Impact of Vertical Velocity Profile Shape in Lightning Data Assimilation

To improve the skill of short-term forecasting for severe convection, this study addresses the key challenge of constructing appropriate convective vertical dynamical structures within lightning data assimilation (LDA). We proposed a LDA observation operator based on Bayesian method and compared its effectiveness against a traditional fixed-profile approach. The fixed-profile experiment, which provided only spatially uniform upward motion, functions similarly to "moisture assimilation", primarily influencing forecasts by gradually enhancing moisture convergence. In contrast, the core advantage of the Bayesian method lied in its ability to retrieve heterogeneous vertical velocity profiles that evolve with convective intensity and incorporate differing downdraft locations. This successfully constructed a more realistic secondary circulation structure. The coupling of these differing profiled facilitated the rapid formation of graupel particles, altered the distributions of the wind and moisture fields, and produced an effect akin to "graupel assimilation". Quantitative verification scores indicated that the Bayesian method yields particularly significant improvement in forecasting skill for heavy precipitation (≥30 mm). Furthermore, both LDA experiments retained the advantage of vertical velocity adjustment, effectively suppressing spurious convective activity through modifications to the dynamical field.