Short paper available: "Model Data Interoperability for IOOS"
rsignell
It's embarrassing to realize that I haven't posted anything here for
months, giving the impression that not much is going on. Quite the
opposite is true. I'll try to rectify this with a flurry of
posts! ;-)
Here's a short paper called "Model Data Interoperability for US IOOS"
that will appear in the Proceedings of the 11th Coastal and Estuarine
Modeling Conference outlining the approach we are taking and some
lessons learned:
http://groups.google.com/group/ioos_model_data_interop/web/signell_ecm11_mar8.doc
Basically this is a summary of what I worked on with many of you last
year. There is a very long list of acknowledgments!
First, although my formal detail to NOAA IOOS ended last November and
I'm back officially at USGS, I am continuing to work with many of you
on model data interoperability issues, and there are several projects
taking place that should help us build out the existing model data
interoperability infrastructure (which I'll post on shortly)
-Rich
John Maurer
Really liked this paper. Thanks for posting it. A clear summary that
helped organize and solidify my thoughts about the various data,
servers, and clients here at HiOOS (and elsewhere in IOOS). I have a
few questions I was hoping you could answer:
1. You mentioned that WCS does not support curvilinear grids. Do you
happen to know what OGC's plans are for supporting them in the future?
2. I didn't realize that the TDS NetCDF Subset Service (NSS) doesn't
support curvilinear grids either--do you know what Unidata's plans are
for adding support for these in the future? However, I just gave it a
shot on one of our own curvilinear datasets and it didn't complain and
seems to have worked... so maybe this is no longer an issue? (We're
using the latest stable release: TDS 4.1.2.) I notice that one glitch
is that since we have defined approximate 1-D lat/lon coordinate
variables with NcML, these get used in place of the more accurate 2-D
auxiliary lat/lon coordinate variables in the NSS output. But if I
remove the 1-D variables, NSS includes the 2-D lat/lon variables with
no (apparent) problem.
3. Lastly, you said that TDS lagged behind ncWMS and did not yet
support curvlinear grids: do you know if this has changed yet and if
so what version of TDS? I have looked at the WMS output from TDS and
cannot get imagery when the WMS version is 1.3.1 but do get results
when switching this to 1.1.1: don't know if this is related.
4. Have you also looked at Live Access Server (LAS) and ERDDAP? And if
so, do you prefer Godiva2? Just curious since we're dabbling in all
three right now and I'm about to look into Godiva2 again. I'm not sure
ERDDAP supports curvlinear grids, though, now that I think about it--I
need to double-check with Bob Simon; but LAS added support in V7.
Thanks again!
Cheers,
John Maurer
Hawaii Ocean Observing System (HiOOS)
Roy Mendelssohn
-Roy
> --
> You received this message because you are subscribed to the Google Groups "IOOS Model Data Interoperability Working Group" group.
> To post to this group, send email to ioos_model_...@googlegroups.com.
> To unsubscribe from this group, send email to ioos_model_data_i...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/ioos_model_data_interop?hl=en.
>
**********************
"The contents of this message do not reflect any position of the U.S. Government or NOAA."
**********************
Roy Mendelssohn
Supervisory Operations Research Analyst
NOAA/NMFS
Environmental Research Division
Southwest Fisheries Science Center
1352 Lighthouse Avenue
Pacific Grove, CA 93950-2097
e-mail: Roy.Men...@noaa.gov (Note new e-mail address)
voice: (831)-648-9029
fax: (831)-648-8440
www: http://www.pfeg.noaa.gov/
"Old age and treachery will overcome youth and skill."
"From those who have been given much, much will be expected"
Rich Signell
> Really liked this paper. A clear summary that
> helped organize and solidify my thoughts about the various data,
> servers, and clients here at HiOOS (and elsewhere in IOOS).
I'm glad you found it helpful!
> 1. You mentioned that WCS does not support curvilinear grids. Do you
> happen to know what OGC's plans are for supporting them in the future?
I think they want to support curvilinear and unstructured grids in the
future, but there is nothing that I know of being developed in the
near term. Those of you who know the OGC side better, please speak
up if this is incorrect!
>
> 2. I didn't realize that the TDS NetCDF Subset Service (NSS) doesn't
> support curvilinear grids either--do you know what Unidata's plans are
> for adding support for these in the future? However, I just gave it a
> shot on one of our own curvilinear datasets and it didn't complain and
> seems to have worked... so maybe this is no longer an issue?
I just tried subsetting a curvilinear grid using the NetCDF Subset
Service with TDS 4.1.2 and to my surprise, it worked fine! The
only glitches were that the 3D variables had lost some of the
parameters for the vertical coordinate transform in the process.
I'll post this to the netcdf-Java group.
>
> 3. Lastly, you said that TDS lagged behind ncWMS and did not yet
> support curvlinear grids: do you know if this has changed yet and if
> so what version of TDS? I have looked at the WMS output from TDS and
> cannot get imagery when the WMS version is 1.3.1 but do get results
> when switching this to 1.1.1: don't know if this is related.
I have only tried 1.1.1, but the curvilinear feature handling on the
backend of the ncWMS server related to the WMS request, so it
shouldn't matter what version of the WMS request you are using.
>
> 4. Have you also looked at Live Access Server (LAS) and ERDDAP? And if
> so, do you prefer Godiva2? Just curious since we're dabbling in all
> three right now and I'm about to look into Godiva2 again. I'm not sure
> ERDDAP supports curvlinear grids, though, now that I think about it--I
> need to double-check with Bob Simon; but LAS added support in V7.
ERDDAP has a lot of it's power because it uses regular grids. If for
your intended application you are okay with representing your
curvilinear grid in ERDDAP as an I,J grid, as Roy said. I don't think
they have plans for supporting curvilinear grids because that would be
a whole different data model.
I haven't done much with LAS yet, so I would like to here the views of
others. I think Steve Hankin told me they can do on the fly
regridding now in the F-TDS, which would be very cool.
-Rich
--
Dr. Richard P. Signell (508) 457-2229
USGS, 384 Woods Hole Rd.
Woods Hole, MA 02543-1598
Wen Long
Nice paper and thanks for the summary and most recent update.
A typo: you had "Ruoying He" twice in the acknowledgment list.
Could you please also let me know where the latest
mexnc+netcdf_toolbox is? I really don't like the new netcdf syntax
within
matlab. With the matlab-native netcdf, I can't figure out how to fetch
a subset of an array easily. With the mexnc, yet we
can just provide index range after the {}.
Thanks,
On Jun 3, 12:01 pm, rsignell <rsign...@gmail.com> wrote:
> Folks,
>
> It's embarrassing to realize that I haven't posted anything here for
> months, giving the impression that not much is going on. Quite the
> opposite is true. I'll try to rectify this with a flurry of
> posts! ;-)
>
> Here's a short paper called "Model Data Interoperability for US IOOS"
> that will appear in the Proceedings of the 11th Coastal and Estuarine
> Modeling Conference outlining the approach we are taking and some
> lessons learned:
>
John Maurer
> > 3. Lastly, you said that TDS lagged behind ncWMS and did not yet
> > support curvlinear grids: do you know if this has changed yet and if
> > so what version of TDS? I have looked at the WMS output from TDS and
> > cannot get imagery when the WMS version is 1.3.1 but do get results
> > when switching this to 1.1.1: don't know if this is related.
>
> I have only tried 1.1.1, but the curvilinear feature handling on the
> backend of the ncWMS server related to the WMS request, so it
> shouldn't matter what version of the WMS request you are using.
Cheers,
John
Richard Signell
No, it doesn't appear that the lastest WMS server capabilities in the
TDS 4.2 handle curvilinear coordinates yet, so I guess the TDS WMS is
still lagging behind.
This request to ncWMS works:
http://geoport-dev.whoi.edu/ncWMS-dev/wms?REQUEST=GetMap&VERSION=1.3.0&STYLES=&CRS=CRS:84&WIDTH=256&HEIGHT=256&FORMAT=image/jpeg&TRANSPARENT=false&LAYERS=coawst22/temp&BBOX=-101.74989476701279,11.88846735222873,-53.25308436246699,48.463925392207926
This request to TDS 4.2 WMS *should* work:
But returns with the error message:
Unexpected error of type java.io.IOException:
ucar.ma2.InvalidRangeException: Number of ranges in section (4) must
be = 5
I'm cc'ing John Caron, Ethan Davis and John Blower for their comment.
-Rich
> --
> You received this message because you are subscribed to the Google Groups "IOOS Model Data Interoperability Working Group" group.
> To post to this group, send email to ioos_model_...@googlegroups.com.
> To unsubscribe from this group, send email to ioos_model_data_i...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/ioos_model_data_interop?hl=en.
>
>
--
Richard Signell
Interesting. But as you can see in the URL that didn't work, I *was*
using the "best time series" URL, which does not have two time
dimensions. But perhaps somehow the 2 time dimesion data sets is
getting called anyway by the WMS. Anyway, I'll try a simpler case
that is not an FRMC and see if it works.
-Rich
On Tue, Jun 29, 2010 at 11:32 AM, Jon Blower <j.d.b...@reading.ac.uk> wrote:
> Hi Rich, all,
>
> I've never actually tested this in TDS - the capability for curvilinear
> grids certainly hasn't been disabled in TDS, but I'd expect TDS to
> generate the necessary look-up tables (a background process that can
> take minutes) on the first request to the dataset since a server
> restart. Thereafter the tables should be cached in memory. This is not
> a good user experience unfortunately as the first user of a dataset will
> probably see a server timeout.
>
> The IOException you're seeing is probably another issue - the
> THREDDS-WMS doesn't correctly handle the two time dimensions in an FMRC
> aggregation; this is largely because there are no WMS conventions for
> such a dataset yet (the OGC MetOcean Domain Working Group is working on
> this). You might have more luck with the best-estimates timeseries.
> (See http://www.resc.rdg.ac.uk/trac/myocean-tools/ticket/15.)
>
> Cheers, Jon