22
Input2012 - EVALUATING POST-ACCIDENT NUCLEAR RISK BY COUPLING GIS AND ROUGH SET THEORY Salem Chakhar University of Laval, Canada Clara Pusceddu University of Sassary – Faculty of Architecture of Alghero, Italy Ines Saad, University of Picardie, France 1 - Cagliari 11 May 2012

Chakhar, Pusceddu & Saad - input2012

Embed Size (px)

DESCRIPTION

Salem Chakhar, Clara Pusceddu and Ines Saad on "Evaluating Post-Accident nuclear risk by coupling GIS and rough set theory"

Citation preview

Page 1: Chakhar, Pusceddu & Saad - input2012

Inp

ut2

01

2 -

EVALUATING POST-ACCIDENT

NUCLEAR RISK BY COUPLING GIS

AND ROUGH SET THEORYSalem Chakhar University of Laval, Canada

Clara Pusceddu University of Sassary – Faculty of Architecture

of Alghero, Italy

Ines Saad, University of Picardie, France

1

-C

aglia

ri 11

Ma

y 2

01

2

Page 2: Chakhar, Pusceddu & Saad - input2012

INTRODUCTION

� The management of the consequences of a major nuclear

accident necessarily involves the consideration of multiple

criteria in order to ensure sustainable development in

areas that might be affected.

Inp

ut2

01

2 -

Ca

glia

ri 11

Ma

y 2

01

2

� Furthermore, the management of the consequences of a

major nuclear accident requires a multidisciplinary

approach to produce a sustainable response to the

environmental, economic and social problems linked to the

various local intricacies.

2

Ca

glia

ri 11

Ma

y 2

01

2

Page 3: Chakhar, Pusceddu & Saad - input2012

OBJECTIVE

� Propose a Multicriteria Evaluation approach for

characterizing the different districts of the affected area in

terms of their vulnerability levels while taking into account

multiple stakeholders with contradictory objectives and

priorities.

Inp

ut2

01

2 -

Ca

glia

ri 11

Ma

y 2

01

2

priorities.

� The proposed approach is composed of 4 phases:

1. identifying the stakes involved,

2. identification of representative criteria,

3. quantifying criteria scores, and

4. group multicriteria classification. 3

Ca

glia

ri 11

Ma

y 2

01

2

Page 4: Chakhar, Pusceddu & Saad - input2012

FOCUSING ON 4 PHASE…. (CHAKHAR…)

� It requires the use of an adequate technique to combine the

perspectives of different stakeholders.

� We adopted the output-oriented strategy (Dias and

Inp

ut2

01

2 -

Ca

glia

ri 11

Ma

y 2

01

2

� We adopted the output-oriented strategy (Dias and

Climaco, 2000) to combine these perspectives. This strategy

works as follows:

1. first, each stakeholder performs her/his individual

classification; then

2. an appropriate aggregation operator is used to

combine the individual classifications into a collective

one. 4

Ca

glia

ri 11

Ma

y 2

01

2

Page 5: Chakhar, Pusceddu & Saad - input2012

A DECISION SUPPORT SYSTEM DESIGN….

� A decision support system supporting the proposed

approach has been developed by coupling GIS technology

and Rough set theory (Pawlak 1991).

Inp

ut2

01

2 -

Ca

glia

ri 11

Ma

y 2

01

2

� The approach is validated using real-world data relative to

a nuclear risk management decision problem in the

southern France.

5

Ca

glia

ri 11

Ma

y 2

01

2

Page 6: Chakhar, Pusceddu & Saad - input2012

IN THIS PRESENTATION ….

1. Introduction of the proposed impact evaluation approach

2. Presentation of the case study with some conclusion

Inp

ut2

01

2 -

Ca

glia

ri 11

Ma

y 2

01

2

2. Presentation of the case study with some conclusion

6

Ca

glia

ri 11

Ma

y 2

01

2

Page 7: Chakhar, Pusceddu & Saad - input2012

1. MULTICRITERIA IMPACT EVALUATION

APPROACHIn

pu

t20

12

-C

aglia

ri 11

Ma

y 2

01

2

Phase 1. Identifying the stakes involved

Phase 1. Identifying the stakes involved

Phase 2: Identifyingrepresentative criteriaPhase 2: Identifying

representative criteria

7

Ca

glia

ri 11

Ma

y 2

01

2

representative criteriarepresentative criteria

Phase 3: Quantifyingcriteria scores

Phase 3: Quantifyingcriteria scores

Phase 4: Group multicriteria classification

Phase 4: Group multicriteria classification

Page 8: Chakhar, Pusceddu & Saad - input2012

1. MULTICRITERIA IMPACT EVALUATION

APPROACHIn

pu

t20

12

-C

aglia

ri 11

Ma

y 2

01

2

8

Ca

glia

ri 11

Ma

y 2

01

2

� Identification of the stakes involved including everything that can be affected by an accident such as zones that are densely inhabited, business activities, and cultural and environmental assets.

� Then one or more adverse effects have to be linked to each stakeso that they represent the consequences of an accident in various sectors.

Page 9: Chakhar, Pusceddu & Saad - input2012

1. MULTICRITERIA IMPACT EVALUATION

APPROACHIn

pu

t20

12

-C

aglia

ri 11

Ma

y 2

01

2

9

Ca

glia

ri 11

Ma

y 2

01

2� Once the various factors and adverse effects have been selected, the criteria that characterize them have to be identified.

� Formally, a criterion is a function qj, defined on a set of decision objects U (which are districts in our case), taking its values in an ordered set, and structuring the stakeholder's preferences according to some points of view.

� The evaluation of an object u in respect to criterion qj is denoted qj(u).

� We denote by Q={q1, …,qm} the set of m evaluation criteria.

Page 10: Chakhar, Pusceddu & Saad - input2012

1. MULTICRITERIA IMPACT EVALUATION

APPROACHIn

pu

t20

12

-C

aglia

ri 11

Ma

y 2

01

2

10

Ca

glia

ri 11

Ma

y 2

01

2

� This involves evaluating the consequences on each district in respect to each criterion.

� The output of this phase is an evaluation matrix where rows represent the districts and columns represent the evaluation criteria.

� Each box then contains the corresponding value of the criterion for the district in question. In terms of this phase, each district u will be associated with the vector (q1(u),…,qm(u)) which represents the evaluations of u with respect to the criteria in Q.

Page 11: Chakhar, Pusceddu & Saad - input2012

1. MULTICRITERIA IMPACT EVALUATION

APPROACHIn

pu

t20

12

-C

aglia

ri 11

Ma

y 2

01

2

11

Ca

glia

ri 11

Ma

y 2

01

2

� The aim of group multicriteria classification phase is to assign the different districts of the study area to different risk classes while taking into account the perspectives of multiple stakeholders.

� A multicriteria classification model, called Dominance-based Rough Set Approach (DRSA) (Greco et al., 2002) an extension to rough sets theory (Pawlak 1991) to multicriteria classification, will individually be used by the different stakeholders.

� Some appropriate aggregation rules are then used to coherently combine the outputs of different stakeholders.

� The DRSA is then used once again to obtain the final classification in terms of vulnerability/risk levels of the districts.

Page 12: Chakhar, Pusceddu & Saad - input2012

4. CASE STUDY: NUCLEAR RISK MANAGEMENT

DECISION PROBLEM

� The problem considered here concerns the management of post-

accident nuclear risk in the southern France region.

� This problem has been conducted during the PRIME project,

which is supervised by the French Institute for Radioprotection

Inp

ut2

01

2 -

Ca

glia

ri 11

Ma

y 2

01

2

which is supervised by the French Institute for Radioprotection

and Nuclear Safety.

� A full description of the project is available in Mercat-Rommens et

al. (2010).

� The study zone covers a radius of some fifty kilometers around

three nuclear sites in the lower Rhône Valley (the Cruas,

Tricastin-Pierrelatte and Marcoule sites). 12

Ca

glia

ri 11

Ma

y 2

01

2

Page 13: Chakhar, Pusceddu & Saad - input2012

4. CASE STUDY: NUCLEAR RISK MANAGEMENT

DECISION PROBLEM

� The objective of the PRIME is to develop, conjointly with the

experts, the stakeholders and representatives of the territory, a

multicriteria evaluation approach permitting to analysis and

characterize the contaminated territory that will be useful for the

managers of the risk.

Inp

ut2

01

2 -

Ca

glia

ri 11

Ma

y 2

01

2

� Practically, the evaluation approach should associate to each

district of the study area a degree representing the risk on this

district of a nuclear accident resulting in releases into the

atmosphere.

� For this purpose, a scale of six from 0 (for a situation described as

normal) to 5 (in the event of a major and long-lasting negative

impact) has been adopted by PRIME working team. 13

Ca

glia

ri 11

Ma

y 2

01

2

Page 14: Chakhar, Pusceddu & Saad - input2012

4. CASE STUDY: NUCLEAR RISK MANAGEMENT

DECISION PROBLEMIn

pu

t20

12

-C

aglia

ri 11

Ma

y 2

01

2

The table describes the vulnerability measurement scale

14

Ca

glia

ri 11

Ma

y 2

01

2

Page 15: Chakhar, Pusceddu & Saad - input2012

4. CASE STUDY: APPLICATION. PHASE 1.

IDENTIFYNG THE STAKES INVOLVEDIn

pu

t20

12

-C

aglia

ri 11

Ma

y 2

01

2

The stakes are organized into 3 groups:

(i) radioecological consequences which are related to the

contamination of urban, agricultural, costal and natural and

forest areas; Rhône River and ground water;

15

Ca

glia

ri 11

Ma

y 2

01

2

forest areas; Rhône River and ground water;

(ii) economic consequences related to contamination and damage on

companies, tourism activity, real estate and employment;

(iii) population reactions.

Page 16: Chakhar, Pusceddu & Saad - input2012

4. CASE STUDY: APPLICATION. PHASE 2.

CHOOSING REPRESENTATIVE CRITERIAIn

pu

t20

12

-C

aglia

ri 11

Ma

y 2

01

2

� Based on the stakes identified in the previous phase, a

comprehensive list of criteria has been identified by the different

stakeholders (see Mercat-Rommens et al., 2010).

� For the purpose of the present paper, only a subset of criteria, will

be used for illustration.

16

Ca

glia

ri 11

Ma

y 2

01

2

be used for illustration.

Page 17: Chakhar, Pusceddu & Saad - input2012

4. CASE STUDY: APPLICATION. PHASE 3.

QUANTIFYING THE CRITERIA SCORESIn

pu

t20

12

-C

aglia

ri 11

Ma

y 2

01

2

� The quantification of criteria required the federation of available radio-ecological data (field data, modeling, experimental results), as well as territory data.

� The assessment method of the radiological sensitivity indicators invoked classic impact calculation models for radionuclides used at the IRSN: CASTEAUR code for river discharges (see Duchesne et

17

Ca

glia

ri 11

Ma

y 2

01

2

the IRSN: CASTEAUR code for river discharges (see Duchesne et al., 2003), ASTRAL code for forest ecosystem and food chain contamination following accidental radioactive pollution (see Renaud et al., 1999; Calmon and Mourlon, 2005), integrating the spatial variability of parameters.

Page 18: Chakhar, Pusceddu & Saad - input2012

4. CASE STUDY: APPLICATION. PHASE 3.

QUANTIFYING THE CRITERIA SCORESIn

pu

t20

12

-C

aglia

ri 11

Ma

y 2

01

2

An extract of the obtained evaluation matrix that represents a common

information table for all the involved stakeholders for the 4 phase.

18 districts (x1,…..x18)

have been carefully

selected (from 491

18

Ca

glia

ri 11

Ma

y 2

01

2

selected (from 491

districts) by PRIME

working team: these

districts are chosen to

be as representative as

possible by including

urban, industrial as

well as rural districts.

Page 19: Chakhar, Pusceddu & Saad - input2012

4. CASE STUDY: APPLICATION. PHASE

4. GROUP MULTICRITERIA CLASSIFICATIONIn

pu

t20

12

-C

aglia

ri 11

Ma

y 2

01

2

� Individual classification

� Given the evaluation of the 18 selected districts in respect to all criteria, each

stakeholder is called to classify each of them on the global vulnerability scale.

� The responses of the stakeholders are then used to define the values of the

decision attributes E1, E2 and E3 associated with the 3 stakeholders considered

in this paper.

19

Ca

glia

ri 11

Ma

y 2

01

2

in this paper.

� Then each stakeholder should apply the DRSA on each decision table to get its

own classification.

� Aggregation of the individual classification

� Final classification

Page 20: Chakhar, Pusceddu & Saad - input2012

4. CASE STUDY: APPLICATION. PHASE 3.

QUANTIFYING THE CRITERIA SCORESIn

pu

t20

12

-C

aglia

ri 11

Ma

y 2

01

2

20

Ca

glia

ri 11

Ma

y 2

01

2

Page 21: Chakhar, Pusceddu & Saad - input2012

4. CASE STUDY: APPLICATION, 4. GROUP

MULTICRITERIA CLASSIFICATIONIn

pu

t20

12

-C

aglia

ri 11

Ma

y 2

01

2

Aggregation

� In this step, we first apply the aggregation procedure to

construct a common decision table with common Condition

attributes (Criteria) and Decision Attributes.

21

Ca

glia

ri 11

Ma

y 2

01

2

Final classification

� Next, the Dominance-based Rough Set Approach (DRSA) is applied

to the common decision table to classify the districts of the

study area.

Page 22: Chakhar, Pusceddu & Saad - input2012

4. CASE STUDY: APPLICATION, 4. GROUP

MULTICRITERIA CLASSIFICATIONIn

pu

t20

12

-C

aglia

ri 11

Ma

y 2

01

2

� The result of classification is shown in Figure 1. The left-hand side of the interface shows the global vulnerability scale with shaded tones. The map on the right-hand side of the interface shows the final classification of the different districts.

22

Ca

glia

ri 11

Ma

y 2

01

2

classification of the different districts.

� It is easy to see that vulnerability decreases relatively concentrically around the Tricastin-Pierrelatte nuclear site, which is the location of the fictive accident considered in this case study.

The obtained risk map represents the main decision support that could be

used by risk managers to effectively and rapidly manage the contaminated

districts by appropriately identifying the required measures for affected

districts