Dataset/Collection exchange

standard description of data collections

Collections are currently too large for OPeNDAP but still considered by most users as a single entity (a half-century 24 member ensemble of GCM runs, six-hourly, daily, monthly, for example).

While Ingrid makes no structural distinction between a dataset and a collection, the DODS mechanism transmits a single dds and das, and Ingrid transmits a set of unique independent variables for netcdf-like clients, so there is a practical limit on how complicated a DODS dataset can be.

There is thus an ad-hoc definition where datasets without independent variables are considered to be collections, e.g. do not offer the option of transmitting with DODS.

example: MIT
example: TAMU

Currently Ingrid uses an XML document to transmit collections, i.e. datasets without independent variables. The key point is that the parts of the collection are transmitted by reference.
For example: returns

<dataset dodsref="dods">
<fullname>NOAA NODC WOA98</fullname>
<description>NOAA NODC WOA98:  World Ocean Atlas 1998, an atlas of objectively analyzed fields of major ocean parameters</description>
<document href=".dataset_documentation.html">dataset documentation</document>
<dataset href=".ANNUAL/index.xml" dodsref=".ANNUAL/dods"><name>ANNUAL</name></dataset>
<dataset href=".MONTHLY/index.xml" dodsref=".MONTHLY/dods"><name>MONTHLY</name></dataset>
<dataset href=".SEASONAL/index.xml" dodsref=".SEASONAL/dods"><name>SEASONAL</name></dataset>
or returns
<dataset dodsref="dods">
<description>CAC:  Climate Analysis Center</description>
<history>Climate Analysis Center SST</history>
<document href=".dataset_documentation.html">dataset documentation</document>
<dataset href=".climatology/index.xml" dodsref=".climatology/dods"><name>climatology</name></dataset>
<dataset href=".smoothed/index.xml" dodsref=".smoothed/dods"><name>smoothed</name></dataset>
<variable dodsref=".sst/dods"><name>sst</name></variable>
<variable dodsref=".ssta/dods"><name>ssta</name></variable>

Soon Ingrid will use a standard XML document to transmit collections -- THREDDS, e.g. SOURCES