[CF-metadata] Conventions for a network of velocity sensors

From: Ateljevich, Eli <eli>
Date: Wed, 9 Feb 2011 11:43:45 -0800

Seth,
I'm very grateful for your opinion. The temporal coverage issue is a great point. I agree that the data can be naturally grouped in files.

There is also one unexploited organizational idea I thought of since I wrote the question: the (HDF5) "group" level. Is CF amenable to the use of this feature? Does it defeat any of the benefits of the metadata standard...like software being able to read files?

Thanks,
Eli


-----Original Message-----
From: Seth McGinnis [mailto:mcginnis at ucar.edu]
Sent: Wednesday, February 09, 2011 10:27 AM
To: Ateljevich, Eli; cf-metadata at cgd.ucar.edu
Subject: Re: [CF-metadata] Conventions for a network of velocity sensors

Hi Eli,

It's just a personal opinion, but I would strongly favor the first approach,
putting each timeseries into a separate file.

It's simple and straightforward, and it reflects the nature of the data source.
 I don't know of any standard nomenclature for sensor names either, but as long
as you're consistent and systematic about it, you should be fine. The only
potential downside to this approach is that the number of files could become
unmanageably large, but that can be mitigated by organizing them into a
directory tree.

I think option 2 is one to avoid. It's a lot more work to write code that
manipulates strings to construct dynamic variable names than it is to write
code that can just look for a variable named "velocity".

Option 3 isn't bad, but as you note there's no easy place to hang the sensor
metadata, plus all of the timeseries must have exactly the same temporal
coverage. I don't think it's a net win.

Cheers,

--Seth

----
Seth McGinnis
NARCCAP Data Manager
Associate Scientist
IMAGe / NCAR
----
On Tue, 8 Feb 2011 23:22:16 -0800
 "Ateljevich, Eli" <eli at water.ca.gov> wrote:
>Hi, I am modeling the San Francisco Bay-Delta an estuary with several
>tributaries including the Sacramento River, San Joaquin River and numerous
>smaller streams. The boundary data are time series of cross-sectional velocity
>profiles across each channel. The cross-sections for different sensors are
>different and each cross-section contains numerous points.
>I would like to retain the information about which velocities come from which
>sensors, and also add some attributes about the sensors. Can anyone comment on
>which of these possibilities is best:
>1. Each sensor is an independent file. Then the variable would be
>water_velocity_x and _y, the transect locations would be included as a
>coordinate and the sensor station name such as sacramento_rsac075 would be an
>attribute. It is unclear to me that there is a standard nomenclature for the
>sensor name.
>2. Each variable is a sensor station-variable pair:
>sacramento_rsac075_velocity. I don't see anything that looks like this in the
>CF examples, but the standard does not address variable names. I would use an
>attribute to identify the sensor name redundantly, and again I am not clear if
>there is a standard for this. The hitch seems to be that each station would
>have a different set of cross-section locations so the file would have a
>location coordinate for each sensor.
>3.The data could be velocity(time,station). The "station" coordinate would be
>associated with lat, long and sensor names. This lumps all the data in one
>variable. The sensor used for each data point can be identified, but there is
>no clear slot for sensor metadata.
>Thanks,
>Eli
>_______________________________________________
>CF-metadata mailing list
>CF-metadata at cgd.ucar.edu
>http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata

Received on Wed Feb 09 2011 - 12:43:45 GMT