A Standard for Units in Heliophysics

A Standard for Units in Heliophysics Bob Weigel, Baptiste Cecconi, Brian Thomas, Doug Lindholm, Andriy Koval, Jeremy Faden, Lee Bargatze, Bobby Candey, and Shing Fung We describe an effort by representatives from the metadata standards groups of HAPI, SPASE, and ISTP Metadata Guidelines to develop a recommendation for the representation of scientific units in metadata. The recommendation is “Some metadata models in Heliophysics do not have a constraint on unit strings (that is, a standard has not been chosen). Our recommendation is that if a standard is used, it should be VOUnits.” 10.5281/zenodo.17353237 ISTP Metadata Guidelines - Part of the International Solar-Terrestrial Program that began in 1980s, guidelines developed in 90s in parallel with CDF file format. SPASE - Space Physics Search and Extract. Began ~2004. More formal metadata model than ISTP and community developed. HAPI - Heliophysics API. Began in 2016. An API and metadata standard for serving time series data via an API. Metadata model is not domain specific - allows linking to domain-specific metadata. Background In the metadata standard for the HAPI API, we have units and unitsSchema attributes; the optional unitsSchema can be used by units-aware software to interpret units. We noted that two Heliophysics metadata standards, ISTP and SPASE, do not have constraints on unit strings (UNITS and Units). However, ISTP has SI_CONVERSION and SPASE has UnitsConversion, which allow automatic unit conversion. Primary questions: 1. Is SI_CONVERSION/UnitsConversion enough? 2. If not, what standard should we use for unit strings? Motivation Question: 1. Is SI_CONVERSION/UnitsConversion enough? Examples of SI_CONVERSION/UnitsConversion: UNITS = Re SI_CONVERSION = 6371200 > m UNITS = eV SI_CONVERSION = 1.602e-19 > J UNITS = eV SI_CONVERSION = 1.60217646E-19 > J UNITS = deg SI_CONVERSION = 0.01745 > rad UNITS = deg SI_CONVERSION = 0.0174532925 > rad UNITS = deg SI_CONVERSION = 1.0 > degree UNITS = deg SI_CONVERSION = 1.0 > degrees UNITS = deg SI_CONVERSION = 1.0 > (degree) Advantages: ●Used in Heliophysics community for many years ●Avoids requirement that data providers use a standard style for unit strings (providers have strong opinions about this) Disadvantages: ●Does not seemed to be used outside of Heliophysics ●Lack of software that uses ●Uneven use by (especially older) missions ●Some inconsistency in implementation across missions, e.g., “degree”, “degrees”, and “(degree)” ●Additional standards needed (e.g., precision on numbers, what terms are valid, e.g., degree or degrees or deg). Current standard states only “... the factor that the variable must be multiplied by in order to turn it to generic SI units” Question: 1. Is SI_CONVERSION/UnitsConversion enough? It could be enough with additional constraints beyond existing specification “... the factor that the variable must be multiplied by in order to turn it to generic SI units”. However, much of the needed addition will overlap with work that has already been done in the development of standards for unit strings. Conclusion: We want to allow data providers to express unit strings in a standard form but still keep the keep the SI_CONVERSION/UnitsConversion option. (Which will be useful for validation.) General agreement that Heliophysics should have a recommendation “... if a standard for unit strings is used it should be X”. Question: 1. Is SI_CONVERSION/UnitsConversion enough? Studied many standards and associated software libraries, when available: NIST SP 811, VOUnits, UDUNITS2, EDS-RFC, QUDT, UCUM, MMS, Cluster, SMDX, …. Also, met with the CODATA working group DRUM (Digital Representation of Units of Measurement) team. Question: 2. If not, what standard for unit strings? Much discussion on why machine readable is important, why “SI strings” is ambiguous, and many other nuances. Historically, what was placed as values for unit attributes in NASA missions has included non-unit information such as documentation, labels, and quantities, e.g., 1/cc (qual only if scan=2=SW) T/F 1/N SUM |B|, nT V-dc Sometimes this was done to enhance axis labels on plots - additional information not in variable name or label could be shown by putting it in units. Question: 2. If not, what standard for unit strings? UDUNITS-2 and VOUnits used on both boundaries of Heliophysics (Earth Science on inner boundary and Astronomy on outer). Much debate on this. Flexibility of UDUNITS-2 syntax was attractive given many existing unit strings follow one of the allowed representations (e.g., m^2, m**2 allowed as is N-m, N.m, and N*M.) Question: 2. If not, what standard for unit strings?