<html>
<head>
<meta content="text/html; charset=windows-1252"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<font face="Times New Roman">Hi Jennifer,<br>
<br>
I didn't mean the table wasn't useful. I meant that the "dataset"
unit reported in the table provides little information other than
distinguishing between "at least partially available" (>0) and
"not available" (=0). That information is, of course, quite
valuable to users ... I agree. That's why we created it.<br>
<br>
Best regards,<br>
Karl<br>
</font><br>
On 12/15/11 7:30 AM, Jennifer Adams wrote:
<blockquote
cite="mid:76CD1298-7F49-4E45-872D-66DB60B544AB@cola.iges.org"
type="cite"><br>
<div>
<div>On Dec 15, 2011, at 10:02 AM, Karl Taylor wrote:</div>
<br class="Apple-interchange-newline">
<blockquote type="cite">
<meta content="text/html; charset=windows-1252"
http-equiv="Content-Type">
<div bgcolor="#FFFFFF" text="#000000"> <font face="Times New
Roman">Hi Jennifer and Mark,<br>
<br>
RE:<br>
</font>
<blockquote type="cite"><a moz-do-not-send="true"
href="http://cmip-pcmdi.llnl.gov/cmip5/esg_tables/transpose_esg_static_table.html">http://cmip-pcmdi.llnl.gov/cmip5/esg_tables/transpose_esg_static_table.html</a></blockquote>
<br>
<font face="Times New Roman">The numbers listed in the table
(default view) show how many "datasets" have been produced
from each model for each experiment. Multiple datasets
are produced for each simulation. Even if you click on
the selection button at the top of the page and select a
single CMOR table, this won't definitively tell you how
many realizations of a simulation have been performed by
each model. [For example, there are two categories of
output -- output1 and output2 -- so the data produced from
one CMOR table can be split across two datasets. Also,
I'm not sure whether the counting eliminates replicated
versions of the datasets (the same data can be found at
more than one data node).<br>
<br>
The bottom line is that this table should only be used to
infer which models have reported (some) results from at
least one realization. The dataset count isn't a
particularly useful quantity for a user. <br>
</font></div>
</blockquote>
I respectfully disagree, since a dataset list for a particular
model and experiment is useful to me. It saves me from having to
go to the gateway and suffer through the painfully slow
mouse-clicking and "Loading, please wait" messages to select
that experiment/model/realm/frequency and discover what the
ensemble member names are. </div>
<div><br>
</div>
<div>--Jennifer</div>
</blockquote>
</body>
</html>