Re: [cf-pointobsconvention] Draft 2



Ethan Davis wrote:
A few comments:

1) The profile observation data CDL needs a "z" dimension:

z = 35 ;

ok


2) For the "Creating Fast Access to Children" contiguous list indices, you should change it to "firstChild(i) and firstChild(i) + numChildren(i) -1" or specify that it is an inclusive/exclusive range.

ok


3) In the list of data types, I don't find the definitions very clear. I think the definitions would benefit from starting with definitions of the base types and working up to the various types of collections for those base types. I think the problem for me is that the connectedness doesn't seem clear when we are talking about collections of the base types. Maybe:

Base types:
- Point data: a set of parameters measured at a time and location (fixed x, y, z, t) - Profile: an ordered set of data points along a vertical line at a single instant in time (fixed x, y, t; connected in z). - Trajectory: an ordered set of data points along a curve in time and space (connected in x, y, z, t; ordered in t).

good


Collections of base types:
- Collection of point data: a set of data points distributed in space and time (unconnected in x, y, z, t)
- Collection of profile data: What is the connectedness between profiles?
   - the profiles are not connected (point collection of profiles?)
- the profiles are all for the same location/station and ordered in time (a time series collection of profiles?) - the profiles are from many stations but all for the same time (station collection of profiles?) - Collection of trajectory data: Same here. What is the connectedness between trajectories?

At the moment, the only "collection types" that make generic sense to me are: 1) the set of events occur at random points, and 2) the set of events occur at some set of locations. Case 1) you might call an "unconnected collection" or "point collection" (not as good?)

Case 2) you might call a "station collection". note that it implies a "time series".