Development of electromagnetic pollution maps utilizing Gaussian process spatial models
Abstract
Development of electromagnetic pollution maps utilizing Gaussian process spatial models Kiouvrekis Y, Zikas S, Katis I, Tsilikas I, Filippopoulos I. Development of electromagnetic pollution maps utilizing Gaussian process spatial models. Sci Total Environ. 2024 Oct 21;955:176907. doi: 10.1016/j.scitotenv.2024.176907. Abstract The rapid proliferation of wireless technologies in everyday environments demands the quick and precise estimation of electromagnetic field distribution. This distribution is commonly depicted through the electric field strength across various geographical areas. The objective of this research is to determine the most effective geospatial model for generating a national-level electric field strength map within the 30 MHz-6 GHz frequency range. To achieve this, we employed five different methodologies for constructing the electric field strength map. Four of these methodologies are based on Gaussian process regression, while the fifth utilizes the classical weighted-average method of the nearest neighbor. Our study focused on a country with a total area of 9251 km2 , using a dataset comprising 3621 measurements. The findings reveal that Gaussian process spatial models, also known as Kriging models, generally outperform other methods when applied to spatial data. However, it was observed that, after excluding some outlier data points, the performance of the classical nearest neighbor models becomes comparable to that of the Gaussian process models. This indicates the potential for both approaches to be effective, depending on the data quality and the presence of outliers. Excerpts For the Total (Fig. 4 and Table 2), the mean stands at 4.78 V/m, with a maximum value of 37.99 V/m, a minimum of 0.28 V/m, and a median of 3.42 V/m.... The present research constitutes the first step of many that can be implemented in the future regarding the construction of systems for mapping EMR exposure. Initially, future research should be conducted in cities, exploring characteristics related to urban infrastructure, such as building materials, density, and height. This novel approach regards not only improving the model's performance but also developing models that consider these characteristics between measurement points to better capture the spatial distribution of EMR within urban and semi-urban areas. Additionally, it would be interesting to incorporate temporal patterns to account for the various varieties of EMR over time, considering factors such as daily, seasonal, and yearly variations. In a subsequent stage, the inclusion of data from environmental sources, such as weather conditions, levels of atmospheric pollution, and population density, could be beneficial to enhance the predictive capability of the models. Furthermore, research into improving the interpretability of the models should be conducted to provide meaningful insights into the factors influencing EMR levels in different environments (urban, semi-urban, and rural). All of the above should be combined with conducting validation studies in collaboration with stakeholders for the effectiveness of the developed models in real-world scenarios. pubmed.ncbi.nlm.nih.gov
AI evidence extraction
Main findings
Using 3621 measurements over a 9251 km2 area, Gaussian process spatial (Kriging) models generally outperformed other mapping methods for predicting electric field strength in the 30 MHz–6 GHz range. After excluding outlier data points, the classical nearest-neighbor weighted-average method showed performance comparable to Gaussian process models.
Outcomes measured
- Electric field strength (V/m) mapping at national level
- Model performance comparison for spatial prediction (Gaussian process/Kriging vs nearest neighbor weighted-average)
Limitations
- Performance depended on data quality and presence of outliers (results changed after excluding outliers).
- No health outcomes assessed
- focus was on exposure mapping/modeling.
View raw extracted JSON
{
"study_type": "exposure_assessment",
"exposure": {
"band": "RF",
"source": "wireless technologies (environmental EMR)",
"frequency_mhz": null,
"sar_wkg": null,
"duration": null
},
"population": null,
"sample_size": 3621,
"outcomes": [
"Electric field strength (V/m) mapping at national level",
"Model performance comparison for spatial prediction (Gaussian process/Kriging vs nearest neighbor weighted-average)"
],
"main_findings": "Using 3621 measurements over a 9251 km2 area, Gaussian process spatial (Kriging) models generally outperformed other mapping methods for predicting electric field strength in the 30 MHz–6 GHz range. After excluding outlier data points, the classical nearest-neighbor weighted-average method showed performance comparable to Gaussian process models.",
"effect_direction": "unclear",
"limitations": [
"Performance depended on data quality and presence of outliers (results changed after excluding outliers).",
"No health outcomes assessed; focus was on exposure mapping/modeling."
],
"evidence_strength": "moderate",
"confidence": 0.78,
"peer_reviewed_likely": "yes",
"keywords": [
"electromagnetic pollution maps",
"electric field strength",
"Gaussian process regression",
"Kriging",
"spatial models",
"nearest neighbor",
"RF exposure",
"30 MHz-6 GHz"
],
"suggested_hubs": []
}AI can be wrong. Always verify against the paper.
Comments
Log in to comment.
No comments yet.
NNEMF synthesis
1. What they did: This is a mapping study, not a cell, animal, or human biology experiment. The team built a national electric-field map for wireless signals from 30 megahertz to 6 gigahertz over a country of 9,251 square kilometers, using 3,621 measurements. They compared five ways to estimate field strength between measurement points: four Gaussian process regression methods, also called Kriging, and one classical nearest-neighbor weighted average. The record does not give a specific absorption rate, a pulsing pattern, a tissue, or a life stage. 2. What they found: For the total map, mean electric field strength was 4.78 volts per meter, the median was 3.42 volts per meter, the minimum was 0.28 volts per meter, and the maximum was 37.99 volts per meter. Gaussian process spatial models generally outperformed the other methods on these spatial data. After some outlier points were excluded, the classical nearest-neighbor method performed about as well as the Gaussian process models. The authors describe the work as a first mapping step and list later additions they did not include here: city form, building materials, density and height, time of day, season and year, weather, air pollution, and population density. 3. How it connects: The measured quantity is the outside radiofrequency electric field and how it changes from place to place. That matches the framework premise that exposure is not one uniform dose: density gating expects a later response to depend on the receiving cell’s machinery and reserve, so the field a place actually receives is the environmental input any such response would meet. This record stops at that input. Field strength here spans more than a hundredfold from the lowest to the highest reported value, which is a non-uniform exposure surface, not a cellular timing, channel, energy, or redox result. The related studies share mapping tools, spatial labels, or exposure description only, and none shares a regulatory step with this measurement, so none is linked as upstream, downstream, parallel, or conflicting biology. 4. Limits: Specific absorption rate, power density, modulation, and any biological endpoint are absent from the record.