Atmospheric and Oceanic Sciences

Abstract Climate change driven increases in temperature, vapour pressure deficit, and drought frequency threaten to enhance peatland wildfire activity and combustion burn severity, potentially emitting vast stores of ancient carbon. Here, we quantify the sensitivity of northern peatland wildfire combustion to projected climate change using an integrated modelling framework. We used the Global Fire Emissions Database burned area product (v5.1) to train random forest models of North America and Eurasia to predict landscape burn frequency as a function of land cover and climatological variables. Future changes in area burned were estimated from multiple general circulation models and across several climate scenarios. Peat burn severity was estimated from simulated peat moisture profiles modelled with Hydrus-1D, which varied initial water table position, potential evapotranspiration rate, drying time, and peat hydrophysical properties. Ensemble results project a moderate increase in northern peatland area burned by the end of the century (3.2 to 3.6 Mha yr -1 across climate scenarios) compared to our contemporary estimate of 2.75 Mha yr -1 . The combined effects of increased area burned and depth of burn result in projected range of peatland smouldering carbon emissions of 47.5–55.3 Mt C yr -1 by the end of the century, compared to our contemporary ensemble estimate of 35.8 Mt C yr -1 . While modelled burn severity was comparable between North America and Eurasia, Eurasia is estimated to contribute approximately 70% of northern peatland smouldering carbon emissions due to a higher average burn rate and larger peatland area. As a result of increased peatland smouldering carbon emissions, our end-of-century peatland combustion findings suggest that, regionally, peatlands have the potential to flip from a net carbon sink to a net carbon source under a high emission scenario As such, incorporating mechanistic peatland combustion processes into Earth system models is essential to reduce uncertainty in future climate projections and to inform mitigation strategies aimed at preserving the global peatland carbon stock and carbon sequestration function.

ABSTRACT Global warming increases the occurrences of compound hot‐dry and hot‐wet extremes that pose amplified socioeconomic and environmental impacts compared to univariate extremes. Compound meteorological hot‐dry extremes have made substantial progress at the global scale, while compound meteorological hot‐wet extremes are still poorly understood. Meanwhile, there remains a research gap on the compound hydrological hot‐dry (hot‐wet) extremes based on high temperature and low (high) runoff, which is more directly related to damages on social and natural systems. This study investigated the spatiotemporal variations and attribution (e.g., associated univariate extremes and their dependence) of compound meteorological hot‐dry (CMHDE) and hot‐wet (CMHWE) extremes and compound hydrological hot‐dry (CHHDE) and hot‐wet (CHHWE) extremes and effects of large‐scale oceanic modes and SST variability on compound extremes across globe during the warm season (i.e., May–September in the Northern and November–March in the Southern Hemispheres) over 1901–2019. Results indicate that the likelihood of compound meteorological and hydrological hot‐dry and hot‐wet extremes showed an increasing trend in more than 90% of global land during 1901–2019. The frequency of compound extremes from high to low was in the order of CHHWE, CMHWE, CMHDE, and CHHDE. The spatial extent of compound meteorological and hydrological hot‐dry and hot‐wet extremes showed an increasing trend in almost all climate regions over the last 119 years, with higher increasing rates in compound hot‐wet extremes. The positive (negative) dependence between meteorological and hydrological hot‐dry (hot‐wet) extremes was detected in 74.1% and 87.1% of global land, with stronger dependence observed for hydrological than meteorological compound extremes. The high temperature contributed to the reduction of return period of compound hot‐dry and hot‐wet extremes in more than 92% of global land, except for western South America, southern Asia, and central‐eastern North America. The contribution of dependence and precipitation/runoff had high spatial heterogeneity, resulting in increases or decreases of return period in regions across global land. The decreases (increases) in runoff under combined high temperature and low (high) precipitation showed in about 95% (70%) of the globe. The effect of high (low) precipitation on increases (decreases) in runoff was higher than that of high temperature on runoff decreases, indicating that precipitation dominates runoff change more strongly than temperature. The conditional probability of low runoff given CMHDE was higher than that of high runoff given CMHWE. La Niña (El Niño) dampened (enhanced) the likelihood of compound hot‐dry extremes in the northern and central South America, southern and central Africa, Sahara, southern Asia, and Australia. The low (high) Dipole Mode Index (DMI) and Atlantic Multidecadal Oscillation (AMO) dampened (enhanced) the likelihood of compound hot‐dry and hot‐wet extremes in more than 70% of global land, indicating that low (high) DMI and AMO were negatively (positively) related to high temperature. The leading modes between global SST and compound hot‐dry and hot‐wet extremes showed positive correlation and explained 75.9%–84.0% of compound extreme variability during 1901–2019. This study contributes to the understanding of characteristics and driving mechanisms of compound extremes at the global scale under climate warming.

Abstract Tropical cyclones (TCs) making landfall often produce extreme precipitation with complex spatiotemporal patterns, posing major challenges to both deep learning models and numerical weather prediction (NWP) systems. This study introduces Rainflow, a conditional flow matching (CFM) based generative model for high‐resolution TC precipitation nowcasting. For the North Atlantic data set, Rainflow predicts precipitation at 0.01 spatial and 10‐min temporal resolution over 6‐hr lead times. By modeling the residual transition from observed to future precipitation sequences, Rainflow jointly captures precipitation motion and intensity evolution within a unified framework. To improve physical consistency, 500 and 850 hPa wind fields from NWP forecasts are incorporated as dynamic constraints. Experiments show that Rainflow substantially outperforms the operational High‐Resolution Rapid Refresh (HRRR) system and the state‐of‐the‐art NowcastNet, especially for extreme precipitation intensities. Additional experiments on Western Pacific TC cases further demonstrate the potential generalizability of Rainflow under a shorter 3‐hr nowcasting setting.

Atmospheric methane (CH 4 ) plays a central role in Earth’s climate, yet the drivers of its decline during the high-dust conditions of glacial periods, such as the Last Glacial Maximum (LGM), remain uncertain. Previous explanations imply source-driven changes, assuming an atmospheric lifetime comparable to that of present day. Recent work shows that interactions between mineral dust and sea salt aerosols produce CH 4 -removing chlorine radicals. In this work, we show that during the LGM, CH 4 lifetime shortened to 7.8 years, 20% lower than that of present day. Chlorine contributed ~15% of global CH 4 loss, fourfold that of present day. Our results reproduce ice core CH 4 isotopic evidence, demonstrating that stronger-than-assumed atmospheric sinks can explain CH 4 variability without invoking substantial source changes, highlighting the overlooked role of chlorine chemistry in the glacial CH 4 budget.

Abstract Surface Water and Ocean Topography (SWOT) provides water surface elevation (WSE) that can constrain river hydraulics through data assimilation (DA), but basin‐scale impacts of assimilating real observations remain uncertain under heterogeneous availability and regulation. We assimilate node‐level SWOT WSE into a large‐scale hydrodynamic model over the Ohio River Basin using an ensemble Kalman filter and test river‐network‐aware localization. Assimilation improves discharge skill at most of the 66 USGS gages, including sites without direct SWOT overpasses, with ∼60% showing higher KGE and ~3% degrading. These improvements suggest that SWOT WSE assimilation extends beyond direct observations, with localization adding benefit. Impacts are governed by transferable constraints, including quality control (QC), residual observation uncertainty, and river‐network context. Comparison with SWOT‐derived discharge at 27 overlap sites provides complementary context and supports using WSE as a direct hydrodynamic constraint. These results show that SWOT WSE assimilation improves discharge estimates under real‐world constraints, supporting water management in data‐sparse and transboundary basins globally.

ABSTRACT This study presents an integrated framework combining process‐based hydrological modelling with advanced deep learning techniques to improve climate‐driven streamflow prediction and flood risk assessment in the Brahmaputra River Basin (BRB) at Bahadurabad. Unlike conventional comparative studies, this work explicitly evaluates the added value of Transformer‐based architectures against traditional hydrological modelling (HEC‐HMS) and recurrent neural networks (LSTM and BiLSTM) under bias‐corrected CMIP6 climate projections. Historical discharge data from 1981 to 2014 were used for calibration, validation, training and testing. Among the evaluated models, the Transformer demonstrated superior predictive capability, achieving an R 2 of 0.94 during testing with PBIAS within ±1%, whereas HEC‐HMS showed comparatively lower accuracy ( R 2 = 0.72) and consistent underestimation exceeding 20%. Future projections for the 2030s, 2050s and 2080s reveal a systematic shift towards earlier monsoon onset and a 10%–25% increase in seasonal discharge, with mean annual flow rising from approximately 22,000 m 3 /s in the baseline period to nearly 30,000 m 3 /s by the late century. By coupling data‐driven predictions with flood frequency analysis using the Gumbel distribution, this study further demonstrates a pronounced intensification of extreme flood events under future climate scenarios. The findings highlight the enhanced capability of Transformer‐based models to capture complex hydroclimatic dynamics and underscore their potential as a robust tool for climate‐resilient water resources planning in large transboundary river basins.

Abstract Glacial lake outburst floods (GLOFs) are an increasing hazard under climate-driven glacier retreat, yet existing assessments remain divided between simplified large-scale screening and physically based but site-specific simulations. Here we bridge this scale gap by presenting a scalable, physically based framework that enables consistent simulation of GLOF-induced flood propagation and downstream exposure across large spatial domains. We apply the framework to Bhutan and simulate a standardized hypothetical outburst scenario for all 567 glacial lakes across the country. The model reasonably reproduces observed GLOF impacts, demonstrating its applicability at large scales. Results reveal strong basin-scale heterogeneity in flood impacts, with inundation depth and frequency varying widely among basins and locally exceeding 80 m. Importantly, flood impacts are not determined by lake size alone: downstream propagation is strongly controlled by basin topography, while societal exposure is spatially decoupled from physical flood magnitude, such that the largest floods do not necessarily produce the greatest societal impacts. These findings show that GLOF risk emerges from the interaction of lake magnitude, topography, and population distribution. The proposed framework enables physically consistent comparison across many lakes and provides a basis for large-scale risk prioritization, offering a pathway toward regional to global GLOF impact assessments.

Long-term, high-resolution monitoring of carbon monoxide (CO) and methane (CH 4 ) is essential for understanding their spatiotemporal variability and supporting climate mitigation strategies. However, satellite observations from instruments such as the TROPOspheric Monitoring Instrument (TROPOMI) are often spatially and temporally incomplete, while existing fusion methods still struggle to achieve both high accuracy and spatiotemporal continuity. Here, we propose a signal-domain fusion approach that combines three-dimensional discrete cosine transform (3D DCT) and singular value decomposition (SVD) to integrate TROPOMI retrievals with GEOS-Chem simulations. A lightweight residual U-Net is further employed to refine the initial reconstruction by learning residual fields from GEOS-Chem simulations and DCT/SVD reconstruction outputs, guided by a masked loss. The method generates global 0.25° and China-specific 0.05° daily gap-free XCO and XCH 4 datasets from 2019 to 2023. In the time-series comparison analysis at representative sites, the fused datasets generally follow the temporal variations observed by the Total Carbon Column Observing Network (TCCON) and TROPOMI. Missing-rate-threshold experiments further show that the fused products perform comparably to original TROPOMI retrievals under low and moderate missing-rate conditions and show improved performance under sparse TROPOMI coverage (MR >0.5), with R 2 values of 0.91 for XCO and 0.83 for XCH 4 , along with reduced biases and standard deviations. The fused datasets also capture regional XCO increases in parts of North America, decreases over eastern China, and widespread XCH 4 growth, wildfire-related enhancements in Chongqing in 2022, and short-term variations over rice-growing regions. These results indicate that the proposed framework can reconstruct missing satellite observations with improved continuity and provide useful fused datasets for studying regional variability, event-related enhancements, and atmospheric composition changes. The generated datasets are publicly available at https://doi.org/10.5281/zenodo.22010891 (An et al., 2026).

Cloud processes constitute one of the key uncertainties for climate change projections. The fourth iteration of the Cloud Feedback Model Intercomparison Project, CFMIP4, contributes to the Coupled Model Intercomparison Project phase 7 (CMIP7), by providing a set of global climate model experiments aiming to enhance our understanding of clouds, circulation and climate sensitivity, thereby informing improved projections of future climate change. CFMIP4 targets four knowledge gaps: (1) Physical mechanisms of cloud feedback and adjustment; (2) Dependence of cloud feedback and adjustment on climate base state and on the nature of the forcing; (3) Coupled mechanisms of the sea-surface temperature “pattern effect”; and (4) Coupling of clouds with circulation and precipitation. CFMIP4 contributes four CMIP7 Assessment Fast Track experiments that are central to the quantification of climate feedback and sensitivity in past, present and future climates, essential for process understanding and model evaluation. Furthermore, CFMIP4 supports the joint analysis of models and observations through a data request that includes process and satellite simulator output.