1. Copyright © 2009, Open Geospatial Consortium, Inc. Time issue : Meteo Domain needs and WMS present means Meteorology DWG Frédéric Guillaud, Marie-Françoise Voidrot 30 September 2009

2. Met requirements in term of time handling

3. Helping the World to Communicate Geographically Copyright © 2009 Open Geospatial Consortium Requirements for forecasted products(1/2) Data sources : –Gridded model data –Forecasted simulated images (Radar, satellite) –Forecasted features (meteorological objects like fronts) –Other products (forecasted accumulations, average …) Data provider point of view –One-dimensional TIME parameter is not sufficient, at least we need Forecast Reference Time * Forecast Period or Forecast Validity Time or Validity Period –Other time information may be mentioned : Statistical Process Period (average, accumulation, ….) Start, End Time of data collection for the model analysis Data CutOff … *CF- terminology

4. Helping the World to Communicate Geographically Copyright © 2009 Open Geospatial Consortium Requirements for forecasted products(2/2) Client point of view …. –Often need the best forecast for a validity time. Problems –The best forecast is not always the last model run. –The best forecast criteria may depend on the client profile : Scientist or model developer, forecaster, mass market …

5. Helping the World to Communicate Geographically Copyright © 2009 Open Geospatial Consortium Requirements for observed and climatology products Data sources : –Remote sensing (radar, satellite, lidar) –Upper / surface air observation –Feature diagnostics (weather objects like fronts, jets, …) –Complex products & climatology (accumulations, averages, …) In most cases, the observation is associated with a time Often a timestamp, sometimes a time range or timestamp list Pixel timestamp (radar or satellite originating data)

6. Helping the World to Communicate Geographically Copyright © 2009 Open Geospatial Consortium Means available into the OGC WMS standard

7. Helping the World to Communicate Geographically Copyright © 2009 Open Geospatial Consortium The time in the OGC WMS specification The issue can be addressed through (at least) 3 mechanisms : Native WMS TIME parameter + –Additional DIMENSIONS –Layer metadata –Metadata layer –Others ideas ? There are slight differences between WMS specification depending on the version (1.1.1 or 1.3.0).

8. Helping the World to Communicate Geographically Copyright © 2009 Open Geospatial Consortium WMS 1.1.1 Dimension + Extent –GetCapabilities E.g : <Extent name="FORECAST_REFERENCE_TIME" default="2009-04-10T12:00:00Z" multipleValues="0" nearestValue="0">2009-03-27T12:00:00Z/2009-04- 10T12:00:00Z/PT6H – GetMap With optional TIME parameter : TIME = And/Or : DIM_xxx = E.g : TIME=2009-04-10T12:00:00Z TIME=2009-03-27T12:00:00Z/2009-04-10T12:00:00Z/PT6H And / Or DIM_FORECAST_REFERENCE_TIME=2009-08-10T00:00:00Z

9. Helping the World to Communicate Geographically Copyright © 2009 Open Geospatial Consortium WMS 1.3.0 Dimension (no more « extent » ) –GetCapabilities More detailed DIMENSION metadata Layer metadata : It is possible to specify one or more metadata URL for each layer. – GetMap With optional TIME parameter : TIME = And/Or : DIM_xxx =

10. Helping the World to Communicate Geographically Copyright © 2009 Open Geospatial Consortium TIME encoding : WMS profile of ISO8601 WMS 1.1.1 Annex B or WMS 1.3.0 Annex D (despite slight differences …) –Timestamp 2009-09-30T12:00:00Z –Period P1Y –Comma separated timestamps 2009-09-30T12:00:00Z, 2009-09-30T15:00:00Z, 2009-09- 30T18:00:00Z –Time range 2009-09-30T12:00:00Z/2009-10-01T12:00:00Z –Time range plus interval 2009-09-30T12:00:00Z/2009-10-01T12:00:00Z/PT3H

11. Helping the World to Communicate Geographically Copyright © 2009 Open Geospatial Consortium Time handling in MET formats (1/2) GRIB 1: –Forecast Reference Time, Forecast Period, Statistical Process Period are encoded in section 1 product definition GRIB 2: –The time definition is more complex but more scalable –Forecast Reference Time is encoded in section 1 product definition. –Forecast Period, Statistical Process Period and other time parameters depends on “templates” –It is possible to describe multiple forecast periods.

12. Helping the World to Communicate Geographically Copyright © 2009 Open Geospatial Consortium Time handling in MET formats (2/2) BUFR: –The O&M sampling Time can be encoded in section1 (identification section) –Many descriptors are available to encode other time characteristics (Reference time for accumulations, extreme data, more complex statistics) See BUFR B13 Table NetCDF: – CF convention for possible time axis : forecast_reference_time, forecast_period … GeoTiff: –In the TIFF 6.0 specification, only one tag (date and time the image was created) –Time dimensions for data are not handled in geoTiff so far !

13. Helping the World to Communicate Geographically Copyright © 2009 Open Geospatial Consortium Possible solutions to be discussed (1/3) We have to define a balance among usage of –Native WMS TIME parameter –DIMENSIONS (compliant WMS 1.3 Annex C) –Layer metadata –Metadata Layer Interoperability goal : WMS TIME semantic is usually implemented for validity time or sampling time for observations. => Other time semantics should be handled with dimensions or metadata

14. Helping the World to Communicate Geographically Copyright © 2009 Open Geospatial Consortium Possible solutions to be discussed (2/3) Criteria DIMENSION parameters versus Layer Metadata –The map request is based on the data characteristic (optional TIME or DIMENSIONS) => it cannot be based on metadata. –Caution : the metadata is linked to a layer, and not a map (case of multidimensional layers) For instance if one run is one layer then one unique metadata for it –More selective requests could be performed through DIMENSIONS (Native TIME plus other time dimensions, e.g : forecast reference time, statistical process period, …) – Problem : dimensions are not always implemented … Criteria Layer Metadata versus Metadata Layer –If the time information has to be a coverage (e.g pixel timestamp), it requires a full metadata layer –Example : Layer metadata ok for numerical model but satellite and radar require metadata layer

15. Helping the World to Communicate Geographically Copyright © 2009 Open Geospatial Consortium Possible solutions to be discussed (3/3) Towards a controlled vocabulary or conventions for additional DIM parameters and metadata layers: –WMO table driven codes based ? –CF standard names based ? (CF-NetCDF metadata) forecast_reference_time, forecast_period … => Mapping GRIB – CF metadata exists (from ECWMF) –SWE O&M based ? samplingTime resultTime –Others ? (HMA, …)

16. Helping the World to Communicate Geographically Copyright © 2009 Open Geospatial Consortium Copyright © 2009, Open Geospatial Consortium, Inc. Thank you for your attention