Date

 

Attendees

Goals

  • Progress meeting to discuss prototype implementation of tile variables in GRIB2.

Discussion items

See the presentation here.

Item

Who

Notes

Presentation of the prototype tile templateMatt Griffith
  • In particular the tileClass references a code table including "simple" tiles as well as tile groupings. This makes it very flexible to be extended in the future
  • Tile coverage will be kept as a key in the GRIB message. This is necessary so that e.g. 2m temp can be encoded on a snow-covered urban tile.
  • Different models' implementations of tiles will be accommodated, but with the first entries of the table representing common tile classes which will need to be agreed upon by all cooperating tile models used in DestinE.

  • A GRIB2 in this form might look something like:

    paramtileClasstileAttribute
    FractionTemperate grasslandSnow covered
    FractionTemperate grasslandFlooded
    FractionTemperate grasslandFlooded and ice covered
    FractionTemperate grasslandUnmodified
    FractionTemperate needleleaf evergreenSnow
    FractionTemperate needleleaf evergreenUnmodified
    2m tempTemperate grasslandSnow covered
    2m tempTemperate grasslandFlooded
    2m tempTemperate grasslandFlooded and ice covered
    2m tempTemperate grasslandUnmodified
    2m tempTemperate needleleaf evergreenSnow
    2m tempTemperate needleleaf evergreen

    Unmodified

  • See the presentation for more details.
CommentsAll
  • The feedback on the template design was generally positive. It was noted that of course the devil is in the detail so now that we have the initial prototype in place we will provide an example of a fully filled out GRIB2 with all the appropriate metadata but with no data for now. Trygve/Ulf and others will work on bundling up some tile variable data for us to then fill this prototype with.

We will communicate more in between, but next official meeting 9am 

UPDATE

Following the meeting, I had a chat with Gianpaolo Balsamo with regard to encoding IFS tile variables with the same template. He was generally happy with how the template looked, and we will run it by some of our Research Department for further comments.

We should also run this by the SURFEX developers. Gianpaolo suggested Patrick Le Moigne, Bertrand Decharme for SURFEX and Valéry Masson for ECOCLIMAP.

Action items

Write a comment…