The objectives of this use case are
- to implement a reproducible method to predict air quality indicators using open source software
- to create annual, monthly, and daily maps of NO2, O3, PM10, PM2.5
This repository stores code for the entire processing chain, including - the retrieval and preprocessing of measurement station data and raster covariates - the prediction of air quality indicators using the Regression Interpolation Merging Mapping (RIMM) method, introduced by Horalek et al. (2023)
For more details on features and progress, visit the task list.
Hourly data from about 7,000 EEA measurement stations throughout the EU serves as basis for interpolation. Supplementary variables such as population density, land cover, and climatology are further inputs for the prediction method. The final maps can be viewed and explored in the OEMC app or downloaded directly via Zenodo.
| Step | File | Description |
|---|---|---|
| 1 | ACS_CAMS_access.R CCS_ERA5_access.R download_nc_job.R rename_nc.R |
Request hourly ERA5 weather and CAMS pollution data. Copy to URLs from the web interface and store in .txt file to iterate over for downloading. Rename files according to metadata. |
| 2 | EEA_stations.qmd |
Create a spatial dataset with all AQ stations and supplement them with static covariates (elevation, population, land cover). |
| 3 | EEA_AQ_data_access.qmd EEA_AQ_data_access_all_countries.R |
Download & pre-process hourly AQ measurements. This includes reading, filtering, and joining up to 5 pollutant time series per station for 2015-2023. |
| 4 | xarray_extract_station_SSR.ipynb EEA_PM25_gapfilling_all_countries.qmd |
Extract hourly Surface Solar Radiation before gapfilling PM2.5 (only where PM10 is measured) using linear regression. |
| 5 | xarray_dask_rel_humidity.ipynb xarray_dask_ws_wd.ipynb |
Process ERA5 wind vectors and temperature data to wind speed & direction and relative humidity. |
| 6 | EEA_AQ_data_temporal_aggregation.qmd xarray_temp_aggregate.ipynb xarray_temp_aggregate_rolling.ipynb |
Aggregate AQ measurements and CAMS/ERA5 hourly data to annual, monthly, daily means/quantiles. |
| 7 | s5p_l3_access.ipynb |
Access Sentinel-5P TROPOMI Level 3 NO2 data and aggregate to coarser temporal resolutions. |
| 8 | AQ_Demo.qmd AQ_interpolation_loop_palma_[daily/monthly/annual].R |
Interpolate AQ data using environmental and socio-economic covariates. Weight and combine individual outputs for rural background, urban background, and urban traffic stations. Return cloud-optimized GeoTiff with prediction and error variables. |
| 9 | AQ_maps_as_Zarr.ipynb |
Convert map outputs to Zarr stores. |
- Horálek, J., Vlasáková, L., Schreiberová, M., Marková, J., Schneider, P., Kurfürst, P., Tognet, F., Schovánková, J., Vlček, O., Damašková, D., 2023. European air quality maps for 2020. PM10, PM2.5, Ozone, NO2, NOx and Benzo(a)pyrene spatial estimates and their uncertainties. (No. ETC HE Report 2022/12).