GlobalHighPM2.5: Daily 1 km Ground-Level PM2.5 Pollution Data (2017–Present)

From 2017 to the present, this dataset provides seamless daily, monthly, and yearly 1 km resolution ground-level PM2.5 concentrations over global land areas. It was generated by Jing Wei and colleagues at the University of Maryland using AI to fuse ground measurements, satellite data, and atmospheric models. The dataset is associated with a 2023 Nature Communications article and is part of the larger GlobalHighAirPollutants (GHAP) series.

Use Cases

• Modeling long-term PM2.5 exposure for epidemiological studies based on the global daily 1 km data.
• Analyzing spatiotemporal variability of air pollution based on the seamless, high-resolution global coverage.
• Validating atmospheric chemistry and transport models based on the AI-derived PM2.5 estimates.
• Assessing environmental justice and pollution inequality based on the detailed 1 km spatial data.

Strengths

• High spatial resolution of 1 km for detailed local analysis.
• Demonstrated high quality with cross-validation R² values of 0.91 (daily), 0.97 (monthly), and 0.98 (yearly).
• Seamless global coverage over land with 100% spatial coverage.
• Data is continuously updated from 2017 to the present.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Data volume limitations are noted, with yearly data files provided separately.
• Last update date is unknown; freshness unverified.

Provenance

Source

University of Maryland, College Park; generated from ground measurements, satellite remote sensing, atmospheric reanalysis, and model simulations.

Collection Method

Generated using artificial intelligence techniques, accounting for spatiotemporal heterogeneity of air pollution.

Time Range

2017 to present

Freshness

Continuously updated to the present, but specific last update date is unknown.

Geography

Global land areas