AIR-01 London Dataset

Dataset information: The Air-01 London dataset is a curated dataset derived from raw public data; which contains outdoor air pollution, meteorological, and land use time series data collected from 1st January 2019 to 31st December 2022. Dataset dimensions: [1461 observations x 74 variables] Date of creation: 15th May 2024 Year of publication: 2026 Table 1. Data specifications DATA SPECIFICATIONS Dataset File: Format: CSV File name: Air-01-London_data.csv Structure: Rows: 1,461 daily time points (dates, format: dd/mm/yyyy) Columns: 74 variables (measurements) Variables: 17 measurements of nitrogen dioxide (NO2) at different air quality monitoring sites in London 6 measurements of particulate matter <2.5μm (PM2.5) at different air quality monitoring sites in London 9 measurements of particulate matter <10μm (PM10) at different air quality monitoring sites in London 4 measurements of meteorological data (air_temp, ws, wd, RH) in London 19 averaged measurements of NDVI (green space remote sensing indicator) 19 averaged measurements of NDBI (built-up/construction remote sensing indicator) Percentage of missing data in dataset: 0% (Original missing values were imputed, see details of missing data % and imputation methodology in B. Galindo-Prieto, A. Papanicolaou & I.S. Mudway, 2026) Table 2. Information of data files DATA FILE CONTENTS USED IN DATA ANALYSIS OF REFERENCE PAPER MISSING DATA METHOD DESCRIBED IN REFERENCE PAPER Air-01-London_data.csv This is the subset used in the data analysis of the publication (reference data paper to cite) "B. Galindo-Prieto, A. Papanicolaou & I.S. Mudway, Partial least squares regression models for outdoor air pollutant forecasting, J. of Chemometrics, 2026". It contains time series air pollution data (NO2, PM2.5, PM10), meteorological data, and land use satellite data. YES YES Air-01-London_O3_ub_data.csv This contains additional data (also part of the Air-01 London dataset) consisting of time series data for the ozone (O3) measurements from three air quality urban background sites. The imputation methodology is the same than in the other missing data imputation described in the dataset reference paper. NO YES Air-01-London_O3_rs_data.csv This contains additional data (also part of the Air-01 London dataset) consisting of time series data of ozone (O3) from one air quality roadside site. The imputation could only be done partially, leaving some NA values inside the O3 roadside data; this was because only one roadside site collected ozone data, which limited the missing imputation in our case. Besides, the imputation methodology is different from the other datasets, please read the description of the missing data imputation of this dataset in the reference data paper. NO YES Table 3. Raw data sources and further specifications SOURCES OF RAW DATA FOR THE CURATED DATASET GENERATION Air pollution time series raw data: Daily time series data of NO2, PM2.5, and PM10 outdoor concentrations collected from monitoring sites located in central London (United Kingdom). Further details related to missing data percentages, type of imputation of missing values, applied filters, etc. are found in B. Galindo-Prieto, A. Papanicolaou & I.S. Mudway, 2026. Data were downloaded from UK air quality networks: London Air Quality Network (LAQN) https://londonair.org.uk/ Automatic Urban and Rural Network (AURN) https://uk-air.defra.gov.uk/networks/ Maximum missing data allowed per monitoring site: 25% Percentage of total imputed data: 8.8% Meteorological raw data: Daily time series measurements representative of London's meteorology (including air temperature, wind speed, wind direction, and relative humidity). Further details related to missing data percentages, type of imputation of missing values, applied filters, etc. are found in in B. Galindo-Prieto, A. Papanicolaou & I.S. Mudway, 2026. Data were downloaded from the NOAA Integrated Surface Database (ISD) https://registry.opendata.aws/noaa-isd/ Percentage of total imputed data: 0.07% NDVI and NDBI satellite raw data: Daily time series images of the regions of interest in London. Further details related to the remote sensing spectral indexes, missing data treatment, applied filters, etc. are found in in B. Galindo-Prieto, A. Papanicolaou & I.S. Mudway, 2026. Resolution: 10 m Bounding box / Buffer: 750 m Data were downloaded from Sentinel Hub https://www.sentinel-hub.com/ Dictionary of variables included in the dataset: See file "Air-01-London_Dictionary.pdf". Citation and reference information: Dataset created for the data analysis of the article: B. Galindo-Prieto, A. Papanicolaou & I. S. Mudway, Partial least squares regression models for outdoor air pollutant forecasting, Journal of Chemometrics, 2026. This article is the main reference for this dataset (please cite it if you use the Air-01 London dataset). The article describes the dataset and its curation process, as well as it shows its application for outdoor air pollutant spatio-temporal prediction. Note: the CSV data files in this Zenodo repository and in the supplementary materials of the main reference are identical (same version and same format). Contact details for questions: Dr Beatriz Galindo-Prieto ([e-mail])