Abstract:Based on the 24 h surface precipitation observation data from 08:00 BT to 08:00 BT of the next day and upper-air sounding data at 08:00 BT and 20:00 BT during 2016-2023 as well as ERA5 reanalysis data from the European Centre for Medium-Range Weather Forecasts (ECMWF), this paper identifies a total of 83 torrential rain days in Beijing during the study period. On this basis, Beijing torrential rain events are classified according to the synoptic situations at 500 hPa, and a statistical analysis is conducted on the specific humidity characteristics at 925 hPa, 850 hPa, 700 hPa, 500 hPa and 200 hPa during the torrential rain days. The results are as follows. The torrential rain events in Beijing are divided into six types: torrential rain induced by Mongolian low vortex and trough type, subtropical high and westerly trough type, Hetao low-vortex type, northeast cold vortex type, tropical cyclone (typhoon) type, and westerly trough and typhoon type. The Hetao low-vortex type and tropical cyclone (typhoon) type tend to produce more extensive extreme torrential rains, while the northeast cold vortex type is dominated by localized torrential rain. In terms of specific humidity, it is markedly higher at all atmospheric levels during torrential rain periods than that during non-torrential rain periods. Among the six types of torrential rain events, tropical cyclone (typhoon) type and Hetao low-vortex type feature the highest specific humidity, which corresponds to the strongest precipitation intensity and the largest rainfall coverage of these two categories. Furthermore, based on hourly precipitation data, the precipitation conditions within one hour before and after the upper-air observation time (08:00 BT and 20:00 BT) are categorized into three grades: torrential rain (≥10 mm), light rain (0.1-9.9 mm) and no precipitation. The specific humidity is the highest at all levels during torrential rain period, followed by that in light rain period, and the lowest specific humidity is found in no precipitation period. In addition, the specific humidity of torrential rain events corresponding to winds from eight directions at different levels is statistically analyzed, and the results indicate that the specific humidity reaches its maximum when the wind direction ranges from 90° to 225° with jet streams, and drops to its minimum when the wind direction is between 270° and 315° accompanied by jet streams. Overall, this study has not only analyzed the specific humidity characteristics below 700 hPa during torrential rain events, but more specially found that the increase in specific humidity at 500 hPa can serve as a significant predictor for torrential rain forecasting.

Abstract:This study conducts a high-resolution numerical simulation and sensitivity experiments on a typical landing convective storm that occurred on 8 August 2022, and investigates the impact of sea surface temperature (SST) variation on the intensity of landing convective precipitation by increasing and decreasing the SST of the Yellow Sea and Bohai Sea. The results show that, compared with the control experiment, the increased SST can enhance the intensity of landing convective precipitation. When SST is raised by 1℃ (2℃), the average accumulated precipitation in the coast increases by 2.6 mm (15.7 mm) and the maximum precipitation increases by 38.2 mm (53.7 mm). When SST is lowered by 1℃ (2℃), the corresponding precipitation decreases by 3.7 mm (8.3 mm), but the maximum precipitation does not show any significant changes. Mechanism analysis indicates that, relative to the control experiment, a rise of 2℃ SST significantly strengthens the coastal boundary layer temperature gradient, wind convergence lifting, as well as the intensity of convective available potential energy and southward-moving cold pools, leading to pronounced reinforcement of the front zone at the cold pool leading edge and dynamic lifting. In contrast, a rise of 1℃ SST does not evidently enhance boundary layer wind convergence and southward-moving cold pool intensity, which results in weaker enhancement of the front zone and dynamic lifting than that under the condition of 2℃ SST increase. When SST decreases by 1℃ and 2℃, the mesoscale environmental conditions and southward-moving cold pool intensity weaken substantially. The front zone at the cold pool leading edge becomes indistinct, and dynamic lifting is greatly weakened or even transformed into downdrafts, ultimately reducing the intensity of landing convective precipitation. The findings of this paper can provide a theoretical references for the intensity forecasting of landing convective precipitation.

Abstract:A rare left-moving supercell generated by storm splitting occurred in central and northern Jiangxi Province from 14:00 BT to 18:00 BT 10 May 2021. Using dual-polarization radar data from the Ji’an CINRAD/SC radar, upper-air and surface observations and ERA5 reanalysis data, this paper analyzes the environmental conditions, structural characteristics, and dynamic mechanisms of this event. The results show that the left-moving supercell was generated in an environment with high convective available potential energy (>3000 J·kg-1) and strong 0-6 km vertical wind shear (>20 m·s-1). The vertical wind shear vector of 3.0-4.9 km exhibited counterclockwise rotation with height, generating significant negative horizontal vorticity, which was a key factor for the selective intensification of the left-moving supercell. During its mature stage, the storm displayed a “dual-vortex” structure at mid-to-upper levels, with a mesoanticyclone located in the leading (northern) sector corresponding to a strong updraft, and a mesocyclone in the trailing (southern) sector associated with the forward-flank downdraft (FFD). Its overall appearance was approximately mirror-symmetrical to the structure of classic right-moving supercell. Dual-polarization parameters demonstrate that the large differential reflectivity factor (ZDR) zone in the lower levels was related to particle sorting under strong vertical wind shear, while the ZDR ring and ZDR column at mid-levels indicated strong rotation and updraft within the meso-anticyclone, respectively. The formation of the mesoanticyclone was attributed to the tilting of negative horizontal vorticity into negative vertical vorticity by the robust updraft. The leftward motion of the storm was closely related to the location of FFD and the Magnus effect resulting from the pressure gradient force induced by the asymmetry between the northern and southern vortices. This study provides a valuable case for understanding the development, structure, and microphysical characteristics of left-moving supercells in China.

Abstract:On 4 August 2024, heavy rainstorm occurred at Wucheng Station in Dezhou, Shandong Province, breaking the historical records for both daily precipitation and maximum hourly precipitation at the station. Utilizing the observation data from automatic weather stations, Doppler weather radar and wind-profiling radar, as well as ERA5 reanalysis data and GDAS data, combined with the HYSPLIT model, this study reveals the multi-scale physical mechanisms that enabled the long-lasting backward propagating convective system and the extreme precipitation it brought. The results indicate that this event occurred under a large-scale circulation pattern characterized by a rearward-tilted trough confronting the subtropical high. Low level jet (LLJ) and ultra-low-level jet (ULLJ) persistently transported abundant water vapor from the South China Sea and the Bay of Bengal to northwestern Shandong, establishing an extremely unstable environment with high energy, high temperature and high humidity within the warm sector ahead of surface front. The cold pool outflow generated by precipitation interacted with the ambient warm, moist airflow, forming a quasi-stationary convergence line. Coupled with temporal and spatial pulsations of the LLJ, this interaction continuously drove new convective cells to propagate southwestward, directly causing the extreme heavy precipitation. The dynamic reconstruction of energy was the key to maintaining the system. Rainfall consumed unstable energy, but the persistent warm, moist advection transported by ULLJ allowed convective available potential energy to be rapidly rebuilt and maintained at its peak. Thus, a positive feedback self-sustaining mechanism of “energy transport-convection consumption-energy reconstruction” was formed, and it was this mechanism that led to the sustained occurrence of the back-propagating heavy precipitation.

Abstract:Based on hourly precipitation data from high-density national and regional stations in Guizhou Province during the flood season (April-September) from 2010 to 2023, as well as Shuttle Radar Topography Mission (SRTM) elevation data and ERA5 reanalysis data, this study systematically investigates the fine-scale spatio-temporal characteristics of rainstorms and their relationships with terrain in Guizhou Province by using multivariate terrain combination analysis and a geographically weighted regression (GWR) model. The results show that rainstorms in Guizhou Province are distributed unevenly with three rainstorm centers. The maximum frequency and intensity of rainstorms are concentrated in the southeastern edge area and extremely heavy rainstorms are scattered. The rainstorm proportion is higher in the northeastern and central-southern parts of Guizhou Province. Rainstorm days show an increasing interannual variability, peaking in 2020-2021. Rainstorms occur mainly in June and July, often accompanied by short-time heavy precipitation and lasting for a long time, which poses high disaster risks. Diurnal variation of precipitation is unimodal and dominated by nocturnal rain (22:00 BT-08:00 BT), and the rainfall intensity often reaches its peaks around 03:00 BT. Short-time heavy precipitation exhibits a pronounced nocturnal peak, and the nighttime precipitation is roughly twice the daytime precipitation, mainly concentrated in the west and south of Guizhou Province and featured with eastward propagation. The Guizhou Province rainstorm-prone areas correspond to the topographic uplift zones of the Beipan River, Leigong Mountain, and Fanjing Mountain, reflecting terrain-wind coupling. Single terrain factors are weak to be directly correlated to precipitation. However, through the analysis of multiple terrain combinations, the explanatory power of precipitation distribution has been enhanced. Slope, aspect, elevation, and relief exert nonlinear effects. Rainstorm stations are the most in south aspects. Rainstorm frequency increases when slope is <20°, elevation ranges in 800-1000 m, and relief is <200 m, but decreases beyond these thresholds. The combined analysis indicates that the steep/moderate slope, south aspect, medium elevation, large/medium relief, and windward slope has the strongest triggering effect. The GWR model diagnosis suggests that there is obvious spatial heterogeneity in the influence of topographic factors. Elevation is the dominant factor in most areas of Guizhou Province, and the local influence of relief, slope and aspect is strong, mostly concentrated in the vicinity of large relief or near rivers and mountains.

Abstract:In view of the characteristics of X-band phased array radar, which offers high spatio-temporal resolution but limited observation coverage and various errors that affect the reliability of data, this study proposes a method of high-quality data fusion between X-band phased array radar and S-band operational radar. First, the optical flow method is applied to extrapolate the motion trends of S-band radar data, enhancing its temporal resolution from 6 min to 1.5 min, synchronized with X-band radar. Subsequently, the optimal interpolation algorithm and pyramid transform algorithm are employed for the spatial fusion of dual-band radar data, and the corresponding products SXnet-O and SXnet-K are obtained. Statistical analysis based on nearly 5000 radar volume scans from the Guangdong-Hong Kong-Macao Greater Bay Area radar network from May to June 2022 demonstrates that the time-matching results extrapolated by the optical flow method outperform traditional persistence forecasting in terms of forecasting scores and correlation coefficients in the temporal dimension. In the spatial dimension, the fused data effectively compensate for the detection blind zones of a single radar, increasing the low-altitude (500 m) echo coverage by approximately 2.5 times compared to the S-band radar. Comparative results indicate that SXnet-O significantly outperforms SXnet-K in system bias control (92% of deviations within ±2 dB), parameter consistency, and boundary stability. Compared with the single-band network data, SXnet-O greatly reduces the mean deviation from the S-band reference. Moreover, the structural precision of X-band and the observational stability of S-band get effectively integrated.

Abstract:Weather radar systems operating continuously over long periods are often affected by equipment failures, external interference, and other factors, which may lead to abnormal echoes and limit their effectiveness in monitoring, warning, and forecasting severe weather.In this study, two types of abnormal echoes, radial and annular echoes, were manually selected and labeled from historical radar mosaic data. A weather radar abnormal echo dataset for radar mosaics, WRAED, containing 20 000 images, was then constructed using data augmentation techniques.An improved model was developed based on the semantic segmentation model DeepLabV3+. A simplified ResNet50 architecture was adopted as the backbone network to reduce computational cost. A simple attention mechanism, SimAM, was added to the backbone feature extraction layers to help the model focus on important echo features. In addition, an intermediate layer was introduced into the decoder to incorporate more detailed and contextual background information.The improved model, named DeepLab-ARER, was designed for abnormal echo recognition in radar mosaics. Experimental results show that DeepLab-ARER achieves good performance in identifying abnormal echoes, with a mean pixel accuracy of 96.75% and a mean intersection over union of 93.95%, representing a clear improvement over DeepLabV3+.The DeepLab-ARER model provides effective technical support for the automatic identification of abnormal echoes in radar mosaic.

Abstract:To address the issue that existing radar-based precipitation extrapolation techniques can not adequately characterize the evolution of thunderstorms and cause systematic biases in heavy precipitation forecasts, this paper proposes a heavy precipitation extrapolation and correction method which integrates the evolution features of thunderstorms. Based on the storm cell identification and tracking (SCIT) algorithm, a thunderstorm development trend discrimination model is constructed for different thunderstorms by combining key factors including strong gradient zones, newly generated cells, and echo area changes. By utilizing radar reflectivity factor, optical flow vectors and dynamic Z-I relationship, we implement evolutionary constraints and corrections during the extrapolating process of heavy precipitation. Verification on the precipitation forecasts in the main flood season of 2025 shows that in the 0-1 h forecasts, the TS scores of the corrected products at the level of ≥20 mm·h-1 and ≥50 mm·h-1 are both higher than those of the optical flow system, with the advantage increasing with precipitation intensity. In the 1-2 h forecasts, the TS scores at ≥20 mm·h-1 and ≥50 mm·h-1 are still significantly improved compared with those of the optical system, but the score at ≥50 mm·h-1 is only 0.004, indicating very limited operational guidance value. Further analysis of typical cases of heavy precipitation demonstrates that the correction technique can capture precipitation characteristics of convective systems rapidly moving and developing to a greater extent and effectively alleviate the problems of underestimating intensity and coverage that exist inherently in traditional extrapolation methods.

Abstract:The accuracy of typhoon location and intensity estimation, as well as the accuracy of typhoon track and intensity forecasts, is verified for the 26 named typhoons over the Northwest Pacific and the South China Sea in 2024. The forecast methods that are verified include subjective forecast methods, numerical weather prediction (NWP) models, statistical forecasting methods, dynamical-statistical forecasting methods, multi-model ensemble forecasting methods, and AI-based weather prediction (AIWP) models. The results show that the overall mean error of typhoon location estimation by the five official typhoon forecasting agencies was 22.3 km, and the overall mean absolute error of typhoon intensity estimation was 2.5 m·s-1. The performance of 48 h and 72 h typhoon track forecasts by the National Meteorological Centre (NMC) of China Meteorological Administration reached the highest level in records. Among global NWP models, ECMWF-IFS exhibited the best performance in typhoon track forecasts. The AIWP model “Fengqing” achieved track forecast performance comparable to ECMWF-IFS within 96 h lead time, and slightly outperformed ECMWF-IFS with 60 h to 96 h in advance. The intensity forecast performance of NWP models and dynamical-statistical forecasting methods was superior to that of ensemble prediction systems and AIWP models. Specifically, AIWP models systematically underestimated the typhoon intensity.

Abstract:In June 2026, the polar vortex in the Northern Hemisphere exhibited a dipole distribution with the primary vortex located over the Arctic Ocean near the pole, showing stronger-than-normal intensity. The Eurasian mid-high latitudes featured a two-trough and one-ridge circulation pattern, and the average geopotential height ridge from West Siberia to northern Xinjiang was abnormally strong. The western Pacific subtropical high was zonally distributed, with greater intensity and a more westward and northward position than normal. The national average temperature was 20.7℃, 0.3℃ higher than the normal value. The national average precipitation was 99 mm, 3.7% below the climatological average. However, cold vortices behaved actively in North China and Northeast China resulting in frequent convective activities. In southern China, heavy precipitation processes occurred frequently, with highly overlapping affected-areas and locally extreme precipitation. During this month, there were five heavy precipitation events and six severe convection events. Regions such as the middle and lower reaches of the Yangtze River, South China, and eastern Southwest China experienced regional heavy precipitation and local extreme precipitation. Thunderstorm, gale and hail events occurred more in Northeast China, North China and the Huanghuai Region. Two typhoons were generated over the western North Pacific and the South China Sea, but neither made landfall in China. In addition, this paper briefly analyzes the evolution of circulation patterns, causes and forecasting difficulties of the extreme precipitation in South China from 14 to 18 June and the first spell of heavy precipitation in the Meiyu season over the middle and lower reaches of the Yangtze River from 19 to 22 June.