Related Experiment Video
Updated: Jul 26, 2025
Use of Principal Components for Scaling Up Topographic Models to Map Soil Redistribution and Soil Organic Carbon
Published on: October 16, 2018
Estimating the daily average concentration variations of PCDD/Fs in Taiwan using a novel Geo-AI based ensemble mixed
Chin-Yu Hsu1, Tien-Wei Lin2, Jennieveive B Babaan3
1Department of Safety, Health and Environmental Engineering, Ming Chi University of Technology, New Taipei City, Taiwan; Center for Environmental Sustainability and Human Health, Ming Chi University of Technology, New Taipei City, Taiwan.
This study introduces a novel Geo-AI model to predict polychlorinated dibenzodioxins and dibenzofurans (PCDD/Fs) concentrations. The advanced model accurately estimates spatial-temporal fluctuations, aiding pollution control and health studies.
Area of Science:
- Environmental Science
- Geospatial Analysis
- Artificial Intelligence
Background:
- Polychlorinated dibenzodioxins and dibenzofurans (PCDD/Fs) pose significant risks to human health.
- Extensive field research is crucial for understanding and mitigating PCDD/F exposure.
- Accurate spatial-temporal monitoring of PCDD/Fs is essential for effective environmental management.
Purpose of the Study:
- To develop and validate a novel geospatial-artificial intelligence (Geo-AI) based ensemble mixed spatial model (EMSM).
- To predict spatial-temporal fluctuations in PCDD/F concentrations across Taiwan from 2006 to 2016.
- To identify key factors influencing PCDD/F distribution and temporal variations.
Main Methods:
- Integration of multiple machine learning algorithms and geographic predictor variables using SHapley Additive exPlanations (SHAP) values.
- Development of EMSMs incorporating kriging, five machine learning models, and ensemble techniques.
- Utilizing daily PCDD/F I-TEQ levels, meteorological data, geospatial predictors, and social/seasonal factors for model construction and validation.
Main Results:
- The developed EMSM demonstrated superior performance compared to other models, achieving an 87% increase in explanatory power.
- Temporal fluctuations in PCDD/F concentrations were significantly correlated with weather conditions.
- Geographical variations in PCDD/Fs were linked to urbanization and industrialization patterns.
Conclusions:
- The novel Geo-AI based EMSM provides accurate estimations of spatial-temporal PCDD/F variations.
- Findings support targeted pollution control strategies and epidemiological research on PCDD/F health impacts.
- This approach offers a robust framework for environmental monitoring and risk assessment of persistent organic pollutants.
