Open rsignell-usgs opened 9 years ago
dpinte
commented
9 years ago @rsignell-usgs Thanks for the suggestion. Iris is in the Canopy repository and we could easily switch to it.
Why would you suggest Iris in favour of netcdf4 we are using now? I definitely not aware of what Iris provides.
johntyree
commented
9 years ago @dpinte no we aren't. I think the THREDDS server was set up before we actually got started, we're just doing the visualization of the existing data. I think @hetland @kjordahl and I will have to get on hangouts and talk about using something like this, although I'm not sure it's directly in scope right now.
rsignell-usgs
commented
9 years ago @dpinte, what Iris provides is the ability to access any CF-compliant model without model specific code. So instead of writing ROMS-specific reading code using NetCDF4, which involve special ROMS-specific routines to compute the vertical coordinates, specify specific variable that contain lon and lat in the dataset, etc, you can just do:
cube = iris.load_cube(model_url,'sea_water_temperature') and get back an object which can unambiguously retrieve the vertical coordinates, horizontal coordinates, time, etc.
This means the user could load not only load ROMS, but also Delft3D, Hycom, POM, NCOM, etc without any modifications -- just letting Iris do the work. See this example: http://nbviewer.ipython.org/github/rsignell-usgs/notebook/blob/master/IOOS_Models.ipynb
rsignell-usgs
commented
9 years ago I'm sure it will be (and should be) out of scope for the current work, but something to think about in the future.
hetland
commented
9 years ago I guess I don't know Iris well enough to know what the advantages and disadvantages are. I guess the one clear advantage of being able to load results from any model might be useful. Would there be other advantages? Disadvantages?
On Wed, Jan 14, 2015 at 3:33 PM, Rich Signell notifications@github.com wrote:
I'm sure it will be (and should be) out of scope for the current work, but something to think about in the future.
— Reply to this email directly or view it on GitHub https://github.com/enthought/tabs/issues/59#issuecomment-69996132.
Prof. Rob Hetland Texas A&M Univ. – Dept. of Oceanography http://pong.tamu.edu/~rob
rsignell-usgs
commented
9 years ago Advantages:
Disadvantage:
Before making a switch, it would be important to see if there are significant performance issues in using Iris over NetCDF4 (Iris uses NetCDF4 behind the scenes to load the data, but also knows about CF conventions), as there is some extra cycles in loading the metadata, etc. It may be inconsequential, but we need to check the performance.
dpinte
commented
9 years ago @rsignell-usgs binary distribution of such package is mandatory. I haven't checked what Anaconda provides. On the Canopy end, it's on the six platforms.
rsignell-usgs
commented
9 years ago The British Met Office folks who lead the development of Iris have their own binstar channel here: https://binstar.org/SciTools/iris where they have the latest Iris (and supporting packages) for 5 platforms.
On Tue, Jan 20, 2015 at 6:39 PM, Didrik Pinte notifications@github.com wrote:
@rsignell-usgs https://github.com/rsignell-usgs binary distribution of such package is mandatory. I haven't checked what Anaconda provides. On the Canopy end, it's on the six platforms.
— Reply to this email directly or view it on GitHub https://github.com/enthought/tabs/issues/59#issuecomment-70757534.
Dr. Richard P. Signell (508) 457-2229 USGS, 384 Woods Hole Rd. Woods Hole, MA 02543-1598
The British Met Office is developing Iris (http://scitools.org.uk/iris/) as a kick-ass package for reading CF-Compliant model output. It's got 4 full time developers working on it: https://github.com/SciTools/iris/graphs/contributors It would be nice to use that as a reader and have this GUI operate on Iris objects, no? @hetland, what do you think?