A global land daily 10-km-resolution surface ozone dataset from 2013-2022

Area of Science:

• Environmental Science
• Atmospheric Chemistry
• Data Science

Background:

• Global ozone pollution is rising, but current datasets lack the temporal resolution needed for comprehensive health impact assessment.
• Understanding ozone's health effects requires more granular data than monthly or annual averages provide.

Purpose of the Study:

• To develop a high-resolution global surface ozone dataset (0.1° daily maximum 8-hour average) for the period 2013-2022.
• To improve the understanding of global ozone pollution and its associated health risks.

Main Methods:

• Constructed an efficient Light Gradient Boosting Machine model integrating diverse data sources: surface ozone observations, satellite products, model simulations, reanalysis data, and emission inventories.
• Implemented rigorous quality control for surface ozone observations, particularly from China and Europe, while accounting for temporal and spatial data characteristics.
• Utilized spatial-based 10-fold cross-validation to assess dataset accuracy.

Main Results:

• The developed dataset achieved a coefficient of determination (R²) between 0.79-0.88 and a root mean square error (RMSE) of 11.32-13.26 μg m⁻³.
• The dataset's accuracy is comparable to current state-of-the-art surface ozone datasets.
• The daily temporal resolution provides significantly more granular information than previous datasets.

Conclusions:

• The new daily global surface ozone dataset offers critical support for air quality management strategies.
• This high-resolution dataset is vital for advancing research on the health impacts of ozone pollution.
• Enhanced data granularity facilitates a more precise assessment of ozone-related health issues.