Below are listed brief summaries of proposals for new conventions dealing with
coordinate systems in netCDF datasets. Some of these proposals are more comprehensive,
dealing with other conventions as well. The proposals are listed in reverse chronological
order, so that the most recent appear first. Proposals are named with the author's
last name and date of the proposal, except that proposals from multiple authors
use an acronym formed from each author's last name.
Gregory, Drach and Tett provide an update to their comprehensive proposal
for conventions for climate data.
Proposes using ":coordinates" attributes naming coordinate functions to define
coordinate systems. Proposes a mechanism for keyword aliasing, to specify
another keyword than "coordinates", for example. Specifies geodetic coordinate
systems. Includes links to revised motivating examples
and more formal definitions.
Comments: Gregory, Caron, Gregory,
Sheldon, Caron, Davies,
Gregory, Caron, Weibel,
Davies, Caron, Caron
- Eaton-1997-07-14: Proposes a more general definition
for coordinate variables, permitting multidimensional coordinate variables.
Also defines "dependent variable". Needs no attribute conventions.
- Davies-1997-07-14: Comprehensive proposal, including
new conventions for names, coordinate variables, cyclic (modular) coordinates,
arithmetic progressions in coordinate variables, units, motivating examples,
global attributes, physical quantity of a variable, and missing values for
coordinate variables. Also includes comments on and comparisons with GDT proposal.
Argues against multidimensional coordinate variables.
Comments: Granger, Caron
- Granger-1997-07-12: Proposes global map attributes,
referential attributes that list the variables or mappings that constitute
a coordinate system. Supports multiple coordinate systems per variable. Applies
this proposal to many of the examples presented by others.
Comments: Caron, Granger,
- Caron-1997-07-11: Provides formal definitions for
dimension, index domain, projection, variable, vector function, coordinate
system, location, and various special kinds of coordinate systems. Presents
11 motivating examples that can be used to compare proposals. Superseded by
Comments: Davies-1997-07-14, Caron
- Emmerson-1997-07-11: Presents field data model
and its coordinate systems for manifold, base, and world. Defends use of referential
attributes. Presents spiral wire example of a 1D manifold and 3D base.
Comments: Sheldon, Emmerson,
Eaton, Sheldon, Eaton,
Sheldon, Emmerson, Sheldon,
- Sheldon-1997-07-09: Defends referential attributes
for coordinate conventions. Proposes meaning for referential attributes in
which variable attribute has same name as one of the variable dimensions.
- Rew-1997-06-20: Proposes global dimension attributes
(with same name as dimension) as an alternative to multidimensional coordinate
variables or referential attributes. Presents layer example.
Comments: Granger-1997-07-12, Davies-1997-07-14
- WW-1997-06-17: Walker and Waring propose multidimensional
coordinate variables and new variable attributes "coordinate_type" and "coordinates"
to associate dimensions with coordinates.
Comments: Rew, Walker,
Gregory, Rew, Gregory,
Caron, Rew, Caron,
Sheldon, Caron, Sheldon,
- GDT-1997-06-13: Gregory, Drach and Tett provide
a comprehensive proposal for an update to COARDS conventions for climate data.
Note: a newer version
of this proposal is now available.
Comments: Hankin, WW-1997-06-17,
Rew, Sheldon, Caron,
Gregory, Caron, Gregory,
Davies, Sheldon, Gregory,
- Caron-1997-06-06: First draft of formal definitions
needed for coordinate convention discussions (superseded by later Caron-1997-07-11). Proposes use of both multidimensional
coordinates and referential attributes for specifying coordinate systems in
Comments: Rew, Caron
- Sheldon-1997-03-26: First mention of "referential
attributes" as an alternative to multidimensional coordinate variables, and
presentation of an example that uses referential attributes.
Comments: Caron, Sheldon,
Hankin, Caron, Eaton,
Sheldon, Eaton, Sheldon,
Denham, Weibel, Caron,
Eaton, Hankin, Packman
User's Guide for C description of coordinate variables: The current "official"
conventions that provide for only one-dimensional coordinate variables and
no referential attributes.
Comments: Eaton-1997-07-14, Davies-1997-07-14
- Pepke-1994-01-27: Conventions for sigma coordinates
and curvilinear grids. In particular, includes a convention for recognizing
component dimensions in addition to space and time dimensions. Proposes "location_"
attributes naming corresponding coordinate variables.
- Treinish-1993-08-23: Describes referential attribute
conventions for handling irregular arrays, series data, data in separate files,
compact specifications for regular dimensions, partially regular grids, and
irregular surfaces. In particular, new variable attributes include "field",
"connections", "positions", "component", and "attributes".
- Signell-1992-10-20: Proposes using AVS field model
as a basis for conventions. Proposes new "components" dimension and "independent_variables"
attribute to specify coordinate variables.
Comments: Rew, Caron,