Convection initiation (CI) and the subsequent upscale convective growth (UCG) at the coast of South China in a warm-sector heavy rainfall event are shown to be closely linked to a varying marine boundary-layer jet (MBLJ) over the northern South China Sea (NSCS). To elucidate the dynamic and thermodynamic role of the MBLJ in CI and UCG, we conducted and analyzed convection-permitting numerical simulations and observations. Compared to radar observations, the simulations capture CI locations and the following southwest–northeast-oriented convection development. The nocturnal MBLJ peaks at 950 hPa and significantly intensifies with turning from southwesterly to nearly southerly by inertial oscillation. The strengthened MBLJ promotes mesoscale ascent on its northwestern edge and terminus where enhanced convergence zones occur. Located directly downstream of the MBLJ, the coastal CI and UCG are dynamically supported by mesoscale ascent. From a thermodynamic perspective, a warm-moist tongue over the NSCS is strengthened by the MBLJ-driven mesoscale ascent as well as by a high sea-surface temperature. The warm moist tongue further extends northeastward by horizontal transport and arrives at the coast where CI and UCG occur. Near the CI location, rapid development of a low-level saturated layer is mainly attributed to the mesoscale ascent and low-level moistening associated with the MBLJ. In addition, subsequent CI happens on either side of the original CI along the coast due to the delay of low-level moistening, which partly contributes to linear convective growth. Furthermore, ensemble simulations confirm that a stronger MBLJ is more favorable to CI and UCG near the coast.

Deep moist convection often occurs in South China with severe weather. Accurate and timely prediction of deep moist convection initiation (CI) is important in providing an earlier warning than that provided by rapidly developing thunderstorms, which may help save lives and property. Therefore, CIs in South China have received increasing scientific attention in recent decades (e.g., Du & Chen, 2019a;Wang et al., 2014;Wu & Luo, 2016). However, given the widely varying possible forcing and complex underlying surface effects in South China, simulating or predicting when and where the individual CI is precisely triggered remains a challenge. Many previous efforts have been made to clarify the mechanisms of CI. Dynamic lifting in the boundary layer driven by synoptic weather systems is a strong favorable forcing for CI, such as shear lines/trough lines (e.g.,

Convectively generated cold pools play a crucial role in the convection initiation and development, but observations of their vertical structure are insufficient. In this study, quantitative vertical evolutions of cold pools during a high-impact heavy rainfall event near the south coast of China were examined using observations from a 356-m high Shenzhen Met-Tower, and their potential impacts on heavy rainfall were further discussed through high-resolution surface station network and radar observations. On 11 April 2019, heavy precipitation occurred near metropolitan Shenzhen, lasting for 50 min at its southern downtown and resulting in 11 deaths. During this event, a shallow cold pool was first observed by the tower and yielded a long-lasting cooling of 2.6 K. Approximately one hour later, another deeper cold pool accompanied by a squall line was added from the west. This addition resulted in a more abrupt and intense surface temperature deficit (5.1 K) and stronger gusty winds (23 m/s). When the two cold pools collided near Shenzhen, the low-level winds converged at their intersection, dynamically enhancing the heavy-rain-producing squall line. Moreover, the collision of the two cold pools reduced the temperature gradient at the northern edge of the merged cold pool, which could inhibit development of the squall line. The area south of the squall line became a relatively favorable environment for convection initiation, given the warm and moist oceanic environment. Consequently, the squall line turned northeast–southwest, forming a training line mode that was nearly parallel to the eastward movement. This training line mode prolonged the precipitation duration in the southern downtown area.