Atmospheric and Oceanic Sciences

Abstract Space weather poses an important but under‐quantified threat to society. While severe geomagnetic storms are recognized as potential global catastrophes, their socio‐economic impacts remain poorly quantified. We present a novel physics‐engineering‐economic framework that links geophysical drivers to power grid geoelectric fields, transformer vulnerability, and macroeconomic consequences. Using the United States as an example, we estimate daily economic losses for a 250‐year geomagnetic storm from transformer thermal heating of 1.81 billion USD (95 percent confidence interval: 1.65 to 1.96 billion USD), disrupting power for approximately 5.1 million people and 135,000 businesses. These estimates are conservative lower bounds, reflecting only transformer thermal heating effects and excluding voltage collapse, cascading failures, and restoration costs. The true societal risk could be substantially higher. Nonetheless, this contribution provides the first nationwide end‐to‐end coupling from space physics to potential macroeconomic loss, with quantified uncertainties. Our results demonstrate that coupled socio‐economic modeling of space weather is feasible and essential, and the framework is scalable and transferable, offering a template for assessing space weather risk to critical infrastructure in other countries.

Abstract Tropical deforestation driven by agricultural expansion to meet growing food demand poses a rising threat to the global carbon cycle. While previous studies quantified agriculture‐related carbon losses, the role of cropland expansion remains poorly resolved, partly because widely used agricultural driver data sets are too coarse to detect fine‐scale forest‐to‐cropland conversion, especially in topographically complex regions. Here we integrate multiple high‐resolution satellite observations to show that tropical forest‐to‐cropland conversion reached 1.08 Mha yr −1 during 2004–2019, accelerating by 0.35 ± 0.07 Mha yr −1 per four‐year period ( p < 0.05). These conversions shifted toward steeper terrains, with mean slope rising by 0.16 ± 0.05° per period. Annual forest carbon loss nearly tripled from 46 to 130 TgC yr −1 , increasing at 30 ± 6 TgC yr −1 per period ( p < 0.05). Our findings reveal a systematic shift of agricultural frontiers toward marginal, carbon‐rich tropical forests, underscoring the need for more targeted land‐use governance to support global climate mitigation goals.

Abstract NORi is a machine learning (ML) parameterization of ocean boundary layer (BL) turbulence that is physics‐based and augmented with neural networks. NORi stands for neural ordinary differential equations Richardson number (Ri) closure. The physical parameterization is controlled by a Richardson number‐dependent diffusivity and viscosity. The neural ODEs are trained to capture the entrainment through the base of the BL, which cannot be represented with a local diffusive closure. The parameterization is trained using large‐eddy simulations in an a posteriori fashion, where parameters are calibrated with a loss function that explicitly depends on the actual time‐integrated variables of interest rather than the instantaneous subgrid fluxes, which are inherently noisy. NORi conserves tracers by design, uses realistic nonlinear thermodynamics, and demonstrates excellent prediction and generalization capabilities in capturing entrainment dynamics under different convective strengths, background stratifications, rotation, and wind forcings. NORi is shown to simulate the seasonal evolution of the BL at Ocean Weather Station Papa with similar performance to the state‐of‐the‐art two‐equation closure. When implemented in a double‐gyre simulation, it is numerically stable for at least 100 years, despite only being trained on 2‐day horizons, and can be run with time steps as long as 1 hr. Combining highly expressive neural networks with a physically grounded base closure proves to be a robust paradigm for designing parameterizations for climate models: data required and training cost are drastically reduced, inference performance can be directly optimized as a primary objective, and numerical stability is implicitly promoted through training.

Abstract Understanding the coastal zone of the Antarctic Ice Sheet (AIS), where it interacts with the Southern Ocean and warmer air masses, is crucial for predicting Antarctica's influence on the global climate and sea level. This region has multiple tipping mechanisms that could trigger large, rapid, and potentially irreversible changes in the AIS, the Southern Ocean and their global connections in the coming centuries. The AIS remains the largest source of uncertainty in future sea‐level projections. Bed topography beneath the ice shelves and the coastal ice sheet is not yet well documented, and is a major source of this uncertainty. This review assesses current knowledge of the coastal zone and highlights methods to investigate it, including aerogeophysical surveys, ground‐ and ship‐based measurements, satellite observations, and computer modeling. An ensemble analysis of published bed topography data sets identifies significant data gaps and their regional distribution, framed in the context of current ice‐sheet behavior and potential instability. We propose scientific priorities and guidelines for future aerogeophysical surveys, advocating for a comprehensive, coordinated international effort to build a next‐generation data set of Antarctic bed properties. Such an initiative would significantly advance understanding of the role of coastal processes in ice‐sheet dynamics, reducing uncertainties in sea‐level rise projections and improving predictions of future ocean and climate changes.

Biomass burning (BB) significantly disturbs ecosystems and threatens regional and global climate, air quality, and human health through the massive emission of pollutants. Carbon monoxide (CO) and nitrogen dioxide (NO2) generated from these fires are key components in atmospheric chemistry, revealing combustion processes and efficiency. Over the past two decades, low-Earth-orbit (LEO) platforms have played a dominant role in trace gas monitoring; however, their snapshot sampling capabilities are unable to capture the rapid diurnal evolution of fire emissions, leading to systematic uncertainties in emission inventories. In this study, we integrate observations from the Geostationary Interferometric Infrared Sounder (GIIRS) onboard FY-4B and the Geostationary Environment Monitoring Spectrometer (GEMS) onboard GK-2B to monitor biomass burning over Southeast Asia. Validation against TROPOMI and ground-based networks (TCCON and Pandora) demonstrates the reliability of this combined dataset. Time-series analysis (July 2022–June 2025) shows that regional CO and NO2 variations exhibit high consistency with fire radiative power (FRP). Focusing on the representative fire hotspot of Northern Laos, we observe a bimodal diurnal NO2 pattern driven by the interplay of emissions, photochemistry, and meteorology. Specifically, we identified a nonlinear response of NO2 growth to fire intensity. Observational evidence suggests that under extreme burning conditions, the conversion of NOx is constrained by limited atmospheric oxidative capacity. We further quantified the intraday dynamics of combustion efficiency (indicated by the enhancement ratio, ER = ΔNO2/ΔCO), revealing significant temporal fluctuations. This pronounced diurnal variability confirms that single-overpass LEO observations introduce a systematic estimation bias in emission factors. This study provides observational constraints for refining emission inventories and demonstrates a framework for applying next-generation global geostationary satellite constellations to fire monitoring.

Abstract Accurate near-real-time (NRT) air quality forecasting is critical for protecting public health but remains challenged by uncertainties in emissions, meteorology, and initial conditions within traditional chemical transport models (CTMs). We present an observation-driven machine learning framework that integrates multisource satellite products, including VIIRS aerosol optical depth and TEMPO NO2 with U.S. EPA AirNow measurements into a machine-learning surrogate of the NOAA Unified Forecast System (DeepAQM). A ConvLSTM-ResNet fusion module reconstructs spatially continuous surface PM2.5 and O3 fields from sparse observations, providing updated initial conditions for 72 h forecasts across the continental United States. DeepAQM reproduces UFS-AQM spatial patterns with strong agreement (R2 > 0.9 for O3 and up to 0.80 for PM2.5) while maintaining normalized mean bias within ±10%. Observation-updated initial conditions improve early forecast skill, increasing site-level R2 during the first 10 forecast hours from approximately 0.10–0.15 to 0.40–0.50 for PM2.5 and from approximately 0.40 to 0.50 for O3. Observation-based fine-tuning provides additional improvements throughout the 72 h forecast period, increasing overall R2 to 0.41 for PM2.5 and 0.82 for O3 during the independent evaluation period. These results demonstrate that integrating satellite and ground observations within DeepAQM can enhance NRT air quality forecasting and provide a scalable pathway toward globally deployable observation-driven prediction systems.

Abstract Quasi-stationary convective bands over Kyushu, Japan, frequently trigger rainy-season disasters, and hours with ≥50 mm h −1 rainfall are increasing. However, skillful nowcasts beyond 3 h remain limited. This study presents FlowsNet, an observation-based multisensor fusion model that learns directly from radar/rain gauge-analyzed precipitation, surface variables from ground stations, geostationary satellite imagery, and satellite-derived precipitation context. The model targets category-4 (C4; ≥50 mm h −1 ) rainfall and incorporates two attention mechanisms: a channel-wise module that weights informative modalities and a spatial module that aligns features with banded structures at multi-hour leads. Training uses a tail-aware ordinal loss that couples focal reweighting with an Earth Mover’s Distance-based ordinal penalty to emphasize rare extremes. FlowsNet maintains a non-zero C4 Critical Success Index through 6 h. From 4 to 6 h, it matches or exceeds the Japan Meteorological Agency’s very-short-range forecast, and it outperforms a leading extrapolation-based method and state-of-the-art deep learning nowcasting models. Case studies show preserved band geometry and corridor placement at long lead over complex terrain. Ablation experiments identify satellite water vapor context and near-surface humidity as key for long-lead C4 prediction; combining satellite context with surface observations stabilizes placement and reduces false alarms. By avoiding numerical weather prediction model state and objective analyses/reanalyzes, the approach reduces latency and hardware demand, improves portability and resilience when model cycles degrade, and offers a practical route to earlier and more transferable warnings for extreme-rainfall events.

Floods degrade road networks at the same time as demand for emergency medical services (EMSs) rises, yet national EMS command systems rarely receive any information on flood-induced road barriers. This paper presents Ambulance STARS (SaTellite-assisted Ambulance Routing System), a service-oriented framework that links satellite observation with ambulance dispatch. A cloud-based flood detection service derives flood extent from Sentinel-1 SAR amplitude change detection executed in a cloud-based Earth observation data and compute backend and translates it into road passability layers. A routing engine then maintains an in-memory road graph whose travel times are calibrated with empirical ambulance speed models built from four years (2020–2023) of GPS records of an EMS fleet in southern Poland, with separate speeds for driving with and without emergency signals (61.8 and 37.2 km/h, respectively). An API gateway with single-file tile delivery, a replicated relational data tier, and an observability stack complete the architecture, and a web client offers dispatchers live routing and multi-unit incident simulation. The framework was tested on the September 2024 flood in the Municipality of Nysa, Poland. The SAR module delineated 665 ha of inundation and marked 8.5 km of the 656.7 km routing network as impassable (508 barrier points), and the same procedure applied to a reference optical mask of 18 September yielded 17.3 km and 1006 points. Because the SAR and optical acquisitions captured different phases of the flood wave, agreement on the rare impassable-road class was low, and the two products were, therefore, used to bracket operational uncertainty rather than to define a single ground truth. Applied without local retuning to Lewin Brzeski, the same flood detection workflow showed consistent performance against the CEMS reference product. The routing module produced statutory 8/15/20 min accessibility maps in 12–34 s under warm-cache benchmark conditions. With SAR-derived barriers, the share of the network reachable within 15 min fell from 88% to 80%, and 2 villages with 938 inhabitants lost road access to EMS entirely. With barriers derived from the optical mask, the 15 min share fell to 39.8% and seventeen settlements lost road access entirely, underlining how strongly the barrier source shapes the operational picture. Post-acquisition processing completes in under one minute under warm-cache conditions with road data preloaded, and satellite-derived road passability is fast enough to support near-real-time decision-making, subject to the constellation revisit time and to integration with EMS command systems.

Abstract Marine heatwaves (MHWs) perturb air–sea CO 2 fluxes, but their cumulative impact on the global coastal carbon sink remains poorly understood. Using four observation-based CO 2 flux datasets, with priority given to a global coastal product, we find that MHWs enhance net CO 2 uptake in global shelf seas by 11.0 ± 1.6% from 1985 to 2020, in contrast to reduced uptake in the open ocean. The enhancement is largely attributable to polar and subpolar shelf seas, where frequent MHWs coincide with sea-ice loss and non-thermal reductions in dissolved inorganic carbon (DIC), outweighing reduced uptake or enhanced outgassing in lower-latitude, warming-dominated regions. Analysis with a global ocean biogeochemical model suggests that the observed DIC reductions are largely linked to enhanced biological carbon fixation. Our findings reveal region-specific MHW impacts on coastal carbon dynamics, highlighting a critical role of high-latitude shelf systems in the global carbon budget under ongoing climate change.

Abstract A long-standing question in atmospheric science concerns how, and to what extent, climate variabilities across different timescales interact to shape global weather and climate extremes. Here we address this issue through a quantitative framework that integrates both large-scale climate controls and upscale feedback processes. Coherent co-variability across multiple timescales appears throughout the global monsoon (GM) domain. Land precipitation on the interannual timescale exhibits a coherent in-phase relationship across all the sub-monsoon systems, largely governed by El Niño-Southern Oscillation (ENSO). Meanwhile, Madden-Julian Oscillation (MJO)-related precipitation accounts for about 30–65% of total intraseasonal precipitation variability across the sub-monsoon regions. Significant positive correlations are found between the interannual variability (IAV) and the intensity of the intraseasonal oscillation (ISO) and between the ISO and the intensity of synoptic-scale variability (SSV) across the entire GM domain, indicating a large-scale control of lower-frequency modes to higher-frequency variabilities. It is further demonstrated that higher-frequency variabilities (e.g., SSV and ISO) exert marked upscale feedback to lower-frequency modes through nonlinear rectification of condensational heating. About 90% of extreme precipitation days occur during the active phase of the ISO across the GM domain, laying a foundation for sub-seasonal prediction of weather and climate extremes weeks and even months ahead.