Introduction
The regional air quality production of the Copernicus Atmosphere Monitoring Service (CAMS) is based on an ensemble of 11 state-of-the-art numerical air quality models developed in Europe : CHIMERE from INERIS (France), EMEP from MET Norway (Norway), EURAD-IM from Jülich IEK (Germany), LOTOS-EUROS from KNMI and TNO (Netherlands), MATCH from SMHI (Sweden), MOCAGE from METEO-FRANCE (France), SILAM from FMI (Finland), DEHM from AARHUS UNIVERSITY (Denmark), and GEM-AQ from IEP-NRI (Poland), MONARCH from BSC (Spain) and MINNI from ENEA (Italy).
The regional air quality models provide 4-day, daily forecasts of the main atmospheric pollutants concentrations, in the lowest layers of the atmosphere for the European domain (east boundary=25.0° W, west=45.0° E, south=30.0° N, north=72.0°). Their horizontal coverage is 0.1° (approximatively 10 to 20 km), allowing background air pollution levels and medium-range atmospheric composition to be monitored and forecasted. To get insight into your local area, please liaise with your national or local air quality agency.
Alongside forecasts, the models perform daily analyses of pollutants near the surface by assimilating, i.e merging, 1-day old observations with the models. For this purpose, each system of the ensemble rely on their own data assimilation system. Surface observations are split into two sets used in the data assimilation process and for statistical assessments of forecasts and analyses. They are collected on a daily basis from the European Environmental Agency (EEA), and more precisely from its near-real-time (NRT) service for air quality measurements over Europe.
The forecasts and analyses from all partner models are combined in calculating the median value of the individual outputs, which currently gives the best estimate from the ensemble. In that case the median value is designated as the ENSEMBLE output.
Beside this near-real time production, validated and interim reanalyses of previous years are produced with the same ensemble of different models. Reanalyses products provide a revised version of the analyses using more observations, a reanalysis weather modeling and much better quality control (thus the term re-analyses).
Please note that access to the numerical data produced by the production chains of the individual models and the ENSEMBLE output is subject to your prior acceptance of the Licence agreement and registration (read more here for licence and here for registration).
Air quality models and data assimilation system
The eleven air quality models use the meteorological parameters settings coming from the 00:00 UTC operational ECMWF IFS forecast from the day before, the boundary conditions for chemical species coming from the CAMS IFS-TM5 global production, the emissions coming from CAMS emission (for anthropic emissions over Europe and for biomass burning). The domain covers the European area: 25W/30N/45E/72N since 12th June 2019. Before this date, the domain coverage was smaller i.e. 25W/30N/45E/70N.
Table 1: Air quality models (Source: Regional Production, Updated documentation covering all Regional operational systems and the ENSEMBLE (01-2021) document)
Upgrades of the CAMS global forecasting system
Upgrade | Date | Development items |
U0 | Feb. 2019 | Integration of hourly GFAS emissions |
U1.1 | June 2019 |
|
U1.2 | Oct. 2019 | Addition of two new state-of-the-art numerical air quality models to form an 9-member ENSEMBLE:
|
U2 | Feb. 2020 |
|
U3 | Feb. 2020 |
|
U4 | June 2020 |
|
U5 | Mar. 2022 |
|
U6 | June 2022 | Addition of two new state-of-the-art numerical air quality models to form an 11-member ENSEMBLE:
|
U7 | Nov. 2022 | Addition of folmaldehyde (HCHO) and glyoxal (CHOCHO) in the model output species |
U7b | June 2023 |
|
U8 | November 2023 |
|
Data access
Data is available for download from the CAMS Atmosphere Data Store (ADS). CAMS ADS registered users can access the available data interactively through the CAMS European air quality forecast ADS download web interface and/or programmatically using the API as per instructions detailed here.
From 1st July 2020, the access method to the European air quality Web Map Service (WMS) is through a new server at ECMWF which hosts the Copernicus Atmosphere Monitoring Service. Details are available here.
The Web Coverage Service (WCS) was shutdown on 30 June 2020. CAMS European air quality data retrieval for a specific geographical area and/or level is now served through the Atmosphere Data Store.
Data availability (HH:MM)
The CAMS European air quality forecasts is a 3-year rolling archive in the ADS for NRT product, as other/better product become available in the form of reanalyses. For traceability of research results making use of this data, we invite users to keep the files that they have downloaded and used for their research.
The production of the forecast and analysis packages take place at different times of the day : Forecast D0 (0-24h) at 05:50 UTC , D1 (25-48h) 05:55 UTC, D2 (49-72h) 07:30 UTC, D3 (73-96h) 08:00 UTC and 11:00 UTC for Analysis (-24H-1H)).
Ensemble analyses :
Data availability guaranteed by 12:00 UTC
Ensemble forecast :
0-24h forecast data availability guaranteed by 08:00 UTC
25-48h forecast data availability guaranteed by 08:00 UTC
49-72h forecast data availability guaranteed by 10:00 UTC
73-96h forecast data availability guaranteed by 10:00 UTC
It is possible that the data will be available earlier but without guarantee.
Please see the following page for the missing data for each model for the full archive period:
Spatial grid
Horizontal resolution of ENSEMBLE forecasts and analyses is on a 0.1°× 0.1° regular latitude-longitude grid.
The eleven individual models have different native spatial grids (for details see Regional Production, Updated documentation covering all Regional operational systems and the ENSEMBLE (01-2021) document) and they are interpolated on the ensemble common regular grid for each time step of the daily forecasts.
Temporal frequency
The daily ensemble forecast, run from 00 UTC, with a Time resolution of 1 hour and Forecasts period from step 0h to step 96h.
Data format
Data are available in NetCDF format for the individual models outputs, and in NetCDF and Grib-Edition2 format for the ENSEMBLE outputs.
Care should be taken when decoding regional CAMS grib files. The page How to identify fields in European air-quality forecast grib files describes how they should be handled.
Level listings
Vertical levels: Surface, 50m, 100m, 250m, 500m, 750m, 1000m, 2000m, 3000m, 5000m
Pollen data are available only at the surface.
Product listings (including pollen)
Model forecasts are not suitable for clinical trials.
The following species could be considered experimental and without assessment and to be used with caution:
- SIA
- PANs
- NMVOC
- PM10_WF
- Anthropogenic total elementary carbon
- Anthropogenic residential elementary carbon
- NH3
- DUST
- Sea Salt Dry
- Total Organic Matter
The land-sea mask field used to produce the regional air quality data is attached here below, already interpolated to a regular lat/lon grid of 0.1°x0.1°:
name | units |
---|---|
land-sea mask (netCDF4) | (0-1) |
Variable Name | NetCDF Units | Grib Units | Variable name in ADS | Analysis | Forecast | Note |
---|---|---|---|---|---|---|
Alder pollen | grains m-3 | grains m-3 | alder_pollen | × | Forecast available from 1st January to 30th June (implemented in November 2020) | |
Ammonia | µg m-3 | kg m-3 | ammonia | × | × | |
Birch pollen | grains m-3 | grains m-3 | birch_pollen | × | Forecast available from 1st March to 31st July (implemented in November 2015) | |
Carbon monoxide | µg m-3 | kg m-3 | carbon_monoxide | × | × | |
Dust | µg m-3 | kg m-3 | dust | × | × | Dust is fraction in PM10 |
Grass pollen | grains m-3 | grains m-3 | grass_pollen | × | Forecast available from 1st March to 31st August (implemented in November 2016) Grass pollen species are not distinguishable at microscopic analysis, i.e., the grass source term has been parameterized to meet the “general-grass” observations. | |
Mugwort pollen | grains m-3 | grains m-3 | mugwort_pollen | × | Forecast available from 1st June to 31st October (implemented in June 2021) | |
Nitrogen dioxide | µg m-3 | kg m-3 | nitrogen_dioxide | × | × | |
Nitrogen monoxide | µg m-3 | kg m-3 | nitrogen_monoxide | × | × | |
Non-methane volatile compounds (VOCs) | µg m-3 | kg m-3 | non_methane_vocs | × | × | Mass concentration expressed as carbon in air |
Olive pollen | grains m-3 | grains m-3 | olive_pollen | × | Forecast available from 1st January to 30th June (implemented in November 2016)
| |
Ozone | µg m-3 | kg m-3 | ozone | × | × | |
Particulate matter d < 10 µm (PM10) | µg m-3 | kg m-3 | particulate_matter_10um | × | × | |
Particulate matter d < 10 µm, wildfires only (PM10_WF) | µg m-3 | kg m-3 | pm10_wildfires | Forecast available from March 2020 | ||
Particulate matter d < 2.5 µm (PM2.5) | µg m-3 | kg m-3 | particulate_matter_2 | × | × | |
Anthropogenic total elementary carbon | µg m-3 | kg m-3 | pm2.5_anthropogenic_total_carbon | × | × | Total elementary carbon is fraction in PM2.5, Forecast available from June 2021 |
Anthropogenic residential elementary carbon | µg m-3 | kg m-3 | pm2.5_anthropogenic_residential_carbon | × | × | Residential elementary carbon is fraction in PM2.5, Forecast available from June 2021 |
Peroxyacyl nitrates (PANs) | µg m-3 | kg m-3 | peroxyacyl_nitrates | × | × | |
Ragweed pollen | grains m-3 | grains m-3 | ragweed_pollen | × | Forecast available from 1st July to 31st October (implemented in June 2019). In principle, the season starts in midsummer: the 2 main driving parameters are the temperature and the photoperiod. A minimum of daily mean temperature is required and too long photoperiod slows down the phenological development. So, the shortening of photoperiod is used in models as a threshold for the season start (end of July in Eastern and South-eastern Europe). The models will only deliver datasets from 1st July till 31st October.
| |
Secondary inorganic aerosol (SIA) | µg m-3 | kg m-3 | secondary_inorganic_aerosol | × | × | SIA is fraction in PM2.5. Data available from 12 June 2019 |
Sulphur dioxide | µg m-3 | kg m-3 | sulphur_dioxide | × | × | |
Glyoxal | µg m-3 | kg m-3 | glyoxal | × | × | The models DEHM, LOTOS-EUROS, MONARCH and SILAM deliver methylglyoxal as glyoxal for a limited period of time until glyoxal has been added to their chemical mechanisms. |
Formaldehyde | µg m-3 | kg m-3 | formaldehyde | × | × | |
Sea Salt Dry | µg m-3 | kg m-3 | pm10_sea_salt_dry | × | × | Sea salt dry is fraction in PM10 |
Total Organic Matter | µg m-3 | kg m-3 | pm2.5_total_organic_matter | × | × | Total organic matter is fraction in PM2.5 |
In Situ Observations
In-situ observations at surface level are collected from the European Environmental Agency (EEA) on a daily basis. These observations are filtered to keep only measurement sites representative of background air pollution. This is performed by selecting background stations that are classified from 1 to 7 according to the Joly and Peuch classification 2023 version (https://opensource.umr-cnrm.fr/projects/air_quality-classification/wiki/Wiki).
Observations datasets are essential for the Regional air quality Near-Real Time (NRT) products for two reasons:
• For the assimilation by the models producing NRT analyses (to constrain the regional forecast models)
• For the evaluation of model outputs (both forecasts and analyses)
Consequently, special care is taken to deliver 2 independent datasets:
- Station set 1 is used to select stations that will provide observations for the NRT analysis of individual models.
- Station set 2 is used to select stations that will provide observations for analysis evaluation.
This split into 2 sets for assimilation and analysis evaluation only applies to O3, NO2, PM2.5 and PM10. Special cases are CO and SO2, for which all stations that are classified are used by models in the assimilation. For these two species, the evaluation is made also against those measurements that have been assimilated. The stations defined for verification are used for quarterly assessments/reports and visualisation.
For the evaluation of the forecast, all stations of those two sets are used, without split.
Once a year, in March, these sets of stations are updated.
Validation reports
Validation Reports for the CAMS reanalysis can be found on the CAMS Quality Assurance website.
Individual and ENSEMBLE model evaluations are available at station level and aggregated by country through an interactive visualization platform for Forecast and Analysis.
Known issues
- Ozone CHIMERE data are not available for timestep 96 for 2019-07-07 and timesteps 69 to 96 for 2019-07-12.
- Data gap for EMEP forecast data on the following dates:
- 31-03-2019
- 12-06-2019
- 17/18-07-2019
- 25/26-08-2019
Data gap on 22 and 23 April 2018,
Guidelines
How to acknowledge, cite and refer to the data
All users of data uploaded on the Atmosphere Data Store (ADS) must provide clear and visible attribution to the Copernicus programme and are asked to cite and reference the dataset provider.
(1) Acknowledge according to the licence to use Copernicus Products.
(2) Cite each dataset used:
(3) Throughout the content of your publication, the dataset used is referred to as Author (YYYY) e.g. for Ensemble data: METEO-FRANCE et. al (2020)
References
- A regional air quality forecasting system over Europe: the MACC-II daily ensemble production.(11 Mar 2015)
- Regional Production, Updated documentation covering all Regional operational systems and the ENSEMBLE (01-2021) document)