Jiangsu is a province in China and has the highest frequency of tornado occurrences. Studying the meteorological background and mechanisms of tornado formation is crucial for predicting tornado events and preventing the resulting disasters. This paper analyzed the meteorological background, instability mechanisms, and lifting conditions of the two Enhanced Fujita Scale level 2 (EF2) and above tornadoes that occurred in southern Jiangsu on 14 May 2021 (“5.14”) and 6 July 2020 (“7.06”) using ERA5 reanalysis data. Detailed analyses of the internal structure of tornado storms were conducted using Changzhou and Qingpu radar data. The results showed that (1) both tornadoes occurred in warm and moist areas ahead of upper-level troughs with significant dry air transport following the cold troughs. The continuous strengthening of low-level warm and moist advection was crucial in maintaining potential instability and triggering tornado vortices. The 14 May tornado formed within a low-level shear line and a warm area of a surface trough, while the 6 July tornado occurred at the end of a low-level jet stream, north of the eastern section of a quasi-stationary front. (2) The convective available potential energy (CAPE) and K indices for both tornado processes were very close (391 for “5.14” and 378 for “7.06”), with the lifting condensation level (LCL) near the ground. The “5.14” showed greater instability and more favorable thermodynamic conditions, with deep southwesterly jets at the mid-level shear line producing rotation under strong convergent action (convergence center value exceeding −1 $\times {10}^{-4}$${\mathrm{s}}^{-1}$). In contrast, the “7.06” was driven by super-low-level jet stream pulsations and wind direction convergence under the influence of the Meiyu Front (convergence center value exceeding −1.5 $\times {10}^{-4}$ ${\mathrm{s}}^{-1}$), resulting in intense lifting and vertical vorticity triggered by a surface convergence line. (3) The “5.14” tornado process involved a supercell storm over a surface dry line experiencing tilting due to strong vertical wind shear, which led to the formation of smaller cyclonic vortices near a hook echo that developed into a tornado. The “7.06” developed on a bow echo structure within a mesoscale convective system formed over the Meiyu Front, where dry air subsidence, entrainment, and convergence of the southeast jet stream triggered a “miniature” supercell. The relevant research results provide a reference for the prediction and early warning of tornadoes.

Hourly precipitation is a standard variable for hydrological applications, flood warning, and quantitative assessment of precipitation-related hazards. Weather radar provides high-frequency observations, but these sub-hourly measurements must be linked to hourly rain-gauge accumulations for radar-based quantitative precipitation estimation (QPE). This study develops a machine-learning-based hourly radar QPE framework using intra-hour temporal features from dual-polarization observations. Each hourly gauge accumulation was matched with ten consecutive 6 min radar observations. Six temporal features were derived from dual-polarization observations, including the mean, sum, standard deviation, linear trend, maximum, and skewness. The results show clear rainfall regime dependence in the contribution of polarimetric variables. Evaluation of machine-learning models confirms the value of intra-hour temporal features, with the best-performing model achieving an RMSE of 3.96 mm h⁻¹, compared with 6.42 mm h⁻¹ for the linear baseline. The experiments further show that the complete temporal feature set improves QPE performance. Diagnostic evaluation also indicates that model bias is sensitive to the representation of heavy rainfall samples. These findings suggest that ML-based radar QPE should emphasize physically meaningful intra-hour polarimetric features and rainfall-imbalance-aware sampling.