Yunying Li

Local short-duration heavy-rainfall warning requires early assessment of hazardous rainfall within a specified neighborhood. Geostationary meteorological satellites provide frequent, spatially continuous observations of cloud systems and storm evolution, but only indirectly reflect surface rainfall. Rain gauges measure surface accumulation directly, but only at irregular locations. Deep learning fusion methods commonly convert gauge observations to gridded representations before combining them with satellite observations. This treatment merges gauge measurements and local gauge support into a single representation, limiting how local surface conditions inform warning decisions and increasing the likelihood that local heavy-rainfall risk is underestimated or overestimated. We present PLUME (Precipitation Nowcasting with Late-Stage Residual Updates from Rain-Gauge Measurements for Short-Duration Heavy-Rainfall Events), a staged fusion model with gridded and point-origin gauge pathways. PLUME first combines a gauge-derived gridded rainfall background with FY-4A AGRI observations to form a base rainfall representation. The point-origin pathway uses local gauge features and observation support to update this representation through signed residuals, which the station decoder then maps to neighborhood-event probabilities. This staged design preserves the complementary roles of continuous spatial context, local rainfall measurements, and gauge availability. We evaluated PLUME for 0–3 h local short-duration heavy-rainfall warning over central and eastern China using an independent test set from May to September 2023. Across Barnes analysis, inverse distance weighting, and kriging, PLUME consistently improved the critical success index (CSI) over matched satellite–grid baselines. Relative CSI gains were 2.6–3.3% under complete gauge input and 3.6–5.9% under simulated gauge missingness. PLUME also reduced the associated CSI loss by 45.8–67.4%. Analysis of the variant with frozen batch normalization (BN) statistics showed that, under all three backgrounds and both input conditions, the largest positive probability updates were concentrated among baseline misses and converted some to hits, whereas the largest negative probability updates were concentrated among baseline false alarms and converted some to correct negatives. The dominant conditional CSI contribution shifted from false-alarm suppression at 0–1 h to missed-event recovery at 1–3 h. By using gridded and point-origin gauge information at different fusion stages, PLUME improves local heavy-rainfall warning through lead-dependent bidirectional refinement.

Organized convective systems produce a disproportionate fraction of warm‐season rainfall, yet satellite‐based analyses generally neglect propagation‐relative microphysical organization. We develop a propagation‐relative partitioning framework using FY‐4A geostationary satellite observations to characterize sector‐resolved cloud effective radius–temperature profiles for 162 eastward‐propagating convective systems over China. Results reveal pronounced horizontal and vertical heterogeneity: the frontal sector sustains a deep warm‐rain growth layer, while the rear sector exhibits larger glaciated particles at colder temperatures. Prior to surface rainfall intensification, these profiles evolve from near‐monotonic growth to a distinct unimodal structure with strengthening frontal‐to‐rear contrasts. These findings identify propagation‐relative microphysical organization as an observational precursor to rainfall intensification in propagating convective systems.

Next-generation geostationary satellite FY-4A provides cloud data with a temporal resolution of 15 minutes, while shallow convection may evolve into deep convection in less than 30 minutes. This temporal sparsity limits the satellite’s ability to capture rapid convective processes, thereby hindering effective weather monitoring and early warning. Temporal super-resolution technology is used to address this limitation, but existing temporal super-resolution methods often directly apply video frame interpolation paradigms designed for RGB images, resulting in transfer errors. To address such issues, a novel spatiotemporal self-supervised temporal super-resolution framework is proposed, which leverages the spatiotemporal consistency between the regional and full-disk observations of the FY-4A satellite to establish a temporal super-resolution mapping relationship without the need for manually labeled data. Additionally, a Multi-channel Encoding and Temporal-flow Evolution Network for Optimized-observation Reconstruction (METEOR-Net) is proposed, which addresses the challenges of multi-channel asynchrony and complex nonlinear evolution in geostationary satellite cloud data through multi-channel collaborative encoding, temporal-flow estimation and evolution, and evolution-aware feature fusion mechanisms. To enhance the focus on convective regions, a new convective region-weighted loss function and evaluation metrics are designed. Experimental results demonstrate that this method achieves state-of-the-art performance in visual quality and convective process reconstruction accuracy in the FY-4A satellite cloud data temporal super-resolution reconstruction task, successfully enhancing the temporal resolution of FY-4A cloud data from 15 minutes to 5 minutes. This provides high-resolution satellite cloud data for future convective weather monitoring and early warning systems.

The seasonal variation of the sea‐land breeze (SLB) circulation is investigated in a typical coastal city Qingdao, China, using continuous 2‐year observation data from Doppler LiDAR. The frequency of SLB varies seasonally, peaking in July and bottoming in January. The establishment time of the sea breeze is primarily concentrated between 08:00 LST and 12:00 LST, with its onset occurring approximately 1 hr earlier in spring and summer compared to autumn and winter. The top of the land‐breeze circulation is lower than that of the sea‐breeze circulation in all seasons and shows a decreasing trend over time. The top height of the sea‐breeze circulation, apart from a gradual decline in spring, exhibits complex curvilinear changes in other seasons. It should be noted that a negative sea‐land temperature difference (SLTD) is not a necessary condition for the formation of SLB in this study. Instead, the inversion layer structure is commonly found during the nocturnal hours of SLB days, resulting in a significant reduction in the atmospheric boundary layer height (ABLH) compared to the average, with a decrease of about 50% in winter and 10%–30% in other seasons. This stable ABL structure suppresses turbulent activity and enhances the wind speed difference between the lower and upper layers during the day, as well as the momentum transfer, which can be used to explain the development mechanism of the SLB circulation.

During mission navigation, the foremost concerns for Unmanned Surface Vehicle (USV) formations are obstacle-avoidance safety and cooperative consistency. To address these issues, this paper proposes an integrated “safety-domain quantification – motion control – cooperative decision” framework that fuses dynamic fuzzy ship-domain theory, leader – follower intelligent control, A* path planning and artificial potential fields. First, an individual dynamic fuzzy safety domain for each USV is constructed by extending classical ship-domain concepts to account for own-ship dynamics, target-ship behaviour and meteorological/oceanographic disturbances. Next, a five-USV formation scenario is designed in which a cooperative obstacle-avoidance decision mechanism and motion-control strategy are developed in full compliance with the International Regulations for Preventing Collisions at Sea (COLREGs). Extensive simulation results demonstrate that the proposed method maintains formation safety and stability under various environmental conditions, achieves an obstacle-avoidance success rate above 95 % in complex obstacle fields, and satisfies COLREGs compliance requirements, thereby providing technical support for safe USV formation navigation.

With the upgrading of geostationary meteorological satellites, their capabilities in Convective Initiation (CI) identification have been enhanced. To improve the applicability of the ARGI-based CI algorithm in central and eastern China, this study uses Fengyun-4A data, integrates radar and precipitation data to construct a True_CI dataset, and defines False_CI events (satellite-identified events without radar or precipitation signals) for comparative analysis. The results show that True_CI events tend to have longer durations, larger cloud cluster areas, and lower central cloud-top brightness temperature (BT) during development. They exhibit distinct features such as reduced differences between water vapor and infrared channels, increased cloud optical thickness, and ice-phase transformation 30 min before CI occurrence—features absent in most False_CI events. Based on these comparative findings, a new satellite-based CI definition is proposed with a set of reference thresholds, which should be adjusted for different latitudes and seasons. The evaluation of the Defined_CI events (defined using the CI definition) via True_CI events indicates that the CI definition on satellite cloud imagery proposed in this study is reliable, and suggests that further research on the pre-CI environmental conditions of weak convection is needed. Supported by hyperspectral data or numerical model products, such research will help clarify which cloud clusters are prone to developing into convective weather.

The surface rainfall regions of convective systems do not always align spatially with the overlying cold‐cloud shields, posing challenges to satellite quantitative precipitation estimates. A novel index of convective rainfall deviation is defined in this study to quantify the spatial discrepancy between surface rainfall and cold‐cloud shields, utilizing the FY‐4A brightness temperature data and surface rainfall observation data from across China. The results show that systems with larger deviations are more prevalent in southern China. For small‐deviation systems, surface rainfall closely coincides with the coldest center of the cloud shields. However, in medium‐deviation systems, rainfall occurs southwest of the coldest center, with relatively expanded but weakened rainfall beneath the cloud shields. Retaining a pronounced southwestern deviation signature, large‐deviation systems also present a secondary northeastern rainfall center. This complex spatial pattern is associated with increased cloud tilting driven by the configuration of environmental updrafts and low‐level southwesterly wind shears.

The melting of sea ice in the Arctic has sparked research into how clouds form in exposed ice‐free zones. This study delves into the seasonal distribution of stratocumulus clouds over the Norwegian Sea‐Barents Sea region and presents a conceptual model for the underlying mechanisms. In spring, autumn, and winter, increased surface water vapor flux provides water vapor from the sea surface. Concurrently, warm air enters the area through the southwest wind of the Norwegian Strait. Under conditions of unstable stratification, the water vapor is lifted to the condensation level, promoting the formation of stratocumulus clouds. Conversely, in summer, evaporation is suppressed due to stable stratification, which hinders vertical vapor transport. Enhanced dry advection reduces the transport of water vapor, leading to a decrease in the amount of stratocumulus clouds during summer. This research enriches our understanding of the physical mechanisms influencing cloud formation and cloud‐climate feedbacks in the Arctic region.

The vertical structure of clouds plays a critical role in atmospheric radiative transfer processes and is a major source of uncertainty in satellite-based retrievals of cloud optical thickness (COT) and cloud effective radius (CER). This study develops a retrieval model for COT and CER based on a random forest framework coupled with spatial gradient features, using multispectral observations from the FY4A/AGRI (Advanced Geostationary Radiation Imager) and simulations from Advanced Radiative Transfer Modeling System (ARMS) over central and eastern China during June–August 2018. The retrieval results agree well with MODIS, with correlation coefficients of 0.87 and 0.91 for COT and CER, respectively. To assess the impact of vertical cloud structure, ten sensitivity experiments varied water and ice content in different cloud layers. The results indicate that upper-level ice clouds significantly mask reflectance from lower clouds, reducing total reflectance by approximately 50 %, leading to lower retrieved values than those of single-layer clouds. For CER < 20 μm, the mean COT increase due to low- and mid-level water cloud variations in single-layer clouds exceeds that in double-layer clouds by about 24 %, primarily due to the masking effect of upper-level ice clouds in double-layer structures. This masking also contributes to retrieval biases in three-layer cloud systems. Furthermore, increased mid-level liquid water enhances the nonlinear relationship between COT and CER, increasing retrieval uncertainties. This study highlights the importance of considering multi-layer cloud structures in remote sensing algorithms and radiative transfer models.

The vertical structure of clouds plays a critical role in atmospheric radiative transfer processes and is a major source of uncertainty in satellite-based retrievals of cloud optical thickness (COT) and cloud effective radius (CER). This study develops a retrieval model for COT and CER based on a random forest framework coupled with spatial gradient features, using multispectral observations from the FY4A/AGRI (Advanced Geostationary Radiation Imager) and simulations from Advanced Radiative Transfer Modeling System (ARMS) over central and eastern China during June–August 2018. The retrieval results agree well with MODIS, with correlation coefficients of 0.87 and 0.91 for COT and CER, respectively. To assess the impact of vertical cloud structure, ten sensitivity experiments varied water and ice content in different cloud layers. The results indicate that upper-level ice clouds significantly mask reflectance from lower clouds, reducing total reflectance by approximately 50 %, leading to lower retrieved values than those of single-layer clouds. For CER < 20 μm, the mean COT increase due to low- and mid-level water cloud variations in single-layer clouds exceeds that in double-layer clouds by about 24 %, primarily due to the masking effect of upper-level ice clouds in double-layer structures. This masking also contributes to retrieval biases in three-layer cloud systems. Furthermore, increased mid-level liquid water enhances the nonlinear relationship between COT and CER, increasing retrieval uncertainties. This study highlights the importance of considering multi-layer cloud structures in remote sensing algorithms and radiative transfer models.