Hazard and living environment: combining industrial risk and landscape representations

Abstract

In the context of the French estuary of the Seine River (Normandy), around the urban area of Le Havre, this paper studies the determinants of industrial risk perception of the resident population. More precisely, to what extent the presence of components in the industrial landscape may influence this risk perception. Several complementary methods were combined to evaluate risk perception, assess surrounding landscapes or measure the distance to landscape components. Qualitative, quantitative, and spatial data were collected, pooled and treated in a geographic information system in order to arrive at two main results. First, risk sensitivity depends on various factors including the landscape dimension and the visibility of industrial components. Second, mental maps drawn by people allow a better understanding of industrial risk sensitivity; it appears that areas of risk are more precisely delineated by people who are less worried about risk.

Keywords:

• industrial
• hazards
• perception
• mental maps
• geostatistics

Notes

1. The European Council Directive 96/82/EC on the control of major-accident hazards, the so-called Seveso directive, requires that member States and companies identify the risks associated with certain hazardous industrial activities and take the necessary measures to deal with them.

2. A PPI [Plan Particulier d'Intervention] is a French emergency/intervention plan designed in case of industrial accident.

3. The survey covered 725 people but only 650 records were included in the spatial database that was used for further analysis. The 75 other records were missing workable location information.

4. This variable is very strongly correlated with the age variable. The explanatory strength of this variable being better than the one of age and the variable for retired informants was retained in the final model.

5. The direction of the effects on the dependent variables is derived from the sign of the coefficient in the logistic regression and confirmed by the computation of marginal effects that are not presented in this paper.

6. Once again the quality of the model was evaluated through the AIC.