1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm...

17
1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm induced Loss in Europe ENSEMBLES GA 2006 – 20th Nov – 24th Nov 2006, Lund Prof. Dr. U. Ulbrich Dr. G. Leckebusch M. Donat

Transcript of 1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm...

Page 1: 1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm induced Loss in Europe ENSEMBLES GA 2006 – 20th Nov –

1ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov

RT6, WP 6.2 Meeting

Estimates of Windstorm induced Loss in Europe

ENSEMBLES GA 2006 – 20th Nov – 24th Nov 2006, Lund

Prof. Dr. U. UlbrichDr. G. LeckebuschM. Donat

Page 2: 1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm induced Loss in Europe ENSEMBLES GA 2006 – 20th Nov –

2ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov

Economic loss Insured loss

Economic and insured loss: Germany 1970 - 1998

Introduction:Storm damages in the past

Tim‘s question 1.      What are the main objectives of our study?

Page 3: 1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm induced Loss in Europe ENSEMBLES GA 2006 – 20th Nov –

3ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov

Model theory

• Loss depends on - local gust wind speed- insured property or amount of forest in the area

• insured property values can roughly be estimated from population density

• Loss increases with wind speed above a threshold.Different storm-loss functions have beenproposed, a frequent one is: loss ~ v3.

Estimation of future changes in climate extremes and their relation to property damage

Following the “multi model approach”direct use of GCM/RCM output in the impact model

Page 4: 1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm induced Loss in Europe ENSEMBLES GA 2006 – 20th Nov –

4ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov

This wind speed is approx. equal to the 98th percentile of wind speeds at regular (non-coastal, no mountain) stations in Germany

• Germany: Insurance companies pay when wind speeds exceed Bft 8 = 17.2 – 20.7 m/s

For property damages:

Model theory

Page 5: 1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm induced Loss in Europe ENSEMBLES GA 2006 – 20th Nov –

5ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov

Model theory

Loss ≈

regions

year

days regionv

dayregionvregionpopc

3

98

max 1)(

),(*)(* for 98max vv

„normalized cubic wind“

3

98

98

v

vvfor 98vv

Approach based on:

Klawa, M. und U. Ulbrich, 2003:

A model for the estimation of storm losses and the identification of severe winter storms in Germany.

Natural Hazards and Earth System Sciences, Vol. 3, 725-732.

Page 6: 1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm induced Loss in Europe ENSEMBLES GA 2006 – 20th Nov –

6ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov

Model improvements in ENSEMBLES

2: GIS (ArcGIS) - including global population distribution data on 1x1 degree grid- including interpolation of forestry data to model grid via GIS

(at present: nearest neighbour)- Calculation of accumulated damage potential for different time

slices and/or regions

1: Calculation of „normalized cubic wind“ from input data (e.g. ERA40) per year

Model structure

3: Fitting the calculated values per year and region to observed losses

year

days regionv

dayregionv3

98

max 1)(

),(

Tim‘s 3.      What have we achieved so far?

Page 7: 1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm induced Loss in Europe ENSEMBLES GA 2006 – 20th Nov –

7ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov

Model improvements in ENSEMBLES

c165/166 c49

1970-2000 0,78 0,83

Correlation with insurance data (GdV):

Loss Ratio based on momentary wind values vs. daily maximum gust from ERA40

0,000

0,100

0,200

0,300

0,400

0,500

0,600

0,700

los

s r

ati

on

[‰

]

GdV c165/166 c49

Input parameter: Wind gusts (Forecasts!)

Overestimation in 1993Underestimation in 1990

Further investigation with respect to the kind

of exceedance

Page 8: 1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm induced Loss in Europe ENSEMBLES GA 2006 – 20th Nov –

8ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov

Model improvements in ENSEMBLES

• 1993 more weak events than 1990• 1990 more extreme exceedances of 98th Percentile than 1993

Approach 2 („dynamic“):Loss limit individually adjusted after loss events

Approach 1 („static“):Loss limit consistently increased

GERMANY: Exceedance of 98th Percentile (1971-2000) in ERA40

Page 9: 1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm induced Loss in Europe ENSEMBLES GA 2006 – 20th Nov –

9ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov

Model improvements in ENSEMBLES „Dynamic Approach“

Idea: Individual variation of loss limit at each grid cell after loss events, depending on time since the last event

Page 10: 1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm induced Loss in Europe ENSEMBLES GA 2006 – 20th Nov –

10ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov

Application of loss model on climate simulations

Loss Ratio in ECHAM5/OM1, 20C, run 1

0

0,2

0,4

0,6

0,8

1

1971

1972

1973

1974

1975

1976

1977

1978

1979

1980

1981

1982

1983

1984

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

Lo

ss R

atio

[‰

]

jährl. Werte Mittelw ert

Loss Ratio in ECHAM5/OM1, A1B, run 1

0

0,2

0,4

0,6

0,8

1

20

71

20

72

20

73

20

74

20

75

20

76

20

77

20

78

20

79

20

80

20

81

20

82

20

83

20

84

20

85

20

86

20

87

20

88

20

89

20

90

20

91

20

92

20

93

20

94

20

95

20

96

20

97

20

98

20

99

21

00

Loss

Rat

io [

‰]

jährl. Werte Mittelwert

ERA 40

(1971-2000)

EH5/OM1, 20C

(1971-2000)

EH5/OM1, A1B

(2071-2100)

Mean value 0,1395 0,1284 0,1507

Std. deviation 0,0894 0,0707 0,1494

+ 17 % + ~110 %

Page 11: 1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm induced Loss in Europe ENSEMBLES GA 2006 – 20th Nov –

11ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov

Future Plans

• completion of sensitivity tests

• application on all available GCM- / RCM-Simulations

• Estimation of robust climate change signalfollowing the Multi Model Approach

Tim‘s 4.      Which of the WP 6.2 tasks, milestones and deliverables (see overleaf) do we plan to contribute to, by when and in what form?

Del. 6.8: Preliminary Report on changes in climate extremes and their relation to flood risk, agriculture, forest and property damage and human health

Page 12: 1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm induced Loss in Europe ENSEMBLES GA 2006 – 20th Nov –

12ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov

5.      What are our main questions requiring discussion in this meeting?

When and where are RCM data available ?

Tim’s 5th question:

Page 13: 1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm induced Loss in Europe ENSEMBLES GA 2006 – 20th Nov –

14ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov

Deliverables to be fulfilled in the next months (up to month 42)

D6.7 Preliminary report on a comparative study of response surface and multiple scenario approaches to assessing risks of impacts using selected impact models. Month 30, Feb.

D6.8 Preliminary report on changes in climate extremes and their relation to flood risk, agriculture, forest and property damage and human health.

Month 30, Feb.07, UEA

D6.13 Methodological report on the linking of preliminary probabilistic projections from the Ensemble Prediction System to impact models.

Month 42 (SYKE?, possibly in co-operation with RT2B)

18 months period, month 25-42

Session 3: Planning and timetabling

Page 14: 1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm induced Loss in Europe ENSEMBLES GA 2006 – 20th Nov –

15ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov

Session 3: Planning and timetabling

Questions to be discussed on this meeting and later on: DISAT (Marco Bindi):

Collection of data to be used for the construction of response surfaces

DIAS (Tove Heidmann): How and when do we get climate data

UREADMM (Tom Osborne):How do we get hold of the data with password only? RCM/GCM??

UNIK (Martina Weiß, Uni Kassel)When ENS projections become available, river basins,

format of response surface?Problem: common data based used by all partners to achieve response surfaces

18 months period, month 25-42

Page 15: 1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm induced Loss in Europe ENSEMBLES GA 2006 – 20th Nov –

16ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov

Session 3: Planning and timetabling

Questions to be discussed on this meeting and later on: SMHI (Phil Graham): a) seasonality in the res. surf. approach?b) what is the proper level of detail for critical thresholds?c) What will we get actually from GCMs?d) direct simulations: how to choose which transient model simulation?

PAS (Malgorzata Szwed): ENS of climate scenarios are not availableProblem with real data availability (classified data e.g. by insurance comp.)scarcity of data on extremes (e.g. 1997 flood in Poland)

SYKE (Stefan Fronzek): (local permafrost)Are joint probabilities of several climate variables possible?How big will be the sample size? (RT2B?)Addressing impact model uncertainty?

18 months period, month 25-42

Page 16: 1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm induced Loss in Europe ENSEMBLES GA 2006 – 20th Nov –

17ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov

Session 3: Planning and timetabling

Questions to be discussed on this meeting and later on: NAO (Christos): Common output format? Question by Marco!

18 months period, month 25-42

Is it possible to be more precise in formulating what we (RT6.2) will need to have from other WP’s? (RT2B, RT3, RT2A,…)

Aim: Preparation of the discussion on Thursday, afternoon (14-14:30 entitled: From regional scales to impacts;

a talk given by Clare Goodess (RT2B) and Tim Carter (RT6.2)

Page 17: 1 ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov RT6, WP 6.2 Meeting Estimates of Windstorm induced Loss in Europe ENSEMBLES GA 2006 – 20th Nov –

18ENSEMBLES 3. GA 2006, Lund – WP 6.2 Meeting 20th Nov

Next meeting?

Any suggestions?

Finland <-> Berlin

When?? A) Mid-April 2007 (before EGU: 15-20.04.07)B) …