Atmospheric fronts are crucial for weather forecasting as they are often associated with intense weather phenomena. Current machine learning methods for identifying atmospheric fronts directly input multiple meteorological element data into networks for training and prediction, without considering feature conflicts among these elements caused by special weather conditions or geographical factors. Such conflicts can increase training difficulty and reduce identification accuracy. To address this, this study proposes the adaptive fuser (AF)‐UNet model, which integrates an adaptive fusion module (AFM) to resolve feature conflicts in multielement meteorological data and employs a dual loss function during training to mitigate severe class imbalance in labels. Experimental results demonstrate that AF‐UNet outperforms existing methods in accuracy and robustness, achieving a critical success index (CSI) of 69.8%, probability of detection of 79.8%, and accuracy of 84.6%. Our work tackles the challenge of feature conflicts in multielement meteorological data that impair network training and recognition accuracy in current machine learning approaches, providing an effective solution to significantly improve atmospheric front identification.

Rapidly developing convection (RDC) serves as a precursor to severe weather events. Timely detection of RDC (DRDC) through satellite remote sensing technology is critical for advancing nowcasting capabilities. With the operational deployment of the rapid scanning mode on China’s FengYun‐2F (FY‐2F) meteorological satellite, it provides multichannel rapid scanning data with a temporal resolution of 6 min and a spatial resolution of 5 km, making it possible to detect RDC operationally using satellites. Leveraging the characteristics of FY‐2F multichannel data, this study proposes a novel method for RDC detection. The method employs multichannel identification of the cloud‐top cooling rate (CTC) combined with a three‐test CTC filter. After filtering, local brightness temperature (Tb) minima are identified for each RDC, and a flood‐fill algorithm is applied to surrounding pixels to delineate rapidly developing convective bodies. This study introduces the innovative application of multichannel RDC identification and a CTC filter that incorporates three tests. Experimental validation using August 2014 scanning data (106° E–113° E, 27° N–31° N) demonstrates robust performance, with a probability of detection (POD), false alarm ratio (FAR), and critical success index (CSI) of 0.88, 0.16, and 0.76, respectively, confirming the method’s accuracy in identifying RDC.

On September 19, 2023, a rare autumn tornado outbreak, consisting of three successive tornadoes (including one EF3), occurred in northern Jiangsu, China, causing significant damage. This study presents a comprehensive analysis of this event using multisource data, including Doppler radar, ERA5 reanalysis, and surface observations. We find that pretornadic mesoscale vortices were triggered and intensified during the eastward movement of the Jianghuai cyclone, along low‐level shear lines on the warm side of a quasi‐stationary mesoscale front. The synoptic environment was characterized by an anomalously strong and persistent subtropical high, which provided record‐breaking moisture (precipitable water >66 mm) and instability (CAPE >2200 J kg −1 ), coupled with extreme low‐level wind shear (>18 m s −1 in 0–1 km). These conditions created a classic yet extreme setting for tornadic supercells. At the storm scale, the tornadoes formed during the merging and intensification of multicell thunderstorms. Radar analysis revealed that the two strongest tornadoes (EF2 in Nancai and EF3 in Funing) exhibited clear supercell characteristics, including hook echoes and tornado vortex signatures (TVSs). Tornadogenesis in both cases coincided with the peak values of low‐level differential velocity (LLDV exceeding 40 m s −1 ). The Funing EF3 tornado was further confirmed by a prominent tornado debris signature (TDS) in dual‐polarization radar data. A key advance presented here is the identification of a hybrid tornadogenesis mechanism. While the storms displayed classic supercell dynamics (tilting and stretching of vertical vorticity), the process was significantly enhanced by the amplification of pre‐existing horizontal vorticity along successive, evolving meso‐β‐scale convergence centers. This mechanism, which shares characteristics with both supercellular and landspout processes, explains the sequential development of this “tornado cluster.” Our results underscore that in high‐shear, high‐instability environments, detailed analysis of mesoscale boundaries are crucial for anticipating tornadogenesis. This study enhances the understanding of autumn tornado outbreaks in eastern China and provides insights for improving monitoring and warning strategies for such rare but high‐impact events.

Utilizing ERA5 monthly averaged data from 1981 to 2022, ERA5 hourly data from 2018 to 2022, and observed hourly precipitation data at the airports of the Civil Aviation Flight University of China (CAFUC), the variation characteristics of short‐duration heavy rainfall (SHR) and the environmental conditions of each synoptic category were studied by classifying large‐scale synoptic circulations related to SHR. The results indicated the following: Synoptic circulation was classified into four categories for the 71 SHR selected based on the criteria: vortex (14 cases), low‐pressure trough (29 cases), subtropical high (26 cases), and other categories (2 cases). The average precipitation amount (PA) and precipitation frequency (PF) increased from 00:00 on the day of the SHR to 00:00 the next day, peaking at night, with the largest values in the vortex category and peaking earlier than the other two. The average precipitation intensity (PI) was largest in the subtropical high (9.3 mm/h), smallest in the low‐pressure trough (6.8 mm/h), and closed in the daytime and nighttime in the vortex. For the four thresholds of ≥0.1, ≥10, ≥20, and ≥50 mm/h, the average PA and PF of nighttime were larger than the daytime, the vortex was the largest, and the peak time was shifted forward between the thresholds, while shifted backward in the low‐pressure trough. Under the three synoptic categories, the increase in PA was mainly due to the contribution of increased PF, whereas the impact of PI was relatively small. Before the SHR, the vortex category had better moisture and dynamic uplift conditions than the low‐pressure trough and subtropical high categories, reflecting the dynamic nature of the vortex. Atmospheric conditions were the greatest at the subtropical high, indicating the thermal nature of the subtropical high. A K index greater than 31 was necessary for an SHR event at the CAFUC airports.

Extensive research has demonstrated that during the mature phase of El Niño, the anomalous southerly wind induced by the western North Pacific anomalous anticyclone (WNPAC) weakens the South China Sea winter cold tongue (SCSCT). This study further examines the decadal effects of the Pacific Decadal Oscillation (PDO) on the weakening of the SCSCT during the mature phase of El Niño. Results suggest that, compared to the positive PDO phase, the weakening of the SCSCT was more pronounced during the negative PDO phase. Upper mixed layer heat budget analysis unveils that the net surface heat flux term dominated the abnormal warming in the northern part of the cold tongue, while the abnormal warming in the southern part is ascribed to both the horizontal heat advection and net surface heat flux terms. The differences between the negative and positive PDO phases (N‐P) were that cooler (warmer) sea surface temperature anomalies (SSTAs) in the tropical eastern (central) Pacific induced weaker (stronger) WNPAC (Southeastern Tropical Indian Ocean anomalous anticyclone [SEIOAC]) through anomalous Walker circulation. Subsequently, for the N‐P, the anomalous Hadley circulation between the SEIO and the SCS led to cross‐equatorial southerly wind anomalies, which were the fundamental cause of the anomalous warming of the SCSCT. Combining the effects of both the tropical Indian Ocean and Pacific Ocean to explain the decadal differences in the interannual relationship between El Niño–Southern Oscillation (ENSO) and the SCSCT, this work provides new insights for predicting the climate variability in East and Southeast Asia.

The Tarim basin located in the western northwest China is the largest inland basin in the world and the core component of the arid region of the Central Asia. In recent years, the frequency of heavy rain and extreme precipitation events in the Tarim basin has increased substantially. It is imperative to enhance the understanding of the triggering and development mechanisms of heavy rain in this region. This study analyzes the atmospheric instability and triggering mechanisms of a short‐duration local heavy rainstorm event in the Tarim basin based on the moist potential vorticity (MPV) criterion. The results show that convective instability positively contributes to precipitation generation, while conditional symmetric instability assumes a pivotal role in its intensification. The cold air from northern short‐wave trough with moist air from southern short‐wave trough converge with the warm and dry air in the low layer of the Tarim basin, triggering the rapid release of convective available potential energy and the initiation of precipitation. Subsequently, the release of condensation latent heat and moist baroclinicity amplify the conditional symmetric instability in the lower atmosphere. The westerly wind over the mountains and the easterly wind in lower layers of the basin form intense vertical shear of horizontal wind. This vertical shear triggers the release of conditional symmetric instability energy, which stimulates further development of vertical motion and causes a sudden increase of precipitation. Although there are no obvious convective cloud systems, the rapid and intensive release of instability energy results in strong negative temperature variation, humidification, and wind shear, and their centers are overlapped. Such synoptic configurations are responsible for the local heavy rainstorm. The findings of this study contribute to a better understanding on the causes of short‐duration local rainstorms in arid areas and offer valuable insights for improving rainstorms predictability in the region.

Accurate forecasting of local heavy rainfall distribution over Northwest Vietnam is challenging due to the regions’s complex rainfall processes and topography. In this study, effect of the Hoang Lien Son (HLS) Mountain range on synoptic‐dynamic processes associated with the heavy rainfall were analyzed to quantify the relationship with the local rainfall distribution. The self‐organizing map (SOM) was applied to the normalized 850 hPa wind and geopotential height data from all heavy rainfall events from 1983 to 2020 to classify the patterns into four groups (nodes). The two largest groups are characterized by the southward movement of low‐level pressure, triggered by the deepening of an upper‐level trough in high‐latitude regions. The second group illustrates the combined effects of a cold surge and quasi‐geostrophic lifting associated with the upper‐level trough along the subtropical jet stream. The third group presents the deepening of the monsoon trough and the westward propagation of tropical disturbances. Further analysis on vertical sections of zonal and vertical wind reveals that the high mountain ranges play a vital role in enhancing upward motion, which induces heavy rainfall in the western side of the mountain ranges. These results are important for meteorologists to assess the likelihood of local heavy rainfall and provide a basis for evaluating the performance of numerical models in simulating heavy rainfall in regions with complex topography.

Thermodynamic profiles measured from the ground‐based microwave radiometer (MWR) have been widely utilized to enrich the atmospheric vertical observations, which however, are characterized by certain retrieval errors. Temperature and humidity retrievals from an in‐situ MWR are assessed against the corresponding radiosonde observations. Calibration experiments are conducted using methods of Kalman filter (KF), random forest (RF), neural network (NN), and time‐involved NN (TiNN), which are subsequently examined across various metrics. The significant errors of MWR retrievals vary with heights and are reduced to certain extents, via these calibration models. KF and RF both offer considerable calibrations on the MWR retrievals but perform inconsistently across different atmospheric layers. The TiNN model is demonstrated with the optimal performance, producing stable and accurate retrievals with root mean square errors (RMSEs) of no greater than 2.0 K and 20% across all heights for temperature and humidity, respectively. Corresponding improving magnitudes reach upto 60% relative to the raw MWR. Such superiority of the TiNN model demonstrates potential advantages of integrating associated auxiliary features, such as time tags into calibration models. The challenges in retrieving relative humidity compared to temperature might be linked to the lower signal‐to‐noise ratios in the moisture retrievals and greater sensitivity to environmental conditions due to indirect and complex nature of the measurements.

This study focuses on modeling and predicting extreme rainfall based on data from the Southern Highlands region, the critical for rain-fed agriculture in Tanzania. Analyzing 31 years of annual maximum rainfall data spanning from 1990 to 2020, the Generalized Extreme Value (GEV) model proved to be the best for modeling extreme rainfall in all stations. Three estimation methods–L-moments, maximum likelihood estimation (MLE), and Bayesian Markov chain Monte Carlo (MCMC)–were employed to estimate GEV parameters and future return levels. The Bayesian MCMC approach demonstrated superior performance by incorporating noninformative priors to ensure that the prior information had minimal influence on the analysis, allowing the observed data to play a dominant role in shaping the posterior distribution. Furthermore, return levels for various future periods were estimated, providing guidance for flood protection measures and infrastructure design. Trend analysis using p value, Kendall’s tau, and Sen’s slope indicated no statistically significant trends in rainfall patterns, although a weak positive trend in extreme rainfall events was observed, suggesting a gradual and modest increase over time. Overall, the study contributes valuable insights into extreme rainfall patterns and underscores the importance of L-moments in identifying the best fit distribution and Bayesian MCMC methodology for accurate parameter estimation and prediction, enabling effective measures and infrastructure planning in the region.

This study analyzed the deviation of the gust factor (GF) method in gust estimation and the differences in the vertical distribution of upper‐level wind speed and temperature when varying deviations occurred using observed data from 11 stations in central‐eastern China from January 2017 to December 2020. A unified upper‐level gust impact model was developed through multiple regression (GF‐L) and machine learning (GF‐M) methods based on data from these stations to improve gust estimation accuracy. The effectiveness of the GF, GF‐L, and GF‐M methods was compared using data from January 2021 to October 2022. Results showed that integrating the upper‐level gust impact model into the GF method reduced underestimation and improved accuracy, with the GF‐M method showing the most significant improvement. The GF‐L method reduced the root mean square error by 5.1% on average compared to the GF method, while the GF‐M method achieved a 9.2% reduction. The study confirmed the applicability of the unified upper‐level gust impact model across different stations, with the GF‐M method demonstrating better results.