Related Experiment Videos
In search of spatial extremes.
A Stein1, K F Turkman, P Bermudez
1Department of Environmental Sciences, Wageningen Agricultural University, The Netherlands.
Novartis Foundation Symposium
|May 8, 1999
Summary
Identifying environmental extremes requires advanced methods. The peaks over threshold method excels for uncorrelated data, while conditional simulations are best for spatially dependent pollutant data in industrial areas.
Area of Science:
- Environmental Science
- Geostatistics
- Pollution Monitoring
Background:
- Environmental studies increasingly focus on identifying spatial extremes, defined as locations with unusually high pollutant concentrations.
- Industrial estates are complex sites influenced by multiple contamination sources, necessitating robust analytical approaches.
- Understanding pollutant distribution is crucial for environmental risk assessment and remediation strategies.
Purpose of the Study:
- To evaluate and compare different statistical methods for identifying spatial extremes of environmental pollutants.
- To determine the efficacy of extreme value theory, conditional simulations, and disjunctive kriging in characterizing pollutant hotspots.
- To analyze the distribution of heavy metals, polyaromatic hydrocarbons, and mineral oil in a southern Netherlands industrial estate.
Main Methods:
- Application of extreme value theory, specifically the peaks over threshold method.
- Utilizing conditionally simulated fields to model spatial dependence in pollutant concentrations.
- Employing disjunctive kriging as a geostatistical interpolation technique.
- Analyzing pollutant data from a large industrial estate influenced by river inundation, material deposition, and industrial activities.
Main Results:
- The peaks over threshold method proved effective for identifying extremes in spatially uncorrelated environmental variables.
- Conditional simulations demonstrated particular utility in areas exhibiting spatial dependence of pollutant concentrations.
- Different methods showed varying degrees of success depending on the spatial characteristics of the pollutants and the study area.
Conclusions:
- The choice of method for identifying spatial extremes should be guided by the spatial dependency of the environmental data.
- Conditional simulations are recommended for complex industrial sites with spatially correlated pollutant distributions.
- The peaks over threshold method offers a valuable alternative for uncorrelated pollutant data, simplifying extreme value identification.