- 3 -
WAFSOPSG/5-IP/3/ WAFSOPSG/5-IP/3
9/6/09
WORLD AREA FORECAST SYSTEM OPERATIONS GROUP (WAFSOPSG)
FIFTH MEETING
Paris, France, 16 to 18 September 2009
Agenda Item / 6: / Development of the WAFS6.1: / Improved GRIB 2 forecasts for convective clouds, icing and turbulence
Follow-up on WAFSOPSG-4 Conclusion (C 4/17 b)) Implementation of EDR turbulence reporting
(Presented by the World Meteorological Organization)
1. INTRODUCTION
1.1 The Group will recall that, in connection with the task given to the World Area Forecast Centers (WAFCs) to provide verification of the new gridded forecasts of icing, turbulence and cumulonimbus cloud, the World Meteorological Organization (WMO) member of the group was requested to provide the group with an update on the status of automated air reports of turbulence based on the WMO aircraft meteorological data relay (AMDAR) programme. Objective ground truth of aircraft turbulence is a prerequisite for meaningful verification of turbulence forecasts. Subjective PIREPS have been deficient in numbers and coverage, and they are expected to be biased towards “positive” encounters, whereas nil reports are rare. Furthermore, subjective PIREPS are based on the effects of turbulence on the airframe, which is dependent on aircraft type, load factor and stage of flight.
1.2 The WMO AMDAR programme has undertaken great strides to create a global data source for automated aircraft observations, and following an ICAO recommendation, software for the reporting of turbulence based on wind data (eddy dissipation rate (EDR)) was developed by participating national programs. There is, however, an ongoing debate on the relative merits of accelerometer –based DEVG versus EDR-based data. The former is directly related to the aircraft response, whereas the latter is an atmospheric quantity that could be measured by meteorological equipment and directly output from numerical weather prediction models. The WMO member thus attempted to obtain an update on the current situation on a global basis.
2. Current Status of development
2.1 Overview of global situation
2.1.1 The 10th Meeting of the AMDAR Panel in November 2007 was presented with a number of national reports including activities on the turbulence and EDR question, whereas the 11th meeting held in the fall of 2008 in Kuala Lumpur did not entertain any meaningful debate on the subject. In the 2007 Report, only the US reported in detail on several sub-projects using EDR, whereas other national reports, if and when they discussed turbulence, used different formulations based on Derived Equivalent Vertical Gusts (DEVG).
2.2 US Situation
2.2.1 There is a problem with the TAMDAR turbulence data, and AirDat is expected to provide an updated version of the software soon. No recent information is available to date. About 70 Delta Airlines aircraft are reporting regularly EDR-based turbulence, subject to an on-board QC procedure. The legacy code for EDR used by United Airlines should have been updated in 2008, pending confirmation All these data are being used operationally for the new Graphical Turbulence Guidance (GTG) product developed by UCAR and GSD. Data are being archived and used for near-real-time and delayed-mode verification studies over the CONUS. It should be noted that both in the US and globally, the major emphasis of the ADMAR development has been to increase the number, accuracy and reliability of the Water Vapour Sensors given their significant impact on regional, high-resolution NWP, and the EDR issue seems to have taken a secondary role in development. A recent spot check on the GSD web site was unsuccessful in locating a significant number of EDR reports over the US.
2.3 Other national AMDAR programs
2.3.1 Currently, the AMDAR panel has been informed of 2 national programs providing turbulence information, (France and Australia), but these are based on DEVG-methodology. Again, no new information was received since 2007. A very detailed study of DEVG measurements by the Airbus A318/319/320 family of Air France for aircraft couples over-flying the territory of France undertaken by Y Lemaire, M. Stoll and D. Bouchara shows significant differences between different avionics ( e.g. he MU –equipped A320 NEVER reported moderate or greater turbulence throughout the study period (6 weeks). Correlations between pairs of aircraft observations even over a very narrow time/space windows (less than 100m vertical, 10-60min and between 0.1 and 1 deg lat/long) reveal extremely low correlations (typically below 5% even eliminating all NIL cases) .
2.3.2 This would suggest that before any further comparisons are undertaken, we need a much better understanding of the typical “structure functions” of turbulence (Auto-correlation R = R(dT, dX, dY, dZ)). This also puts any verification results of turbulence forecasts against such reports into perspective. For instance, for terrain-induced turbulence at the Hong Kong International Airport, the maximum EDR1/3 in the last few nautical miles before touchdown as analyzed from aircraft data could vary by more than one time within a period of about 10 minutes.
2.3.3 Australia has undertaken some investigations of DEVG data, but no published results have become available to us so far. A private communication by Mr. Dean Locket was received mentioning the problems of doubtful DEVG measures obtained during aircraft manoeuvres during final approach and departure, where tight turns, flaps setting and other pilot actions are seen to affect the measurement.
2.3.4 A collaborative study has been established between the Hong Kong Observatory and the National Aerospace Laboratory in the Netherlands to prepare a piece of software (called WINDSTURB) to calculate EDR profiles from Quick Access Recorder (QAR) data of the aircraft from local as well as overseas airlines. The EDR data so obtained could be used as “sky truth” for further tuning of the turbulence alerting algorithm of the Wind Shear and Turbulence Warning System. They are also used for developing the turbulence detection algorithms based on the measurements of remote-sensing meteorological equipment at the Hong Kong International Airport, such as Sodars and LIDARs. Moreover, initial trials have been made to use high resolution numerical weather prediction models (at a spatial resolution of several tens to a couple of hundred metres) to simulate EDR field around the airport and compare the results with actual measurements.
3. Conclusion
3.1 Not so unexpectedly, aircraft turbulence continues to prove elusive not only to forecasting, but also to measuring and observing. In the absence of well-established, robust structure functions of turbulence, stratified by season and root cause phenomena ( jet-related, gravity-wave, convection-related) , disputes about the methodology of measurement-derived turbulence values , be it EDR ( suffering badly from the marginal sampling frequency and accuracy of wind components by a moving platform that is itself influences the micro-scale of flow around the aircraft) or DEVG, which would require a number of additional parameters ( e.g. current load factor as function of fuel burn, flaps setting and trim) to provide comparable results, remain almost futile and explain the very modest number of studies into the phenomenon.
3.2 Data optimization methods applied by EUCOS and other national/regional AMDAR programs, while being an excellent tool for providing cost-effective data sets, are severely limiting the availability of densely spaced on-line reports allowing the determination of structure functions, as they are effectively eliminating any “close pairs” of observations. In the light of this, ICAO may consider relaxing its insistence on EDR as sole basis for “objective” turbulence measurements, encourage the creation of a comprehensive test data set including the necessary meta-data (aircraft type, load, configuration, trim…) that would allow for the definition of robust structure functions as a basis for further verification work for NWP-derived turbulence forecasts. The demise of the METLINKSG may have limited the scope of further work in this field, and a transfer of this responsibility to the AMOFSG could be considered an option.
3.3 Unless we improve the availability, scientific understanding and robustness of a ground truth of turbulence, the results of the evaluation of gridded forecast products (necessary for user confidence, but also as a yardstick to judge the effects of future model developments) will remain questionable as a basis for strategic decisions concerning further product development, use and application.
4. ACTION BY THE WAFSOPSG
4.1 The group is invited to note the information contained in this paper.
— END —