pure experimental zone

32
PURE Experimental Zone PURE Probability, Uncertainty and Risk in the Environment http:// pureexperimentalzone.org/ Katherine Royse, Andrew Hughes John Hillier, Lei Wang and Anubha Singh

Upload: aren

Post on 22-Feb-2016

61 views

Category:

Documents


0 download

DESCRIPTION

PURE Probability, Uncertainty and Risk in the Environment. PURE Experimental Zone . Katherine Royse, Andrew Hughes John Hillier, Lei Wang and Anubha Singh. http://pureexperimentalzone.org/. PURE P robability, U ncertainty and R isk in the E nvironment. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: PURE Experimental Zone

PURE Experimental Zone

PUREProbability, Uncertainty and Risk in the Environment

http://pureexperimentalzone.org/

Katherine Royse, Andrew HughesJohn Hillier, Lei Wang and Anubha Singh

Page 2: PURE Experimental Zone

Aim: To increase the impact of NERC’s Natural Hazard research, and to take a leadership role in changing how uncertainty and risk is assessed and quantified across the Natural Hazards

Two Parts: Research Programme Research and Knowledge Exchange Network

Goals :1. Improve the assessment and quantification of uncertainty and risk in

NH2. Stimulate good practice guidance and standardisation of the

assessment and quantification of uncertainty and risk across the NH community.

PURE

PUREProbability, Uncertainty and Risk in the Environment

http://pureexperimentalzone.org/

Page 3: PURE Experimental Zone

Why Now?

Mythe water treatment works, 2007, threaten drinking water supplies to 350,000 people

Page 4: PURE Experimental Zone

Global Economic losses related to natural hazards are increasing

2010 : US$130 Billion, Earthquakes Haiti, Chilli, China Iceland, floods Pakistan

2011: US$380 Billion ,Earthquakes Japan, New Zealand, Floods Thailand, Wind Storms USA

2012: US$160 Billion, Hurricane Sandy and drought USA, Earthquake in Italy

Hurricane Andrew 1992 - Katrina 2005 : US$31.3bn -US$40-60bn insured losses

PUREhttp://pureexperimentalzone.org/

Page 5: PURE Experimental Zone

Catastrophe ModelingWhat is it ?

A tool that quantifies risk

How does it do it?Examines insured values that are exposed to catastrophic

perils such as hurricanes, earthquakes and terrorism

Why do we need it?Aids management decision making on

Pricing and underwriting Reinsurance buying Rating Agencies Portfolio management

PUREhttp://pureexperimentalzone.org/

Page 6: PURE Experimental Zone

Types of Catastrophe: Natural & Man Made

Earthquake, volcanic eruption

Tropical storms, hurricanes

Wind storms

Other natural catastrophes:Floods (river and coastal)

TsunamisHeat waves and droughts

WildfiresHail storms

Landslides, avalanches

Man-made catastrophes:Fires, Explosions

Road traffic and rail disastersAviation and space catastrophes

Mining accidencesShipping accidences

Collapse of buildings / bridgesTerrorism

World hazard map: Munich Re (2009)

e.g. 7 bn USD in 2008 (Guy Carpenter,2009) PURE

http://pureexperimentalzone.org/

Page 7: PURE Experimental Zone

Limitations Catastrophe models provide a guide to the risk not answers to precise calculations of the Risk

• Catastrophe models are inherently limited by limitations in our scientific knowledge

• Every model is based on simplifying assumptions and there is significant uncertainty around many of those assumptions

• Catastrophe model results are very sensitive to certain assumptions which is why the model loss estimates vary so significantly for different, scientifically valid assumptions

PUREhttp://pureexperimentalzone.org/

Page 8: PURE Experimental Zone

PURE

What is it ?Why do we need it ?How do I use it ?

http://pureexperimentalzone.org/

The Experimental Zone

Page 9: PURE Experimental Zone

What is the Experimental Zone ?

PUREhttp://pureexperimentalzone.org/

1. A Platform which allows for the sharing and linking of data, models, and analytical and visualisation tools (in real-time) across the internet

2. Provides an environment and the technology to enable users to test model compositions and demonstrate the effects of model component choice on end results

3. Facilitate collaboration between research groups and enable cross-disciplinary models to be linked

4. Allow for the integration of commercial and open source data and model components

5. Simplify sensitivity and uncertainty analysis by allowing for the easy substitution of one model component with another

Page 10: PURE Experimental Zone

PUREhttp://pureexperimentalzone.org/

Why do we need it ? 1. New European regulation under Solvency II requires that firms

understand the assumptions made in all models used as part of their solvency calculations and management activity

2. It will lower the barriers to access to current research developments and the scientists themselves. The Zone will allow for a greater access to a range of models in an open system

3. Allow the exploration of uncertainty due to different compositions – model comparison, model critique and understanding model limitations

4. Moving towards “Plug and Play” component driven system which means that model components can be swapped in and out of the compositions and that everyone can bring in ideas

Page 11: PURE Experimental Zone

The Website

PUREhttp://pureexperimentalzone.org/

Page 12: PURE Experimental Zone

Developing a Groundwater Catastrophe model for the

Marlborough and the Berkshire Downs

http://pureexperimentalzone.org/PURE

Page 13: PURE Experimental Zone

Background to GW flooding

Problem definition & issues

Recent events 1993/94, 2000-01 & 2002-03

Insurance claims est. £M50-100 per year

Brighton 2000/01 event £800k not including damage to railway

Most vulnerable properties (~383,000) located on exposed Chalk aquifers of SE England, although other places do flood!

http://pureexperimentalzone.org/PURE

Page 14: PURE Experimental Zone

The Black Box

PUREhttp://pureexperimentalzone.org/

Page 15: PURE Experimental Zone

Catastrophe Model Components

Stochastic Module Hazard Module Vulnerability Module Financial Module

• Based on historical data• Large catalogue of

simulated events generated from historical parameters & scientific predictions

• Local intensity of hazard parameter for each cat event

• Geological & topographical features of a region included

• Converts hazard parameter to loss: “damage functions”

• Includes factors or “modifiers” eg Sum insured, Construction, Occupancy, Year built

• Applies deductibles and limits• Calculate losses on different

financial perspectives

Source: RMS

PUREhttp://pureexperimentalzone.org/

Page 16: PURE Experimental Zone

How will IEM techniques help ?

1. Lack of openness has restricted the ability of the insurance community to exploit new research / willingness to try new things

2. New European regulation under Solvency II requires that firms understand the assumptions made in all models used as part of their solvency calculations and management activity

3. Allow the exploration of uncertainty due to different compositions – model comparison and understanding limitations

4. Moving towards “Plug and Play” means that models can be swapped in and out of the composition

5. Allows greater access to a range of models – an open system

6. Scenario management – “Manager components”

PUREhttp://pureexperimentalzone.org/

Page 17: PURE Experimental Zone

• Used OpenMI to link models together

• Fluid earth : – Software Development Kit (SDK)

– Pippistrelle to provides an interface for linking OpenMI compliant models –

PUREhttp://pureexperimentalzone.org/

Page 18: PURE Experimental Zone

PURE

Study area

http://pureexperimentalzone.org/

Page 19: PURE Experimental Zone

How it works - 1• Groundwater model is run

for 30 years to provide location of groundwater flooding (GW head above land surface)

• Ranking of GW flood events produced (equal basis)

• Maps of GW flooding produced

PUREhttp://pureexperimentalzone.org/

Page 20: PURE Experimental Zone

How it works - 2• Distribution of houses

matched against flood extent for each event identified

• Total cost for each event produced by calculating flood depth and relating that to cost of damage

• All events are ranked by total cost and OEP curves produced

PUREhttp://pureexperimentalzone.org/

Page 21: PURE Experimental Zone

www.fluidearth.netPURE

http://pureexperimentalzone.org/

Page 22: PURE Experimental Zone

Components are interchangeable: potentially anywhere in the world!

Video of setting up the GW Flooding composition is available on PURE Expt Zone

Modelling Co.

CAT Modelling

Ins. Co. Portfolio

PURE

Page 23: PURE Experimental Zone

Results

Occurrence exceedance probability curve is the cumulative loss

distribution for any one occurrence in a given year i.e. the probability that losses from a single event will

exceed a given amount

Flood depth- Damage Curves

PURE

Page 24: PURE Experimental Zone

• Other perils can be modelled

• All these components are available from the Experimental Zone

• Videos for constructing the CAT model are on the Zone too

• Technology and help to make models linkable are available – via FluidEarth SDK

PUREhttp://pureexperimentalzone.org/

Results

Page 25: PURE Experimental Zone

PUREhttp://pureexperimentalzone.org/

What next ? We are now in the process of developing phase 2

Our vision for the future:

1. Developing a ‘playground’ to trial out new ideas

2. Open and closed zones

3. Online Catalogue of available data, models, software, visualisation and analytical tools

4. Web-enabled software tools (Pipistrelle)

Page 26: PURE Experimental Zone

PUREhttp://pureexperimentalzone.org/

Developing a ‘Playground’ to trail out new ideas

1. Quick to set up new compositions; particularly useful for un-modelled perils

2. Allows users to test out a variety of different model compositions

3. Direct contact with researchers and model developers

4. Provide access to a world wide community

Page 27: PURE Experimental Zone

PUREhttp://pureexperimentalzone.org/

Open and Closed areas 1. The experimental zone will be

developed on an open access / open source principle as far as possible

2. Closed areas will be developed for those users working on more sensitive projects

3. Provide users with the ability to choose how open there individual components are

Page 28: PURE Experimental Zone

PUREhttp://pureexperimentalzone.org/

Lowering the barriers to use

1. Make the zone accessible and easy to find by all users

2. Improve and develop user manuals and videos

3. Provide a catalogue of linkable components

Page 29: PURE Experimental Zone

PUREhttp://pureexperimentalzone.org/

Web-enabled Software tools1. Going forward we will remove

the need to download tools onto your own PC

2. Access to tools e.g. Pipistrelle will be via the Experimental zones own server

3. Develop the zone so that it has an elastic infrastructure to meet the demands of the user community

Page 30: PURE Experimental Zone

PUREhttp://pureexperimentalzone.org/

Online Catalogue & Toolkit

1. Community generated content

2. Case studies

3. Linkable model and data components

4. Publically/commercially available data and toolkits

5. Visualisation and analytical tools

Page 31: PURE Experimental Zone

PUREhttp://pureexperimentalzone.org/

Initiating discussions with the user community

1. The Experimental zone needs to work for a very varied community

2. Before the development of part 2 starts we will work with a small user group to understand how they could use to zone

3. If your interested please get in contact

Page 32: PURE Experimental Zone

Summary

PUREhttp://pureexperimentalzone.org/

The Experimental Zone

1. It is a Platform which allows for the sharing and linking of data, models, and analytical and visualisation tools (in real-time) across the internet using OpenMI

2. It will simplify sensitivity and uncertainty analysis by allowing for the easy substitution of one model component with another

3. It will Facilitate collaboration between research groups and enable cross-disciplinary models to be linked

4. It will allow for the integration of commercial and open source data and model components