Development of electromagnetic pollution maps utilizing Gaussian process spatial models

Authors: Kiouvrekis Y, Zikas S, Katis I, Tsilikas I, Filippopoulos I

Year: 2024 Oct 21

Category: Environmental Science, Geospatial Modeling

Journal: Sci Total Environ

DOI: 10.1016/j.scitotenv.2024.176907

URL: https://pubmed.ncbi.nlm.nih.gov/39442732/

Abstract

Overview

The rapid proliferation of wireless technologies in everyday environments demands the quick and precise estimation of electromagnetic field (EMF) distribution. Typically, this distribution is visualized through the electric field strength measurements across different geographic regions. Understanding and mapping these fields is essential because there is a documented connection between EMF exposure and potential health risks, making national EMF monitoring important for public safety.

Methodology

  • The study set out to identify the most effective geospatial strategy to generate a national-level electric field strength map for frequencies within 30 MHz-6 GHz.
  • Five mapping methodologies were compared: four based on Gaussian process regression (Kriging models) and one on the classical weighted-average nearest neighbor approach.
  • The research was conducted in a country covering an area of 9251 km², utilizing a robust dataset of 3621 electric field measurements.

Findings

  • Gaussian process (Kriging) models generally outperformed the traditional nearest neighbor method for spatial data mapping.
  • After excluding outlier data, the nearest neighbor and Gaussian process models performed comparably, indicating model choice can be data-dependent.
  • Typical measured values: mean electric field strength of 4.78 V/m, maximum 37.99 V/m, minimum 0.28 V/m, and median of 3.42 V/m.

Implications for Future Research

The study advocates further development of EMR mapping systems, particularly:

  • Applying models in urban contexts, accounting for building materials, density, and height to capture spatial variability in EMR.
  • Incorporating temporal patterns (daily, seasonal, yearly) and environmental data (weather, pollution, population).
  • Improving model interpretability to better understand environmental factors affecting EMF distribution.
  • Validating models in real-world scenarios through stakeholder collaboration, given the established connection between EMF exposure and public health concerns.

Conclusion

This research provides foundational insights and tools for developing detailed maps of electromagnetic pollution, which are important for ongoing EMF risk assessments, public health monitoring, and EMF safety policy.

← Back to Stats