Present included: John Schattel , Ben Domenico, Sylvie Thepaut, Sandor Kertesz, Michal Weis, Marie-Françoise Voidrot, Frédéric Guillaud,

In the OGC Met Ocean DWG Teleconference on 16 November 2009, the main ideas were that: No remarks on previous Teleconference Minutes. No feedback by now on the action to check WMO available vocabulary that could fit to the concepts of DIM_RUN_START_TIME or DIM_FORECAST_OFFSET (Action Chris)

Out of the email echanges, a new use case was submitted by Benjamin Chartier : how to give access to datasets (coming from climatology statistics for instance) described by an incomplete time value in the GetMap request (for example a time wihout the year element which doesn't seem to be compliant ISO 8601) example : average temperature over many years for a specific day of the month.

If ISO 8601 doesn't allow this, should we make it possible into the TIME parameter or dedicate a DIM_CLIMATOLOGICAL_TIME which could allow partial times to keep the main TIME parameter ISO compliant. Michael suggests a WPS to stick to ISO

For Observation time handling a first step proposed is to identify what the SWE, SOS.. standards recommend to check if we cover all our use cases with it. We wondered if expectation time is an issue. (for polar orbital satellites products for instance the user might wish to know the time of the next available image). Maybe this has to be a metadata. It is a DIMENSION if you might do a request based on it but for instance the time attached to each pixel of a satellite image is metadata. We should check if the TIME value attached to the satellite images by the upstream providers are always done with the same convention (the beginning or the end of the scan)

We should also check the works done on time by the WCS working group.

In an email Jeff recommends to account for the absence of any TIME for all products. Definition of Defaults is necessary. It can be for instance the most up to date for the current time (up to a thredhold of "age") or the best Run if the last in not the best.

Do we need a VALIDITY_TIME_WINDOW to handle the time steps of the model concept or is it metadat attached to the model?

-- MarieFrancoiseVoidrotMartinez - 21 Nov 2009