Issue #239 has been updated by Jitsuko Hasegawa.
- Status changed from ET discussion to Validated
- Target version set to InterCBS2016
There are a number of cases where it is desirable to allow the "forecast time" to be negative. These include, but are not limited to, accumulations which start before the reference time during a 4-D variational analysis, which was raised by DWD as agenda item 2.1.2 at IPET-DRMM-II in 2014.
Dr Weiqing Qu, Australian Bureau of Meteorology (BoM), emphasized the risk that some GRIB edition 2 software assumes that the forecast time must be positive regardless of the fact that the regulations do not specify it must be positive. In particular, both wgrib2 and GRIB-API assume that the forecast time is positive and behave incorrectly when a negative value is encoded or decoded.
The meeting recognized the issue critical and reviewed the proposal by BoM. Accordingly, it agreed the new Regulation 92.6.3 to be introduced in GRIB edition 2. A change of wgrib2 and GRIB-API software will be coordinated by the respective responsible developers to conform with the added regulation.
On the other hand, the amendments to each product definition template were not preferred, because regulation is applied to each template and confirmation will be needed when a new template is developed.
The meeting agreed the proposal as shown in the Annex to this paragraph to validate the impact of the new regulation and take the procedure for adoption between CBS sessions, because the risk of comparatively low.
DRC #239: GRIB encoding of the beginning of overall time interval in the past
- Author: Eizi TOYODA
- Status: Validated
- Priority: Normal
- Assignee: Eizi TOYODA
- Category: GRIB2 structure
- Target version: InterCBS2016
2.1.2 GRIB encoding of the beginning of overall time interval in the past
DWD needs to encode in GRIB a product which is obtained by a statistical process (accumulation) over a continuous time interval. GRIB Product definition template (PDT) 4.8 seems to be most appropriate for this purpose.
The product is generated at a point in time which is denoted as the reference time, encoded in GRIB Section 1, octets 13-19. The end of the time interval can be encoded using octets 35-41 in PDT 4.8. However, the beginning of the time interval may not be individually defined.
The Note 2 of PDT 4.8 says: The reference time in section 1 and the forecast time together define the beginning of the overall time interval.
The beginning of the time interval may lie in the past with respect to the reference time. In such a case, a negative value must be defined as forecast time (octets 19-22 in PDT 4.8), which is not permitted.
The meeting agreed on necessity to develop this type of representation for negative forecast time and to set up a sub-group. JMA, UK Met Office (UKMO), European Centre for Medium-Range Weather Forecast (ECMWF), DWD and Meteorological Service of Canada (MSC) contribute to the discussion.
Document to the meeting:
No comments :
Post a Comment