[Seavox] To business....

[Luis Bermudez]

John,

3b is a very important issue. Discovery metadata that could be added  
automatically is the way to go. So if the metadata contains a summary  
of the depth, a computation may be performed by the data aggregator  
or data portal to categorize the data set by the data aggregator's  
own view; which can be different than the data producers view. This  
way also data producer don't have to think of another metadata element.

Luis

---------------------------------------
Luis Bermudez Ph.D.
Software Engineer
MMI Liaison - http://marinemetadata.org
bermudez at mbari.org
Tel:  (831) 775-1929
Monterey Bay Aquarium Research Institute
7700 Sandholdt Road, Moss Landing CA 95039-9644, USA





On Nov 7, 2006, at 11:12 AM, John Graybeal wrote:

> Roy,
>
> Thanks, this is helpful.
>
> The following is verbose, for only the Really Interested -- I'm  
> seeing if Roy and I are on the same page. (Conclusions at the bottom.)
>
> At 4:13 PM +0000 11/7/06, Roy Lowry wrote:
>> 3) It provides a neat way of discovering near-surface and near-bed  
>> data sets.  To search for these using absolute depths needs  
>> measurement depth, water column depth and the desired relationship  
>> between them to be incorporated into the query in some manner,  
>> which I've never managed to do satisfactorily.  Keywords plus a  
>> guideline algorithm to generate keywords from data labelled with  
>> absolute values is the most satisfactory way I've found
>
> Here's how my thought process works, with respect to previous  
> emails about determining near-bed. I think this mostly matches your  
> own, except for (4).
>
> 1) It would be really valuable to have an indication of near-bed  
> data.  (Ideally on a per-record basis, but for this application,  
> let's make it on a per-file basis.)  I assume the existence of the  
> keyword will mean "at least some points in this data set are in the  
> indicated vertical range."
>
> 2) I agree with your experience: There is no elegant system to  
> enable the derivation of distance from sea bed (can I call this  
> 'altitude', for this thread only?) from other parameters.  At MBARI  
> we've done systematic analyses of large numbes of extremely well- 
> documented data sets, where we have both pressure/depth/lat/lon  
> data, and detailed bathymetric profiles (= water column depth,  
> roughly),  to identify algorithmically whether the measurement was  
> "on the bottom."  We succeeded, but only to an approximation, and  
> it required great effort.  This is not a realistic computation for  
> most situations, especially in that it must be performed at every  
> data point.
>
> 3) Given there are no great answers, we have three choices:
>  a) Get someone to manually characterize the data set based on  
> their knowledge of it.
>  b) Take advantage of the necessary values to compute these  
> results, which can either be "measured altitude" (from a sounder),  
> or "water column depth - measurement depth."  If the values are  
> available, calculate the results for every point and set the  
> keyword accordingly.
>  c) A combination of (a) and (b).
> My preference is (c) -- the flag/keyword gets set if any of the  
> above conditions indicate the altitude is less than X for any  
> measurement in the data set.  (Unfortunately a single X can not  
> meet all science uses, but we'll have to live with that. Human  
> judgment could perhaps override the precise number, in that case.)   
> In which case, we specify that meaningful data ranges (for  
> altitude, in the case of this particular keyword) are an ideal  
> source for setting this flag, but manual characterization of the  
> data set is needed otherwise.  (And we understand that the absence  
> of the keyword could just mean a bad or missing characterization  
> process.)
>
> 4) For the many scientific purposes to which "near-bed" data can be  
> put, the distance from the sea bed is arguably more important than  
> the percentage of the water column. In shallow waters, the fact the  
> measurement is 3 meters from the floor is more significant than the  
> fact it is 0.5 meters from the air.  This rule of thumb is not 100%  
> true, but I claim it is more than 50% true.  And intuitively I  
> don't want "near-bed" in 6000m of water to be twice the size of  
> near-bed in 3000m of water.  Anyone who considers using the data  
> will have to consider the reality of the data set, and it's better  
> to be inclusive (include some that may not apply) than exclusive  
> (exclude some that may apply).
>
> CONCLUSIONS
>
> So if I got to set X, I might ignorantly prefer a slightly smaller  
> number like 3 meters, but I wouldn't worry about a different value  
> for shallow waters.  It's still near the sea bed and many similar  
> processes will typically apply.
>
> John
>
> P.S. I think what makes this seem complicated is that we're  
> conflating the concepts of "depth" and "altitude" into a single  
> keyword field.  From a data management perspective, one set of  
> depth keywords and another set of altitude keywords would be  
> better.  But as long as you get to use multiple keywords in a  
> description, algorithmically it is equivalent.
> -- 
> ----------
> John Graybeal   <mailto:graybeal at mbari.org>  -- 831-775-1956
> Monterey Bay Aquarium Research Institute
> Marine Metadata Initiative: http://marinemetadata.org   ||  Shore  
> Side Data System: http://www.mbari.org/ssds
> _______________________________________________
> Seavox mailing list
> Seavox at mailman.nerc-liv.ac.uk
> http://mailman.nerc-liv.ac.uk/mailman/listinfo/seavox
>
> -- 
> This message (and any attachments) is for the recipient only. NERC
> is subject to the Freedom of Information Act 2000 and the contents
> of this email and any reply you make may be disclosed by NERC unless
> it is exempt from release under the Act. Any material supplied to
> NERC may be stored in an electronic records management system.
>