Daily PM2.5 Maps for East Asia at 6km Resolution, 2011-2019

Daily surface fine particulate matter concentrations inferred from satellite data over South Korea, eastern China, and Japan. The dataset provides continuous 6km x 6km resolution maps from 2011 to 2019, generated by a random forest algorithm trained on national monitoring data. Drew C. Pendergrass from Harvard University created this resource, which shows high correlation with ground observations (R² = 0.89 for daily values).

Use Cases

• Analyzing long-term air pollution trends based on the 2011-2019 daily time series.
• Modeling population exposure to PM2.5 based on the continuous 6km x 6km spatial coverage.
• Validating other air quality models or satellite products based on the high reported cross-validation accuracy (R² = 0.89).
• Studying the relationship between meteorological conditions and PM2.5 levels based on the use of weather variables as model predictors.

Strengths

• High predictive accuracy for daily PM2.5 values, with an R² of 0.89 reported from cross-validation.
• Covers a nine-year time range from 2011 to 2019, enabling longitudinal analysis.
• Provides spatially continuous data at a 6km x 6km resolution over a large region of East Asia.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset's spatial coverage is limited to South Korea, eastern China, and Japan, which may not be suitable for global studies.

Provenance

Source

Harvard University

Collection Method

Generated by applying a random forest algorithm to gap-filled GOCI satellite AOD data and meteorological variables, trained on national PM2.5 monitoring networks.

Time Range

2011 to 2019

Geography

East Asia (South Korea, eastern China, Japan)