[CF-metadata] point observation data in CF 1.4
Ute.Broenner at sintef.no
Wed Oct 13 05:35:02 MDT 2010
Thanks for pointing that out for me. This is exactly what I need.
Consider the environment before printing
From: rsignell at gmail.com [mailto:rsignell at gmail.com] On Behalf Of Rich Signell
Sent: Mittwoch, 13. Oktober 2010 13:31
To: Ute Brönner
Cc: John Caron; Jonathan Gregory; cf-metadata at cgd.ucar.edu
Subject: Re: [CF-metadata] point observation data in CF 1.4
> 2. I think trajectory is when you follow a set of "things", boats, a person. But at each time step they are identical, maybe not the same number because of missing data. I could assume that I have a trajectory but actually I can't be sure if my particles are the same as before. Therefore I chose not to take that convention.
Let me attempt to clarify what I think Ute is talking about.
Typical producers of this kind of data are numerical particle tracking
models. These codes step through time, following the (x,y,t) or
(x,y,z,t) trajectories of individual particles. At each time step,
more particles may be introduced to be tracked, while other particles
stop being tracked because they leave the domain, hit the boundary, or
Thus at each time step there are a known number of particle being
tracked, but this varies as the run proceeds in time. So each
particle trajectory may have different start and end times. There may
be thousands or tens of thousands of particles, so it's not feasible
to write each trajectory into a separate file.
We want a featureType that will allow us to write the entire
collection of particles at each time step into a single file, and that
will allow us to extract all the particles at a single time step, as
well as extract individual particle trajectories by their ID. It
didn't seem like the existing pointFeature types work for this case.
Dr. Richard P. Signell (508) 457-2229
USGS, 384 Woods Hole Rd.
Woods Hole, MA 02543-1598
> I'll be happy to contribute to the discussion as well as good as I can!
> Best regards,
> Ute Brönner
> Consider the environment before printing
> -----Original Message-----
> From: John Caron [mailto:caron at unidata.ucar.edu]
> Sent: Dienstag, 12. Oktober 2010 19:31
> To: Rich Signell
> Cc: Ute Brönner; Jonathan Gregory; cf-metadata at cgd.ucar.edu
> Subject: Re: [CF-metadata] point observation data in CF 1.4
> Hi Ute:
> 1. im not sure what this means: " I now finally find out that the new
> approaches of Netcdf 4 are not implemented in the Java API, yet."
> 2. I would have thought "3D particle tracking" would be a trajectory.
> how is it different?
> On 10/11/2010 4:24 AM, Rich Signell wrote:
>> I was thinking that you could use the proposed convention in 9.3.2 as
>> a workaround, with the "station" being each record. But I see now
>> that the coordinate variables for lon, lat need to be a function of
>> station, so as you say, that won't work.
>> Clearly there is a need for another Point Convention type to handle
>> the output from particle tracking models like this. I can think of at
>> least four models that would benefit from this convention right now,
>> including the NSF RAPID grant we are working on for 3D particle
>> tracking using LTRANS for the Deepwater Horizon Spill.
>> @Jonathan, Caron& Hankin: Can we revive your discussion? I'd be
>> happy to participate.
>> On Mon, Oct 11, 2010 at 4:05 AM, Ute Brönner<Ute.Broenner at sintef.no> wrote:
>>> Hei Rich,
>>> I found that convention before, this was what I mentioned to Jonathan.
>>> But first of all, this is not convention, yet, and secondly I have no stations but a varying set of observations per timestep (neither stations nor trajectories). I now write the data with redundant time as a limited dimension, and records(time, latitude, longitude) and have
>>> mass (record), radius(record) etc.
>>> Thanks anyway,
>>> Ute Brönner
>>> Consider the environment before printing
>>> -----Original Message-----
>>> From: rsignell at gmail.com [mailto:rsignell at gmail.com] On Behalf Of Rich Signell
>>> Sent: Freitag, 8. Oktober 2010 13:59
>>> To: Ute Brönner
>>> Cc: Jonathan Gregory; cf-metadata at cgd.ucar.edu; John Caron
>>> Subject: Re: [CF-metadata] point observation data in CF 1.4
>>> On this page:
>>> It appears that your case *might* be handled by:
>>> 9.3.2 Ragged array (contiguous) representation
>>> I'm pretty sure that this "ragged_row_count" feature *is* included in
>>> NetCDF-Java, but John Caron (cc'd here) could confirm.
>>> Please report back to this group if you find success (or perhaps even
>>> if you don't!)
>>> On Thu, Oct 7, 2010 at 3:15 AM, Ute Brönner<Ute.Broenner at sintef.no> wrote:
>>>> thanks for your answer! My troubles were related to shape and dimensions.
>>>> I now finally find out that the new approaches of Netcdf 4 are not implemented in the Java API, yet.
>>>> I now use a record dimension which is unlimited and a limited time dimension. Hope that works, otherwise, I have your address now :-)
>>>> Best regards,
>>>> Ute Brönner
>>>> Consider the environment before printing
>>>> -----Original Message-----
>>>> From: Jonathan Gregory [mailto:jonathan at met.reading.ac.uk] On Behalf Of Jonathan Gregory
>>>> Sent: Mittwoch, 6. Oktober 2010 19:13
>>>> To: Ute Brönner
>>>> Cc: cf-metadata at cgd.ucar.edu
>>>> Subject: [CF-metadata] point observation data in CF 1.4
>>>> Dear Ute
>>>> You are right, the convention for timeseries of different lengths being
>>>> contained in one variable is not yet agreed. Some months ago John Caron, Steve
>>>> Hankin and I discussed it at length but did not quite manage to finish it,
>>>> unfortunately. So there isn't a CF convention for it at the moment.
>>>>> but I have some trouble in writing the data.
>>>> What kind of trouble?
>>>> Best wishes
>>>> CF-metadata mailing list
>>>> CF-metadata at cgd.ucar.edu
>>> Dr. Richard P. Signell (508) 457-2229
>>> USGS, 384 Woods Hole Rd.
>>> Woods Hole, MA 02543-1598
More information about the CF-metadata