1 / 16

Update: AFWA ensemble development and findings

Update: AFWA ensemble development and findings. J. Hacker / E. Kuchera Collaborators: C. Snyder, J. Berner , S.-Y. Ha, M. Pocernich , J. Schramm, WRF developers. Guiding interests. Primarily lower atmosphere (winds, shear) Multiple time scales (0-60 h) and fine spatial scales

hada
Download Presentation

Update: AFWA ensemble development and findings

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Update: AFWA ensemble development and findings J. Hacker / E. Kuchera Collaborators: C. Snyder, J. Berner, S.-Y. Ha, M. Pocernich, J. Schramm, WRF developers

  2. Guiding interests • Primarily lower atmosphere (winds, shear) • Multiple time scales (0-60 h) and fine spatial scales • Desire to run decision support algorithms using the output • Limited in-house NWP model development • Complement and augment global NWP ensembles

  3. Development and testing • Multi-parameters in a single set of physics (Param) • Stochastic backscatter in a single set of physics (Berner) • Multi-scheme/parameterization WRF (Phys) • Up to 20 configurations; typically run 10 members • Began with WRFv2.2 and ongoing with WRFv3.2 • Limited (3) multiple physics configurations, chosen for independence and low individual errors • Perturbations to land-use tables • Perturbed observations in independently cycling WRF-3DVar (PO) • Ensemble Transform Kalman Filter (ETKF) • Ensemble Kalman Filter (EnKF; Snyder/Ha) Model uncertainty IC uncertainty All runs use the GEFS ensemble (Ensemble Transform) for lateral boundary conditions.

  4. Physics Configurations AFWA Operational configuration

  5. Testing/verification over two domains CONUS: Nov 2008 – Feb 2009, 00/12Z initialization every other day (cycling as needed). Korea: Oct 2006, 00/12Z initialization every other day (cycling as needed).

  6. Should we include obs error? • Rank histograms of 24-h, 10-m wind speed • Obs errors from N (0,so) • NCEP so is 1.1-1.3 ms-1 (dependent on pressure) Chose not to include observation error because it makes interpretation ambiguous.

  7. Effect of land-use perturbations • Land use perturbations (Eckel and Mass 2005) have a small positive impact on mean error in the PBL; effect is smaller for probabilistic metrics. Temperature Wind speed Solid Includes LU perturbations

  8. Score against direct dynamical downscaling • Baseline is AFWA control on GEFS • 1 – difference; positive is better • Clear advantage of multiple physics 2-m T 700-mb T 10-m WS 700-mb WS Multiple PhysicsMultiple Physics / Perturbed Obs (3DVar) Perturbed Parameters

  9. Score against solely multiple physics 2-m T 700-mb T • Baseline is multi-physics on GEFS • 1 – difference; positive is better • Typically an advantage for obs use in very short range 10-m WS 700-mb WS Multiple Physics / Perturbed Obs (3DVar) ETKF (GEFS mean) Hybrid (3DVar on mean) - All ensemble use same multiple physics configuration

  10. Score against solely multiple physics • Baseline is multi-physics on GEFS • 1 – difference; positive is better • Fewer physics schemes with stochastic streamfunction perturbations is promising StochasticThree Physics+Perturbed ParametersStochastic+LimitedPhysics+Parameters - More recent results (Berner) show advantages of Stochastic+Multi-physics

  11. Concurrent to R&D: • AFWA proceeded with deploying a simple multi-physics ensemble • Uses more LBC sources than used in R&D • Pushed to higher resolution • Deployed in sparsely observed locations • Developed operational products aimed at USAF users • Include back-end models

  12. 4 km SWA ensemble12 Apr 2010 Iraq IR satellite loop from 09-18Z 50 knot wind gust probability at 19Z 58 knots observed at 1911Z Black contour=where individual ensemble member forecasted 40 knots sustained Lightning probability loop from 09-18Z 00Z 12 Apr 2010 ensemble run Black contours=where individual ensemble member forecasted intense lightning

  13. 4 km SWA ensemble27 Apr 2010 Iraq “One thing to take away from this was the success of the Ensembles” 28 OWS storm review for 27 April thunderstorm event • Keys to forecast success • Convective scale ensemble members (4 km) • Direct diagnosis of supercells in WRF (updraft helicity) • Good ensemble agreement (high forecast confidence) 15Z satellite and radar 15 hour supercell forecast

  14. 10 June 2010 Wake low from MCS—24 hour forecast

  15. Closing thoughts • Logistical need to get away from multiple models and physics schemes • User needs often introduce constraints that are not recognized in current research; relocatable/rapidly deployable domains is an example • Observation error desirable to include in verification, but generally applicable values are not available • Multiple physics shows a clear advantage over doing nothing else • Using obs in perturbations and/or via data assimilation shows advantage in very short ranges; LBC a strong constraint • Stochastic perturbations (backscatter) showing promise; may gain from combining with other methods for PBL forecasting

  16. Ongoing: C&V probability • Ceiling and Visibility are key forecast parameters for USAF ops • Evaluating potential for probabilistic predictions from current ensembles • Evaluating potential for ensemble augmentation techniques to fill out pdf of visibility

More Related