[Go-essp-tech] Extending the DRS syntax to observations

[Cinquini, Luca (3880)]

Hi Renata,
I'm glad you are experimenting with the same issues...

I actually also experimented with using the ESG publisher to publish AIRS data that was organized in a DRS-like fashion like I mentioned:

<root directory>/cmip5/observations/nasa/aqua/airs/mon/atmos/ta/l3/vYYYYMMDD/<files>

I was able to succeed by using the following snippet of code in esg.ini:

[project:AIRS]
category_defaults =
        experiment | airs_exp
        model | airs_model
dataset_id = %(activity)s.%(product)s.%(agency)s.%(mission)s.%(instrument)s.%(time_frequency)s.%(realm)s.%(variable)s.%(level)s
dataset_name_format = AIRS Level 3 Monthly Data (NetCDF)
directory_format = /esg/data/%(activity)s/%(product)s/%(agency)s/%(mission)s/%(instrument)s/%(time_frequency)s/%(realm)s/%(variable)s/%(level)s/%(version)s
handler = esgcet.config.netcdf_handler:NetcdfHandler
parent_id = nasa.airs
variable_per_file = true
las_configure = true
maps = las_time_delta_map

In other words, I instructed the publisher to use fictitious values for experiment and model, but I also instructed it to read all the relevant DRS fields from the directory structure itself. That seemed to work.
Maybe Bob can comment if this seems reasonable to him.

And I agree that a CMOR table for observations needs to be developed. Could we open up that process so we can include all relevant values not just from ORNL, but also NASA, NOAA and any agency who will contribute observations ?

thanks a lot,
Luca

On Jan 26, 2011, at 4:50 PM, Renata McCoy wrote:

Hi Luca,

I was experimenting with setting up a publishing script for the ARM cmbe observational data following the proposal below. I see few issues:
I understand (I am cc-ing Bob Drach to check on that) the publisher needs at least those 3 fields for the "CMIP5' project:
project (which is 'CMIP5'), experiment (that could be specified as 'none' in command options, but maybe should be 'observations' ?),  and model
A 'model' is an important field and I was trying to set it up so as 'cmbe' would be my 'model', which seems to be equivalent to 'airs' for NASA AIRS data

The other problem is that we need a specific table (CMOR table) to establish the controlled variable list and data frequency that is obs specific.
We (with ORNL team) are trying to create an obsCfSites table, that would be mostly a copy of cfSites table with obs specific time definition (for cmbe it's 1 hour average, for CDIAC data - 30 min average, and for 3D data with vertical axis - we would like to specify that any pressure level could be reported). I think it would be good if we could encompass all the observational data needs in one observational table.

I did not try to test publish the data yet, I am working on rewriting the data with the appropriate metadata first, so I am not sure what other problems may pop up.

Greetings,
Renata

------------------------------------------------------------
Renata B. McCoy, Ph.D
Program for Climate Model Diagnosis and Intercomparison(L-103)
Lawrence Livermore National Laboratory
P.O. Box 808
Livermore, CA 94551

(925) 424-5237 (voice)
(925) 422-7675 (fax)
mccoy20 at llnl.gov<mailto:mccoy20 at llnl.gov>
------------------------------------------------------------


On Jan 26, 2011, at 2:50 PM, Cinquini, Luca (3880) wrote:

Hi all,
apologies for cross-posting...

I'd like to start a discussion on how the DRS specification for CMIP5 model output (http://cmip-pcmdi.llnl.gov/cmip5/docs/cmip5_data_reference_syntax.pdf)
can be applied to observational datasets that will be made part of the same archive.  A proposal based on recent workshops with PCMDI is detailed on this wiki page:

http://oodt.jpl.nasa.gov/wiki/display/CLIMATE/Data+and+Metadata+Requirements+for+CMIP5+Observational+Datasets

As an example, the directory structure for the NASA AIRS dataset would look like this:

<root directory>/cmip5/observations/nasa/aqua/airs/mon/atmos/ta/l3/vYYYYMMDD/<files>

where:

"observations"=<product>, same as DRS "output1" or "output2"
"nasa"=<agency>, same as DRS "institute"
"aqua"=<mission>, replaces DRS "model"
"airs"=<instrument>, replaces DRS "experiment"
"l3"=<processing level>, replaces DRS "ensemble member"
vYYYYMMDD=the dated version, same as specified by the DRS

The values for the various fields <agency>, <mission>, <instrument> and <processing level> would need to be selected from a controlled vocabulary similar to the one established for models.

Any comment or insight on the matter is appreciated.... The idea is to try to finalize the specification relatively quickly, let's say a couple of weeks, so that we can start preparing
and publishing these observations into the CMIP5 archive.

thanks in advance,

Luca



-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/go-essp-tech/attachments/20110126/2bacf6bd/attachment.html