The topics discussed were:
1) We stressed the importance of the history global attribute. This attribute is particularly important in the case of satellite data, where the re-mapping history relates directly to the quality & precision of the data.
2) We decided that for un-re-mapped satellite data, the primary navigation parameterization should be the exact one for the given satellite type (i.e. GOES-8, GOES-7, GMS, etc.). These variables should be grouped together with a "nav" dimension, in the same way that gridded data navigation information is grouped together. Each variable containing navigable data should have a navigation attribute set to the "nav" dimension.
3) Multiple navigation parameterizations can exist in the same netCDF file, but the navigation variables must be grouped separately by different "nav" dimensions. In this case, variables containing navigable data should have the navigation attribute set to the appropriate "nav" dimension.
4) Multiple instances of a single navigation parameterization can exist in the same netCDF file (for example, satellite images following the path of a hurricane, or a floating grid following storm development). In this case, the "nav" dimension must be greater than 1, and while the parameterization of the navigation remains constant, the values of the navigation variables may change as a function of image or grid. In this case, each navigable variable must be dimensioned by the "nav" dimension (rather than just having a navigation attribute).
5) If including the generalized orthographic navigation in the un-re-mapped satellite netCDF is desired, the author of the netCDF may choose to include it, although it is not required by the conventions. If the author chooses to include it, it should follow the navigation parameterization rules defined in the gridded data conventions. Namely, the navigation variables should be grouped together by a "nav" dimension. The content of the navigation parameterization is determined by TABLE C in the document GRIB Edition 1, by John Stackpole.
6) We also decided to abandon the weak conventions of _dim and _var (for example, navigation_dim in the grid definition) and to discourage the use of hyphens in variable names, dimensions, and attributes. We also agreed that for navigable data types such as gridded data and satellite data, we will need a key word to define the type of navigation used (such as GOES-7, or GRIB1, etc). This key word will need to be described by a convention. Please see #3 in the topics for our next meeting, above.