Application of validation data for assessing spatial interpolation methods for 8-h ozone or other sparsely monitored...

Area of Science:

• Environmental Science
• Atmospheric Chemistry
• Geostatistics

Background:

• Adverse health effects of high ground-level ozone concentrations are known.
• Estimating ozone exposure is difficult due to sparse urban monitoring networks.
• Network sparseness hinders validation of interpolation techniques, increasing overfitting risk.

Purpose of the Study:

• To evaluate simple spatial interpolation techniques for 8-hour ozone exposure estimation.
• To identify the most reliable method for sparsely monitored urban areas.
• To assess the feasibility of validating interpolation techniques with limited data.

Main Methods:

• Tested various spatial interpolation techniques using thousands of random data subsets.
• Utilized data from two urban areas with dense monitoring networks for true validation.
• Employed ordinary kriging with an exponential variogram, calibrating only the range parameter.

Main Results:

• Ordinary kriging demonstrated superior performance compared to other tested methods.
• The method yielded reliable confidence intervals for ozone exposure estimates.
• Sufficient information for range parameter calibration exists even with low Moran I p-values.

Conclusions:

• Ordinary kriging is a robust method for estimating ozone exposure in sparsely monitored urban environments.
• The technique provides reliable confidence intervals, crucial for risk assessment.
• The study offers an R script to apply this methodology to other sparsely monitored substances.