Solar irradiance forecast aims to accurately estimate future solar irradiance based on historical data, playing a vital role in energy production and grid management. While ground-based station measurements provide local accuracy, geostationary satellites offer much broader environmental contexts, such as cloud coverage, which serves as a key factor for accurate forecasting. However, effectively integrating these multimodal observations remains a challenge, with existing methods suffering from inflexibility and high computational costs. To address this problem, we propose SatSolarCast, a flexible and efficient multimodal framework that introduces a memory alignment learning mechanism to integrate geostationary satellite data and historical irradiance observations. By preserving and recalling long-term spatiotemporal patterns from a specialized satellite memory bank, SatSolarCast enables effective guidance for both short- and long-term prediction. Additionally, SatSolarCast offers plug-and-play compatibility and can be incorporated into various forecasting architectures. Extensive experiments across four ground stations demonstrate that SatSolarCast substantially improves forecasting performance compared to prior methods with much lower computational costs.

Assimilating accurate water vapor observations is an effective way to enhance the rainfall forecasting performance of numerical weather prediction models. In this study, using the three-dimensional variational data assimilation (3DVAR) method, Precipitable Water Vapor (PWV) data observed by two kinds of powerful water vapor instruments, i.e., the Advanced Microwave Scanning Radiometer 2 (AMSR2) onboard the Global Change Observation Mission 1st-Water (GCOM-W1) satellite and the Global Navigation Satellite System (GNSS), are assimilated into the Weather Research and Forecasting (WRF) model to enhance its rainfall forecasting capacity. The assimilation impacts of AMSR2 and GNSS PWV are comprehensively examined for the extreme rainfall events occurred in South China during the period of April 01 to 30, 2024. Assessed by the rainfall observations from the densely distributed automatic weather stations, our results indicate that assimilating AMSR2 and GNSS PWV improves rainfall forecasting performance in terms of both rainfall forecast scores and spatial patterns. For the intense-precipitation days, with a threshold of 150 mm accumulated rainfall within 24-h, assimilating AMSR2 PWV improves the equitable threat score (ETS) by up to 0.020 (26.3 % of improvement), while assimilating GNSS PWV improves the ETS by 0.021 (27.6 % of improvement). When both AMSR2 and GNSS PWV are assimilated, the ETS score increases by up to 0.030 (39.5 % of improvement).

We introduce Integrated Relative Humidity (IRH), a layer-resolved diagnostic of tropospheric saturation, to investigate moisture-precipitation interactions in the subtropical coastal environment (Hong Kong). Using 16 years (2005-2020) of co-located radiosonde and rain gauge data, we classify observations into four precipitation lifecycle stages: No Rain (NR), Before Rain (BR), During Rain (DR), and After Rain (AR). IRH is computed from the surface to nine standard pressure levels (1000-150 hPa), enabling layer-specific saturation diagnostics across these stages. IRH probability density functions (PDFs) reveal stage-specific vertical structures with saturation peaks consistently near 850 hPa. BR shows increasing lower-tropospheric saturation, DR exhibits the highest deep-column saturation, and AR displays reduced mid-level saturation, reflecting dynamic redistribution across the precipitation lifecycle. Bimodal PDF structures in BR, DR, and AR stages are consistent with transitions between shallow and deep convective regimes. Reversed cumulative distribution function (RCDF) for IRH = 0.90 captures low/mid-level buildup in BR and DR, while RCDF for IRH = 0.80 highlights upper-level dissipation in AR. Crucially, IRH thresholds of 0.82 at 700 hPa and 0.70 at 150 hPa correspond to 84% and 82% of DR events, respectively, while both exclude 76% of NR cases. Furthermore, the threshold of 0.70 at 150 hPa is also exceeded in 72% of BR events with the same NR exclusion. These findings demonstrate IRH as a layer-resolved metric for moisture-precipitation interactions, providing a quantitative diagnostic framework that may inform the development of nowcasting tools and aid in model evaluation in subtropical environments.

Tropical cyclones (TCs) are one of the most destructive natural disasters. It is important to document long-term changes in TC properties to better prepare for the damages associated with TCs. TC lifespan is one of the little-studied TC characteristics. Here we show that annual mean TC duration has been decreasing at rates of -13 ± 6 and -7 ± 6 hours per decade, corresponding to approximately 56 and 30 h shorter average lifespans, in the Northeast and Northwest Pacific, respectively, from 1982 to 2024. The decreasing trend is primarily due to shorter TC tracks associated with poleward migration of TC genesis latitude. The TC intensification (weakening) rate before (after) the first (last) lifetime maximum intensity has increased, a phenomenon in which environmental conditions and internal convective structure may both play a role. The shorter TC lifespan over the open ocean before entering coastal zones compresses the time available for weather forecasting and disaster preparedness.

Antarctic sea ice expanded pre–20151–3 and declined rapidly post–20154, a regime shift generally attributed to a combination of factors, including sudden Southern Annular Mode transition5 and tropospheric teleconnections associated with the Pacific4,6–8. Yet, mechanisms responsible for the decadal variability remain unresolved. We demonstrate that stratosphere–troposphere coupling can potentially contribute to transmitting late austral winter Pacific Decadal Oscillation (PDO) forcing to subsequent summer Antarctic sea ice. The positive PDO phase excites dual pathways: (1) tropospheric Rossby waves strengthening both the Amundsen and Dumont d’Urville Sea Lows (ASL/DSL); and (2) planetary waves modulating the stratospheric polar vortex, whose downward influence amplifies ASL anomalies. The strengthened ASL and DSL generate warm advection anomalies that ultimately promote subsequent summer sea ice melt. Furthermore, analysis of the Coupled Model Intercomparison Project Phase 6 (CMIP6) ensembles supports the role of stratospheric pathways in capturing the observed PDO–ASL linkages. Thus, accurately representing stratosphere–troposphere interactions is beneficial for simulating Antarctic sea ice’s decadal variability.

Sea ice variability in the Laptev and East Siberian Seas (LESS) notably affects Arctic climate and maritime safety. While El Niño-Southern Oscillation (ENSO) winter sea surface temperature anomalies influence global climate, their effect on subsequent autumn LESS ice remains unclear. This study reveals a post-2000 intensification of winter ENSO's impact on subsequent autumn LESS ice, driven by accelerated ENSO phase transitions compared to the pre-2000 era. Rapid phase transitions of El Niño after 2000 generate persistent cold anomalies in the tropical central-eastern Pacific during the subsequent autumn, strengthening and displacing the Western North Pacific anticyclone (WNPAC) northward. This WNPAC triggers Rossby waves establishing an Arctic anticyclone, warming, and moistening the LESS atmosphere, thereby driving substantial ice loss. In contrast, pre-2000 slower El Niño decay exhibited weaker tropical-Arctic connectivity. These results identify ENSO phase transitions rate as a critical regulator of Arctic sea ice variability, with important implications for seasonal forecasting.

Mesoscale convective systems (MCSs) are known to induce heavy rainfall and pose significant weather hazards in tropical and mid-latitude regions. Understanding the environmental conditions of MCSs is essential for improving model simulations and operational precipitation forecasts. In this study, we analyze 929 MCS precipitation events, including 120 events with maximum rain rate exceeding 20 mm/h, over South China from April-June 2007 to 2020. We also separately investigate 15 warm-sector heavy rainfall MCS (WMCS) events from the ordinary MCS (OMCS), as they are often mis-predicted in current weather forecast systems. Our results indicate that moisture and instability are two key factors for MCS precipitation over South China. Water vapor shows a rapid increase 6 hr before the precipitation onset, peaking 1 hr before the maximum precipitation. Under the southwesterly wind, warm and moist air flows from the South China Sea to the coastal regions of South China. The air is lifted to upper levels (at around 450 hPa) due to the strong atmospheric lifting during MCS passage. The WMCS precipitation events exhibit higher moisture conditions and stronger atmospheric lifting than those of the OMCSs. Furthermore, we find that the favorable environment for MCS (e.g., lifting and convergence) is disrupted by mountains when the systems move across mountainous terrain. Consequently, the relationship between precipitation and thermodynamics is weaker for the leeside area of mountains compared with the windward area.

Past studies have presented different findings regarding how the impact of Arctic sea ice melt on Eurasia's winter cold extremes (WCE) is changing, with some suggesting a strengthening effect while others indicate a weakening influence. In this study, we demonstrate that interannual variability of Arctic sea ice is closely linked to climatological mean sea ice concentration (SIC). In low ice regimes (mean SIC <0.2), interannual variability declines with sea ice loss, resulting in a weakened Rossby wave response and consequently a reduced impact on WCE in mid-high latitudes. In contrast, high ice regimes (mean SIC >0.8) exhibit increased interannual variability with sea ice loss, extending its influence on WCE into mid-latitudes in Asia through amplifying Rossby waves. As Arctic sea ice loss progresses from the southern edge of the low ice regime to the interior of the high ice regime, WCE in mid-latitude Asia increase. These insights are critical for improving seasonal predictions of Eurasian WCE and enhancing adaptive capacity to cope with extreme weather events.

Forecasting geostationary infrared brightness temperature sequences from historical observations is a significant and challenging task. By analyzing these predictions, cloud evolution, convective activity, and atmospheric radiative states can be revealed in advance, offering high potential value in domains such as weather nowcasting, energy management, and disaster monitoring. Recently, artificial intelligence techniques have provided valuable insights into this task. However, as a nascent research area, the lack of a standardized, high-quality benchmark has significantly impeded progress. Moreover, training existing deep learning models for this task remains computationally expensive due to the complexity of their network architectures and modeling mechanisms. To address these challenges, we introduce a new benchmark, FY4ABT, and propose a lightweight prediction model, WavePredNet. Specifically, FY4ABT comprises three sub-datasets designed to respectively evaluate prediction performance under short-term, medium-term, and long-term scenarios. Meanwhile, WavePredNet effectively captures multi-scale dynamics, including both low- and high-frequency components with low computational costs while delivering exceptional performance.

Forecasting rapid intensification (RI) of tropical cyclones (TC) is a mission known for large errors. One under-researched factor that affects TC intensification is salinity, which is important for density stratification in certain ocean regions and can affect the surface enthalpy flux under a strengthening hurricane. To investigate the impact and efficacy of using salinity information in state-of-the-art forecasting, we use a statistical model consisting of a variety of machine learning (ML) methods. For salinity data, we use satellite measurements of pre-storm sea surface salinity (SSS) as a proxy for the salinity stratification. We train and test the model on various ocean basins, including the Atlantic, eastern North Pacific and western North Pacific. A calibrator is trained on top of the ML models to correct and enhance probability forecasts. The calibrator significantly improves probability forecasts relative to recent works. The ML model performance is improved with the addition of SSS in the Eastern North Pacific, western North Pacific, and the Caribbean subregion of the North Atlantic, and the overall model performance is better than previous studies. SSS decreases model skill for a model trained on the full Atlantic basin. In the Indian Ocean, SSS is also notably correlated with RI occurrence, but the TC samples are not sufficient to train ML models.

Recent research suggests atmospheric cloud radiative effect (ACRE) acts as an important feedback mechanism for enhancing the development of convective self‐aggregation in idealized numerical simulations. Here, we seek observational relationships between longwave (LW) ACRE and the spatial organization of mesoscale convective systems (MCSs) in the tropics. Three convective organization metrics that are positively correlated with the area of MCS, that is, convective organization potential, the area fraction of precipitating MCS, and the precipitation fraction of MCS, are used to indicate the degree of convective organization. Our results show that the contrast in the LW ACRE inside and outside an MCS is consistent across different MCS precipitation intensities throughout the life cycle of an MCS, typically 90–100 W/m 2 , and provides important positive feedback to the circulation of the given MCS. However, the LW ACRE inside and outside an MCS as well as their difference are not strongly related to the degree of organization, suggesting that the LW cloud radiative feedback may be supportive of MCS formation and maintenance without necessarily being a dominant factor for spatial organization of MCSs. The domain average vertical velocity does tend to be related to the measures of convective organization, suggesting that factors that favor large‐scale low‐level convergence may exert a leading effect in creating an environment favorable for mesoscale organization of deep convection.

Urbanisation significantly alters the interaction between land surface and the lower troposphere, impacting occurrences of natural hazards. The influence of urbanisation on natural hazards like heatwaves, hailstorms, and flooding remains debated. However, it is well established that impervious surfaces in urban areas can lead to flooding amplification. Singapore, amidst rapid urbanisation, experiences frequent flooding, exacerbated by its tropical-monsoon climate and climate change. Utilising high-temporal-resolution rainfall data from 2017 onwards, we examined the dynamics of urban flooding in Singapore. In total, 108 flooding events were reported for the period 2017-2023, all of a transient nature, primarily linked to cloudbursts. Based on the unique precipitation characteristics associated with urban flash flooding, the term 'burst flooding' is introduced to refer to urban floods caused by intense, short-duration rainfall events. A notable increase in cloudburst occurrences in November and December during La Ni & ntilde;a years emphasises the role of global climate phenomena in local weather.

The Soil Moisture Active Passive (SMAP) radiometer can provide dependable ocean wind measurements for tropical cyclones (TCs), almost without being affected by rain. However, SMAP cannot resolve the eyes of TCs because of its coarse spatial resolution. Therefore, it is a challenge to independently identify the storm centers and the wind structures. In this study, we develop a reliable method for estimating TC metrics - center locations, intensities, and wind radii (R34, R50 and R64) - purely from SMAP, without any external data input. Due to SMAP's relatively coarse spatial resolution, a simple vortex model is used to estimate storm center locations from SMAP-measured ocean winds. To validate our methodology, a large dataset is extracted from SMAP measurements. Spanning 663 TC scenes, the dataset covers all storm intensity categories that have occurred in the northern Hemisphere during 2015-2022, and includes 93 cases where stepped frequency microwave radiometer (SFMR) and synthetic aperture radar (SAR) measurements are coincident, within 30 min. SMAP-estimated TC metrics are then comprehensively compared with best-track reports, and when available, SFMR and SAR observations. For collocated SFMR and SAR measurements, the average differences for storm center locations are approximately 36 and 51 km, respectively. The root-mean-square errors (RMSEs) between SMAP-estimated TC intensities and those obtained from SFMR and SAR are 6.18 and 7.51 m/s, respectively. For wind radii, the RMSEs are 42.93 km (R34), 35.96 km (R50) and 32.23 km (R64), respectively. In addition, the average difference between our model and best-track reported TC center locations is 49.2, 32.4 and 23.5 km for tropical storms, category 1-2 storms, and major storms, respectively. The RMSE between SMAP-estimated storm intensities and best track reports is 6.82 m/s. For wind radii, the RMSE values are 47.16 km(R34), 34.94 km (R50) and 21.25 km (R64), respectively.

High-resolution observations of ocean surface wind (OSW) are crucial for analyzing and forecasting tropical cyclones; however, current data are insufficient. While data sets such as the Cross-Calibrated Multiplatform (CCMP) and ERA5 provide gap-free wind fields, their 25 km resolution and tendency to underestimate cyclone wind speeds limit their utility. In contrast, synthetic aperture radar (SAR)-derived OSW offers high-resolution observations but suffers from sparse spatiotemporal coverage. To bridge this gap, we develop a new deep learning-based method to fuse CCMP and globally sourced SAR-derived OSW, combining the strengths of both data sets. SAR-derived OSW from Sentinel-1, RADARSAT Constellation Mission, and RADARSAT-2 satellites are used as ground truth, with corresponding CCMP data as input for training. We evaluate deep learning architectures including UNet, DeepLabV3+, and TransUNet, finding that TransUNet offers superior performance. TransUNet reduces CCMP errors by 50% in root mean square deviation for wind speeds above 20 m/s, 51% in maximum wind speed, and 62% in the radius of maximum wind (RMW) compared with SAR wind. Against IBTrACS data, TransUNet decreases CCMP errors in maximum wind speed by 47% and in RMW by 74%. Our method accomplishes three objectives simultaneously: downscaling CCMP, filling SAR's spatial and temporal gaps, and correcting CCMP biases. By producing uniformly distributed, gap-free, high-accuracy, and fine-resolution OSW for tropical cyclones, our method can benefit multiple areas of tropical cyclone research, such as rapid intensification, hindcasting, and forecasting. Furthermore, this method can be adapted to downscale other geophysical variables, paving the way for high-resolution climate modeling and analysis.

PM 2.5 affects air quality, therefore, understanding the mechanism of PM 2.5 growth is essential to figure out mitigation measures. Hourly real-time concentrations of water-soluble inorganic ions (WSIIs), including anions and cations, in fine particulate matter (PM2.5) were measured in Baoji, northwest China. During the winter monitoring period, the concentrations of PM 2.5 and most WSIIs exhibited similar trends. Mass proportions of SNA [i.e., sulfate (SO42- ), nitrate (NO3-), ammonium (NH4+)] in PM 2.5 gradually increased with air deterioration, while equivalent ratios of anions to cations also increased. The heterogeneous aqueous reactions and/or gas-phase homogeneous reactions promoted the formation of secondary inorganics, especially during the haze events. Rapid transformations of primary gaseous precursors to secondary pollutants could lead to the substantial formation of SO42- and NO3-. In terms of particle growth rate, the mass proportions of SNA in PM 2.5 decreased from General Growth (GG) to Explosive Growth (EG) events. Furthermore, the particle growth rates did not coincide with the pollution levels, while it occurred most frequently during the Transition Period, instead of the Polluted Period. The diurnal variation of SNA at different PM 2.5 growth rates has been discussed. The results of the Random Forest (RF) model showed that RH was an important factor for EG of PM2.5, while low RH was a reliable reason for the relatively low mass proportion of SNA. The results of this study could advance our understanding of particle growth and provide scientific evidence to support the establishment of unique air quality control measures under different pollution scenarios in Fenwei Plain, China.

Weakly absorbing reactive trace gases play important roles in the atmospheric environment and usually have short lifetimes ranging from seconds to days. HIRAS-II, the second hyperspectral infrared atmospheric sounder aboard the world's first civilian meteorological satellite in dawn-dusk orbit, FengYun-3E (FY-3E), can theoretically detect more than a dozen weakly absorbing reactive trace gases and make important contributions to global trace gas mapping by filling the gap for diurnal variation. This study uses state-of-the-art weak absorber thermal infrared spectral feature quantification and identification methods to detect weak absorbers from FY-3E/HIRAS-II and successfully capture 14 species from 35.4 million FY-3E/HIRAS-II clear-sky measurements in July 2023. We map the reliable global distribution of spectral features from nine routine reactive gases and find that these gases originate from scenes that are usually of special concern, including densely populated areas, vegetation, and biomass burning. This study confirms the capability of FY-3E/HIRAS-II in detecting weak absorbers and serves as a stepping stone for subsequent research in concentration retrieval. The case of the ammonia column over wildfires retrieved using neural network technology initially demonstrates that FY-3E/HIRAS-II can improve our understanding of the diurnal variation of trace gases by complementing measurements at dawn and dusk.

Tropical marine low cloud feedback is key to the uncertainty in climate sensitivity, and it depends on the warming pattern of sea surface temperatures (SSTs). Here, we empirically constrain this feedback in two major low cloud regions, the tropical Pacific and Atlantic, using interannual variability. Low cloud sensitivities to local SST and to remote SST, represented by lower-troposphere temperature, are poorly captured in many models of the latest global climate model ensemble, especially in the less-studied tropical Atlantic. The Atlantic favors large positive cloud feedback that appears difficult to reconcile with the Pacific—we apply a Pareto optimization approach to elucidate trade-offs between the conflicting observational constraints. Examining ~200,000 possible combinations of model subensembles, this multi-objective observational constraint narrows the cloud feedback uncertainty among climate models, nearly eliminates the possibility of a negative tropical shortwave cloud feedback in CO2-induced warming, and suggests a 71% increase in the tropical shortwave cloud feedback. Atlantic and Pacific low-cloud variability is difficult to simultaneously capture in climate models. Here, multi-objective optimization reconciles cloud feedbacks in the two basins, constraining tropical cloud feedback toward higher values.

Downward surface solar radiation (DSSR) is critical for the Earth system. It is well-known that DSSR over land has fluctuated on decadal timescales in the past. By utilizing a combination of station observations and the latest CMIP6 simulations, here we show that DSSR had a global consistent decline during 1959-2014, with comparable contributions from greenhouse gases (GHGs) and anthropogenic aerosols. The role of GHGs is even more important in the satellite period. The contribution from GHGs comes through rising temperature, which reduces the DSSR by increasing water vapor but is partly offset by reduced cloud. Future changes of DSSR are heavily dependent on climate change scenarios, which can be predicted well by global mean surface temperature (GMST) and aerosol concentrations. The sharp aerosol reduction and weak temperature rise in the SSP245/SSP126 scenarios will limit or stop the long-term decline of DSSR thus leading to a brighter future.

The reduction of Arctic sea ice concentration (SIC) is a key indicator of global warming. In September 2012, SIC reached its lowest recorded value. Since then, sea ice melt has slowed down, showing a linear trend of only -0.4±6.8%/decade from 2012 to 2023, compared to -11.3±3.3%/decade from 1996 to 2011. Here, we demonstrate that the recent slowdown in September sea ice melt is closely coupled with the multi-decadal variability of the preceding summer North Atlantic Oscillation (NAO), which has transitioned from the lowest point of its negative phase in the early 2010s to a positive phase. During this shift, decreased heat and moisture, along with reduced downward longwave radiation, have contributed to offsetting the long-term decline, leading to a slowdown in Arctic sea ice melting. Additionally, the Atlantic Multidecadal Oscillation plays a primary role in driving the interdecadal variability of the NAO and Arctic sea ice by modulating wave-mean flow interactions.

Processes at the air-sea interface govern the climate mean state and variability by determining the exchange of momentum, heat, and water between the atmosphere and ocean. Traditional climate models compute those exchanges across the air-sea interface by assuming an ocean surface with roughness determined by atmospheric wind and stability conditions, essentially assuming ocean surface waves are in equilibrium states. In reality, that is rarely the case. Such effects have been emphasized in numerical weather predictions for weather systems like tropical cyclones. An accurate representation of ocean surface waves requires a prognostic ocean surface wave model. The addition of WAVEWATCH III to the Community Earth System Model version 2 (CESM2) makes it possible to parameterize the impacts of ocean surface waves on the momentum and energy exchange. This study documents the implementation of a sea-state-dependent surface flux scheme in CESM2. It considers the effects of waves on ocean surface roughness and those of sea spray on sensible and latent heat. It is found that the new scheme significantly impacts mean atmospheric circulation and the upper ocean. The errors in mean atmospheric circulation and surface temperature patterns are reduced. The modified surface flux lowers the eddy-driven jet speed and weakens the Hadley circulation. Global sea surface temperature (SST) warm bias is reduced due to the cooling of the Southern Ocean and eastern boundary currents. In particular, some parts of eastern and central Pacific exhibit a weak cooling trend in the simulation for recent decades, reducing the existing SST trend bias in CESM2.