Hourly and daily gridded air quality surface concentrations derived from in situ low-cost sensor measurements over Europe from 2018 to 2024

This dataset provides surface-level particulate matter (PM2.5 and PM10) concentrations derived from in situ low-cost sensor (LCS) measurements, produced by the European Centre for Medium-Range Weather Forecasts (ECMWF) in the context of the Horizon Europe All Data for Green Deal project. This dataset differs from existing air quality datasets by combining crowdsourced observations with a robust correction framework, producing high-resolution, reference-aligned gridded products. It is particularly useful for researchers, public health analysts, and citizen science initiatives seeking timely and spatially detailed information on air pollution.
The dataset spans the years 2018 to 2024, with both hourly and daily temporal resolution, on a gridded spatial domain covering Europe at approximately 0.01° horizontal resolution. The main dimensions of the dataset are time, latitude, longitude, and pollutant species. Uncertainty estimates and validation results are included in the dataset documentation (available on the project’s website and deliverables).
Observations from the Sensor.Community project were combined with meteorological reanalysis data from the fifth generation of the European Centre for Medium-Range Weather Forecasts (ECMWF) atmospheric reanalyses (ERA5) and air quality analysis fields from the Copernicus Atmosphere Monitoring Service (CAMS) European model. To improve the reliability of the raw LCS measurements, a machine learning correction model was trained using meteorological and sensor data as input features, with quality-controlled reference observations from the European Environment Agency (EEA) as the target. This approach enhances consistency with regulatory-grade measurements.
The corrected values were spatially interpolated onto a regular latitude–longitude grid using two approaches:

Ordinary Kriging over the full European domain using only corrected LCS data; and
Universal Kriging over central Europe, integrating CAMS model outputs as covariates.

Any new versions of the dataset will be selectable as a variable as the data generation pipeline is modified or bug fixes are applied. Bug fixes will be released using a minor revision (i.e. v1.1), while major changes will be released as major revisions (i.e. v2.0). Bug fixes will be published to the Known Issues area on the Documentation.