Summary

Air pollution is the leading environmental threat to global public health [1]. As a result, accurately characterizing air quality levels is a priority. The city of Barcelona (Spain) faces a persistent NO₂ pollution problem [2,3], mainly driven by high vehicle density [4], a compact urban morphology, and its coastal location. 

We present a database consisting of daily and annual NO₂ surface concentrations with associated uncertainty estimates from 2019 to 2024 in Barcelona city, as summarized in Table 1. The database is obtained by combining high-resolution outputs from the CALIOPE-Urban air quality model [5] with multiple observational datasets using a data-fusion method based on Universal Kriging (UK) [6].  Exceedance probability maps are also provided by combining concentration levels with their associated uncertainty. These maps are based on daily and annual NO₂ thresholds from European Air Quality Directives 2008/50/EC [7] and 2024/2881 [8], as well as the 2021 WHO guidelines [1], and serve as a valuable tool for policymakers and regulatory assessments by offering information directly linked to specific NO₂ limits.

The first dataset corresponds to annual averages of the variables in high-spatial resolution maps, at a grid of 25m x 25m. The NO₂ annual thresholds for computing the probability maps are (i) 10 µg/m³, the WHO recommended limit [1], (ii) 20  µg/m³, the European annual limit value set by the current air quality Directive 2024/2881 [8], and (iii) 40  µg/m³, the European annual limit value set by the air quality Directive 2008/50/c [7] (the one in force at the period of the database). 

The second dataset provides daily and annual averages of the variables aggregated at the census tract level. The exceedance probability maps are calculated using the NO₂ levels and uncertainty already expressed in the census tracts. The annual limits are the same as above, while the NO₂ daily thresholds for computing the probability maps are (i) 25 µg/m³, the WHO-recommended limit [1], and (ii) 50  µg/m³, the European daily limit value set by the air quality Directive 2024/2881 [7]. 

For more information on the methodology, see reference [6]. For the code repository, refer to [12]. For more details about this database, refer to [13].

Table 1: Scheme of the two datasets provided, showing air quality variables, temporal and spatial resolutions, formats, and
temporal coverage.

Structure

The Zenodo repository is organized based on the structure presented in Table 1. Each dataset is a .zip file, composed of folders based on the different formats.

Dataset 1

This dataset is composed of annual values in high-spatial resolution (25m x 25m). The zip file is Dataset1.zip.

1) SHP format

The folder here is named Dataset1_SHP. The files are named as shp_reso_$variable$_annualmean_$year$.shp, where $variable$ can be NO2, sdrel or exc_together; and $year$ is a number from 2019 to 2024. In addition to the .shp extension, the corresponding .shx, .prj, and .dbf files are also present.

The files are processed using the R package sf [9], and each contains approximately 250,000 features as geometry type MULTIPOLYGON. The projection is EPSG:4326, using the geodetic coordinate reference system WGS 84. In addition to the geometry, the attribute fields vary depending on the variable:

• For the NO2 variable, the field no2 represents the NO₂ annual mean concentration, with units of µg/m³.

• For sdrel, the field sd_rel indicates the relative uncertainty of the NO₂ annual mean, expressed as a %.

• For exc_together, the fields exc_40, exc_20, and exc_10 correspond to the probability of exceeding (expressed as % probability):

  • 40 µg/m³ (the annual limit value set by the European Air Quality Directive 2008/50/EC [7]),

  • 20 µg/m³ (the annual limit value set by the European Air Quality Directive 2024/2881 [8]), and

  • 10 µg/m³ (the annual limit value recommended by the WHO 2021 guidelines [1]).

These exceedance probabilities are calculated by combining the annual mean NO₂ values with their associated uncertainties and selecting a NO₂ threshold.

2) Raster format

The folder here is named Dataset1_raster. The files are named as raster_together_annualmean_$year$.tif, where $year$ is a number from 2019 to 2024.

The files are processed using the R packages raster [10] and terra [11]. Each file contains approximately 600,000 cells, as a mask is applied to the city of Barcelona. Of these, approximately 250,000 cells have valid (non-NA) values. The projection is EPSG:4326, using the geodetic coordinate reference system WGS 84. The rasters contain five bands, with the following names:

• no2 represents the NO₂ annual mean concentration, with units of µg/m³.

• sd rel indicates the relative uncertainty of the NO₂ annual mean, expressed as a %.

• The bands exc40, exc20, and exc10 correspond to the probability of exceeding (expressed as a % probability):

  • 40 µg/m³ (the annual limit value set by the European Air Quality Directive 2008/50/EC [7]),

  • 20 µg/m³ (the annual limit value set by the European Air Quality Directive 2024/2881 [8]), and

  • 10 µg/m³ (the annual limit value recommended by the WHO 2021 guidelines [1]).

These exceedance probabilities are calculated by combining the annual mean NO₂ values with their associated uncertainties and selecting a NO₂ threshold.

Dataset 2 (A)

This dataset is composed of annual values in the census track aggregation. The zip file is Dataset2A.zip.

1) SHP format

The folder here is named Dataset2A_SHP. The files are named as shp_census_annualmean_$year$.shp, where $year$ is a number from 2019 to 2024. In addition to the .shp extension, the corresponding .shx, .prj, and .dbf files are also present.

The files are processed using the R package sf [9], and each contains 1,068 features as geometry type MULTIPOLYGON. The projection is EPSG:4326, using the geodetic coordinate reference system WGS 84. In addition to the geometry, the files contain a total of 7 fields:

• OBJECTID, a number identification from 1 to 1,068.
• ID_census, the official census tract identifier in Spain, based on the classification by the Instituto Nacional de Estadística (INE).

• no2 represents the NO₂ annual mean concentration, with units of µg/m³.

• sd_rel indicates the relative uncertainty of the NO₂ annual mean, expressed as a %.

• Fields exc_40, exc_20, and exc_10 correspond to the probability of exceeding (expressed as a % probability):

  • 40 µg/m³ (the annual limit value set by the European Air Quality Directive 2008/50/EC [7]),

  • 20 µg/m³ (the annual limit value set by the European Air Quality Directive 2024/2881 [8]), and

  • 10 µg/m³ (the annual limit value recommended by the WHO 2021 guidelines [1]).

These exceedance probabilities are calculated by combining the annual mean NO₂ values with their associated uncertainties and selecting a NO₂ threshold.

2) CSV format

The folder here is named Dataset2A_CSV. The files are named as csv_census_annualmean_$year$.csv, where $year$ is a number from 2019 to 2024. The files contain 1,068 rows and, in addition to a column called X (which represents the row numbers), they include the same information as the SHP files mentioned above, but without the geometry column. To associate each row with its corresponding spatial area, please refer to the geometry in the SHP files from the same dataset.

Dataset 2 (B)

This dataset is composed of daily values in the census track aggregation. The zip file is Dataset2B.zip.

1) SHP format

The folder here is named Dataset2B_SHP. The files are named as census_daily_$year$_$variable$.shp, where $variable$ can be no2, sd_rel, exc_25, or exc_50; and $year$ is a number from 2019 to 2024. In addition to the .shp extension, the corresponding .shx, .prj, and .dbf files are also present.

The files are processed using the R package sf [9], and each contains 1,068 features as geometry type MULTIPOLYGON. The projection is EPSG:4326, using the geodetic coordinate reference system WGS 84. In addition to the geometry, the files contain a total of 367 fields (368 for 2020):

• OBJECTID, a number identification from 1 to 1,068.
• ID_census, the official census tract identifier in Spain, based on the classification by the Instituto Nacional de Estadística (INE).

• From $year$0101 to $year$1231 (i.e., 20190101, 20190102, ... 20191231). The meaning of the dates columns varies depending on the $variable$ specified in the file:

  • If $variable$ is no2, they represent the daily mean NO₂ value for that day, with units of µg/m³.
  • If $variable$ is sd_rel, they represent the relative uncertainty of the daily mean NO₂ value for that day, expressed as a %.
  • If $variable$e is exc_25, they correspond to the probability of exceeding 25 µg/m³, the daily limit value recommended by the WHO 2021 guidelines [1], expressed as a % probability.
  • If $variable$ is exc_50, they correspond to the probability of exceeding 50 µg/m³, the daily limit value set by the European Air Quality Directive 2024/2881 [8], expressed as a % probability.

These exceedance probabilities are calculated by combining the annual mean NO₂ values with their associated uncertainties and selecting a NO₂ threshold.

2) CSV format

The folder here is named Dataset2B_CSV. The files are named as csv_daily_$year$_$variable$.csv, where $variable$ can be no2, sd_rel, exc_25, or exc_50; and $year$ is a number from 2019 to 2024. The files contain 1,068 rows, and they include the same information as the SHP files mentioned above, depending on the $variable$, but without the geometry column. To associate each row with its corresponding spatial area, please refer to the geometry in the SHP files from the same dataset. Note that the date columns start with X, from X$year$0101 to X$year$1231.

uncertAIR platform

In addition, the datasets are accessible through the uncertAIR platform (https://earth.bsc.es/shiny/uncertAIR/).  uncertAIR is a research project funded by the Barcelona City Council. The project emphasized the importance of addressing uncertainty in air quality simulations and aimed to create an accessible tool for data dissemination and visualization.

Legal Notice

Note: for more information on the methodology, see reference [6]. For the code repository, refer to [12]. For more details about this database, refer to [13].

How should I use the information and products available in this repository or the uncertAIR platform?

1. Data, images, and other products from this repository or the uncertAIR platform available on this server may be used solely for research and education purposes. Products may not be used for commercial purposes.
2. In case of establishing links to the contents of this website, kindly inform us via email alvaro.criado@bsc.es
3. The authors cannot guarantee that the data are correct in all circumstances. Neither do accepts any liability whatsoever for any error or omission in the data, or any loss or damage arising from its use.
4. Numerical data must not be supplied as a whole or in part to any third party without prior authorization.
5. Articles, papers, or written scientific works of any form, that are based in whole or in part on data, images or other products supplied by this database should contain an acknowledgement, giving credit to the Center every time data/images/products are used. Please use the reference [12] as well as the citation to this repository.