Accurate forecasting of heavy precipitation in central China is still a challenge, within which a key issue is our still incomplete understanding of the convective systems (CSs) responsible for such events. In this study, through use of an iterative rain-cell tracking algorithm, the macroscale characteristics (scale, intensity, duration, etc.) of the CSs that produced 595 short-term heavy precipitation events in Hunan Province, central China, are quantitatively analyzed, based on radar reflectivity, echo top, and rainfall observations at 1-km and 6-min intervals in April–September of 2016–2018. The results show that CSs present significant seasonal and diurnal features. Spring CSs usually cover a larger echo area with stronger convective cores and thus generate more precipitation than summer CSs, though summer CSs develop more vigorously and frequently. CSs initiated at 1400–1600 local time are characterized by the strongest convection and a smaller spatiotemporal scale, causing violent and transient showers with typical areal precipitation of 0.5–1 mm km⁻², but less total precipitation. Further analyses of the relationships among the scale, intensity, duration, and total precipitation of CSs reveal that the convective intensity is linearly correlated to the spatiotemporal scale of CSs, with the duration increasing on average by 0.0372 h dBZ⁻¹; the echo area is significantly correlated to the total precipitation, and the duration and rainfall amount are connected with the area expansion rate (AER) of CSs: when the AER exceeds 50%, CSs expand rapidly with increasing total precipitation, but the duration is shorter. These findings provide a helpful reference for the forecasting of short-term heavy precipitation induced by CSs in central China.

In this study, driven by ERA5 reanalysis data, the Weather Research and Forecasting (WRF) version 4.0 was used to investigate the eastward propagation of the Madden–Julian oscillation (MJO) in the tropical atmosphere during December–February (DJF) of 2007/2008. The experiment with 11 cumulus parameterization schemes respectively shows that the Grell 3D scheme is one of several worse ones in describing MJO activities. In addition, still by use of the Grell 3D scheme, four nudging assimilation experiments for water vapor in all model vertical layers (Ndg_all), lower layers (Ndg_low), middle layers (Ndg_mid), and upper layers (Ndg_upp) were conducted. It is found that when the water vapor in the model approaches to the observed value, the model performance for MJO activities is improved greatly. Among the four nudging simulations, Ndg_all certainly performs best. Although Ndg_mid is important for the MJO-filtered profiles related to moisture, Ndg_low and Ndg_upp exhibit superiority to Ndg_mid in simulating MJO eastward propagation. Ndg_low has advantages when MJO features are represented by zonal wind at 850 hPa and precipitation because the lower-level MJO-filtered moisture is conducive to the existence of lower-level condensational heating to the east of the MJO convective center. Ndg_upp performs better when describing the MJO eastward propagation features by outgoing longwave radiation (OLR) since it can capture the moisture and cloud top temperature of deep convection associated with MJO, as well as front Walker cell. These results suggest that the lower-level moisture is more important in regulating the MJO eastward propagation, and the observed maximum MJO-filtered moisture in the middle troposphere might be a phenomenon accompanying the MJO deep convection, but not a factor controlling its eastward propagation.