Dear CF,
A small group (Mark Hedley, Jonathan Gregory, David Arctur, Sean Arms, Tim Whiteaker, and I) met this morning.
There was generally agreement on:
> 1) There is general consensus that the "simple geometry with linked time series" use case is worth pursuing in CF.
> 2) We agree that we need to support all of the simple multigeometry types and their non-multi variants. This includes holes in polygons.
> 3) We seem to agree that a solution will use the existing contiguous ragged array (CF 1.6 9.3.3 <http://cfconventions.org/cf-conventions/v1.6.0/cf-conventions.html#_contiguous_ragged_array_representation>) with some attribute names and relationships changed to better suit geometries.
Two issues were discussed in some depth, brief notes regarding our discussion on these points follow:
> 4) We are not sure if node sharing (which supports explicit declaration of topology of geometries and can reduce file sizes) is needed in a CF specification for simple geometries.
A lack of node sharing doesn?t really block us moving forward with use cases that use shared nodes since software can infer it.
In the spec for this, we should leave the door open and make a note about how to handle (infer) topology explicitly.
We will state that we put it out of scope for now and that it may be included in a future version of CF.
Side Note: We need to make sure that the assumption that there is a straight line between nodes _in the coordinate reference system the nodes are defined in_ is described clearly.
> 5) We are not sure if this new content is more appropriate as an extension of the point/timeseries DSG feature types or a variant of cell bounds to be introduced in a new section of the spec.
What else do we get from discrete sampling geometries? ? Probably not going to handle moving polygons.
We can already do polygons with cell bounds but we can?t do multis or holes.
Maybe we should create a general geometry encoding. It could be used via the cell bounds / auxiliary coordinates and be attached to cells or DSG points.
Need to be careful that the DSG specs' limitations on one featureType per file isn?t a problem.
From this, I?ll draft up a new proposal and circulate it for the group to review.
Best,
- Dave
> On Mar 7, 2017, at 7:28 AM, David Blodgett <dblodgett at usgs.gov> wrote:
>
> Dear CF,
>
> This is a reminder that we will be getting together on a google hang out in about thirty minutes to discuss the way forward on simple geometries in NetCDF-CF. I will write up a brief summary of this call and share with the list later today.
>
> Brief review:
> 1) There is general consensus that the "simple geometry with linked time series" use case is worth pursuing in CF.
> 2) We agree that we need to support all of the simple multigeometry types and their non-multi variants. This includes holes in polygons.
> 3) We seem to agree that a solution will use the existing contiguous ragged array (CF 1.6 9.3.3 <http://cfconventions.org/cf-conventions/v1.6.0/cf-conventions.html#_contiguous_ragged_array_representation>) with some attribute names and relationships changed to better suit geometries.
> 4) We are not sure if node sharing (which supports explicit declaration of topology of geometries and can reduce file sizes) is needed in a CF specification for simple geometries.
> 5) We are not sure if this new content is more appropriate as an extension of the point/timeseries DSG feature types or a variant of cell bounds to be introduced in a new section of the spec.
>
> Proposed Agenda:
> Affirm agreement on 1-3
> Discuss the alternatives in 4. Make a decision or craft a strategy to resolve disagreement.
> Come to consensus on 5 and discuss the outline for draft addition to the spec.
>
> Talk soon,
>
> Dave
>
> NetCDF - Simple Geometries Discussion
> Scheduled: Mar 7, 2017, 8:00 AM to 9:00 AM (CT)
> Hopefully google hangouts will work. Please use this url:
> https://plus.google.com/hangouts/_/calendar/ZGJsb2RnZXR0Lmgyb0BnbWFpbC5jb20.c192p3bskticchduvubks6ug44 <https://plus.google.com/hangouts/_/calendar/ZGJsb2RnZXR0Lmgyb0BnbWFpbC5jb20.c192p3bskticchduvubks6ug44>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <
http://mailman.cgd.ucar.edu/pipermail/cf-metadata/attachments/20170307/690d0010/attachment.html>
Received on Tue Mar 07 2017 - 08:31:57 GMT
