Jon,
I completely agree that the remaining steps are many and they will not be easy.
And Roy is correct that getting as far as we have was more difficult than it
looks in hindsight. But it's important to keep in mind that all the hard work
is not thrown away if we don't accomplish the ultimate goals completely and
quickly. For instance, I think it will be extremely valuable for our community
to agree on CF conventions for some of the non-gridded data types we have
identified. This will facilitate the exchange of other types of data within
our own community with added semantics that we agree on. This can provide a
building block for future OGC/ISO extension standards, but it will be valuable
and useful in a very practical way strictly on its own merit.
Another important class of benefits I refer to as "collateral successes." These
are valuable outcomes that were not among the initial explicit goals. A great example
of this is going on in another GALEON discussion thread on parametrized projections and
EPSG codes. This will help us be more explicit about the coordinate reference systems we
use. On the other side of the coin, it is clear that many members of the traditional GIS
community are developing a much better understanding of 3D time varying data. And some
have implemented code for reading and writing CF-netCDF directly. These are very
valuable steps forward even though they don't completely solve the interoperability
problem in themselves and there is still a huge amount of challenging work to be done.
I'll close with the quote from Thelonius Monk that I have at the top of the whiteboard in
my office: "Simple Ain't Easy."
-- Ben
On Thu, Oct 9, 2008 at 2:08 AM, Jon Blower <jdb@xxxxxxxxxxxxxxxxxxxx> wrote:
Hi Ben,
This is a really useful email, thanks.
This is a significant win in two ways. First it actually works and some
members of traditionally GIS-oriented communities (e.g. the US hydrology
community) are using it to access metoceans data from our servers for use in
GIS applications.
Excellent! This is exactly what this is all about if you ask me.
This demonstrates that there's a lot of value even in the "most basic"
use case of regularly-spaced grids. My main worry is that the value
(i.e. return on investment) will sharply decrease as we try to
incorporate more complex cases.
We've taken nearly
all the steps for those metoceans CF conforming data collections that
contain regularly spaced grids. One by one, we need to pick off the other
data types (including the unstructured meshes), develop CF conventions, map
to a standard coverage data model where appropriate.
I think this is where we need to take care. It's going to be a long,
tough road and the law of diminishing returns will hit us hard. The
number of communties that need regularly-spaced grids is going to be a
lot larger than those that really need full-complexity data through an
OGC interface. Personally, I would rather spend time on rolling the
GALEON1 approach much more widely so we can build a base of really
good-quality, well-defined, services that people can really use. I
know it's an interesting piece of research to keep pushing the
envelope, but we need to be clear about what we're buying.
Cheers, Jon
