Multiscale Characteristics of Convection-Allowing Ensemble Perturbation Evolution in Warm Season Precipitation Forecasts

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 4 February 2014
Hall C3 (The Georgia World Congress Center )
Aaron T. Johnson, CAPS/Univ. of Oklahoma, Norman, OK; and X. Wang, M. Xue, F. Kong, G. Zhao, Y. Wang, K. W. Thomas, K. Brewster, and J. Gao

Convection-allowing precipitation forecast perturbations resulting from various initial condition (IC) and physics perturbations, are examined for two forecast cases and systematically over 34 forecasts out to 30h. Different sources of perturbations and various spatial scales of IC perturbations are compared on different scales using Harr Wavelet scale decomposition of precipitation forecast perturbations.

For a case where the background forecast during the forecast period was driven primarily by a synoptic scale baroclinic disturbance, small scale IC perturbations resulted in little forecast perturbation energy on medium and large scales, compared to larger scale IC and physics (LGPH) perturbations after the first few forecast hours. For a case where the background forecast had ongoing convection at the initial time which grew upscale into a Mesoscale Convective System (MCS) which then influenced the subsequent convection, small scale IC perturbations resulted in similar forecast perturbation energy as LGPH perturbations on all scales after about 12 h. Averaged over 34 forecasts, after a few hours the impact of small scale IC perturbations was at least as large as the LGPH perturbations on the small scales and was comparable to, but smaller than, the LGPH perturbations on the medium scales. The spatial structure of small scale IC perturbations also affected the evolution of (mainly medium scale) forecast perturbations.

These results motivate further research on how to sample errors of various sources and spatial scales in the optimal design of convection-allowing ensembles. Preliminary results of such research, using a nested (in both time and space) ensemble data assimilation method, will be briefly discussed.