PM 2.5 Levels (µg/m³)
This real-time map has been developed by the HealthDEEP IRL. PM 2.5 levels are calculated with a spatio-temporal random forest model, which is trained on PM 2.5 sensor measurements as well as weather data, fire hotspots and topographic data. The model's accuracy is assessed with Mean Squared Error (MSE) and R². The map updates every hour.
About this Project
This study aims to provide a very high-resolution, real-time PM 2.5 map in an area affected by agricultural fires. Small spatial variations in PM 2.5 concentrations can be detected thanks to this fine granularity. The international research lab HealthDEEP, which support this work, studies the relationship between biodiversity, health, and urbanization. It has conducted many works in the area, chosen for its urban-to-rural gradient. The goal is to study the relationship between air quality and forest. Do forests help reducing air pollution?
Methodology
- Sensor data is collected every hour from AirGradient PM2.5 sensors
- Fire hotspots, weather data, and topographic data are incorporated into the model
- Predictions are made using a spatio-temporal random forest approach
- Map and visualizations are updated every hour
Contact
For more information, visit HealthDEEP
AirGradient Sensor
Installation in Village 7
Installation in Phatho PCU