Q2XAFS2024 Data Formats, MetaData

This page contains information from the Working Group on Data Formats, Metadata, and Databases.

Presentation (comments welcome)

Q2XAFS2024_data.pdf

Main Recommendations:

  • use XDI for a single XAS spectrum: a plaintext file with clear column labels and organized metadata
  • NeXuS (using HDF5) is the preferred choice for multi-spectral data files.
  • For communicating XAS data, the main data to present is “edge-step normalized” mu(E).
  • Required metadata are:
    1. Symbol of absorbing element.
    2. Absorption edge of absorbing element (“K”, “L3”, etc).
  • Strongly encourage metadata include:
    1. d-spacing used (preferred) or crystal cut (at least) for monochromator used to select X-ray energy, so that energies can be recalibrated with high accuracy.
    2. name or abbreviation of laboratory, facility, and/or beamline used.
    3. date of data collection.
    4. name of person uploading or collecting this data.
    5. mode of data collection (transmission, fluorescence, etc)

Example NXxas data and XDI files:

Examples of NXxas data and XDI files, converted from existing data can be found at

NXxas_examples.zip

This file contains 3 example data sets:

  1. Fe_XDIFiles: 3 XDI files (all Fe K edge, transmission XAFS) are put into a single nexus file.
  2. KEK_PF: 1 data file from Photon Factory, KEK. Fluorescence XAFS, that was recorded as a function of angle.
  3. MultiElementFluor: 12 data files (XDI-ish) for V K-edge fluorescence XAFS using a 4-element detector. The original files have columns saved for “V Ka” and a “Dead-time Correction Factor” for MCAs 1 through 4. The script sums the deadtime-corrected chanbels to make “ifluor”, and puts these all into V_XANES_nexus.h5

Suggested Discussion Topics

  • How can we encourage usage of XDI and NXxas?
  • How can we help beamlines and data producers to convert data for uploading to Journals, Data Portals, etc?
  • Can the data you take at your favorite beamline(s) be converted to XDI and/or NeXuS XAS format?
  • Is sharing “edge-step normalized mu(E)” too big of a burden for producing or consuming XAS data?
  • Are there other pieces of metadata that should always be required or strongly encouraged?
  • Should on-line databases be tied to beamlines, facilities, and institutions? Is it OK to copy data from existing facility-specific databases to xaslib.xrayabsorption.org?

Some resources: