Implementation of High Resolution GRIP 2 Wafs Upper-Air Forecasts

1

WORLD AREA FORECAST SYSTEM OPERATIONS GROUP (WAFSOPSG)

FIFTH MEETING

Paris, France, 16 to 18 September 2009

IMPLEMENTATION OF HIGH RESOLUTION GRIB 2

WAFS UPPER-AIR FORECASTS

(Presented by the WAFC Provider States)

1.INTRODUCTION

1.1The group will recall that, at its fourth meeting, it formulated two conclusions related to the planned introduction of GRIB 2-coded WAFS upper-air forecasts; they called for the WAFC Provider States to:

a)provide two additional levels at FL 320 (275 hPa) and FL 360 (225 hPa); replace the current thinned grid with a regular grid with a resolution of 1.25 degrees; and change the temporal resolution from 6 to 3 hours, i.e. provide forecasts at time steps of T+6, 9, 12, 15, 18, 21, 24, 27, 30, 33, 36 hours from the analysis time; and study, in coordination with the WAFSOPSG Member from IATA, the desirability of improving vertical resolution below FL 300 to address the specific needs of flight operations over polar regions. (Conclusion 4/12 refers); and

b)develop the higher resolution upper air forecasts, as specified, in time for the WAFSOPSG/5 Meeting (Conclusion 4/13 refers); and

It will also be recalled that the group endorsed the detailed implementation plan for the transition from the GRIB 1 to GRIB 2 code form within the WAFS(Decision 4/18 refers).

1.2This working paper outlines the progress that has been made by the WAFC Provider States towards making GRIB 2 code-form WAFS forecasts available, and expresses some further issues for the consideration of the group.

2.discussion

2.1The group will be pleased to note that, since the WAFSOPSG/4 Meeting, the two WAFCs have worked collaboratively towards developing upper-air gridded WAFS forecasts for wind/temperature/humidity/tropopause and icing/turbulence/CB in the WMO GRIB 2 code form, to fulfil their obligations to the WAFSOPSG.

2.2Traditionally, operational WAFS upper-air gridded forecasts have been in the WMO GRIB 1 code-form, and only comprised forecasts of upper wind (including maximum), upper-air temperature and humidity, geopotential altitude of flight levels, and flight level and temperature of tropopause, as per Annex 3 — Meteorological Service for International Air Navigation,Chapter 3 ptrovisions. Forecasts in the GRIB 1 code-form for icing, turbulence and CB have only been available since 2006/07 for trial and evaluation purposes on FTP servers, they have never been disseminated across satellite distribution system for information relating to air navigation (SADIS)or international satellite communications system (ISCS)satellite broadcasts.

2.3It may be noted that no changes have been, or will be, made to the existing GRIB 1 data sets available on the SADIS and ISCS services.

2.4The development of WAFS Forecasts in the GRIB 2 code form by the two WAFCs began in 2008 following the formulation of WAFSOPSG Conclusion 4/13. Noting the expressed wish of Conclusion 4/12 part d), in consultation with the WAFSOPSG member from IATA, it was agreed that the WAFCs would include additional flight level information at FL270 (350hPa) during the development of the GRIB 2 data.

2.5The specification of WAFS forecasts in the GRIB 2 code form is presented in theappendix to this paper. This specification is also available in an abridged form within Annex 4 of the SADIS User Guide (

2.6The group may wish to note that, by April 2009, WAFC London was in a position to produce the requisite 1078 GRIB 2 WAFS bulletins on a daily basis for internal evaluation – i.e. they were not distributed beyond WAFC London. These introductory data sets were uncompressed and represented approximately 50MB worth of data per model run – compare this with an existing GRIB 1 data set that amounts to approximately 10MB per model run. The underlying reason for the five-fold increase in data volume is attributable to the higher spatial and temporal resolution offered by the GRIB 2 data set, as well as the inclusion of icing/turbulence/CB forecast information. The 50 MB data volume is in accordance with earlier expectations.

2.7Due to the volume/size of the uncompressed GRIB 2 WAFS data sets, particularly when compared to existing GRIB 1 data, the WAFCs recognised from an early stage that an appropriate compression utility would need to be applied to the data set prior to transmission. Dissemination trials conducted by the SADISProviderState over the SADIS 2G satellite broadcast in late-April 2009 (using uncompressed GRIB 2 WAFS data), resulted in a transmission time of 1 hour 55 minutes to send all 1078 GRIB 2 bulletins. This is a five-fold increase in the transmission time compared to an existing GRIB 1 data set across SADIS 2G at present (circa 22-25 minutes). Given that a GRIB 2 data set contains five-times the volume of data compared to GRIB 1, these transmission times were in line with expectations.

2.8WAFC Washington is yet to undertake GRIB 2 data trials on the ISCS satellite broadcast. Nevertheless, a test plan is being developed, the scope of which is expected to be presented to the WAFSOPSG/5 Meeting in the form of a WAFC Washington information paper.

2.9By using an appropriate compression algorithm, the WAFCs expect to achieve a compression ratio in the region of 2:1 for the GRIB 2 WAFS grids. This will result in the net data volume falling to around 25MB per model run. Transmission times are similarly expected to fall to just under 1 hour on the SADIS 2G broadcast.

2.10At the time of writing (June 2009), WAFC London is continuing to develop a compressed version of the GRIB 2 WAFS data. Development is expected to be completed by August 2009; whereupon further transmission tests on the SADIS 2G broadcast are expected in order to validate the predictions shown in the paragraph 2.9 above. If conducted in time, the results of these additional trials will be shared with the WAFSOPSG/5 Meeting.

3.Conclusions

3.1In accordance with the GRIB 1 to GRIB 2 transition plan endorsed by the WAFSOPSG/4 (Decision 4/18 refers), the WAFC Provider States were tasked to make available the GRIB 2 data sets on ISCS and SADIS FTP services, in parallel with GRIB 1, by September 2009. As expressed in the WAFS long term plan to be presented to the WAFSOPSG/5 Meeting under Agenda Item 7, the WAFCs expect to complete this task between September 2009 and March 2010, subject to further endorsement by the group.

3.2The dissemination of GRIB 2 WAFS data on the ISCS and SADIS satellite broadcasts is yet to be formally agreed. The WAFS and SADIS Operations Groups will need to consider whether the transmission performance demonstrated by the GRIB 2 trials conducted over the satellite links are acceptable, including compressed versus uncompressed, and the downstream implications of providing either form of data (e.g. time taken to decompress data on end-user systems).

3.3It should also be noted that when distributing the products over the satellite broadcasts, there would be no way of labelling (encoding) the data as “operational” or “developmental”, and thus no way of managing how they are interpreted or applied. Indeed, on the FTP services, the only way to delineate the difference would be by placing them in “operational” or “developmental” sub directories – it is not possible to encode this information into the bulletins themselves. Therefore, careful consideration should be given to how this information is distributed beyond the WAFCs if it is to be considered (partly or wholly) as developmental (or experimental) rather than operational.

3.4The group may wish to agree that only the upper wind/temperature/humidity/ tropopause forecasts in the GRIB 2 code form should be available on the ISCS and SADIS broadcasts, until such a time that the entire complement of icing/turbulence/CB forecasts have reached a standard deemed acceptable for flight planning applications. To achieve this, the group is invited to formulate the following decision:

4.ACTION BY THE WAFSOPSG

4.1The WAFSOPSG is invited to:

a)note the information contained in this working paper; and

b)decide on a draft decision for the groups consideration.

APPENDIX

GRIB 2 WAFS product specification

Total number of GRIB 2 bulletins per model run = 1078 from each WAFC.

4 model runs (issues) per day based on 0000, 0600, 1200 and 1800 UTC synoptic data time.

Example 1: Bulletin YUXC85 EGRR is WAFC London’s u-component wind forecast at 850hPa (FL050) valid at T+6 from data time.

Example 2: Bulletin YLXM15 KWBC is WAFC Washington’s mean CAT forecast at 150hPa (FL450) valid at T+36 from data time.

— END —