Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Towards an Open Global Air Quality Monitoring Platform to Assess Children's Exposure to Air Pollutants in the Light of COVID-19 Lockdowns

Christina Last, Prithviraj Pramanik, Nikita Saini, Akash Smaran Majety, Do-Hyung Kim, Manuel García-Herranz, Subhabrata Majumdar

TL;DR

This work tackles the challenge of global inequities in air quality monitoring for children amid COVID-19, proposing an open framework that combines ground PM2.5 data with satellite-derived inputs to predict exposure at global and regional scales. A stacked ensemble model using base learners such as LightGBM, XGBoost, and Random Forest, with a linear meta-model, achieves a global $MAE$ of $5.60~\mu g/m^3$, $RMSE$ of $22.8~\mu g/m^3$, and $R^2=0.49$, performing best in Myanmar and poorest in Ghana. The approach emphasizes stakeholder involvement (UNICEF and Solve for Good) and incorporates a Mapbox story map to communicate pre/post-lockdown exposure shifts, while expert feedback identifies data scarcity, need for explainability, and plans to incorporate auxiliary data sources. The contribution lies in a concrete, collaborative pathway for developing regionally tuned, data-driven air quality insights to inform child health policies in the post-COVID era. This work lays a foundation for equitable global air quality monitoring by integrating open data, multi-source features, and iterative stakeholder feedback to guide future enhancements.

Abstract

This ongoing work attempts to understand and address the requirements of UNICEF, a leading organization working in children's welfare, where they aim to tackle the problem of air quality for children at a global level. We are motivated by the lack of a proper model to account for heavily fluctuating air quality levels across the world in the wake of the COVID-19 pandemic, leading to uncertainty among public health professionals on the exact levels of children's exposure to air pollutants. We create an initial model as per the agency's requirement to generate insights through a combination of virtual meetups and online presentations. Our research team comprised of UNICEF's researchers and a group of volunteer data scientists. The presentations were delivered to a number of scientists and domain experts from UNICEF and community champions working with open data. We highlight their feedback and possible avenues to develop this research further.

Towards an Open Global Air Quality Monitoring Platform to Assess Children's Exposure to Air Pollutants in the Light of COVID-19 Lockdowns

TL;DR

This work tackles the challenge of global inequities in air quality monitoring for children amid COVID-19, proposing an open framework that combines ground PM2.5 data with satellite-derived inputs to predict exposure at global and regional scales. A stacked ensemble model using base learners such as LightGBM, XGBoost, and Random Forest, with a linear meta-model, achieves a global of , of , and , performing best in Myanmar and poorest in Ghana. The approach emphasizes stakeholder involvement (UNICEF and Solve for Good) and incorporates a Mapbox story map to communicate pre/post-lockdown exposure shifts, while expert feedback identifies data scarcity, need for explainability, and plans to incorporate auxiliary data sources. The contribution lies in a concrete, collaborative pathway for developing regionally tuned, data-driven air quality insights to inform child health policies in the post-COVID era. This work lays a foundation for equitable global air quality monitoring by integrating open data, multi-source features, and iterative stakeholder feedback to guide future enhancements.

Abstract

This ongoing work attempts to understand and address the requirements of UNICEF, a leading organization working in children's welfare, where they aim to tackle the problem of air quality for children at a global level. We are motivated by the lack of a proper model to account for heavily fluctuating air quality levels across the world in the wake of the COVID-19 pandemic, leading to uncertainty among public health professionals on the exact levels of children's exposure to air pollutants. We create an initial model as per the agency's requirement to generate insights through a combination of virtual meetups and online presentations. Our research team comprised of UNICEF's researchers and a group of volunteer data scientists. The presentations were delivered to a number of scientists and domain experts from UNICEF and community champions working with open data. We highlight their feedback and possible avenues to develop this research further.

Paper Structure

This paper contains 14 sections, 6 figures, 1 table.

Table of Contents

  1. Introduction
  2. Motivation
  3. Project Workflow
  4. Data
  5. Methods
  6. Model
  7. Stakeholder involvement
  8. Expert Feedback
  9. Data scarcity:
  10. Links to COVID-19 infection rates:
  11. Local heterogeneity of data and transparent predictions:
  12. Technical feedback:
  13. Auxiliary data sources:
  14. Conclusion

Figures (6)

  • Figure 1: Global distribution of (a) Ground Level Air Quality Sensors, and (b) Child Population, Data source: silent.
  • Figure 4: The project process 1. Identifying the problem, 2. Scoping a potential solution, 3. Developing a scientific methodology, and 4. Receiving feedback from project partner (UNICEF) and update project process.
  • Figure 5: Top panel illustrates the "local" extraction of AOD from a 75m buffer around a point. Bottom panel illustrates the "city-level" extraction of AOD for each city globally. AOD = Aerosol Optical Depth is a well-known proxy for PM2.5 Kumar.
  • Figure 6: Flow diagrams for (a) Data Preparation, and (b) Modelling Process.
  • Figure 7: Comparative analysis of (a) Hà Nôi (an urbanised location in northern Vietnam) and (b) Đ iên Biên (a rural region in northern Vietnam) and their predicted PM2.5 concentrations
  • ...and 1 more figures