National-scale NMF source apportionment of PM2.5 across the Czech Republic, 2021-2025: data, model outputs and analysis code
Year: 2026
Type: dataset
Abstract: Data, model outputs and the complete analysis code underlying: Seibert, R., Volná, V., Hladký, D., Krejčí, B. (2026). National-scale NMF source apportionment reveals a west-to-east asymmetry in PM2.5 concentrations associated with transboundary transport. Environmental Pollution. Non-negative matrix factorization (ns-NMF, 9 factors) source apportionment of chemically speciated filter samples from 33 air-quality monitoring sites across the Czech Republic, collected during annual monitoring campaigns in 2021-2025 (4384 samples, 57 variables), together with 96 h HYSPLIT/ERA5 back-trajectories and the directional-contrast analysis of transboundary transport at eight border stations. Contents README.txt – full description of every file, the data treatment, the model settings and the changes since version 1 nmf_input.xlsx – the exact sample-by-variable matrix used for the factorisation site_campaign_metadata.csv, pm25_pm10_ratios.xlsx, pm25_pm10_ratios_applied.csv – site, campaign and PM2.5/PM10-ratio metadata nmf_profiles_abs.xlsx, nmf_profiles_perc.xlsx, nmf_series.xlsx, nmf_r2_per_species.xlsx – the factor profiles, the per-sample factor contributions and the reconstruction quality scripts.zip – all analysis code: the production NMF model, the trajectory computation, every derived analysis, and every figure and table of the article and its supplementary material, with the execution order (README.md), the random seeds (SEEDS.txt) and the R and package versions (sessionInfo.txt) bootstrap_confidence_intervals.xlsx, R01_*.xlsx … R27_*.csv – the analysis workbooks behind every table and figure, including the bootstrap confidence intervals and every sensitivity analysis R02_directional_bootstrap_draws.zip – the 2000 bootstrap replicates behind every directional-contrast interval, one row per contrast and one column per replicate trajectories_96h.zip, trajectories_README.txt – the 96 h HYSPLIT back-trajectories and their configuration The PM2.5 column carries the mass rescaled to a PM2.5 basis at the 12 PM10-sampled sites, using site- and season-specific PM2.5/PM10 ratios; the individual species concentrations remain PM10 concentrations at those sites. The factorisation was fitted to the raw ng m-3 matrix, and PM2.5 mass was not part of it. The data treatment, the model settings and the limits of the deposit are described in full in README.txt.
Institution: Czech Hydrometeorological Institute
Language: English
Subfield: Atmospheric Science
Field: Earth and Planetary Sciences
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7204193682