Abstract This study investigates the impact of climate change on the extreme 2020 Meiyu over the middle and lower reaches of the Yangtze River (middle‐lower Yangtze River (MLYR)) through global variable‐resolution ensemble subseasonal hindcasts. Results reveal that post‐1980 climate change enhanced the 2020 extreme Meiyu rainfall over the MLYR region by approximately 14.67% at monthly scale, while simultaneously decreasing light and moderate precipitation frequency but intensifying heavy and extreme precipitation occurrences. Climate change intensified the low‐pressure over northern China and southern China while weakening the Western Pacific subtropical high and the low‐pressure over the Indian Peninsula. The circulation pattern results in significant shear between northeasterly and northwesterly winds in the southern MLYR region, contrasting with the high‐pressure dominance in the northern MLYR region. This configuration suppressed convergence, vertical motion, and precipitation in the northern MLYR while enhancing these processes along its southern. Compared with frequently re‐initialized simulations, subseasonal hindcasts better preserve the continuous evolution of the Meiyu system and the consistency between precipitation and circulation responses, making them a more suitable framework for attribution of 2020 persistent rainfall Meiyu season. Overall, this study highlights the value of global variable‐resolution subseasonal simulations for attributing changes in Meiyu precipitation and its associated large‐scale circulation background under climate change. Future studies would benefit from improved subseasonal forecasting capabilities to enhance attribution reliability.

Lake-land thermal contrasts significantly modulate regional air quality, yet the coupling mechanisms by which inland lakes regulate the diurnal evolution of PM2.5 and its components remain poorly understood. This study conducts high-resolution (1 km) WRF-Chem simulations over Lake Chaohu and the adjacent megacity of Hefei, China, during spring to elucidate these interactions. Results reveal a distinct diurnal reversal effect. During daytime, the lake presence facilitates PM2.5 increases of predominantly 0-10 & micro;g m-3 both over the lake and in surrounding urban areas by suppressed planetary boundary layer height, weakened vertical mixing, and reduced dry deposition velocities, which collectively transform the lake into "storage zone" that prolongs PM2.5 lifetimes. This accumulation is dominated by secondary PM2.5, as the cooler and more humid lake air thermodynamically favors the ammonium nitrate formation. Furthermore, convergence zones where lake breezes meet background winds create localized stagnation traps that intensify shoreline pollution. At night, while the lake surface maintains higher PM2.5 concentrations than surrounding land, its impact on the city reverses, exerting a purification effect with urban PM2.5 decreasing by predominantly 0-10 & micro;g m-3 as land-breeze circulation enhances vertical mixing and facilitates primary pollutant dispersion. Sensitivity experiments reveal that failing to distinguish lake surfaces in emission inventories can significantly amplify daytime pollution. These findings emphasize that lakes act as complex dual regulators of urban air quality, with identified mechanisms likely applicable to other urban-lake systems globally. This study highlights the necessity of high-resolution meteorological modeling and precise surface characterization for improved air quality forecasting in lake-adjacent megacities regions.

This study establishes a non-hydrostatic integrated atmospheric model across scales (iAMAS-Mars) and validates its reliability through comparison with various observational datasets. The model performance is systematically evaluated using three sets of experiments: one with a global uniform resolution of 240 km (U240) and two with variable resolutions ranging from 240 km to 60 km (V60) and 120 km to 5 km (V5). The results demonstrate that iAMAS-Mars successfully reproduces the main characteristics of the Martian atmosphere: reasonable simulation of seasonal variations in surface pressure and temperature; good representation of the vertical distribution of atmospheric thermal structure and seasonal evolution of zonal wind fields, including winter polar jets and equatorial easterlies; successful simulation of the seasonal transition of Hadley circulation and perihelion enhancement effects; satisfactory reproduction of the CO2 seasonal cycle; successful capture of the primary seasonal features of dust activity, including the aphelion clean period and perihelion dusty period structure. We find that in all three resolution cases, within the selected time range, as the resolution increases, the simulation fidelity gradually improves. Notably, the 5-km high-resolution variable-mesh simulation successfully captures small-scale dust phenomena in the Utopia Planitia region, revealing the modulation of complex topography on local wind fields and dust-lifting processes. This model provides an effective tool for investigating multi-scale atmospheric processes on Mars.

Abstract. Atmospheric composition is governed by multiscale processes spanning global transport to local chemistry. Simultaneously capturing these scales is essential, yet conventional modeling trades global coverage for regional detail restricted by artificial boundaries. Global variable-resolution (VR) modeling provides a unified alternative, yet online gas-phase chemistry within nonhydrostatic VR frameworks remains limited, and how mesh refinement alters coupled physical–chemical responses is not fully understood. Here, we develop and evaluate an online gas-phase chemistry framework in the VR model iAMAS (v2.5.1). Operating on a unified unstructured mesh without lateral boundaries, the modular framework couples KPP-generated SAPRC-99 tropospheric chemistry, Linoz stratospheric ozone, Fast-J photolysis, emissions, transport, and deposition. Global baseline simulations at ~60 km resolution (U60km) for July 2019 validate large-scale tracer distributions (O3​, NO2, HCHO, CO) and surface ozone diurnal cycles across China. Regionally refined simulations at ~4 km over North China (V4km) demonstrate how online VR chemistry resolves cross-scale physical–chemical coupling over complex terrain. V4km recovers the observed ridge-high/valley-low nighttime ozone pattern, reversing spatial correlation from R=−0.40(U60km) to R=+0.46 by resolving slope circulations and heterogeneous NOx​ emissions within a shallow nocturnal boundary layer. Refinement also reorganizes urban–rural and day–night ozone responses: higher urban NOx suppresses daytime ozone peak under VOC-limited conditions, whereas reduced coarse-grid NOx dilution weakens rural nighttime titration and sustains higher ozone. These results demonstrate that mesh refinement in a global VR model alters the coupled dynamical–chemical state rather than merely sharpening concentration gradients, establishing iAMAS as a modular platform for seamless global-to-regional atmospheric-composition studies.

Abstract Atmospheric reanalysis underpins weather and climate research, AI model training, and numerical-model initialization. However, the temporal evolution of reanalysis variables is not continuously smooth, contrary to the expectation of temporal continuity that many users implicitly hold. Because reanalysis blends observational records with short-range model forecasts within finite assimilation windows, state discontinuities arise inevitably at the seams between successive windows. Whether these discontinuities leave a downstream forecast-skill signature distinguishable from ordinary lead-time dependence remains unclear. Using the July 2023 North China extreme rainfall event as a test case, we conduct two 24-member initialization sweeps at approximately 4- and 16-km resolution over the refinement zone, initialized from consecutive hourly ERA5 analyses. Forecast behavior changes abruptly across the assimilation-window boundary. At convection-permitting resolution, neighborhood-based verification shows that, where a boundary-related skill change is detected, its magnitude exceeds the full distribution of non-boundary adjacent-hour changes. Circulation and counterfactual analyses link the 09:00–10:00 UTC transition to changes in the moisture corridor and rainfall placement, shifting a forecast from failure to success with only one hour apart in lead time. These results show that ERA5 window boundaries can introduce structured differences among otherwise adjacent initializations. Reanalysis-driven forecasts, sensitivity experiments, and time-lagged ensembles should therefore account explicitly for the assimilation-window position of their initial conditions.

Abstract. Tracer transport is a critical computational bottleneck in high-resolution atmospheric chemistry models, where tens to hundreds of species are advected. The iAMAS model (v2.6.3) on spherical centroidal Voronoi tessellations (SCVTs) currently employs a scheme (2H1FCT) that performs two steps of third-order transport followed by one step of flux-corrected transport (FCT), in which the FCT correction step dominates computational cost. This study develops TVD-FFSL, a computationally efficient hybrid tracer transport scheme. Horizontally, a total variation diminishing (TVD) flux operator with the KOREN limiter is employed. Vertically, a flux-form semi-Lagrangian (FFSL) operator based on piecewise parabolic method reconstruction handles cells with CFL > 1unconditionally. Together with a second-order TVD Runge-Kutta time integration, the scheme ensures monotonicity without a separate FCT step. Idealized 2D tests demonstrate that TVD-FFSL achieves robust shape preservation. Although its errors are slightly higher than 2H1FCT, superior convergence rates render the accuracy gap negligible at finer resolutions (∆x ≤ 30 km). Realistic 3D dust simulations on a 16–60 km variable-resolution grid confirm its long-term stability and accuracy comparable to 2H1FCT. Performance benchmarks show that TVD-FFSL achieves over 2× speedup in standalone transport tests and exceeds 3.75× speedup in long-term atmospheric dust simulations, significantly reducing the computational overhead of numerous tracer transport. The design principles of TVD-FFSL could be transferable to other unstructured meshes, offering a pathway toward accelerating high-resolution atmospheric chemistry simulations.

Simulating accurately the South Asian summer monsoon is crucial for food security of several South Asian countries yet challenging for global climate models (GCMs). The GCMs suffer from some systematic biases including dry bias in mean monsoon rainfall over the India subcontinent and excessive equatorial light rain between which the relationship was rarely discussed. Numerical experiments are conducted for one month during active monsoon with global quasi-uniform resolution of 60 km (U60 km) and 3 km (U3 km) separately. Evaluation with observations shows that U3 km reduces the dry bias over northern India and excessive light rain over the equatorial Indian Ocean (EIO) that are both prominent in U60 km. Excessive light rain in U60 km contributes critically to stronger rainfall and latent heating over the EIO. A Hadley-type anomalous circulation is thus induced, whose subsidence branch suppresses updrafts and reduces moisture transport into northern India, contributing to the dry bias. The findings highlight the importance of constraining excessive light rain for regional climate projection in GCMs.

Air pollution in cities impacts public health and climate. Turbulent mixing is crucial in pollutant formation and dissipation, yet current atmospheric models struggle to accurately represent it. Turbulent mixing intensity varies with model resolution, which has rarely been analyzed. To investigate turbulent mixing variations at multiple resolutions and their implications for urban pollutant transport, we conducted experiments using the Weather Research and Forecasting model coupled with Chemistry (WRF-Chem) at resolutions of 25, 5, and 1 km. The simulated meteorological fields and black carbon (BC) concentrations are compared with observations. Differences in turbulent mixing across multiple resolutions are more pronounced at night, resulting in noticeable variations in BC concentrations. BC surface concentrations decrease as resolution increases from 25 to 5 km and further to 1 km, but they are similar at 5 and 1 km resolutions. Enhanced planetary boundary layer (PBL) mixing coefficients and vertical wind flux at higher resolutions reduce BC surface concentration overestimations. The 1 km resolution parameterized lower mixing coefficients than 5 km but resolved more small-scale eddies, leading to similar near-surface turbulent mixing at both resolutions, while the intensity at higher altitudes was greater at 1 km. This caused BC to be transported higher and farther, increasing its atmospheric lifetime and column concentrations. Variations in mixing coefficients are partly attributed to differences in land use and terrain, with higher resolutions providing more detailed information that enhances PBL mixing coefficients, while grid size remains crucial in regions with more gradual terrain and land use changes. This study interprets how turbulent mixing affects simulated urban pollutant diffusion at multiple resolutions.

Abstract. Aerosols have significant impacts on regional climate, which has been widely investigated with numerical experiments. However, uncertainties of simulated aerosol impact due to long-standing chaotic effect remain unclear. Here we propose a diagnostic method based on large ensemble simulations and random sampling algorithm to unveil the chaos-induced uncertainties in simulated aerosol climatic impacts that is overlooked in previous studies. Taking dust impacts on Indian summer monsoon system as a demonstration, our findings reveal that, while dust generally enhances the large-scale summer monsoon circulation consistently among ensemble members, its impacts on regional systems, such as monsoon depressions, exhibit significant chaotic effect: the simulated aerosol impacts on precipitation from individual ensemble member differ substantially, even inversely. Through quantitative analysis, we demonstrate that the magnitude of these chaotic effects diminishes following a N-½ relationship with ensemble size N. Furthermore, our results indicate that statistical significance testing alone may be insufficient for robust attribution of dust impacts, as even small ensembles can yield statistically significant yet contradictory results. This study emphasizes the necessity of employing adequate ensemble sizes to capture reliable physical impacts of aerosol on regional climate.

Aerosols have significant impacts on regional climate, which has been widely investigated with numerical experiments. However, the uncertainties of simulated aerosol impact due to the long-standing chaotic effect remain unclear. Here we propose a diagnostic method based on large ensemble simulations and a random sampling algorithm to unveil the chaos-induced uncertainties in simulated aerosol climatic impacts that have been overlooked in previous studies. Taking the dust impacts on the Indian summer monsoon system as a demonstration, our findings reveal that, while dust generally enhances the large-scale summer monsoon circulation consistently among ensemble members, its impacts on regional systems, such as monsoon depressions, exhibit significant chaotic effect: the simulated aerosol impacts on precipitation from individual ensemble members differ substantially, even inversely. Through quantitative analysis, we demonstrate that the magnitude of these chaotic effects diminishes following a N-12 relationship with ensemble size N. Furthermore, our results indicate that statistical significance testing alone may be insufficient for the robust attribution of dust impacts, as even small ensembles can yield statistically significant yet contradictory results. This study emphasizes the necessity of employing adequate ensemble sizes to capture reliable physical impacts of aerosol on the regional climate.

Regular latitude-longitude grids in global simulations encounter polar singularities in the Arctic and Antarctic regions. In contrast, unstructured meshes have the potential to overcome this issue; however, so far, the performance of unstructured meshes in polar areas has barely been investigated. This study investigates the efficacy of unstructured meshes over Antarctica using the integrated Atmospheric Model Across Scales (iAMAS, v1.0) with multi-source observations. Four mesh configurations of the iAMAS model were assessed, varying in resolution (120, 60, 16, and 4 km) over the Antarctic region. The study evaluates the performance of the iAMAS simulation for both the surface layer and the upper meteorological fields (temperature, pressure, specific humidity, and wind speed), comparing simulations with data from the fifth-generation ECMWF reanalysis (ERA5) and measurements from automatic weather stations and radiosondes. The results indicate that the iAMAS model does not exhibit the polar singularity issue observed in ERA5, where the ERA5 with regular latitude-longitude grids significantly underestimates wind speeds at the polar grid center. In the relatively flat region of East Antarctica, all four iAMAS experiments at various resolutions demonstrate comparable and even superior performance in simulating temperature and wind speed compared to ERA5. In regions with complex terrain, such as near the Transantarctic Mountains, the iAMAS model (particularly at coarse grid resolutions like 120 km) exhibits a cold bias and stronger wind speeds, consistent with biases identified in other Antarctic simulations using regional models. In particular, mesh refinement at 4 km in complex terrains significantly enhances iAMAS's accuracy in simulating the meteorological fields for both the surface layer and upper atmosphere, suggesting that a grid resolution of 4 km (or even higher) is optimal in such regions. In contrast, in flatter areas, such as the high East Antarctic Plateau, increases in grid resolution yield minimal improvements in simulation accuracy, and a 60 km grid resolution appears sufficient.

Typhoons pose significant threats to coastal and inland regions,with their impacts exacerbated by climate change and population growth[1].Recent studies have shown increased frequency of powerful storms,slower translation speeds,northward shifts in the Western North Pacific(WNP)basin,and a near tripling of glo-bal exposure to typhoons since 1970[1-4].These changes present new challenges for forecasters and researchers,particularly in pre-dicting impacts on inland and high-latitude populations lacking prior exposure or resilience to typhoon effects.

Extreme rainfall events are becoming increasingly severe under a warming climate. North China has experienced several catastrophic rainfall events, of which the rainstorm in 2023 was particularly severe inducing unprecedented damage. Since 1980, the neighboring Mongolian Plateau (MP) has been warming at a rate three times the global average, faster than the surrounding regions. Whether a link exists between extreme rainfall in North China and the fast MP warming is unknown. Here, using global variable‐resolution atmospheric model with convection‐permitting capability over North China, we find the rapid warming trends, particularly over the MP, are highly conducive to extreme rainfall over North China. In the 2023 case, the fast MP warming induced an anomalous terrestrial high, which in the Western North Pacific Subtropical High created a strong high‐pressure system over North China. This system obstructed northeastward movement of Typhoon Doksuri, concentrating moisture supply which prolonged and intensified the extreme.

Forecasting uncertainties among meteorological fields have long been recognized as the main limitation on the accuracy and predictability of air quality forecasts.However,the particular impact of meteorological forecasting uncertainties on air quality forecasts specific to different seasons is still not well known.In this study,a series of forecasts with different forecast lead times for January,April,July,and October of 2018 are conducted over the Beijing-Tianjin-Hebei(BTH)region and the impacts of meteorological forecasting uncertainties on surface PM2.5 concentration forecasts with each lead time are investigated.With increased lead time,the forecasted PM2.5 concentrations significantly change and demonstrate obvious seasonal variations.In general,the forecasting uncertainties in monthly mean surface PM2.5 concentrations in the BTH region due to lead time are the largest(80％)in spring,followed by autumn(～50％),summer(～40％),and winter(20％).In winter,the forecasting uncertainties in total surface PM2.5 mass due to lead time are mainly due to the uncertainties in PBL heights and hence the PBL mixing of anthropogenic primary particles.In spring,the forecasting uncertainties are mainly from the impacts of lead time on lower-tropospheric northwesterly winds,thereby further enhancing the condensation production of anthropogenic secondary particles by the long-range transport of natural dust.In summer,the forecasting uncertainties result mainly from the decrease in dry and wet deposition rates,which are associated with the reduction of near-surface wind speed and precipitation rate.In autumn,the forecasting uncertainties arise mainly from the change in the transport of remote natural dust and anthropogenic particles,which is associated with changes in the large-scale circulation.

An unprecedented heavy rainfall event in China ("21.7" extreme rainfall event) was simulated using the global variable-resolution model (MPAS-Atmosphere) across the scales (4, 8, 16 and 50 km). Although almost all experiments at different resolutions reproduce the spatiotemporal characteristics of precipitation, the simulated precipitation intensity from high to low is 16, 8, 50, and 4 km, with the 16 km simulation being closest to the observations. Precipitation magnitude is prominently influenced by the difference in simulated large-scale circulation across a range of grid spacings. Further analysis revealed that the differences in latent heating across scales affect the geopotential height and wind field by altering temperature. The latent heating in 4 km simulation is the minimum while the 16 km simulation is maximum. More latent heating release leads to the low-level pressure depression, amplifies the water vapor flux convergence, produces stronger upward motion and more clouds, and ultimately results in stronger precipitation. The sensitivity experiments for turning off latent heating tendencies during the event showed that the latent heat release has positive feedback on the "21.7" heavy rainfall event. This study highlights the importance of scale-awareness of latent heat at different resolutions and suggests that the difference in simulated latent heat release during the event is the main reason for simulated different atmospheric circulation and precipitation across scales. Due to the advancements in the numerical computing power, numerical weather prediction models (NWP) are operating at horizontal grid spacing ranging from 1 to 10 km. This study evaluates the simulation performance of the global variable-resolution model MPAS-A for a heavy precipitation event and investigates the influence of scale-aware convective parameterizations on modeling rainfall at horizontal resolutions across the scales (4, 8, 16 and 50 km). The results show that 4 km simulation exhibits the weakest precipitation while the simulated precipitation at 16 km is the strongest. The difference in precipitation simulated by four experiments across scales derives from the variations in large-scale circulation simulated with different refined resolutions. Further analysis revealed that the differences in latent heat at different resolutions ultimately lead to the changes in geopotential height and large-scale wind field by altering temperature. 4 km experiment simulates less latent heat than other lower-resolution experiments. Owing to the positive feedback of the latent heat release on precipitation, 4 km experiment simulates the minimum precipitation. This work highlights that latent heat plays a vital role in simulation of synoptic condition and precipitation. Precipitation is influenced by the difference in simulated atmospheric circulation across a range of grid spacings The differences in simulated latent heat release across scales affect the geopotential height and wind field by altering temperature The latent heat release has positive feedback impact on the "21.7" heavy precipitation event

In 2020 early summer, a historically severe rainy season struck East Asia, causing extensive damage to life and property. Subseasonal forecast of this event challenges the limits of rainy season predictability. Employing the integrated atmospheric model across scales and the Sunway supercomputer, we conducted ensemble one-month forecasts at global 3 km, variable 4-60 km, and global 60 km resolutions. The global convection-permitting forecast accurately captures the rainband, while other forecasts exhibited northward and weaker shifts due to the northward shifts of the atmospheric rivers over Japan, attributed to intensified Western North Pacific Subtropical High (WNPSH). Further, the double-ITCZ-like tropical rainfall pattern in Western Pacific in global convection-permitting forecast contributes to a more accurate WNPSH and rainband. In contrast, other forecasts show a single-ITCZ-like pattern in Western Pacific, leading to a northward-shifted WNPSH and rainband, advocating the importance of accurately representing tropical convections, as they can significantly affect mid-/high-latitude weather and climate.

The Tibetan Plateau (TP) is one of the most climate-sensitive regions around the world. Aerosols imported from adjacent regions reach their peak during the pre-monsoon season and play a vital role in the TP environment. However, the strong interannual variation in aerosols transported to the TP has not been fully understood. Here, we show that the interannual variability of pre-monsoon aerosols transported to the TP is influenced more by rainfall over the southern Himalayas than near-surface wind. Rainfall modulates fire events and biomass burning emissions and reduces aerosols over the TP by wet scavenging. Contrary to the role of wind in increasing aerosol transport, the positive correlation between wind and aerosols in the TP reported in previous studies is contributed by the negative interannual correlations between wind and rainfall and between rainfall and fire events over the southern Himalayas. This study highlights the co-variability of wind and rainfall and their confounding impacts on aerosols in the southern Himalayas and over the TP. With pre-monsoon rainfall projected to increase in adjacent regions of southern TP, aerosol transport to the TP may be mitigated in the future.