Aerosol particles in near-surface and elevated air often exhibit distinct physicochemical characteristics, but their differences remain insufficiently characterized. Aerosol samples were collected under clear, fog, and dust conditions at a coastal site (5 m a.s.l.) and at the top (1330 m a.s.l.) of Mt. Unzen in southwestern Japan. Single-particle analyses using electron microscopy with energy-dispersive X-ray spectroscopy revealed bimodal size distributions under clear conditions, characterized by submicron mineral particles and supermicron sea salt particles. In contrast, unimodal size distributions were observed under fog and dust conditions, with S-rich particles dominating the submicron range and Si-rich particles prevailing in the supermicron range. The average sulfur content in mineral particles under fog conditions reached 36% by weight, significantly higher than under clear conditions (15%), although sulfur levels remained comparable between the two sites in both clear and foggy air. During dust events, sulfur accumulation was more pronounced at the coastal site (16.8%) than at the mountaintop (4.9%), likely due to slower air mass movement near the surface as indicated by backward trajectories. These results highlight the combined effects of altitude and transport dynamics on aerosol aging and the need to incorporate near-surface altitude-resolved mechanisms in atmospheric models for coastal East Asia.

In the Kanto region in Japan, cold-air damming (CAD) is known to occur approximately 13 times a year and to be linked to snowfall and heavy rainfall, making its accurate forecasting critically important. This study evaluated the forecast accuracy of CAD in the Japan Meteorological Agency's operational mesoscale model (MSM) over the period 2007-2024. The results revealed that 12-h forecasts showed a warm bias exceeding +0.8 K in the lower atmosphere over southern Kanto and an underestimation of the southwestward cold-air flow. These biases increased with longer forecast lead times and tended to extend southwestward over the ocean south of the Tokai region. Analysis of the position of the isentropic surfaces revealed that the thickness of dammed cold air and its seaward extent tended to be underestimated. Although the warm bias exhibited a gradual decreasing trend (−0.02 K/year), a warm bias persisted even after the MSM was updated to the new non-hydrostatic model.

Data-driven weather prediction models show promising performance and are continuously advancing. In particular, diffusion models represent fine-scale details without spatial smoothing, which is crucial for mesoscale predictions, such as heavy rainfall forecasting. However, the applications of diffusion models to mesoscale predictions remain limited. To address this gap, this study proposes an architecture that combines a diffusion model with Swin-Unet as a deterministic model, achieving mesoscale predictions while maintaining flexibility. The proposed architecture trains the two mod-else independently, allowing the diffusion model to remain unchanged when the deterministic model is updated. Comparisons using the Fractions Skill Score and power spectral analysis demonstrated that incorporating the diffusion model improved accuracy compared to predictions without it. These findings highlight the potential of the proposed architecture to enhance mesoscale predictions, including heavy rainfall, while maintaining flexibility.

This study investigates the statistical characteristics of Baiu frontal depressions (BFDs), which occasionally accompany heavy rainfall in Japan. Using a high resolution (5 km) regional reanalysis dataset in June and July from 1991 to 2020, we objectively detected about 7000 BFDs (more than 1,000 distinct events) by combining methods to identify isolated depressions and the Baiu front. Our analysis shows that BFDs occur widely around Japan, particularly in western areas, and BFDs with smaller radii are more prone to intense precipitation. Although smaller BFDs are more common in the western areas, both small and large BFDs also occur in the eastern areas. Composite analyses indicate that small BFDs exhibit heavy rainfall near their centers and have an upright or slightly eastward tilted vertical structure, suggesting the essential role of diabatic heating. In contrast, BFDs with larger radii exhibit maximum precipitation east of their centers and have a westward tilted structure, suggesting a greater contribution from baroclinic processes.

This study investigates the probabilistic forecasting of a heavy rainfall event over Kyushu, Japan during July 2022 based on a series of data assimilation and forecast experiments using 1000-member ensembles. The deterministic forecast successfully reproduced the overall intensity and pattern of the heavy rainfall, but faced challenges in predicting the exact timing and location of the rainfall because of the strong nonlinearity within the atmospheric system. Based on the ensemble forecasts, the probability forecast indicated a low likelihood of heavy rainfall. Substantial uncertainties were observed, particularly with respect to the timing and spatial distribution of the rainfall, reflecting the inherent difficulties of forecasting such extreme and localized events. However, by allowing for some discrepancies in the timing and location of the rainfall, the probability of heavy rain was increased. This approach highlights the value of incorporating spatial and temporal tolerances when interpreting ensemble outputs. Such adjustments can enhance the reliability of probabilistic forecasts, providing more actionable information for those involved in disaster preparedness and risk mitigation.

Mesoscale convective systems (MCSs) frequently produce heavy rainfall and trigger disasters, yet accurately forecasting them remains a challenge. This study aims to provide valuable insights into predictability of an MCS that produced record-breaking rainfall over northern Kyushu, Japan, on July 9, 2023. The assimilation of all-sky infrared radiances from Himawari-9 clearly improves the initial conditions, resulting in an accurate ensemble forecast. Ensemble-based correlations reveal close relationships between the precipitation amount in northern Kyushu and low-level water vapor flux, horizontal convergence, and mid-tropospheric humidity. A comparison of better and worse ensemble members in terms of the precipitation amount indicates that the better members exhibit a moist absolutely unstable layer, which may promote the development of the MCS. Moreover, immediately before the onset of the heavy precipitation, the better ensemble members exhibit stronger low-level water vapor flux than the worse members. Notably, such differences are unclear several hours prior to the onset, indicating limited predictability of the target MCS due to rapid forecast error growth.

Using 43-year reanalysis data, the time evolution, three-dimensional structure, and climatology of the Ertel potential vorticity (PV) filaments associated with the stratospheric polar vortex from winter to spring in the Southern Hemisphere are examined. The filaments are formed when a part of the polar vortex extends equatorward, mainly due to the breaking of planetary waves originating from the troposphere. In early winter, the polar vortex occasionally has a structure with multiple PV steps. The filaments are continuously observed during the abrupt disappearance of the lowest latitude PV step and followed by shrinking of the polar vortex. A small-scale barotropic vortex forms at the tip of the filament when the filament is sufficiently zonally elongated. The climatology of the filament occurrence frequency is examined for an 850 K isentropic surface. The filament root, which is robustly determined and used as a reference of the filament location, is mainly distributed over 25°-45°S in the western hemisphere in July–October with a slight eastward movement in October. The momentum flux associated with small-scale disturbances, including vortices arising from the filaments, is largely negative in 30°-50°S, giving westward (eastward) forcing in the equatorward (poleward) side to diminish the PV maximum including filaments.

This study is the first to quantify the precipitation enhancement associated with the merging of mesoscale convective systems in the tropics and to evaluate global storm-resolving models (GSRMs). By classifying events into non-merger single-cell and merger multi-cell convection, we analyze how systems evolve during the DYAMOND Summer Phase in the tropics and subtropics. Both observations and models show that these two storm types form where column water vapor exceeds 50 mm, particularly across Asia and the Intertropical Convergence Zone. While most models capture the spatial distribution of non-merger systems in association with the simulated moisture biases, discrepancies in merger systems, which occur mainly near coastal regions, are more significant and highly variable among models. Observed merger systems are 2.1-2.4 times more intense in maximum rainfall and last longer (> 20 hours) than non-mergers. However, models show a larger spread in enhanced intensity and in the timing of peak rainfall in merger systems. These results underscore the importance of accurately representing both large-scale moisture and convective processes in GSRMs.

The pre-existing environmental conditions favorable for the development of cold-season extratropical cyclones formed in the west of the Japanese Islands are investigated. We focus on the early stage of cyclones to isolate environments unaffected by developing cyclones. Using reanalysis data and a feature-tracking algorithm, we identify and categorize cyclones into “strong” and “weak” based on their maximum intensity. Strong cyclones are characterized by pre-existing anticyclonic southerly winds and consequent warm and moist environments over the formation area. A preceding upper-level trough appears to strengthen these low-level southerly winds by inducing divergence behind it. Further analysis of strong cyclones reveals that stronger cyclones are accompanied by a more extensive intrusion of low-level cold air from the Asian continent, which enhances temperature gradient over the formation area. Correspondingly, the subtropical jet stream intensifies and increases upper-level divergence in its left-exit region above the cyclones. A concurrent upstream trough in the polar-front jet and ridge in the subtropical jet likely contribute to this enhanced upper-level divergence. These findings suggest that differences in early-stage environmental conditions of extratropical cyclones may explain the difference in their maximum intensity.

This study examines the use of the Japan Meteorological Agency's mesoscale ensemble prediction system (MEPS) to improve deterministic forecasts for heavy snowfall events, focusing on the case of a December 2023 heavy snowfall event in Iwamizawa City, Hokkaido. This event involved a convergence band cloud over the northern Sea of Japan that was driven by mesoscale low-level cold air advection. To capture key thermal variability related to this phenomenon, we applied principal component analysis (PCA) to 925 hPa temperature fields from 21 MEPS members. Clustering in the principal component plane identified four representative forecast scenarios. One cluster significantly improved prediction of the location of the convergence band cloud and associated snowfall compared to the operational mesoscale model (MSM). By projecting the mesoscale analysis fields—available six hours after initialization—onto the same phase plane, the most accurate scenario could be selected up to 21 hours before peak snowfall. We validated our method by applying it to three additional heavy snowfall cases and confirmed improvement over the MSM. These results highlight that MEPS-based clustering of mesoscale cold air advection patterns provides a robust approach to enhancing precipitation forecasts and supporting earlier weather warnings in Hokkaido.