The Graphical Turbulence Guidance (GTG) system & recent high-resolution modeling studies Aviation & Turbulence in the Free Atmosphere Royal Meteorological Society Meeting at Imperial College, London 15 Jan 2014 Robert Sharman NCAR/RAL Boulder, CO USA sharman@ucar.edu
Aviation turbulence r&d areas Better observations of aircraft scale turbulence In situ turbulence measurement methods Ground-based and airborne remote sensing techniques, including satellite-based technologies Better nowcasting & forecasting products Forecast products for strategic planning Concentration has been on upper-level CAT, MWT Diagnosis and nowcast products for tactical avoidance Use of observations to nudge short-term forecasts Better understanding of turbulence mechanisms Analyses of high-rate data gathered in field programs (airborne and surface) Case studies using high-resolution simulations 3D Controlled environment Can be used to formulate and test turbulence diagnostics
GTG Forecasting Procedure* Aircraft turbulence is much smaller than present operational NWP model resolutions Therefore cannot directly predict aircraft scale turbulence - only hope is to infer turbulence potential from larger scale inhomogenities Compute turbulence diagnostics from coarse resolution operational nwp model Multiple l causes require multiple l forecasting strategies Graphical Turbulence Guidance Product (GTG) = Ensemble mean of diagnostics Available 24x7 on NOAA s ADDS website Ensemble mean available on Operational ADDS (GTG2.5) (http://aviationweather.gov/adds) 0-12 hr lead times, updated hourly 10,000 ft-fl450 *Sharman et al. Weather & Forecasting 2006
Some common turbulence diagnostics Frontogenesis function (good at upper levels) D D v D v F or Dt Dt Dt p Dutton Empirical Index Dutton 1.25 S 0.25 2 10.5 S H V Unbalanced flow (Koch et al., McCann, Knox et al.) Deformation X shear (Ellrod) 2 R 2 Juv (, ) f u 2 2 1/2 I DEF S, DEF D D V SH ST v u u v D D SH ST x y x y Eddy dissipation rate (ε 1/3 ) computed from second order structure functions of velocity and/or temperature Dq( s) [() qx qx ( s)] 2 D () s C D s) C s 2/3 2/3 q q () s REF ( q () s 2/3
But what are we forecasting? Aircraft scale eddies that affect aircraft Aircraft response is aircraft dependent but this is what pilot reports: light, moderate, severe CANNOT forecast these levels for every aircraft in the airspace Instead need atmospheric turbulence measure (i.e. aircraft independent measure) We forecast EDR (= ε 1/3 m 2/3 s -1 ) ICAO standard Can relate to airborne and remote EDR estimates Can relate EDR to aircraft loads (σ g ~ ε 1/3 ) Convenient scale 0-1 For reference ICAO standard d thresholds h (2001,2010 ) for mid-sized aircraft are EDR=0.10, 0.3, 0.5 for light, moderate, severe, resp. EDR=0.10, 0.4, 0.7 for light, moderate, severe, resp. EDR PIREP 5
Conversion of diagnostics to EDR Each D is rescaled to an EDR assuming a log- normal distribution of edr log 1/3 a blog D i Where a and b are chosen to give best fit to expected lognormal distribution in the higher ranges EDR=0.1,0.3,0.5 a and b depend on Moderate log and SD log 1/3 1/3 Which must be estimated from climatology Severe DAL in situ EDR data <logε 1/3 >= 2.85 SD[log ε 1/3 ] = 0.57 6
Example diagnostics converted to EDR 3.5-hr forecast 3.3km conus grid* *courtesy J-H Kim, NASA Ames 7
Verification: PIREPs and In situ EDR measurements PIREPs Verbal subjective Mean position error ~ 50 km Automated EDR (ε 1/3 m 2/3 / s) estimates Calculation performed onboard ACMS, data transferred via ACARS @ 8Hz Peak + mean over 1 min. Position uncertainty < 10 km Currently deployed or planned to be deployed on ~ 70 UAL B757-200s ~ 90 DAL B737-800s,737-700s ~ 100 DAL B767-300ERs and -400ERs ~ 400 SWA B737-800s,-700s Currently ~ 5000/hour Compared to 400/day PIREPs Must use both so need to be able to convert from pirep to edr -> simple mapping developed PIREPS Red=severe, blue=mwt 24 hrs of UAL insitu 24 hrs of DAL insitu 03 Jan 2010 8
Current GTG3 RUC-based performance (6-hr fcst ROC curves 12 mos. valid18z) - Binary discrimination i i performance of smooth vs. moderateor-greater (MOG) PIREPS + in situ edr data of predic cting MOG obability Pro GTG3 Ellrod1 EDR UBF, LHFK SGS TKE Individual diagnostics Null-MOG GTG3 AUC=0.812 Probability bilit of false alarms High threshold Low threshold (Predict no turb) (Predict turb everywhere) 9
Other diagnostic combination strategies* 2010-2011 Cross-Validation ROC Curves 6-hr forecast 13km WRFRAP verified against insitu edr only Edr >03vsedr >0.3 <03 0.3 FL>200 Random forest AUC: 0.85 KNN AUC:0.83 Logistic Regression AUC: 0.80 Current GTG AUC: 0.78/0.79 *Courtesy John Williams
GFS Global GTG UKMET GTG output based on 3 global models for the same case. Contours are EDR (m 2/3 s 1 ) at FL290 31 Dec 2011 6 hr fcst valid 18Z All models used native vertical grids All models used same number of diagnostics All models used same thresholds ECMWF
Characterization studies Use research simulation models to recreate both the large scale forcing and smaller scale turbulence to determine origins of recorded turbulence encounters Can control simulation options to isolate effects Cloud/latent heating, Cloud top radiative cooling Typically use multi-nested approach 12
Simulations to identify relation of cirrus bands to turbulence* *Knox et al. Transverse cirrus bands in weather systems: a grand tour of an enduring enigma Weather 2006
Example cirrus bands/turbulence 15 Nov 2011* FL310 Forecasted turbulent areas *Courtesy MelissaThomas DAL
Simulations of two cirrus bands cases 17 June 2005 Moderate and severe turbulence insitu edr measurements near Transverse (Radial) MCS Outflow Bands over central US - Trier & Sharman MWR 2009 - Trier et al. JAS 2010 9 Sep 2010 : Moderate and severe turbulence reported in vicinity of mid-latitude cyclone over western Pacific Ocean - Kim et al. submitted to MWR 15
Example 1. Out-of-cloud CIT near MCS* In situ edr + wind vectors Note bands are transverse to wind vectors Observed turbulence 0905 UTC 16 June 43 N 40 N Model TKE 37 N Courtesy UW CIMSS *Trier and Sharman, MWR, 2009 *Trier et al JAS 2010 ARWRF simulation =3 km
MCS CIT mechanism No cloud With cloud With cloud No cloud Strong vertical shear at flight levels due almost entirely to MCS outflow on north side Vertical shear at flight levels on south side weaker because easterly outflow winds and shear are opposed by their westerly environmental (adiabatic) counterparts
Observations and WRF-Simulations of the 17 June 2005 MCS Case 600m nest IR Satellite at 0950 UTC 17 June 2005 Simulated Cloud Top Temperature (0950 UTC, t = 5.8 h ) Observed Turbulence @ 37 Kft (0936-0957 UTC) L=light, M=Moderate Turbulence UA 776 Trier and Sharman (2009, Mon. Wea. Rev.) Observations and MCS-Scale Simulations Trier et al. (2010, J. Atmos. Sci.) High-Resolution Simulation and Analysis of Radial Bands
4-hour Loop of Brightness Temperature, 12-km Moist Static Instability N < 0 and 11.5-13-km Vertical Shear from 07-11 UTC 17 June with Dt = 10 min 2 m North East Tb Anvil bands originate within zones of moist static instability Bands are aligned along the anvil vertical shear vector Similar to horizontal convective rolls in boundary layer arising from thermal instability
Case 2: cirrus bands in mid- latitude cyclone 9 Sep 2010* Control (CTL) Simulation No Cloud Radiative Feedback (NCR) Simulation x x x x Approximate Locations of Observed Turbulence Approximate Locations of Observed Turbulence Domain 3 = 3.3 km Domain 3 = 3.3 km *J-H Kim, submitted to MWR 500 km Model-Derived Reflectivity and Sea-Level Pressure at 0300 UTC 9 Sep (t = 9 h)
Relation of wind shear vectors and stability to bands CTL at 2140 UTC 8 Sep 2010 (t = 3.67 h) CTL at 2340 UTC 8 Sep 2010 (t = 5.67 h) Brightness Temperature, 11.75 km Moist Static Stability, 200 km 10.75-12.75 km Wind Shear vectors
Brightness Temperature, and 11.75 km MSL Moist Static Stability Control (CTL) Run No Cloud-Radiative Feedback (NCR) Run 2220 UTC 8 Sep (t = 4.33 h) 2310 UTC 8 Sep (t = 5.17 h) 200 km
Simulation of bands - summary Both cases show bands owe their existence to convective instability in anvil within background shear MCS case: background shear mainly from upper-level outflow Pacific mid-latitude cyclone case: shear mainly from jet stream Both have morphology akin to PBL rolls Since bands originate i from convectively unstable regions of anvil, it is not surprising that those areas are turbulent MCS case cloud-top radiative cooling encourages bands, in Pacific case it is crucial for bands i23
GTG Summary and future work Using an ensemble of turbulence diagnostics (GTG) instead of one diagnostic gives more robust performance Provides EDR Technique can be used with any input NWP model Longer term goals Forecast convective turbulence (CIT) Provide probabilistic forecasts Use of in situ edr provides valuable verification source Numerical simulations Use multi-nested approach to relate large scale to small scales for a variety of environments and turbulence sources Helps understand turbulence genesis which may ultimately lead to formulation of better diagnostics 24