Convective-scale ensemble prediction experiments under different large-scale forcing with consideration of uncertainties in initial and lateral boundary condition

As the first effort to develop a self-adaptive convective scale ensemble prediction system, this study examines two different initial perturbation methods for the simulation of two different types of heavy rain case and their impacts on precipitation prediction under the same lateral boundary perturbations. A suite of ensemble experiments has been conducted: In the first two experiments, both initial perturbation (ICs, include ETKF and DOWN) and lateral boundary perturbation (LBCs) are considered, while in others experiments only ICs or LBCs are considered. Results show that under strong large scale forcing and when both ICs and LBCs are considered, the ETKF method provides more net perturbation energy in the early hours while the DOWN method provides more energy in the later hours of the integration. However, when the large scale forcing is weak, the ETKF method results in more perturbation energy than the DOWN method during the entire forecast period, indicating the importance of consistency between initial and lateral perturbations. Note that initial perturbation is always dominant at the first several hours of integration and LBCs is dominant in the following hours. The effect of initial perturbations is more distinct under strong large scale forcing than under weak large scale forcing. Meanwhile, the synoptic-scale forcing has significant impacts on quantitative precipitation forecast (QPF) and probability forecast skill.