Atmospheric and Oceanic Sciences

The high heterogeneity of resource types in multi-cloud computing environments and the complexity and diversity of billing models make the workflow scheduling problem more complex and challenging when scientific workflows need to be completed within strict deadline constraints. Although existing research has progressed in reducing the completion time of workflow scheduling in multi-cloud environments, many difficulties still exist in minimizing scheduling costs while satisfying deadlines. Therefore, a random forest enhanced particle swarm optimization algorithm (RFPSO) is proposed in this study. The RFPSO algorithm implements intelligent initialization of resource allocation through a random forest model, which improves the efficiency of finding optimal solutions. Moreover, it designs a reflective boundary constraint mechanism and a hierarchical task allocation mechanism based on critical path. This design ensures that critical tasks can prioritize access to higher-performance computing resources, which effectively guarantees that tasks within the workflow are scheduled and completed by their deadlines. In addition, the quality of the optimal solution is improved through a local neighborhood search mechanism. Experimental results on scientific workflow datasets such as Epigenomics and Montage show that, compared with existing state-of-the-art methods, RFPSO reduces execution costs by an average of 57.31%.

Abstract Rapid progress in the field of machine-learning for weather prediction has led to the emergence of algorithms whose forecasting skill can exceed that of traditional physically based models. This development represents an opportunity to improve the quality of forecasting services provided by operational centers, particularly given the speed at which machine-learning based models generate predictions. Despite the clear promise of these systems, questions remain about the ability of the current generation of machine-learning models to generate physically consistent predictions of the full suite of required forecast fields under all conditions. Answering these questions will require careful comparisons between the well-understood physically based models, current state-of-the-art machine-learning models, and the hybrid models that combine elements of these two archetypes. The Weather Prediction Model Intercomparison Project (WP-MIP) is a World Meteorological Organization-supported initiative whose initial goal is to create a centralized database of physically based, machine-learning, and hybrid model forecasts to enable a distributed assessment and evaluation effort. The first instance of WP-MIP focuses on global deterministic predictions using both center-specific and common initializations to facilitate sensitivity studies. Forecasts contributed by institutions across six continents will be used to develop AI-ready verification techniques that highlight the strengths and weaknesses of each class of prediction system, with the goal of establishing best-practice guidance to model developers and national weather centers. The broad engagement of the operational and forecast-evaluation communities in WP-MIP will ensure that the project’s results are highly relevant to the development and deployment of next-generation weather prediction systems.

Abstract. Wildfires in the Amazon, increasingly influenced by climate variability and anthropogenic activities, pose severe environmental and health challenges. While drought events amplify fire activity and emissions, the cascading effects of droughts and deforestation on air quality and health remain underexplored. This study addresses this gap by combining satellite observations of fire activities with the Global Fire Emissions Database (GFEDv4s) and the chemical transport model, GEOS-Chem High Performance (GCHP) to quantify the impacts of droughts and deforestation on fire emissions, air quality, and health risks from 2010 to 2015. “Fire-on” and “fire-off” simulation reveal that biomass burning dominates dry-season (July–November) air quality, contributing 50 % to regional CO and PM2.5 and 33 % for ozone in non-drought years. These contributions increase to 60–80 % for CO and PM2.5 and 50 % for ozone during drought years. Significant correlations between pollutant levels and drought intensity reflect a climate-driven amplification of fire impacts. Using the Global Exposure Mortality Model (GEMM) and exposure-response relations, we estimate that fire-induced PM2.5 and ozone increase premature mortality by 6.0 % and 18.6 % in non-drought years, which rise to 8.9 % and 24.4 % during drought years. These findings underscore the critical roles of droughts in exacerbating fire emissions and health risks, even under stable deforestation rates. This study highlights the urgent need for integrated wildfire management and climate adaptation strategies to protect public health and achieve sustainability goals.

Abstract The Indian Ocean Dipole (IOD) significantly modulates the Australian climate via extratropical Rossby wave teleconnections. However, current Coupled Model Intercomparison Project Phase 6 models generally overestimate the IOD amplitude while underestimating its remote impacts on Australian precipitation and surface air temperature. This historical underestimation stems from a deficient atmospheric response over the western tropical Indian Ocean. Due to the overly strong and westward-extending eastern pole of the IOD, models fail to generate sufficient diabatic heating and upper-level divergence in the west. The western basin is the key upstream region capable of bypassing the reflective waveguide barrier of the Southern Hemisphere subtropical jet. This discrepancy impacts the Rossby wave source to build the equivalent-barotropic dynamic bridge to Australia. Because this historical teleconnection pathway is flawed, future projections are highly uncertain. While models project intensified IOD impacts under high-emission scenarios, this amplification may be related to background jet changes rather than a robust IOD response.Correcting spatial biases in tropical diabatic heating is thus essential for reliable regional climate predictions.

Abstract. Accurate simulation of snowmelt runoff (SMR) is critical for water resource management. However, despite the abundance of global hydrological models, little is known about their SMR performance. This study presents a comprehensive evaluation of SMR across 15 state-of-the-art large-scale models and runoff products by focusing on their biases in first-order indices, i.e., the total volume (Qsum), peak flow (Qmax), and centroid timing (CTQ) of runoff in the snowmelt period. Then by introducing 1455 snow-dominated basins with diverse topography and vegetation complexities, we further proposed a novel model robustness metric to test how different models perform under increasing basin complexity, thereby allowing for a quantification on how they adapt to complex environmental conditions. Our results reveal that (1) most models exhibit underestimated Qsum and Qmax and predict CTQ too early. These biases are particularly pronounced in regions such as the western United States, northern Europe, and northeastern China. (2) Model biases systematically increase with basin complexity, with CTQ exhibiting strong sensitivity to mean elevation and topographic variability, while Qsum and Qmax being shaped more by mean elevation and the diversity of vegetation types in the basin. (3) The robustness assessment further shows that observation-constrained runoff products exhibit the most outstanding performance (i.e., low biases and strong adaptability to stern conditions), followed by the hydrological and land surface models. Notably, while global hydrological models generally exhibit stronger robustness in simulating Qsum and Qmax, land surface models show a clear advantage in simulating CTQ, highlighting their structural strength in capturing melt timing rather than runoff magnitude. This study provides a large-sample benchmark for SMR evaluation and complements existing model assessment approaches by examining model performance across basin complexity gradients, offering useful insights for future model development and uncertainty reduction.

Non-stationary time series are common in many real-world domains, including infectious disease spread, where the underlying relationships between variables evolve over time. However, most existing forecasting methods assume stationarity and fail to capture changing causal dynamics. To address this challenge, we propose the Causal Regime Bayesian (CaReBayes) forecasting framework, which integrates regime detection, causal discovery, and Bayesian forecasting within a unified approach. CaReBayes segments time series into regimes using temporal causal discovery, fits a Bayesian structural autoregressive model for each regime, classifies the current regime, then performs regime-specific forecasting with uncertainty quantification. The framework introduces methodological advances: a grid-search procedure for automated regime-dependent causal discovery, a classification method that assigns future observations to regimes based on learned Bayesian structures, and regime-conditioned Bayesian structural forecasting. Across both synthetic and Ontario COVID-19 time series data, CaReBayes outperforms benchmark models for time series forecasting. In addition to improved forecasting performance, it produces regime-dependent causal graphs that summarize candidate structural relationships in the system, enhancing interpretability.

Transfer of ocean heat toward Antarctica's ice shelves is a key driver of changes in glacial mass balance. Previous work has indicated that transfer of warm, mid-depth waters across the continental slope is elevated close to Antarctica's largest ice shelves, where dense waters form over the continental shelf. Yet, ice shelves in such regions maintain relatively low area-averaged melt rates, suggesting that ocean processes must shield them from the inflowing heat. In this study, a high-resolution simulation is used to isolate the mechanisms of heat transfer toward the Filchner-Ronne Ice Shelf in the southern Weddell Sea. This study shows that ocean eddies divert almost all heat arriving on the continental shelf up to the ocean surface and away from the ice shelf by stirring heat anomalies along tilted density surfaces. These findings highlight the importance of understanding continental shelf eddies in the context of basal melt variability, for example, in driving future heat delivery to Antarctica's largest ice shelves.

Abstract Observations and numerical simulations are used to analyze the environment and processes that caused a record flash flood over central Long Island, New York on 13 August 2014. A ∼20 km wide band of intense rainfall resulted in nearly 330 mm of rain over 3 hours at Islip, New York. The rainfall was enhanced by low-level convergence along a surface trough along with ∼400 J kg −1 of convective available potential energy and a low-level jet that advected moisture into the area. This surface trough amplified as it rotated northward over Long Island due to low-level potential vorticity generated by latent heating within the band. All operational high resolution (< 4-km grid spacing) models and a 31-member ensemble of 3-km Weather Research and Forecasting using different initial conditions and physics underpredicted the precipitation by more than a factor of two. Excessive explicit convection that the model spins-up south of Long Island over the ocean depletes the instability and weakens the jet advecting moisture into the band, resulting in the rain band dissipating too quickly. This is shown by using smaller explicit (3- and 1-km) nests just encompassing Long Island, which resulted in less precipitation south of Long Island and thus more favorable band ingredients and thus a much greater precipitation forecast (∼450 mm in 3 hours). An ensemble of this small domain setup illustrates some of the parameterization sensitivity in the outer 9-km domain, in which the precipitation band failed to develop for some members using a different convective scheme.

Abstract. Ground-based and satellite atmospheric observations are essential for reducing uncertainties in methane (CH4) emissions by atmospheric inversion, particularly in data-sparse regions such as Africa. However, adding new observation sites does not yield linear improvements of emission uncertainties because overlapping transport sensitivities reduces marginal information gain. Here we develop a Bayesian framework to strategically optimize CH4 observation network design for column retrievals from upward-looking Fourier Transform Infrared (FTIR) spectrometers (e.g., EM27/SUN), jointly identifying the optimal number of sites and their spatial configuration. The framework quantifies uncertainty reduction for grid-point (1°) total and sectoral emissions while accounting for transport redundancy, cloud screening, and observational errors. Using January and July as representative months, we find that uncertainty reduction increases rapidly during early network expansion but gradually saturates beyond a certain number of additional sites. An optimized configuration of ten new sites added to the existing network achieves over 65 % reduction in prior uncertainty for total African CH4 emissions in both months, with comparable improvements across fire, wetland, and anthropogenic sectors. Sensitivity analyses indicate that while the optimal number of sites varies with assumptions about cloud filtering, the spatial configuration remains robust, supporting cost-effective observation network design in data-sparse regions.

Thunder generates atmospheric acoustic waves that couple into the ground, producing seismic signals called "thunderquakes." Although widely observed, this atmosphere-solid Earth conversion has rarely been exploited for imaging because the coupled wavefield is complex and its governing physics are poorly constrained. Here, we show that thunderquakes recorded by preexisting telecommunication fiber-optic cables using distributed acoustic sensing (DAS) contain coherent air-coupled Rayleigh waves that can serve as seismic energy sources for near-surface tomography. We validate this mechanism using three-dimensional spectral-element simulations and dispersion modeling of thunderquakes. We analyze 2.5 years of continuous DAS data and a catalog of 458 high-fidelity thunderquakes validated by lightning records. Cross-correlation virtual-source interferometry and stacking yield dispersed surface waves from which we invert shear-wave velocity structure to ∼100-meter-depth beneath an urban karst setting. The resulting tomographic image reveals several previously undetected weak zones, some coinciding with surface deformation measured by Interferometric Synthetic Aperture Radar. The tomographic results are further validated by independent borehole logs and engineering surveys. Our results establish that thunder energy can be converted into dispersive seismic wavefields in the solid Earth and that thunderquakes can act as novel, meteorologically driven sources for seismic imaging in regions with limited access to traditional seismic sources.

Abstract. Causes of model uncertainty in complex modeling systems can be identified using large perturbed-parameter ensembles (PPEs), combined with statistical emulators to increase sample size and enable variance-based sensitivity analyses and observational constraint. In global climate models such as the UK Earth System Model (UKESM), these approaches are typically applied at the global or regional mean scales for a limited set of variables. Accelerating progress in understanding the multi-faceted causes of climate model uncertainty, requires implementing such workflows at the model grid box scale, to enable analyses across variables that reveal how uncertainties propagate and interact spatially. However, this approach requires training millions of Gaussian process (GP) emulators and fitting an equal number of generalized additive models (GAMs) – a major computational bottleneck. We present a high-performance, open-source pipeline that introduces optimisations for this workflow. For GP emulation, we implement task-level parallelism and streamlined data handling on high-performance computing systems. For GAM fitting, we integrate a parallelized pyGAM interface with R's mgcv::bam() back end, using fast fREML estimation with discrete smoothing, memory-efficient batching, and improved input–output routines. These changes reduce GP training time by 97.5 % (6177 s → 154 s) and GAM fitting time by 95.2 % (10 623 s → 511 s), yielding a ∼ 25 × faster end-to-end workflow (96 % total runtime reduction) and cutting peak memory use by a factor of 12. Outputs are numerically identical to the baseline implementation (Pearson correlation = 1.00 for both GP and GAM predictions). We demonstrate the approach using a UKESM PPE comprising 221 members scaled up to 1 million using GP emulators, and GAM fits applied to output for a single target variable, to show that the improved performance enables multi-variable, higher-resolution, and potentially multi-model analyses that were previously impractical. These improvements pave the way for PPE studies to scale in scope without compromising statistical fidelity, enabling more comprehensive exploration of model parameter uncertainty within feasible HPC budgets.

Abstract. The efficient and accurate simulation of grounding line dynamics in marine ice-sheet models remains a challenge, largely due to restrictive time-step limitations. The restrictive time step size of ice-sheet simulations (∼ ∆t = 0.01−0.5 years) has led to the routine use of approximate models that compromise physical complexity compared to full Stokes models. To address the time-step restriction and enhance the applicability of full Stokes simulations, we implement a numerical stabilisation scheme at the ice-ocean interface, namely the Free-Surface Stabilisation Algorithm (FSSA). The FSSA acts by predicting the surface elevation at the next time step, resulting in a reduction in surface oscillations and an increase in the largest numerically stable time step. When applied to the ice-ocean interface, FSSA acts in combination with the sea spring numerical stabilisation scheme, allowing larger time steps to be taken. In order to test the capabilities of the FSSA when applied to the ice-ocean interface, we perform the benchmark simulation of Experiment 3a from the Marine Ice Sheet Model Intercomparison Project (MISMIP). These simulations demonstrate the ability of the model to capture grounding line migration on both prograde slopes (oceanward sloping) and retrograde slopes (inland sloping). We find a time step size of ∆t = 10 years to be numerically stable and accurate in the MISMIP experiment, which is more than an order of magnitude larger than the small time steps traditionally used. In comparison, a time-step size of ∆t = 50 years can maintain numerical stability, but is not capable of capturing the full range of grounding-line motion in the MISMIP experiments. We further demonstrate the applicability of the FSSA to a 3D marine terminating model domain, finding that a time-step size of ∆t = 10 years is numerically stable. The increase in the largest numerically stable time step by greater than an order of magnitude in marine-terminating Stokes ice-flow problems through the inclusion of FSSA broadens the applicability of Stokes models, which have otherwise been deemed too computationally expensive for large-scale applications.

Sea-level rise has large impacts on coastal areas. Many approaches to quantify the net benefits of regional or global coastal adaptation rely on strong assumptions about economically efficient decision-making and may neglect critical uncertainties about future extreme water levels and socioeconomic development. In particular, the deep and dynamic uncertainties associated with future sea levels complicate efforts to adapt coastal areas and infrastructure to account for sea-level rise. Robust decision-making (RDM) provides a way to identify adaptation strategies that perform well across uncertain sea-level futures. In this work, we quantified the impacts of an RDM approach on regional and global coastal adaptation using an economic regret, defined as the difference in cost of a strategy compared to the “optimal” outcome. We modeled decision-making using a regret-based criterion and computed the economic regret of each adaptation decision candidate, which we then compared to classical decision-making approaches. We found that the majority of coastal segments that changed strategies under regret-based criteria opted for a higher level of adaptation, primarily by expanding their retreat elevation. Although the total adaptation costs remained comparable to those under the cost-minimizing criterion, the use of regret led to a 3–50% reduction in flood damage, highlighting the advantages of prioritizing robust outcomes over purely cost optimization. While our analysis assumed immediate implementation of adaptation and evaluated a limited set of decision criteria, the results demonstrate that incorporating regret-based decision-making into coastal impact models can provide policy-relevant insights for designing more robust adaptation strategies under deep uncertainty.

ABSTRACT Changes in precipitation phase and seasonal water supply can modify land‐surface moisture memory, but their implications for summer heat extremes remain insufficiently understood in midlatitude dryland transition zones. Northern China has experienced substantial hydroclimatic changes in recent decades, yet it remains unclear whether these changes have altered the efficiency with which spring snowmelt regulates subsequent summer heat extremes. Using daily observations from 348 meteorological stations during 1980–2019, we examined long‐term changes in precipitation phase and seasonal hydroclimatic conditions and assessed how effective cumulative spring snowmelt ( M AM ) is linked to summer heat extremes. We identified a hydroclimatic transition around 2000, with northern China shifting from a ‘warming‐without‐wetting’ to a ‘warming‐wetting’ regime, mainly associated with increased summer rainfall. This transition was spatially heterogeneous: northwestern China experienced persistent wetting with increases in both snowfall and rainfall, northeastern China became wetter primarily because of increased rainfall, whereas North China retained a warming‐drying tendency. Against this background, response analysis showed that greater M AM was generally accompanied by fewer and weaker summer heat events. The regulatory association of M AM was consistently stronger for T max than for N hot . Pathway analysis further suggested that antecedent upper‐root‐zone soil moisture and surface energy partitioning mediated the snowmelt–heat linkage, with additional modulation by summer rainfall. After 2000, mean effective cumulative spring snowmelt decreased, soil moisture and latent heat flux decreased, while sensible heat flux and summer heat metrics increased. These changes indicate a weakened M AM ‐related regulation of T max and a reduced efficiency of snowmelt‐derived moisture buffering under the post‐2000 hydroclimatic background. These results suggest that recent hydroclimatic change in northern China has altered not only seasonal water availability but also the land‐surface regulation of summer heat extremes, with implications for understanding how climate warming reshapes regional water–heat interactions and amplifies heat risk.

Abstract. Nitrogen oxides (NOx = NO + NO2) in the troposphere form an array of secondary pollutants that are detrimental to air quality, ecosystems, and climate. The family of reactive oxidized nitrogen (NOy) in the atmosphere consists of NOx and its reservoir species (e.g. HNO3, PAN). Our understanding of the processes underlying the transformation of NOy has advanced considerably over recent decades, however, the relative importance of NOy partitioning and loss pathways remain uncertain. In this study, we use the GEOS-Chem global chemical transport model and observations from the ATom flight campaign to assess the simulated global budget of tropospheric NOy, and the production and loss fluxes between key NOy species. Our simulation indicates that the mean global chemical lifetime of NOx is ∼ 23 h and the mean global deposition lifetime of NOy is 5.5 d. The global mean NOx:NOy ratio is 0.23 at the surface (over continents it is 0.34) and is 0.10 at 500 hPa. In addition to the four most prevalent gas-phase species (NO, NO2, HNO3, PAN) that have been central to previous descriptions of tropospheric NOy chemistry, we find that other species play key roles in driving overall chemical cycling. The model representation of organic nitrate chemistry is highly simplified and likely overestimates the importance of hydrolysis as a sink while underestimating deposition. Finally, the photolytic loss of particulate nitrate (pNO3-) to form NO2 and HONO, as represented in our simulations, is comparable to its depositional loss, indicating the importance of further constraining this photolysis sink.

Abstract. Impacts of anthropogenic aerosols on clouds and snowfall during winter precipitation events remain highly uncertain, particularly under heavy pollution. A winter snowfall event over the Guanzhong Basin (GZB) and its surrounding regions (GZBs), China, has been simulated using a cloud-resolving, fully coupled WRF-Chem model to quantify the respective roles of aerosol–radiation interactions (ARIs) and aerosol–cloud interactions (ACIs). The simulated temporal variation and spatial distribution of air pollutants and precipitation generally agree with the observations in the GZB+GZBs. Sensitivity experiments are performed to evaluate effects of ARIs and ACIs by changing the anthropogenic emissions. The precipitation response to ARIs and ACIs exhibits regional contrast in GZB and GZBs due to different aerosol concentrations. In the GZB, exclusion of ARIs leads to a slight increase in precipitation with increasing emissions, mainly associated with enhanced ice-phase precipitation induced by ACIs. ARIs increase the precipitation in the GZB when emissions increase reaches a threshold, caused by ARI-induced enhancement of relative humidity (RH) which increases ice water path and favors survival of falling ice particles. In contrast, precipitation in the GZBs decreases with increasing emissions, reflecting suppression of liquid-phase precipitation by ACIs and reductions in RH caused by ARIs. In addition, changes in anthropogenic emissions exert limited influence on the spatial distribution of precipitation across the combined GZB–GZBs region. These findings provide process-level insight into how ARIs and ACIs regulate snowfall under polluted conditions, with implications for improving aerosol–precipitation coupling in regional climate and weather models.

Abstract Increasingly frequent extreme weather events are heightening the chances of mosquito‐borne diseases outbreaks, posing a notable threat to human health. However, how extreme weather influences the transmission dynamics of mosquito‐borne diseases remains poorly understood. Here, we integrate heat waves and effective precipitation as parameterization schemes into an SEI‐SEIR dynamic model to quantify their driving effects on chikungunya transmission. Based on the 2025 chikungunya fever outbreak in Foshan, the model successfully reproduced the observed transmission dynamics. Attribution analysis revealed that heat waves exerted a dual effect on transmission. Mosquito suitability declined by 9.7%, yet accelerated viral replication and altered human behavior generated a positive forcing that drove a net 76% increase in infected cases. Effective precipitation regulated outbreak scale nonlinearly, dispersed rainfall increased mosquito abundance by 205% and amplified transmission by 10.5%. Our research offered new insights into the impact of extreme weather on mosquito‐borne infectious diseases.

Accurately mapping environmental and social inequalities at fine spatial scales is critical for urban policy, yet the data required are often costly and infrequently updated. Vision foundation models can extract information directly from satellite imagery, offering a rapid and scalable alternative. We compare three modelling pipelines for predicting two contrasting indicators of urban inequality - air pollution and house prices - across England on a fine hexagonal grid: regression models trained on structured features of form and function (census, land cover and urban morphology), models trained on 128-dimensional image embeddings from a geospatial foundation model and a hybrid of the two, each evaluated with and without coarse regional context. Structured features achieve the best overall accuracy, but image embeddings become competitive once regional context is added - most clearly for air pollution, where image-only models reach a median R2 of 0.78 (0.85 with regional context), indicating that the embeddings capture genuine image signal. For house prices the picture is more cautious: image-only models achieve a median R2 of around 0.58; however, much of the embeddings' apparent gain reflects coarse spatial location rather than image content. Off-the-shelf satellite embeddings, while not yet surpassing data-intensive approaches, are a promising low-cost complement, particularly for rapid, large-scale, or repeated analyses and in settings where traditional data are limited.

Abstract Climate change has intensified and increased the frequency of extreme heat and wildfire smoke events, posing growing risks to health in carceral settings where infrastructure limitations and regulatory gaps heighten vulnerability. This study characterized temporal trends and geographic patterns in extreme heat and wildfire smoke exposure across 410 California correctional facilities from 2000-2023. We used ERA5-Land reanalysis data to calculate daily maximum temperature and identified days exceeding the 28°C threshold corresponding to California's Indoor Heat Illness Prevention Standard. Wildfire smoke PM2.5 exposure was estimated using a validated statistical model generating daily ground-level predictions at 10km resolution. Temporal trends were modeled using facility-level linear regression. Mean wildfire smoke PM2.5 concentrations rose from 0.4 µg/m³ (2006-2010) to 1.4 µg/m³ (2019-2023), a 3.5-fold increase, with Northern California experiencing the highest burden and steepest increases. Annual days above 28°C increased from 27.3% to 29.8% statewide, with the Central Valley and desert regions showing the highest current burden. On average, facilities experienced 6.8 dual heat and smoke burden days per year during 2019-2023, with Central Valley and Sierra foothill facilities most affected. These findings underscore the need for region-specific mitigation strategies and updated regulatory protections for people living and working in California correctional facilities.

Abstract Global terrestrial ecosystems exhibit substantial interannual variability (IAV) in net carbon (C) flux. Determining the biogeographic origin of this variability is essential for the understanding and forecasting of global C cycling and carbon-climate feedbacks. Currently, most studies identify either global drylands or moist tropical forests as the dominant source of IAV. Considering this, we investigated whether the use of three different global ecosystem classifications of drylands and moist tropical forests, as well as two alternative geographical scales, could alter which ecosystem is the dominant contributor to terrestrial net C flux IAV. Using the simulation results of 18 dynamic global vegetation models from the TRENDY v11 model intercomparison, we calculated the absolute and area-weighted contributions of net C flux IAV for: individual 0.5° grid cells, global ecosystem classifications, and ecoregions (intermediate scale between grid cells and global ecosystems). For all three of the global ecosystem classification schemes, we found the drylands IAV contributions of 41%, 32%, and 37% were significantly greater than the associated IAV contributions of 20%, 19%, and 24% from the moist tropical forests ( p < 0.001). However, the moist tropical forests had a higher IAV contribution per unit area across all three classification schemes (∼3% versus ∼1%–2% ( p < 0.001)). At the ecoregion scale, this switch between drylands and moist tropical forests was absent; as seven of the ten highest absolute and nine of the ten highest area-weighted contributing ecoregions were drylands. Specifically, we found tropical and subtropical grasslands, savannas, and shrublands to be particularly substantial contributors to global terrestrial net C flux IAV, with the Cerrado’s absolute IAV contribution of 3.52% exceeding all but one of the other 763 global ecoregions IAV contributions (all p < 0.05). Our findings demonstrate that drylands persist as the dominant contributor to global terrestrial net C flux IAV, irrespective of different global ecosystem classifications or geographic scales.

Global streamflow reanalyses such as GloFAS-ERA5 are available everywhere yet lose fidelity in small, snow- and glacier-fed headwaters that feed Central Asia’s transboundary rivers. We present an explainable, calibrated deep ensemble framework that corrects the GloFAS-ERA5 log-residual at gauges of the Syr Darya and Amu Darya systems using the CA-discharge archive. An entity-aware long short-term memory (LSTM) backbone drives a regime-gated mixture of experts trained under a closed-form mixture continuous ranked probability score (CRPS) and augmented with snow physics constraints and a regime-conditional (Mondrian) conformal layer; skill was assessed under temporal holdout, leave-one-basin-out and prediction in ungauged region protocols, with grouped Shapley value attribution. Correction rendered all 74 gauges skillful, raising the median modified Kling–Gupta efficiency (KGE′) from 0.386 (raw) to 0.825 (flagship); on temporal point skill the framework is statistically tied with gradient boosting (paired Wilcoxon p = 0.49). Under-dispersed raw intervals (90% coverage 0.68) were recalibrated to near-nominal coverage (~0.90), and high-flow exceedance decision skill was moderate (Q90 Brier skill score 0.34, ROC-AUC 0.92), while low-flow (Q10) exceedance showed no skill over climatology. Transfer to ungauged, more glacierized catchments was a measured limit that degraded with glacier fraction and basin area. The framework’s value is calibration, an inspectable (supervised) regime structure and regional physical insight, not point skill superiority.

Atmospheric water vapor is a major driver of Earth’s climate, yet despite its vital role in driving extreme weather and informing Numerical Weather Prediction (NWP) models, precise quantification of Precipitable Water Vapor (PWV) remains a challenge due to its high spatial and temporal variability. To provide an upgraded evaluation reflecting the most recent data and next-generation instrumentation, this study evaluates the accuracy, relative to radiosonde measurements, in calculating PWV values across six different instruments: Microwave Radiometer (MWR), NOAA-21, TROPOMI, Pandora spectrometer, AERONET sun-photometer, and GNSS/GPS against the bias-corrected Vaisala RS-92 and RS-41 radiosondes over Beltsville, Maryland, utilizing an updated 2024–2025 database. As a certified GRUAN site, HUBC adheres to strict international observation standards designed specifically to provide reference-quality data and comprehensive corrections for systematic errors. This rigorous framework justifies their application as the definitive ‘referent truth’ benchmark for remote sensing validation. RMSE and bias were the primary metrics used to assess relative accuracy. GPS measurements provided the highest level of relative accuracy, yielding the lowest RMSE (1.50 mm) and a near-unity linear fit (y = 0.98x). NOAA-21 and TROPOMI exhibit higher random noise when compared to ground-based instrumentation, yet both obtain high relative accuracy retrievals with negligible biases. AERONET and Pandora also showed strong performance with low RMSEs and R2 values of 0.986 and 0.988, respectively, while slightly underestimating PWV. In contrast, the Radiometer performed with the lowest relative accuracy, characterized by the highest RMSE (6.19 mm) and a significant negative bias (−5.55 mm). Although all instruments maintained high correlation coefficients (R2 ≥ 0.904), these results indicate that satellite and ground-based remote sensing provide reliable PWV retrievals, while GPS presents the most robust benchmark for high-accuracy PWV retrievals relative to radiosonde measurements. The findings also underscore the need for instrument-specific calibration constants to better align remote sensing retrievals with in situ observations.

Abstract. During the SOUTHTRAC mission in autumn 2019 elevated mixing ratios of carbon monoxide (CO), carbon dioxide CO2, nitrogen oxide (NO) and total reactive nitrogen NOy were observed during a flight at the beginning of October. The potential plume extended over more than 1000 km (15° latitude) east of the Brasilian coast at altitudes of 13 km in the upper troposphere. In-situ measurements showed elevated ozone in this plume (≈ 100 ppbv), being 20–40 ppbv higher than during a previous flight in early September at exactly the same flight route. For the plume flight positive correlations of ozone and pollutants (CO, NO, NOy) indicate ozone production in these pollution layers. Lagrangian Analysis shows, that the observed air masses were strongly affected by biomass burning over Amazonia. A combined analysis of chemical Lagrangian box model and a global chemistry climate model (EMAC) revealed that ozone production from biomass burning predominantly caused the ozone enhancements. The effect is eventually intensified by NOx produced from lightning. Upward transport of the plumes happened ≈ one week before the flight, allowing ozone to be formed and enhanced by 25 % compared to the September flight. Estimates of the climate impact show, that the biomass burning produced ozone has a local effect on the radiative forcing of 50 mWm−2.

Abstract The Labrador boundary current system significantly impacts freshwater pathways and upper-ocean stratification in the subpolar North Atlantic, which in turn influences the state of the Atlantic Meridional Overturning Circulation. In this study we analyze an extensive collection of shipboard hydrographic and velocity sections across the Labrador shelf and slope, occupied in the summer and fall seasons since 1940 along the Seal Island transect. The mean sections provide a quantitative view of the boundary current system and its velocity structure, consisting of the Labrador Coastal Current (LCC), the Labrador Current (LC), and a permanent recirculation associated with Hamilton Bank situated between the two currents. The LCC, which flows adjacent to the coast, is strongly influenced by local wind forcing. It has exhibited increasing volume and freshwater transports since around 2010, while its baroclinic component has progressively strengthened since the early 1970s, largely influenced by increased river runoff from northern Canada. By contrast, the LC, situated at the shelfbreak, is more closely linked to larger scale atmospheric patterns – captured by the North Atlantic Oscillation (NAO) – rather than local winds. Over the past decade the LC has been in a strengthened regime, associated with a positive phase of the NAO, with enhanced equatorward freshwater transport. On the mid-shelf, the recirculation and its associated mixing facilitate a net offshore flux of freshwater.