Atmospheric and Oceanic Sciences

There were errors in Results and discussion, Section 3.2 Influence of extraction methods on the UV-Vis properties of BDOM, paragraphs 1 and 2, page 5. The values of SUVA₂₅₄ correction (shown in bold) appear below: "However, these extracts exhibited much lower SUVA 254 values (Figure 2b, 0.05 L mg C⁻ 1 m⁻ 1 ), suggesting that strong alkaline conditions may have altered aromatic structures through hydrolysis, thereby reducing the apparent aromaticity of the dissolved fraction." "Conversely, water-extracted BDOM showed the highest SUVA 254 value (0.82 L mg C⁻ 1 m⁻¹).""Neutral salt extracts (CaCl₂: 0.64; NaCl: 0.32 L mg C⁻ 1 m⁻ 1 ) exhibited intermediate aromaticity, reflecting their ion-exchange-mediated release mechanism that selectively mobilizes specific aromatic constituents (41).""The acidic extractants (SPLP: 0.11; TCLP: 0.04 L mg C⁻ 1 m⁻ 1 ), and especially HCl (0.02 L mg C⁻ 1 m⁻ 1 ), generated BDOM with progressively reduced aromaticity.There were errors in Results and discussion, Section 3.3 The impact of extraction methods on BDOM fluorescent components, paragraphs 2-5, pages 6 and 7 regarding the reported fluorescence intensity values for the TCLP, water and NaCl extracts. Due to a transcription error, the C1, C2, C3 and C4 values and SUVA₂₅₄ values were incorrectly reported."Protein-like component C1, associated with tryptophan-containing compounds (47, 48), showed the highest fluorescence intensity under high ionic strength (NaCl extraction: 809 [R.U.]) and weakly acidic conditions (SPLP: 761 [R.U.]), which may reflect ionic stabilization of charged biomolecules and mild protonation effects, respectively." "Under strongly acidic conditions, C2 was absent in the HCl extract and detected only at trace levels in the TCLP extract (107 [R.U.]). These results coincide with a pronounced reduction in SUVA 254 (to 0.04), consistent with the sensitivity of fulvic-like fractions to protonation and possible precipitation under acidic conditions." There was a mistake in Table 3, page 7 as published. Due to a calculation error (all SUVA₂₅₄ values were inadvertently multiplied by an extra factor of 100), the SUVA₂₅₄ values in Table 3 were incorrectly reported as being 100 times higher than the correct values.The corrected Table 3 (shown in bold) appears below: The original version of this article has been updated.

Deep convective clouds are frequently associated with extreme weather events such as torrential rainfall and thunderstorms, making their accurate identification critical for improving short-term weather forecasting and enabling effective disaster early warning. Most existing deep learning approaches for cloud classification rely on conventional convolutional neural networks (CNNs), which suffer from limited receptive fields and struggle to capture long-range contextual dependencies essential for robust cloud segmentation. To overcome this limitation, we propose DIF-UNet, a novel U-Net architecture that combines a Swin Transformer and a CNN-based residual network as dual encoders. The model introduces two key components: (1) a Semantic Information Fusion Module (SIFM) that adaptively integrates heterogeneous semantic features from both encoders using learnable parameters and (2) a Cross-Layer Feature Interaction Module (CLFI) that enhances hierarchical feature synergy to preserve fine-grained spatial details while reducing information loss. Evaluated on FY-4A multi-channel satellite observations, DIF-UNet achieves state-of-the-art performance with an intersection-over-union (IoU) score of 77.78% and an F1 score of 87.39%, outperforming baseline models by 3.84% and 2.49%, respectively. These results demonstrate that explicitly modelling both local and global contextual information significantly improves deep convective cloud recognition accuracy.

Abstract On 4 July 2025, rainfall ≥ 254 mm (≥ 10 in.) fell on the Guadalupe River basin headwaters in Kerr County, Texas, causing the deadliest flash flood disaster in the U.S. since the Big Thompson, Colorado, event of 1976. Flooded Locations And Simulated Hydrographs (FLASH) predictions driven by Multi-Radar Multi-Sensor (MRMS) rainfall estimates correctly identified the specific location and magnitude of the flash flooding threat with 1–2 hours of lead time. Addition of storm-scale model precipitation forecasts to drive FLASH could potentially increase lead time, but providing accurate and precise, short lead-time rainfall forecasts is particularly challenging. In a hydrologic context, spatial locations of extreme rainfall are paramount as basin boundaries can segregate incoming rainfall into different watersheds, reducing their combined hydrologic impact, or conversely, amplifying the impact by concentrating rainfall into a single watershed. In this case, rainfall was concentrated within the Guadalupe River basin headwaters. NSSL’s Warn-on-Forecast System (WoFS) targets these short lead times for which predictions may become increasingly precise, but are best described probabilistically owing to the random nature of individual thunderstorms. In this study, ensemble rainfall forecasts from an experimental 1-km grid-spacing WoFS (WoFS-1km) configuration are used to drive an ensemble of FLASH hydrologic forecasts (WoFS-FLASH) for the 4 July 2025 event. The WoFS-FLASH system accurately predicted the location and magnitude of the event in at least 50% of members with more than 5 h of lead time, providing strong motivation to develop a coupled WoFS-FLASH for real-time applications.

Abstract Ocean tides are a major energy reservoir in Earth's climate system, yet global tide models systematically fail to reproduce observed tidal energetics. This limitation arises because existing models do not adequately represent how tide‐generated internal waves in the deep ocean regulate ocean tide energy, accounting only for their initial generation and neglecting the wave stresses associated with their subsequent propagation. These wave stresses can be out of phase with the tidal flow and substantially modify tidal amplitudes. Here we explicitly represent these wave stresses in a global tide model and assess their impact on the energetics of global tides. We show that including these wave stresses is essential for reproducing observed tidal energetics and enables energetically consistent simulations even at coarse spatial resolution. These results highlight a previously missing physical process in tide models and facilitate more accurate and efficient modeling of global ocean tides.

Abstract Urban flooding is one of the most damaging impacts of climate change. The two main causes of changes in rainfall‐driven flooding are altered precipitation and changes in urbanisation. Yet attribution studies only focus on the former. Furthermore, very few event attribution studies examine subdaily rainfall extremes, the timescale at which such extremes are accelerating the most. Here, for the first time, we carry out an impact event attribution study examining the effects of both climate change and urbanisation on a flash‐flooding event in the U.K. city of Leeds. By combining a convection‐permitting climate model with a flood inundation model, we show that the extent of flooding in the urban area of Leeds was increased during this event in 2014 by 49% compared to the potential flooding that would have occurred if a similar rainfall event had taken place 30 years earlier. The increase from urbanisation (29%) is almost twice that from climate change (16%). Both factors combine nonlinearly to increase flood extent by more than the sum of their parts. Our results also show that urban flood risk could be significantly misrepresented if changes in rainfall intensity are used as a proxy for changes in pluvial flooding.

Abstract. Soils across permafrost regions are one of the largest terrestrial pools of mercury (Hg) in the world, storing an estimated 500–1500 Gg of Hg in the top three meters of soil. Ongoing climate-driven thaw threatens to release this legacy Hg into the environment. Efforts to quantify and model this pool have been hindered by a lack of harmonized, spatially resolved observations. To address this, we compiled a database of 117,802 Hg observations collected between 1988 and 2022 from 59 studies across Arctic, sub-Arctic, and alpine permafrost regions of the Northern Hemisphere, including North America, northern Europe, Eurasia and the Tibetan Plateau. The database includes Hg concentration measurements in solid materials – such as soil, leaves, roots, and wood – as well as in water samples from soil porewater, lakes, and rivers across the northern hemisphere permafrost domain. The database enables cross-site synthesis, model calibration and evaluation, and environmental assessments by standardizing and harmonizing data from diverse sources. Data standardization included unit conversion, categorization of observations by type, and quality-control procedures to ensure consistency across studies. Analytical uncertainty was preserved where reported in source studies, and quality control indicators – including range and outlier flags – were applied to support data screening and interpretation. Mercury concentrations vary widely across observations, with lake sediment showing the highest median values (70 ng g−1, IQR: 45–116), followed by soil (50 ng g−1, IQR: 32-90), and vegetation (15 ng g−1, IQR: 9–33). Water observations (total Hg) had a median of 2 ng L−1 (IQR: 2–6). Statistically significant differences in Hg concentrations among observation types were observed at both global and regional scales, generally following the pattern: lake sediment > soil > vegetation, although this ordering is sensitive to regional sampling distribution. These patterns, along with spatial and observation-type biases, highlight the need for improved coverage in underrepresented regions such as Eurasia. The database is freely accessible through Zenodo under the concept https://doi.org/10.5281/zenodo.18300989 (all versions; Olson et al., 2026a), to support ongoing research and model development in Arctic and sub-Arctic Hg cycle studies.

Abstract. Hydroxymethanesulfonate (HMS) is a critical source of particulate sulfur, formed by formaldehyde (HCHO) and sulfur dioxide (SO2) in droplets. Current models relying on bulk aqueous-phase HMS formation only explain ∼ one-third of unexplained sulfate concentrations, leaving gaps in atmospheric sulfur budget, especially in polluted and cold environments. Using Born–Oppenheimer molecular dynamics simulations, we explored HMS and its isomer hydroxymethyl sulfite (HMSi) formation mechanisms across aqueous phase and air–water/ice interfaces. Air–water interfaces enable nearly barrierless HMS formation (0.6 kcal mol−1) via unique stepwise water-mediated proton transfer, preferring HMS over HMSi (0.6 vs. 6.1 kcal mol−1), which contrasts sharply with the competitive pathways observed in the bulk aqueous phase (7.7 vs. 7.6 kcal mol−1). In contrast, protonation of formaldehyde under strongly acidic conditions reverses reaction selectivity, favoring HMSi formation over HMS. Importantly, these reaction mechanisms remain viable at air–ice interfaces in cold environments including polar areas and the upper troposphere, revealing ice surfaces as previously overlooked yet significant sites for atmospheric organosulfate formation. Our findings suggest that interfacial mechanisms may provide efficient pathways for HMS and HMSi formation in both polluted and cold environments, helping to reconcile model-observation discrepancies in the atmospheric sulfur budget.

Abstract Understanding the composition of carbonaceous aerosols, black carbon (BC) and organic aerosols (OA), remains a major challenge in atmospheric science. Using data from two aircraft campaigns with identical instrumentation over Europe and East Asia, we analyze statistical relationships between concentrations of five trace gases (CO, NO 2 , HCHO, O 3 , and SO 2 ) with BC and OA in order to estimate carbonaceous aerosol in urban pollution plumes. We show that across both campaigns, CO is the best proxy for BC ( R 2 ≈ 0.6). In plumes, OA shows statistical links with NO 2 , O 3 , and CO, reflecting the combined influence of emissions, and secondary organic aerosol formation. Linear regressions based on trace gases remain limited, especially for OA, whereas the use of nonlinear machine-learning regression improves the quantification of BC and OA ( R 2 ≈ 0.9 for BC, R 2 ≈ 0.7 for OA). However, the number of flights is limited, the results should not be interpreted as applicable to flights in other regions and seasons. Our findings indicate that co-emitted and co-produced trace gases contain information for quantifying carbonaceous aerosol in urban pollution plumes. This potential is more robust for BC, whereas OA remains more complex to estimate because it depends on multiple predictors.

Abstract Through a suite of global simulations using the Global-to-Regional Integrated forecast SysTem (GRIST) model, this study investigates the impact of different scale-aware convective parameterization schemes (CPS) on typhoon simulation across resolutions from 60 to 3.75 km. Three implementations are examined: scaling convective adjustment time (CTRL), scaling cloud-base mass flux (SA-MB), and scaling entrainment rate (SA-ER). All CPS show improved performance with increasing resolution. CTRL and SA-MB exhibit comparable scale-adaptivity, whereas SA-ER demonstrates stronger CPS suppression and outperforms the other two, reducing mean track errors by 10–45 km and intensity underestimation by 8–47%. Decomposition of the surface pressure tendency budget reveals that SA-ER enhances microphysical processes more vigorously, more than offsetting the reduction of convective component and yielding a net ∼10% increase in precipitation and diabatic heating, thereby deepening the storm. These benefits hold across most resolutions but diminish at the convection-permitting scale for typhoons that are not sufficiently compact and intense, where excessive outer-rainband warming disrupts the secondary circulation and weakens intensification. Explicit convection without CPS also performs poor at this resolution, producing the largest track and intensity errors among all experiments—underscoring that even at convection-permitting scales, scale-aware CPS remains indispensable. The results of this study demonstrate the advantages of scale-aware CPS on typhoon modeling, with the entrainment scaling approach emerging as a particularly effective pathway for improving typhoon intensity forecasts.

Under the background of global climate warming, glaciers in the Three-River Headwaters Region, as a crucial component of the “Asian Water Tower,” exert profound influences on regional water resource security and ecological stability through their mass balance variations. Due to the scarcity of in situ observations caused by the harsh high-altitude environment, long-term monitoring based on remote sensing techniques is urgently required. In this study, the geodetic method was employed, using the SRTM-C DEM acquired in 2000 as the reference, and recent glacier surface DEMs were generated from high-resolution ZiYuan-3 tri-stereo imagery obtained during 2024–2025. Through refined DEM co-registration, differencing, and systematic error corrections, the glacier mass balance in the Three-River Headwaters Region from 2000 to 2025 was systematically estimated. The results indicate that the glaciers in the study area exhibited an overall negative mass balance during the study period, with significant spatial heterogeneity. Among the sub-regions, the Lancang River source region experienced the most pronounced mass loss (−0.70 ± 0.07 m w.e. yr−1), whereas the Yellow River source region showed the lowest mass loss (−0.37 ± 0.09 m w.e. yr−1). Compared with earlier studies, glacier mass loss has accelerated in recent years and is closely associated with regional climatic characteristics. This study provides a scientific basis for understanding glacier changes and their hydrological and ecological impacts in the Three-River Headwaters Region.