13A.2 The ARM implementation of data object design

Thursday, 27 January 2011: 1:45 PM
606 (Washington State Convention Center)
Matt C. Macduff, PNNL, Richland, WA; and B. Ermold, S. Beus, and C. Sivaraman

The Department of Energy's Atmospheric Radiation Measurement (ARM) program has been collecting and storing netcdf data since 1993. For many users of the data the netcdf header is the first step in making use of the data. It has been a long-standing challenge to maintain stability and consistency of structure across time and across many different datasets. The lack of consistent structure requires painful rediscoveries of a datasets history by each user. In 2006, the ARM program established a definition of a Data Object Design (DOD) as a versioned description of the netcdf file structure. Specifically, all variables, dimensions, attribute names and some attribute values were considered to be part of a DOD. Based on this definition, a database, libraries and management tools were developed that are providing consistency to the structure within ARM data. This paper describes the software developed around this definition and their current and future impact. In addition we discuss limitations of the definition, mitigations and future work.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner