Published January 26, 2026 | Version v1
Computational notebook Open

PD-SID and CD calculation for PMF factor profiles from Brezins et al. (2026)

Authors/Creators

  • 1. EDMO icon Aix-Marseille University

Contributors

  • 1. ROR icon Aix-Marseille Université
  • 2. EDMO icon Aix-Marseille University
  • 3. AtmoSud
  • 4. EDMO icon Paul Scherrer Institute PSI
  • 5. ROR icon Université Grenoble Alpes
  • 6. ROR icon Institut des Géosciences de l'Environnement
  • 7. IRD
  • 8. CNRS Senior scientist - Aix Marseille University

Description

This Zenodo repository contains a Jupyter Notebook tool designed to evaluate Positive Matrix Factorization (PMF) profiles dissimilarity using Cosine Distance (CD), Pearson Distance (PD), and Standardized Identity Distance (SID), as defined in Brezins et al. (2026, in prep.). The notebook also allows visualization of PD-SID and CD plots, including thresholds for homogeneous chemical profiles.

The tool can be applied to factors obtained from either Rolling or classical PMF factors. Factors may be compared pairwise, including all pairwise comparisons with mean values and standard deviations, or evaluated against a selected reference factor profile.

Two matrices containing multiple versions of a Xact Biomass Burning factor profile obtained at the MRS-LCP site in 2023, from Brezins et al. (2026, in prep.), through different PMF analyses (static and Rolling), are provided to illustrate how to use the script.

The author acknowledges the use of the ChatGPT AI tool for Python script correction.

Files

Mx_Rolling_Biomass_Burning_YEAR.csv

Files (2.3 MB)

Name Size Download all
md5:8aba0bcb6e624c334cac18dfcde6883e
2.2 MB Preview Download
md5:eebe38d0b4ba50fde2aea7eb8112d5ed
15.3 kB Preview Download
md5:c5e5c281b8200a846edda458c169a176
62.2 kB Preview Download

Additional details

Funding

Agence Nationale de la Recherche
SHIPAIR ANR-21-CE22-0015

Dates

Available
2026-01-26
Submission

Software

Programming language
Python