPI) lshade_5mdem_int IFFI projects (ISPRA) km lue Geological Service of the Friuli Venezia Giulia region High (FVG) : 255 University of Trieste Low : 0 ± 3/28/2015

13 Landslide susceptibility assessment Weights of evidence modelling, Using debris flow data only W + i = W ī = log log e e P { B S} P { B S } i i P { B S} P { B S } i i Only very high areas concidered for run-out modeling Major event susceptibility map Success rate: 82.53% Prediction rate: %

14 Intensity (mm/hr) When do floods and landslides occur in Fella? Floods and landslides occur with higher intensity rainfall compared to normal Intensity duration relationships similar to those from other study areas Sub-daily values are also important but limited timeseries ( < 20 years) Can disaggregate daily to hourly but uncertainty in the precise hourly distribution of the rainfall ID (daily) for Fella River Duration (hrs) Floods and flash floods Debris flows, slides Caine, 1980 ID equation Guzzetti et al 2008 ID equation

15 Timeline of change in landslide risk Return period scenarios were determined with extreme value distribution analysis of daily rainfall and using the debris flow inventory August Disaster Moderate Major Minor

16 Flow-R debris flow run-out model An Empirical GIS model for gravitational hazards at a Regional scale INPUTS: 1. Source areas 2. DEM 3. Parameters Travel angle Maximum velocity OUTPUTS: 1. Probability of flow direction 2. Maximum kinetic energy Horton, P., Jaboyedoff, M., Rudaz, B., and Zimmermann, M.: Flow-R, a model for susceptibility mapping of debris flows and other gravitational hazards at a regional scale, Nat. Hazards Earth Syst. Sci., 13, , doi: /nhess , /28/2015

17 Flow-R source area criteria 1. Only very high susceptibility zones are used. 2. We take gullies and channels into consideration by using a planar curvature threshold (< 4/100 m 1) 3. Original debris flow source areas for each return period 4. A slope angle threshold related to flow accumulation (Horton et al., 2008; Heinimann, 1998; Rickenmann and Zimmermann 1993) Rare fitting Risk Scenario (return period) Travel angle ( ) Velocity (m/s) Major ( yrs) Moderate ( yrs) Minor (10-25 yrs) 17 8 Frequent (1-10 yrs) /28/2015

18 Flow-R probability maps to intensities? Flow-R probability is not a hazard intensity, but just the likelihood of a flow reaching an area.we need intensities!!! Expert based transfer function between probability and impact pressure Legend prbm_13_15 Value High : Impact pressure (KPa) of past events at elements at risk 2. Morphological distribution of Low : debris flow probabilities within a debris flow path 3/28/2015

19 Flow-R probability maps to intensities? Relating Flow-R probabilities with intensities from physical based local scale debris flow models (Flo-2D, MassMov2D) both type of models are using the same DEM MassMov2D (Hussin et al., 2014) Flo-2D (Calligaris et al., 2008, Calligaris and Zini, 2012 Debris flow dam (2003) RAMMS Voellmy Rheology Houses 3/28/2015

20 Flow-R probability maps to intensities? Heavily skewed probability distribution towards the very low probabilities Legend prbm_13_15 Value High : Low : The probability map was classified into 10 Quantile classes to better represent the lower probabilities Relationship Flow-R probability vs max flow height

21 Debris flow impact pressures (KPa) Major event ( year RP) Major event ( year RP), travel angle = 13, velocity = 15 m/s Moderate event ( year RP), travel angle = 15, velocity = 10 m/s Minor event (10 25 year RP), travel angle = 17, velocity = 8 m/s ± 0 5 Kilom km

22 Debris flow intensity flow heights (m) Major event ( year RP) Moderate event ( year RP) Minor event (10 25 year RP) ±

23 Changes in DEM to include mitigation measures Studying the Local scale effects of a regional scale model using an updated DEM 2003 Major event (DEM 2003) Post-2011 Minor event (DEM 2007) ± ±

24 Flashflood modeling 10 yr 25 yr 50 yr 100 yr Middle 138 mm 170 mm 196 mm 225 mm Low limit (95%) Higher limit (95%)

25 Flood depth Flow velocity Flood modelling Hydraulic modelling was performed using HecRAS 4.1 and its GIS-assisted version GeoHecRAS (ArcGIS 10.1) The bathymetry of the river and correspondent topography of the flood plain was obtained from Lidar data at 1m resolution

26 Elements at risk mapping Data gathered: Building occupancy type buildings Building material type Number of floors Abandoned or destroyed Population in normal time Population tourist season Google street view Mapping from the office! In the field

27 Occupancy types

28 Construction types

29 Building value Piano regolatore generale comunale Banca dati immobiliare agenzia delle entrate (Italian Revenue agency) Market value ( /sq m) and Administrative units

30 Maximum building value

31 Population distribution modelling Two scenarios: A: Normal period B: Tourist season

32 Generating vulnerability curves

33 Generating vulnerability curves A. Regional scale vulnerability assessment using dynamic run-out modeling Maximum impact pressure ( yrs.) = 35 Kpa (total destruction of several houses) Maximum impact pressure for all other return periods does not exceed 35 KPa and is considered a cut-off value Debris flow vulnerability curves for impact pressure B. Local scale vulnerability assessment using photo-documentation Assume a shape of the curve: a, b constants (regression line) Debris flow vulnerability curves for debris height

34 Risk calculation Hazard Vulnerability Value Elements at risk Legend MainBasin_To Polygan1 5mdem_int Elevation 2753m ± RISK 426m Hillshade_5mdem_int Value High : 255 Low : 0 Scenario (return period) km Nr. of buildings exposed by impact pressure method Nr. of buildings exposed by flow height method Major ( yrs) Moderate ( yrs) Minor (10-25 yrs) Frequent (1-10 yrs) 7 13 Scenario (return period) Loss estimation using impact pressure method ( ) Loss estimation using flow height method ( ) Major ( yrs) 8,419, ,814, Moderate ( yrs) 2,099, ,534, Minor (10-25 yrs) 733, ,480, Frequent (1-10 yrs) 12, , Piano regolatore generale comunale AAL (IP) = 540, AAL (FLw) = 957,073.00

35 Multi-hazard risk maps

36 Validation with losses occurring in last event

37 Conclusions Change in risk: ,53 million 13,81 million 1,48 million Risk decreases after a major event because also sediments needs to be recharged, risk after 2003 is even smaller than before There is an underestimation of the total loss? Why? Hazard intensities are underestimated. Vulnerability curves from literature possibly not suitable for our area.

38 RiskChanges: a Spatial Decision Support System for the analysis of changing multi-hazard risk, based on possible future scenarios and risk reduction alternatives to analyse the effect of risk reduction planning alternatives on reducing the risk now and in the future, and support decision makers in selecting the best alternatives. Users: Civil protection organizations. Organizations with the mandate to design structural risk reduction measures Planning organizations

39 Modules

40 Data input module Study area Administrative units Hazard data sets: Hazard type Hazard intensity Spatial probability Return period Elements-at-risk dataset Type Value Population

41 Debrisflow (DF) hazard Impact pressure (IP) Flashflood (FL) hazard Water depth (DE) Landslide (LS) hazard Spatial probability (SP) Return period: 20 years Return period: 20 years Return period: 20 years DF_IP_20_A0 FL_DE_20_A0 LS_SP_20_A0 Return period: 50 years Return period: 50 years Return period: 50 years DF_IP_50_A0 FL_DE_50_A0 LS_SP_50_A0 Return period: 100 years DF_IP_100_A0 Return period: 100 years FL_DE_100_A0 Return period: 100 years LS_SP_100_A0

42 Project Alternatives Scenarios Future years

43 Alternatives for risk reduction Alternative 1: Engineering solutions Alternative 2: Ecological solutions Alternative 3: relocation Items related to construction cost Storage basins Slope stabilization Expropriation of land and existing buildings where construction will take place Expropriation of land and existing buildings where construction will take place Slope stabilization Water tank construction Compensation of owners of buildings Expropriation of existing buildings Lawsuit Hazard change s Yes Yes No Elementsat-risk changes No Partly Yes

44 Possible future scenarios Scenario 1 Scenario 2 Name Land use change Climate change Business as Rapid growth without taking into No major change in climate usual account the risk information expected Risk informed Rapid growth that takes into account No major change in climate planning the risk information and extends the expected alternatives in the planning Scenario 3 Worst case Rapid growth without taking into account the risk information Scenario 4 Most realistic Rapid growth that takes into account the risk information and extends the alternatives in the planning New Return Period in Future Year Old Return Period (± 5) 17 (± 6) 14 (± 7) 11 (± 8) Climate change expected, leading to more frequent extreme events Climate change expected, leading to more frequent extreme events 50 (± 10) 45 (± 12) 35 (± 14) 25 (± 16) 100 (± 20) 90 (± 23) 75 (± 26) 55 (± 30) 200 (± 40) 180 (± 44) 150 (± 49) 110 (± 53)

45 Scenario 1: Business as usual

46 Vulnerability curve database Search database: Hazard type Intensity type EaR type Vulnerability type: Physical Population Upload new curves Use existing curves Link curves to units in a map

47 Loss and risk assessment Loss calculation for each combination of: Hazard type Return period Elements-at-risk Select the type of risk analysis: Administrative units Hazard types Elements-at-risk Alternatives Scenario/future years Economic /population

48 Vulnerability Temporal probability Hazard scenarios Return period: 10 years Return period: 50 years Return period: 100 years A B C Elements-at-risk Low Intensity High Vulnerability 1 Risk curve 0.1 A B C Low Costs High Low Intensity High Low High Loss = costs * vulnerability

49 Temporal probability Temporal probability Loss maps Return period: 10 years Return period: 50 years Return period: 100 years A B C Low loss High Risk curve: Area X 0.1 A Average annual risk Risk curve: Area Y 0.1 A Average annual risk Y X B C B C Low Loss High Low Loss High

50 Vulnerability Temporal probability Hazard scenarios Return period: 10 (8-12) Return period: 50 (40-55) Return period: 100 (89-120) Average Intensity Average Intensity Average Intensity STD Intensity STD Intensity STD Intensity Elements-at-risk Average Value STD Value 1 Low Intensity Vulnerability with uncertainty High Risk curves: minimum, avergae and maximum Low Costs High Low Intensity High Low High Loss = costs * vulnerability

51 Cost-benefit analysis Use results of the risk calculation (AAL) as input; Determine investment costs, maintenance costs, and costs Project lifetime. Investment period Calculate indicators: Cost Benefit Ration, Net Present Value, Internal Rate of Return

52 Results Risk Analysis for alternatives and scenarios

53 Multi-criteria evaluation Use risk indicators: AAL, % risk reduction Use cost-benefit indicators: CBR, NPV, IRR Define other indicators: social, economic, environmental Standarization Weighting Ranking Compare ranking of different stakeholders

54 Visualization module Visualize data in database: Input data, loss data, risk data Depends on the study area and project defined.

55 Visualization options

56 Online version of the system

57 Training opportunities Training package on the use of GIS for landslide hazard and risk assessment Open source software Exercises & data Cees van Westen ftp://ftp.itc.nl/pub/westen/safeland

58 Training package on the use of GIS for multihazard risk assessment Textbook Exercise book Data Software Training opportunities Cees van Westen Distance education course: Materials can be obtained from: ftp://ftp.itc.nl/pub/westen/multi_hazard_risk_course

59 DR. C.J. VAN WESTEN ASSOCIATE PROFESSOR THANK YOU UNIVERSITY TWENTE, FACULTY OF GEO-INFORMATION SCIENCE AND EARTH OBSERVATION (ITC) PO BOX 217, 7500 AA ENSCHEDE, THE NETHERLANDS T: F: E: I: DRM I: