Mesoscale convective systems (MCSs) are important producers of severe weather events and are influenced by aerosols that are either locally emitted or remotely transported. Despite widespread recognition of aerosol–MCS interactions, how aerosols from local and distinct sources alter MCS cloud microphysics and precipitation remains inadequately quantified. To address this knowledge gap, this study investigates an extreme-precipitation-producing MCS event in South China affected by concurrent local anthropogenic pollution and remote biomass-burning aerosols. Using the Weather Research and Forecasting-Chem model, four sensitivity experiments are conducted, including clean, local emissions only, remote transport only, and all emissions scenarios. For the investigated MCS, results show that aerosol perturbations increase the convective-region-mean rainfall intensity by around 20%. Local aerosols, characterized by high hygroscopicity and high concentrations in the lower troposphere, promote droplet activation and exert a relatively stronger influence on precipitation during the early analysis period, but are rapidly depleted by wet scavenging. In contrast, remotely transported aerosols are less susceptible to wet scavenging and maintain a more persistent aerosol supply throughout the MCS lifecycle, resulting in a relatively stronger influence during the late analysis period in this event. Moreover, the presence of remote aerosols in the remote-only and full emission experiments may contribute to a larger stratiform fraction by generating smaller snow particles in the stratiform region with slower fall speeds. Our results underscore the necessity for an accurate representation of the sources, vertical distribution, and hygroscopicity of aerosols in numerical weather prediction models.
Thunderstorms and electrified clouds maintain the global electric circuit, whose fair-weather return current contributes to the atmospheric potential gradient (PG) near the surface of the Earth. At land observatories, aerosols, clouds, atmospheric conductivity, boundary-layer dynamics, and site effects modify the measured PG, producing substantial day-to-day variation. Conventional averaging can conceal this variability by merging distinct daily patterns into a single composite. Here we show that clustering corrected Nagycenk Observatory measurements by normalized PG shape alone identifies five daily morphology families, without using meteorology, season, PG magnitude or disturbance, or Carnegie similarity during cluster formation. Using 8642 daily curves from 1962 to 2009, the selected descriptive resolution (K = 5) separates evening-peak, midday-peak, Carnegie-resembling evening-peak, early-morning-peak, and morning-peak families. Post-hoc diagnostics show that the midday- and morning-peak families occur under warmer, drier conditions with stronger solar radiation and lower fractions of highly disturbed days. In contrast, the other evening-peak and early-morning-peak families occur under more humid and cloudy conditions and show greater disturbance. The Carnegie-resembling family has the strongest reference-curve similarity (r = 0.920), but 28.7% of its days have high disturbance scores, showing that morphological similarity alone does not establish global-circuit attribution. By separating morphology identification from physical interpretation, this approach provides a reproducible basis for comparing long-term land PG records without imposing their physical origins.
Reliable day-ahead forecasts of global horizontal irradiance (GHI) are essential for photovoltaic scheduling and power-system risk management. This study evaluates the condition-dependent performance of the Rapid-refresh Multi-scale Analysis and Prediction System-WIND version 3 (RMAPS-WINDV3) and the European Centre for Medium-Range Weather Forecasts (ECMWF) day-ahead GHI forecasts over China in 2024. Ground-based radiation observations, numerical forecasts, Himawari-9 radiation retrievals, and Himawari-9 Cloud Property (CLP) cloud-type labels are integrated within a common station-based verification framework. Forecast errors are first calculated at individual stations and then aggregated by climate sub-region, and RMAPS-WINDV3 minus ECMWF differences are assessed using station-paired bootstrap resampling. The results reveal pronounced metric dependence. RMAPS-WINDV3 has significantly lower station-averaged mean absolute error (MAE) than ECMWF in five of the seven climate sub-regions, whereas its station-averaged root-mean-square error (RMSE) is significantly higher in five sub-regions. Northwest China is the only climate sub-region in which both metrics are significantly lower for RMAPS-WINDV3, with ΔMAE and ΔRMSE values of −50.13 and − 29.88 W m−2, respectively. In contrast, positive ΔRMSE values occur in most other sub-regions and reach +17.43 W m−2 in Tropical South China and + 14.94 W m−2 in Central and South China. Season-hour stratification shows that the higher RMSE of RMAPS-WINDV3 is concentrated during photovoltaic-relevant midday hours, with a mean 10:00–14:00 BJT ΔRMSE of +13.45 W m−2 and a summer mean of +26.32 W m−2. Himawari-9 cloud-type stratification further shows that clear-sky conditions generally produce smaller or negative ΔRMSE, whereas the largest positive cloud-regime ΔRMSE values vary by climate sub-region and range from +30.63 to +87.63 W m−2. These findings demonstrate that the relative performance of RMAPS-WINDV3 and ECMWF depends on region, verification metric, season, hour, and cloud regime, rather than following a universal ranking of the two forecast systems.
Complex mountainous terrain distorts background electrostatic fields and contributes to spatial variations in lightning occurrence. This study develops an electric-field-line mapping framework and defines the Zero-Line Density Index (ZLDI) for terrain-scale assessment. Field lines from uniformly distributed initial positions are traced to the terrain, and ZLDI is constructed from the mean-normalised kernel density of their terminations along terrain arc length. Under two-dimensional Laplacian conditions, flux conservation shows that continuous unsmoothed ZLDI has the same normalised spatial form as the surface-normal electric-field enhancement factor when the initial normal field is approximately uniform. ZLDI is therefore interpreted as a scale-resolved mapping implementation of electric-flux concentration rather than an independent electrostatic quantity. The framework is evaluated using idealised terrains and six transects in Guangdong and Hubei Provinces. At a bandwidth of 850 m, the four Guangdong cases yield mean Pearson correlation and Top-20% coverage values of 0.784 and 77.43%, respectively. For the two Hubei cases, the Pearson correlations are 0.709 and 0.907, and coverage values are 88.24% and 88.06%. ZLDI generally provides stronger spatial agreement than elevation, slope, and raw or scale-matched near-ground field indicators, although elevation outperforms it for the C2 multi-ridge terrain, where directional diagnostics indicate three-dimensional influence. An illustrative leader comparison shows case-dependent agreement. A representative 1000-line ZLDI calculation requires 86.4 s, compared with 3454.9–6705.9 s for simplified leader simulations. ZLDI is positioned as a first-stage terrain-scale screening metric rather than a predictor of lightning attachment.
The meridional mode of the East Asian summer subtropical westerly jet (EASWJ) exerts a pronounced influence on East Asian summer weather and climate. Observational results indicate that, due to the combined influence of the Quasi-Biennial Oscillation (QBO) and El Niño–Southern Oscillation (ENSO), the meridional oscillation of the EASWJ exhibited a significant quasi-biennial signal during the historical period, which played an important role in modulating the meridional distribution of summer water resources in East Asia. However, how state-of-the-art climate models represent the quasi-biennial signal of the EASWJ remains unclear. Here, we analyze ERA5 reanalysis data together with 15 CMIP6 models to examine the quasi-biennial signal in the meridional mode of the EASWJ and to assess the respective and combined influences of the QBO and ENSO. Among all examined models, only EC-Earth3-Veg successfully reproduces both the individual and combined influences of the QBO and ENSO on the meridional mode of the EASWJ. Most models exhibit a quasi-biennial spectral peak in the EASWJ meridional mode, but it is generally weak, reflecting their limited ability to capture the combined modulation of the EASWJ by the QBO and ENSO. The main reason is that most models fail to reproduce the observed QBO influence on the EASWJ, primarily due to biases in the simulated climatological jet position, the QBO amplitude, and the representation of the QBO's impact on tropical convection. In contrast, nearly all models capture the ENSO–EASWJ linkage reasonably well, although in some models the simulated position of the jet response is displaced, possibly owing to biases in the location of ENSO sea surface temperature anomalies.
