The Air Force Weather Agency's mesoscale ensemble: scientific description and recent results

We provide a scientific description and verification of the WRF-based mesoscale ensemble run at the U.S. Air Force Weather Agency (AFWA). The 10 members are AFWA's contribution to the Joint Mesoscale Ensemble (JME), which is in turn the mesoscale component of the Joint Ensemble Forecast System (JEFS). An October 2006 test period on a 45/15-km nested domain provides a basis for development and evaluation of different perturbation methods. All WRF ensembles use NCEP GFS ensemble lateral boundary conditions. Model perturbation methods included are: (1) multiple physics suites, (2) land-surface table perturbations, (3) stochastic tendency perturbations, and (4) stochastic stream-function perturbations. Initial-condition (IC) perturbations considered are (1) simple down-scaled GFS ensemble ICs, (2) perturbed observations in 10 independent WRF-3DVar cycles, (3) the ensemble-transform Kalman filter, and (4) the ensemble Kalman filter. We will describe each of the perturbation methods above, present verification statistics from the October 2006 test period, and discuss the pros and cons of the various ensemble implementations possible within the system.