Initial Perturbation Scheme for Storm-Scale Ensemble Forecast: Evolution of Initial Errors in a Convection System

This paper presents the growth dynamics of initial errors and predictability in convection systems, thus, to provide a significant reference for designing initial perturbation method suitable to storm-scale ensemble forecasting (SSEF), by simulating a typical supercell storm using the Weather Research and Forecasting (WRF) model. The instabilities and releases of latent heat are the dominant factors to affect error growth and spatial distribution. The ability of single deterministic convective forecast is limited by the rapid growth of initial errors in a developing convection system. Error propagation, in the form of wave, is able to inspire quadratic errors in unstable perimeter regions. The new errors generated by original forecast errors are not predictable, and thus restrict system predictability. Ensemble forecast (EF) technique is considered to overcome the limitations and improve the convective forecasts. Whereas, the initial perturbation schemes in synoptic-scale EF are not suitable to SSEF because of different mechanisms in error evolutions. Lack of key convection-dynamic factors in an initial perturbation scheme, the nonlinear effects in convection will result in the rapid increase of correlation coefficients among ensemble members, and become a serious obstacle to improve the ability of SSEF in forecast uncertainty. Therefore, a practical initial perturbation method based on a convection vortex-relocation technique, where two types of instability conditions relative to convection are regarded as perturbation factors, is designed. Preliminary results from ensemble forecast experiments are in general agreement with viewpoints presented here.