Atmospheric and Oceanic Sciences

Abstract Landfall reorganizes the tropical cyclone eyewall boundary layer (BL) through elevated surface roughness and reduced enthalpy fluxes. Using a 100‐m resolution nested large‐eddy simulation of Typhoon Hato (2017), we compare eyewall BL structures before and during landfall. Kilometer‐scale rolls and streaks dominate variance and vertical fluxes in both stages, but become shallower and less vertically coherent during landfall, redistributing turbulent transport toward lower levels. Flux decomposition reveals that while tangential momentum flux weakens, low‐level radial momentum flux is enhanced by its large‐scale component. Simultaneously, increased stability near the mixed‐layer top strengthens coupling between vertical velocity and thermodynamic perturbations, indicating a wave‐like response superimposed on residual turbulence. Cross‐scale diagnostics show that the modulation of sub‐kilometer turbulence by kilometer‐scale coherent motions becomes vertically compressed and increasingly confined to the lower BL. These results highlight how landfall‐induced surface forcing modifies multi‐scale turbulent organization, enhancing our understanding of landfalling typhoon dynamics.

The Brazilian Cerrado, a global biodiversity hotspot, is under increasing pressure from agricultural expansion and native vegetation conversion, underscoring the need for efficient monitoring to support conservation and environmental policies. In heterogeneous landscapes, land use and land cover (LULC) mapping using supervised classification methods faces a major bottleneck: the need for extensive and high-quality training datasets. To address this challenge, we propose a semi-automated, clustering-based methodology for mapping secondary vegetation within previously deforested areas, reducing training-sample requirements and enabling scalable mapping through the clustering of satellite image time series. In the first stage, an unsupervised process integrates graphics processing unit (GPU)-accelerated Self-Organizing Maps and hierarchical clustering with Dynamic Time Warping to produce spectro-temporal clusters. In the second stage, specialists label and refine these clusters by visual interpretation, transferring expert knowledge from individual pixels to grouped spectro-temporal patterns. Applied to 692,000 km2 of previously deforested land in the Cerrado biome, the methodology produced a mapped secondary vegetation area of 81,209 km2 (11.74%). The design-based estimated area was 98,683 ± 10,071 km2, with an overall accuracy of 96.45 ± 1.52%, a user’s accuracy of 96.27 ± 2.40%, a producer’s accuracy of 79.22 ± 7.94%, and an F1-score of 86.90%. The initial cluster labeling accounted for 86.3% of the final secondary vegetation area and limited the interpretation task to approximately 3000 cluster-level decisions. Implemented in the TerraClass Cerrado 2024 cycle, the workflow reduced the secondary vegetation mapping phase from approximately two years to six months while maintaining the thematic accuracy required for large-scale operational monitoring.

The Northeast China Cold Vortex (NCCV), a typical synoptic-scale system in Northeast China, North China, and the Jianghuai region, frequently triggers thunderstorms, strong winds, and heavy precipitation, making it significant for meteorological monitoring and operational forecasting. However, the cloud microphysical properties of NCCV-associated cloud systems remain poorly characterized, as long-term cloud microphysical observations are limited. This study utilizes cloud products from the Moderate Resolution Imaging Spectroradiometer (MODIS) to analyze cloud-type frequencies and four key cloud microphysical properties under NCCV conditions: liquid effective radius (Re_liq), ice effective radius (Re_ice), liquid water path (LWP), and ice water path (IWP). Nearly identical cloud-type compositions are found for the two groups, NCCV and non-NCCV samples with similar cloud fractions on the regional scale, which are dominated by stratocumulus (Sc), altostratus (As), cumulus (Cu), and stratus (St), with Sc accounting for above 40% of total cloud occurrence. Yet microphysical properties differ markedly between these two groups. LWP shows the most contrast and it is evidently larger in NCCV than in non-NCCV cloud systems. As for the spatial structure of cloud microphysics in the NCCV domain, it is found that Sc, As, St, and nimbostratus (Ns) constitute the primary background, and Sc remains the dominant cloud type in almost all spatial sectors. LWP and IWP tend to have stronger spatial heterogeneity than Re_liq and Re_ice. LWP gets notably larger in the northern to northwestern sectors, whereas IWP shows much higher variations in both radial and azimuthal dimensions. These results reveal the statistical microphysical characteristics of cloud systems associated with NCCV from the perspective of satellite observations, providing a reference for a deeper understanding of their unique cloud and precipitation physical processes.

Accurate and transferable estimation of forest carbon is essential for operational forest management and national greenhouse gas reporting, yet it remains challenging because of variation in stand structure and site conditions. Unmanned aerial vehicle-based light detection and ranging (UAV-LiDAR) provides detailed structural information for modelling tree-level carbon, but model transferability across sites is often limited. In this study, we compared three modelling approaches—a linear mixed-effects model (LMM), a generalised additive model (GAM), and Random Forest (RF)—within a unified framework of multi-site, locally post hoc calibrated, and fully local model-fitting strategies. Using data from 20 radiata pine (Pinus radiata D. Don) plantation stands across New Zealand (35,201 trees), a leave-one-site-out (LOSO) framework was used to assess multi-site model transferability and support post hoc calibration, while local models were evaluated using repeated within-site train/test splits. We also evaluated how prediction accuracy changed with increasing local sample size and compared random tree selection with plot-based sampling. Multi-site models showed poor generalisation, with mean relative RMSE ranging from 35.9% to 56.9% and substantial site-level bias. Applying post hoc calibration to the multi-site model using a 50-tree sample reduced prediction error by 30 to 60% (mean relative RMSE 22.8–25.0%) and substantially reduced bias across sites. The fitting of fully local models with the same sample size yielded only modest further improvements (mean relative RMSE 21.9–23.1%). Gains in accuracy were minimal with increasing sample sizes above 50 trees for post hoc calibration and 175 trees for the fully local models, and differences in accuracy between sampling strategies were small. These results show that post hoc calibration of multi-site UAV-LiDAR models with a small local sample provides a practical and efficient approach for tree-level carbon estimation in plantation forests.

Study region The Yellow River Basin, China. Study focus Irrigation plays a critical role in sustaining agricultural development, particularly in water-scarce regions such as the Yellow River Basin (YRB). Yet, the response of irrigation water requirement (IWR) to drought events across different crop types and growth stages remained inadequately characterized. Here, we developed a daily-scale, multi-crop simulation framework that integrated the FAO Penman-Monteith method and the Standardized Precipitation Evapotranspiration Index (SPEI) to assess the spatiotemporal heterogeneity of IWR responses to drought events across four major crops (wheat, maize, rice, and soybeans) in the YRB during 1980–2018, which enabled systematic comparison of drought impacts across crop types and growth stages. New hydrological insights Drought significantly increased IWR for all crops, with basin-average increases of 27.70–31.36% during drought periods compared to non-drought periods. IWR increased consistently with drought severity, corroborated by significant linear correlations between SPEI and IWR anomaly, particularly in the Wei River Basin characterized by intensive irrigation and severe drought conditions. Specifically, rice exhibited the strongest sensitivity to drought across all growth stages, whereas the mid-season stage was identified as the most critical period for drought-induced IWR amplification for most crops. These findings moved beyond aggregate drought-impact assessments to provide an explicit understanding of irrigation sensitivity to drought, offering a scientific basis for targeted water allocation and drought adaptation strategies in water-stressed regions.

Abstract. Understanding the vertical distribution of aerosol and clouds i.s critical for climate modeling, weather forecasting, and air quality monitoring. Lidar observations are central to profiling atmospheric composition, yet signal attenuation in optically thick layers limits the effective retrieval of some important properties above those layers. More complex measurement approaches, using a combination of Lidar and cloud radar systems, can be taken to support more inclusive and accurate inference. In this study, we develop a deep learning framework to address this trade-off and gap in the cost of data acquisition by enabling full-column aerosol and cloud classification using only standard lidar inputs, achieving particularly high skill for aerosol typing while demonstrating robust, physically consistent classification of ice-cloud fields even under conditions of strong lidar signal attenuation, with liquid-cloud uncertainties primarily arising from closely related microphysical classes. The approach is based on a U-Net architecture trained to predict combined aerosol and cloud types from vertical profiles of backscatter and depolarization. Classification targets integrate established aerosol typing from PollyXT with cloud and precipitation categorization from Cloudnet, facilitating a unified scheme. The model achieves high precision, recall, and F1-scores above 95 %. By evaluating numerous complex case studies, we establish the model's ability to exploit information embedded in the lidar signal below attenuating layers, including structural and contextual features, to infer atmospheric conditions at higher altitudes, offering a robust AI-based enhancement to lidar-based atmospheric profiling and target classification. The application of AI in this context closes the gap between the need for vertical cloud maps and the sparse availability of Cloudnet.

Relative radiometric normalization (RRN) is fundamental to multi-temporal remote sensing analysis; however, conventional techniques often struggle with nonlinear distortions, outlier contamination, and heterogeneous land-cover conditions. To address these challenges, we propose a diffusion-based probabilistic framework that models radiometric inconsistency as a combination of deterministic residuals and stochastic perturbations. In this framework, the forward process injects structured noise and stochastic perturbations, while the reverse process restores radiometric consistency through a dual-objective variational formulation. At the core of this framework is a spatial–spectral attention residual network (SSARN), which integrates residual learning with dual attention mechanisms to capture cross-band dependencies and multi-scale spatial context. A preprocessing stage guided by the structural similarity index (SSIM) further enhances robustness by automatically selecting stable pseudo-invariant regions for model training. Comprehensive experiments on multi-temporal Sentinel-2 datasets demonstrate that the proposed method consistently outperforms existing approaches, achieving higher accuracy and enhanced spectral fidelity. Moreover, the framework ensures greater consistency of the normalized difference vegetation index (NDVI) and preserves fine-grained textural details, underscoring its potential as a scalable and resilient solution for large-scale RRN in remote sensing applications.

Landslide susceptibility mapping (LSM) in mountain–basin transition zones remains challenging because conventional approaches rely mainly on historical inventories and static conditioning factors, whereas independent deformation evidence is seldom incorporated to refine susceptibility zonation. This study proposes an integrated LSM framework for the Xining Basin by coupling a Mamba-based model (Mamba-LSM) with SBAS-InSAR-based deformation-informed bidirectional reclassification, with the key innovation lying in the use of independent deformation evidence to refine susceptibility zonation after model prediction. Specifically, Mamba-LSM integrates six-channel neighborhood patches, CNN-based local spatial encoding, and Mamba-based latent feature transformation to improve the representation of local terrain context for landslide susceptibility assessment. Results show that Mamba-LSM achieved the highest AUC among the evaluated models, reaching 0.9011 with an F1-score of 0.7431. After deformation-informed bidirectional reclassification, the high- and very-high-susceptibility classes occupied only 25.31% of the study area but contained 69.84% of the mapped landslides, and were concentrated mainly in valley–mountain transition belts, river-incised slopes, and engineering-disturbed sectors where SBAS-InSAR deformation hotspots were also preferentially distributed. These findings demonstrate that integrating independent SBAS-InSAR deformation evidence can improve both the spatial concentration of landslides in high-susceptibility zones and the physical interpretability of susceptibility zonation.

Abstract. Beaches provide essential ecological functions and support socio-economic resilience, yet accurate mapping is hindered by systematic limitations in global Digital Elevation Models (DEMs). A critical challenge remains in the intertidal zone, where frequent tidal inundation creates extensive data voids, disrupting the continuity of coastal topography. To bridge this fundamental data gap, we present NZ-BeachTopo30 which is a national-scale and full-coverage 30 m beach topography dataset for New Zealand constructed by fusing ICESat-2 photon-counting altimetry with Sentinel-2 multispectral time series. The dataset is available at https://doi.org/10.5281/zenodo.17785546 (Wang, 2025). Using DeltaDTM as a high-precision baseline for the stable backshore, we trained an XGBoost model on ICESat-2 control points and Sentinel-2 spectral-geometric features to reconstruct the missing intertidal topography specifically. SHAP analysis was further employed to interpret the physical driving mechanisms of these predictors. Validation against airborne Lidar confirmed that the dataset accurately recovers elevations in previously void zones with an RMSE of 0.94 m. By integrating these predictions with the DeltaDTM baseline, the final national-scale product achieves robust accuracy with an R2 of 0.75 and an RMSE of 1.17 m. This targeted integration significantly expanded valid topographic coverage by 145.8 % from 79.9 to 196.5 km2. It delivers the first spatially continuous and full-coverage beach topography dataset for New Zealand. Given the global availability of ICESat-2 and Sentinel-2, NZ-BeachTopo30 offers a scalable solution for worldwide applications and provides a robust foundation for inundation modeling and coastal management.

Abstract An accurate and timely assessment of wind speed and energy output allows an efficient planning and management of this resource on the power grid. Wind energy, especially at high resolution, calls for the development of nonlinear statistical models able to capture complex dependencies in space and time. This work introduces a Convolutional Echo State AutoencodeR (CESAR), a spatio‐temporal, neural network‐based model which first extracts the spatial features with a deep convolutional autoencoder, and then models their dynamics with an echo state network. We also propose a two‐step approach to also allow for computationally affordable inference, while also performing uncertainty quantification. We focus on a high‐resolution simulation in Riyadh (Saudi Arabia), an area where wind farm planning is currently ongoing, and show how CESAR is able to provide improved forecasting of wind speed and power for proposed building sites by up to 17% against the best alternative methods.

Reliable flood loss models can support rapid mitigation and recovery decisions, but their value depends on whether relationships learned from one disaster are useful in another. We evaluate tract-level National Flood Insurance Program (NFIP)-insured housing losses across 6065 census tracts affected by the 2016 Tax Day Flood, Hurricane Harvey, and Hurricane Irma. Mean normalized loss ratio (Mean_NLR) and the probability of observed NFIP-insured loss are modeled with a parsimonious hazard-exposure-vulnerability (HEV) specification driven by precipitation, Special Flood Hazard Area (SFHA) share, insurance penetration, population density, social vulnerability, building age, and Community Rating System discounts. We used nested grouped spatial cross-validation to tune Random Forest, XGBoost, and logistic classification models while retaining OLS and spatial lag models as benchmarks. Tuned regression performance is modest, with out-of-fold R 2 peaking at 0.34 for Harvey and lower values for Tax Day and Irma. Binary classification is stronger, with within-event AUC of 0.81–0.90 and Harvey PR-AUC up to 0.71. Transfer is high between the two Texas events, with XGBoost AUC of 0.92–0.95. Although population density and precipitation are consistently influential, AUC declines when Texas-trained models are applied to Irma because hazard mechanisms, exposure patterns, and predictor distributions differ across regions. Because the NFIP claims used to train these models capture insured and claimed building losses with capped payments, the models are appropriate for screening observed NFIP-insured losses. These models should not be used to estimate total physical damage or as universal flood loss transfer functions.

The Environmental Trace Gases Monitoring Instrument-II (EMI-II) onboard the Gaofen-5B satellite provides high-resolution hyperspectral measurements in the ultraviolet range, enabling the retrieval of total ozone column (TOC) at regional to global scales. In this study, an optimized TOC retrieval algorithm for EMI-II is developed and comprehensively evaluated over the Yangtze River Basin. The algorithm integrates several improvements, including a refined spectral calibration with pseudo-absorption cross sections to correct wavelength shifts and stretches, and a region-specific air mass factor (AMF) look-up table generated using the SCIATRAN radiative transfer model. Validation against ground-based Brewer and Dobson spectrophotometer data from the World Ozone and Ultraviolet Radiation Data Centre (WOUDC) shows that the optimized retrieval achieves a correlation coefficient exceeding 0.9 and a mean bias within ±5%. Cross-comparisons with the TROPOspheric Monitoring Instrument (TROPOMI) and Geostationary Environment Monitoring Spectrometer (GEMS) products further demonstrate the reliability and consistency of the EMI-II retrievals. The results confirm that EMI-II provides accurate and stable TOC measurements across diverse surface and atmospheric conditions in China. This study establishes a validated retrieval framework that enhances the scientific application potential of Chinese environmental satellites for atmospheric monitoring and supports the development of future ozone observation missions.

Climate change may intensify the deterioration of river water quality by altering streamflow regimes, precipitation patterns, and organic matter transport pathways. In this study, a Hydrological Simulation Program-FORTRAN (HSPF)-based streamflow and total organic carbon (TOC) water quality model for the Tamjin River Basin, Korea, was developed, and future TOC pollution was evaluated under quantile delta mapping (QDM) bias-corrected Shared Socioeconomic Pathway 5-8.5 (SSP5-8.5) climate scenarios. Unlike previous studies that generally applied climate bias correction, watershed modeling, or pollutant-load assessment as separate procedures, this study links QDM-preserved climate change signals, process-based HSPF simulations, and TOC-specific discharge-load, delivered-load, exceedance-frequency, and load-reduction indicators within a single management framework. The model showed acceptable performance, with Nash–Sutcliffe efficiency (NSE) values of 0.67 and 0.68 for streamflow at Jangheung Dam and Gamcheon Bridge, respectively, and a TOC deviation of volume (DV) of 0.6% at Tamjin5. Under the SSP5-8.5 no-action scenario for the 2040s, the mean streamflow decreased by 33.1%, whereas the mean TOC concentration increased by 76.8% relative to the baseline. The number of days exceeding 4 mg/L TOC increased from 41 to 216 days yr−1, and the Korean TOC-based water quality class deteriorated from Ib to III. In contrast, the 20% and 30% load reduction scenarios offset approximately 33.8% and 67.9% of the climate-driven increase in TOC, respectively, with the 30% reduction scenario showing greater effectiveness during low-flow seasons. Elevated TOC levels may have implications for downstream water treatment because organic matter can increase chemical demand and disinfection-byproduct formation potential. However, these treatment-related effects were not directly evaluated in this study. These results suggest that TOC should be considered as a complementary indicator to conventional biochemical oxygen demand (BOD)-based management when developing climate-resilient water-quality strategies for the Tamjin River Basin.

Abstract Effective risk management of weather and climate hazards requires robust estimates of the likelihood of occurrence. The most common tool for this is extreme value analysis (EVA), but likelihood estimates based on observed data can be highly uncertain due to the relatively short observational record. Substantially larger samples of plausible extreme weather events can be obtained using the UNprecedented Simulated Extremes using ENsembles (UNSEEN) approach, which involves applying EVA to large forecast/hindcast ensembles. While larger sample sizes generally reduce the sampling uncertainty associated with EVA, using seasonal or decadal forecast data introduces additional uncertainties related to bias correction and model diversity. In this study, a multi-model ensemble of hindcast data from the Decadal Climate Prediction Project was analysed to quantify these additional uncertainties in the context of extreme temperature and rainfall across Australia. Factoring in bias correction and model diversity dramatically increased the uncertainty attributed to estimated event likelihoods from the UNSEEN approach. Model diversity tended to be the largest source of uncertainty (typically 50-70\% of the total). Bias correction was also a significant source of uncertainty (30-50\%), while the uncertainty associated with sample size was negligible in comparison. Our results suggest that multi-model analysis should become a standard part of any UNSEEN workflow. The UNSEEN-based approach to estimating the likelihood of climate extremes should be understood as an approach that has different uncertainty characteristics to an observation-based approach, as opposed to less uncertainty.

Multi-source precipitation products exhibit strong regional differences across China’s complex monsoon climates and pronounced topographic gradients, making single-metric evaluations insufficient for product selection. This study evaluates 28 widely used precipitation products over China from four categories: gauge-based, satellite-derived, reanalysis, and multi-source merged products. Product performance is assessed at both grid and seven major climate-zone scales using conventional error statistics, consistency metrics, ETCCDI (Expert Team on Climate Change Detection and Indices) extreme precipitation indices, precipitation detection skill, and SAL (Structure–Amplitude–Location) diagnostics for the intensity, structure, and location of heavy-rainfall events. These indicators are further synthesized within an ensemble multi-criteria decision-making framework to derive national and regional rankings. The results show that most products capture daily precipitation variability reasonably well, but intense rainfall events remain associated with widespread amplitude underestimation and enlarged errors, while extremes also exhibit notable structural distortion and location bias. At the national scale, multi-source merged generally show greater overall robustness. Regional rankings further reveal strong spatial heterogeneity: arid zones and some humid regions tend to favor reanalysis-type products, whereas plateau regions show higher sensitivity to satellite products and greater ranking uncertainty. Overall, this study provides a transparent and application-oriented framework for integrated precipitation product evaluation and ranking over China. The resulting national and climate-zone-specific rankings offer practical guidance for precipitation product selection, candidate-pool construction for multi-source merging, and hydrometeorological risk analyses.

Abstract Atmospheric instabilities that moved against the background mean wind were observed directly on 31 March 2022 from Poker Flat, Alaska using both hydroxyl airglow imaging and sodium resonance LiDAR. These instabilities were extracted from data using a series of Fourier and Morse wavelet transforms to isolate them from background gravity waves (GWs) and therefore minimize contamination. Horizontal wavelength, direction, period, and an estimation of the vertical extent of the unstable layer were measured from ripples in airglow images. Density fluctuations in simultaneous LiDAR data exhibited a similar period and were then used to measure vertical extent and altitudinal location. Together, these observations enabled a set of instabilities to be described in quasi‐3D: horizontal wavelength of 10.6 ± 0.8 km, vertical extent between 1.5–2.5 km, period of about 3 min and traveling 9.0 ± 3.3° north of east. Furthermore, temperature and wind shear profiles derived from LiDAR data showed localized regions where the Richardson number dipped below 0.25, the threshold required for the formation of Kelvin‐Helmholtz instabilities. Gravity waves observed in airglow images propagated 16.9 ± 5.5° west of north with a horizontal wavelength of 34.8 ± 3.3 km and a period of 10 min, and a larger‐scale 2 hr period GW was present in LiDAR data.

Tropospheric delay poses a major limitation to high-precision InSAR observations, particularly in high mountain and canyon regions. To address this issue, this study proposes a combined model (EiMLP) that integrates gross error identification with a multilayer perceptron (MLP) for topography-dependent tropospheric delay correction. The gross error identification module detects unwrapped phase jumps based on phase gradients, followed by an MLP model that reconstructs the atmospheric phase using unwrapped phase and elevation information from a single interferogram. The proposed method is validated in the Baihetan Hydropower Station area and compared with traditional correction methods. Experimental results demonstrate that the proposed method achieves a Structural Similarity Index Measure (SSIM) of 0.970 and a Root Mean Square Error (RMSE) of 0.673 rad for the simulated atmospheric phase. After atmospheric correction, the average phase standard deviation of the interferograms is reduced by 83%, and the topography-related correlation is significantly suppressed. Furthermore, after correction by the proposed method, the signals that are masked by atmospheric errors in the original interferograms can be clearly revealed, indicating the potential for slope instability. These findings indicate that the EiMLP model, operating on a single interferogram, exhibits robust applicability and provides a reliable reference for future InSAR tropospheric delay correction.

Vegetation recovery after forest fires is a vital indicator of ecosystem resilience. However, the specific differences between structural and functional recovery after fire have remained unclear. In this study, we quantified and compared post-fire recovery using two distinct vegetation indicators: the Enhanced Vegetation Index (EVI) for structural recovery and Solar-Induced Chlorophyll Fluorescence (SIF) for functional recovery. We analyzed the spatiotemporal dynamics and drivers of post-fire recovery. A Transformer model was used to simulate pre- and post-fire variations in EVI and SIF, while a Random Forest model was employed to identify the key drivers of recovery. We analyze the spatiotemporal dynamics and drivers of post-fire recovery. A Transformer model simulates pre- and post-fire variations in EVI and SIF, while a Random Forest model identifies key drivers of recovery. Our results show a steep decline in both indicators after fires, with SIF recovering more slowly than EVI. Three years after the fire, about 78% of burned areas regain at least 80% of their pre-fire EVI levels, but SIF recovery reaches only 70%. Bivariate dependency analysis indicates that precipitation and temperature promote recovery, whereas topography and the Differenced Normalized Burn Ratio (dNBR) have the opposite effect. This study advances a phased, analytical approach to post-fire forest vegetation recovery, offering a dual-perspective framework for understanding forest resilience and providing actionable insights for sustainable restoration and management.

Abstract Emissions of nitrogen oxides (NOx) from coal-fired power plants (CFPPs) pose a significant challenge to air quality. In China, although policies like “Promoting the Big and Quashing the Small” and the adoption of ultra-low emission (ULE) technologies have been implemented in power plants, accurately quantifying their impacts on NOx emissions remains difficult due to lack of facility-specific information. This study investigates the influence of CFPP changes on NOx emissions from 2018 to 2024 in North China, by integrating a self-compiled, high-accuracy power plant database (with 517 plants in North China, including 109 retired or newly built) with satellite-derived NOx emission data at a high spatial resolution (0.05°×0.05°). We find that among the locations with no industrial sources, emissions show show substantial reductions (−1.97 kg km−2 h−1 on average; Wilcoxon test p < 0.001; t-test p < 0.001)where CFPPs were retired during this period, compared with the increases (+0.76 kg km−2 h−1 on average; Wilcoxon test p < 0.01; t-test p < 0.01) where CFPPs equipped with ULE technology were newly built. However, when the retired or newly built CFPPs are located in the same 0.05°×0.05° grid cells with industrial sources, their emission signals are often dominated by industrial emission reductions, resulting in emission declines in both cases. These results underscore the rising relative contribution of industrial emissions in regions where power plants and heavy industry are spatially co-located, and highlight the need to strengthen emission monitoring and control efforts directed at the industrial sector.By comparison, existing emission inventories have difficulty in capturing the power plant dynamics, particularly the effect of CFPP retirement. Our results demonstrate the value of satellite remote sensing coupled with detailed facility information for assessing point-source emission dynamics.

Abstract Deforestation remains a critical challenge despite increasing global attention. Voluntary and market-based initiatives have proven insufficient to reverse this trend. In response, the European Union introduced the Deforestation Regulation (EUDR), requiring information for several forest-risk commodities (e.g., palm oil, soy, rubber, cocoa, and coffee) to enforce deforestation-free standards across these supply chains. One important barrier to effective implementation is the limited understanding of farmers operating in forested landscapes. Most existing agri-food system models evaluate deforestation risk at the national scale, overlooking differences between landholding types, limiting targeted policy insights. Here, we evaluate how different farm sizes contribute to the production of EUDR-listed crops within forested landscapes, using spatial datasets on crop distribution, forest cover, and farm size. We find that small-scale farms (<2 ha) are responsible for a large share of forest-linked production: 91% for rubber, 81% for palm oil, 53% for coffee, and 60% for cocoa. These crops are produced mainly in countries where traceability and compliance pose major challenges. In contrast, soybean production in forested areas is dominated by large-scale farms. We identify regions where smallholders may face high risks of exclusion from EU supply chains due to EUDR compliance across Indonesia, Vietnam, Thailand, and Côte d’Ivoire. These findings highlight the need for targeted support to smallholders in these countries, including investment in data collection, certification systems, and land tenure security. Our findings also reveal a misalignment between the EUDR’s country benchmarking classification and actual deforestation exposure, indicating that the current country classification approach of the EUDR needs revision. These findings highlight the importance of understanding which types of farms are affected by policies such as the EUDR and guiding targeted support to ensure that forest conservation initiatives do not come at the cost of smallholder livelihoods.

Abstract. Hailstorms are a damaging weather phenomenon worldwide. In response, several countries – including Switzerland – have implemented hail mitigation strategies, most notably through cloud seeding with ice-nucleating particles (INPs). In this study, we investigate the impact of silver iodide (AgI) perturbations on eight convective storms observed in Switzerland and southern Germany. Our focus is on evaluating the effectiveness of an early seeding strategy and examining its relationship with two key meteorological parameters: Convective Available Potential Energy (CAPE) and 0–6 km wind shear. We also assess how different storm-tracking thresholds influence the interpretation of seeding effects. Simulations were conducted using the Consortium for Small-Scale Modeling Regional Weather and Climate Model (COSMO). AgI particles were introduced as a prognostic variable during the cumulus stage and released into the updraft region near the cloud base at a concentration of 20 cm−3. The results indicate that early seeding increases both the mass and number concentration of ice and graupel, accompanied by stronger updrafts. In contrast, the response of hail mass is ambiguous and varies with the tracking method. Hail size and hail-covered area also show no systematic dependence on CAPE or wind shear. Despite the variability in the hail response, our results show that early seeding increases the mean hail diameter in 80 % of the cases, with a median increase of 7.6 % – corresponding to a 31.3 % increase in kinetic energy – while simultaneously reducing the spatial extent of the hail-affected area by 39.8 % (median), with 92.4 % of simulations exhibiting a decrease in hail area.

Landslides and their associated secondary hazards present substantial threats to both public infrastructure and resident safety. The rapid and accurate identification of large-scale landslide events remains a critical challenge in the field of engineering geology. In this study, a YOLO-based deep learning model is developed for landslide identification relying on a training dataset constructed using the satellite imagery of Longyan City, Fujian Province, in 2024. Adopting the double machine learning model, we examine the causal inference relationships between landslide and causative factors, including rainfall (R), mean Normalized Difference Vegetation Index (NDVI) and Distance to roads (DRoa). A total of 1185 landslides is identified in 2024, covering an area of approximately 31.02 km2. The landslides are predominantly concentrated in Shanghang, Wuping, Changting, and the southern part of Xinluo. The landslides mainly correspond to elevations around 300–500 m, slopes among the interval of [10°, 25°], and annual rainfall intensities ranging from 1600 m to 1700 mm. The top five key factors for landslide occurrence in descending order are NDVI, R, DRoa, Distance to Rivers (DRiv) and Aspect (A), in terms SHAP values. Causal inference analysis reveals that the rainfall in June and July shows significant positive causal effects to landslide, which is consistent with the physical mechanism of rainfall-induced landslide and the landslide data reported by the government. The framework proposed and the findings in this study offer valuable technical and theoretical support for landslide identification and risk assessment in southwestern Fujian.

Maximum latewood density of conifers is the most widely used annual-resolution summer temperature proxy. Regions with few conifers, however, remain underrepresented in global paleoclimate records. Here, we use x-ray micro–computed tomography (micro-CT) to show that latewood density measurements of European beech ( Fagus sylvatica L.) in a temperate lowland forest exhibit a strong summer (May to September) temperature signal ( r = 0.73; 1833 to 2022 CE). Complementary wood anatomical analyses using deep learning segmentation reveal that both vessel and fiber anatomy are key drivers of latewood density variability and its temperature sensitivity. By integrating these anatomical responses, x-ray micro-CT–based latewood density measurements generate a robust and temporally stable summer temperature signal. Our results highlight the untapped potential of broad-leaved tree species for density-based climate reconstructions in temperate regions and open previously unidentified avenues for high-resolution paleoclimatology beyond the use of conifers.

Landslides pose widespread threats to mountainous communities and infrastructure worldwide, yet susceptibility mapping in alpine gorge basins is often constrained by sparse and incomplete local inventories. Whether national-scale landslide information can be transformed into reliable local knowledge remains unclear, particularly where strong topographic and environmental heterogeneity limits the direct transfer of broad-scale models. Here, we use the Parlung Tsangpo Basin on the southeastern Tibetan Plateau as a test case and compare four strategies for introducing national-scale information: local baseline modelling, direct national-model transfer, weighted source–target joint training, and prior-informed local modelling. The experiments use the same conditioning factors, data-processing workflow, spatially independent test set, and evaluation metrics, allowing the transfer strategies to be assessed under controlled conditions. The prior-informed strategy treats the susceptibility probability produced by the national model as a geoscientifically interpretable external prior, which is then relearned and recalibrated by local samples, and achieves the best overall performance. On the independent test set, it reaches an area under the receiver operating characteristic curve (AUC) of 0.901 and reduces the expected calibration error (ECE) to 0.055, outperforming the local baseline, direct transfer, and joint training strategies. Its susceptibility map shows an extensive low-susceptibility background with spatially concentrated high-susceptibility patches, thereby reducing broad-scale overprediction while preserving local landslide-prone zones. These results indicate that national-scale landslide information is more effective when converted into a locally recalibrated probability prior than when transferred directly, providing a practical pathway for susceptibility assessment in data-scarce mountainous basins.