b'Metadata standard for MTTechnical noteadopted promptly. Otherwise MT data will remain in the drawer (Kelbert etal., 2018) and its full potential will not be realised.High-performance data requirements suggest that a preferred solution for an MT data standard should:Be self-describing, keeping metadata with dataHandle big data flexiblyPermit fast input/outputBe platform independent and open formatHandle heterogeneous data.Additionally, for ease of implementation within workflows, the solution should:Provide interfaces and utilities in a range of software languages to accommodate users differing work environmentsHave an established and wide user baseOffer scope for user support in their implementation and use.Figure 1.Examples of metadata for Level 0 and Level 1A and 1B MT time-seriesTable 1 sets out some metadata and data options that have data been developed which may offer solutions to the Australian MT community until an internationally-agreed standard is Currently, there is no widely adopted international standard fordeveloped.MT data or metadata. Notwithstanding, the renaissance of the MT method in Australia and internationally, the expected continuedThe EMERALD scheme for MT data has been developed by acquisition of significant amounts of MT data for the foreseeablethe GFZ German Research Centre for Geosciences (Ritter future, and the growing interest in the availability of FAIR MTetal., 2015). It stores MT data and metadata in file pairsa data, suggest that a solution, even an interim solution, should beusually binary file containing the MT time series, spectra and Table 1.Metadata and data format options for publicly accessible MT data. Rows shown in bold text are the metadata and time-series formats recommended in this article.Reference Summary ConsProsMetadataEMERALD Ritter etal. (2015) ASCII XML metadata files Mature Human readable Metadata files separate to data files XML a dated formatIRIS Peacock and Frassetto (2020) JSON metadata files usable as self- Self-describing JSON an efficientImmature formatdescribing headers in binary dataformatfilesData formatsASCII Wyborn etal. (2020) Mature format Human readable Space inefficient Inefficient I/O Not self-describingSEG EDI Wight (1988) ASCII-based SEG standard format forMature format Human readable Space inefficient Inefficient I/O MT and other EM data Not self-describingEMERALD Ritter etal. (2015) Stores MT data and metadata in fileMT-specific Mature formatpairs; related utilities are in C, C and FORTRAN, with interface functions in Matlab and PowershellHDF5 https://www.hdfgroup.org/ High-performance, big data-capable,Self-describing Mature format WidelyNot MT-specificflexible, n-dimensional scientific dataused to archive large datasetsformatnetCDF4 https://www.unidata.ucar.edu/ HDF5-based data format developedSelf-describing Mature formatNot MT-specificsoftware/netcdf/ Ip etal. (2019) for atmospheric data; programmingUtilities and toolkits widely interfaces for C, Java, Fortran,available Extensive user communityPython, C, IDL, MATLAB, R, Ruby, PerlMTH5 https://mth5.readthedocs.ioHDF5-based data format adapted forMT-specific Self-describing Immature formatPeacock (2018) MT time seriesGeoHDF Wyborn etal. (2020) netCDF4-based generalised containernetCDF4-based Geophysics-specificUnder developmentformat for geophysical time-seriesSelf-describingdataASDF http://seismic-data.org/ KrischerHDF5-based data format adapted forMature format Adaptable to MT UsedSeismic-specific No efficiencies etal. (2016) Duan and Kirkbyseismic data in MTpy Efficiencies where bothwhere seismic data is not (2018) seismic and MT data are managed managedDECEMBER 2021 PREVIEW 62'