Tidal wetlands, located at the dynamic land-sea interface, provide vital ecosystem services, yet face increasing threats from human activities and climate change. Accurate and up-to-date global mapping of tidal wetlands is essential for assessing their status and advancing conservation efforts. However, challenges such as tidal fluctuations and limited data availability lead to inconsistencies across existing datasets. Moreover, previous studies have largely overlooked the adjacent terrestrial environments of tidal wetlands, reducing our understanding of coastal dynamics and associated environmental drivers. To address these critical issues, this study proposes a novel global coastal mapping framework with three key components. First, using the ocean tide model EOT20, we systematically analyzed tidal variations observed in Sentinel-1 and Sentinel-2 imagery between 2019 and 2021, facilitating the development of an adaptive image selection strategy to ensure low-tide coverage. Second, we integrated multi-source global datasets and employed a knowledge-driven, semi-automatic sampling approach to generate training samples for tidal wetlands and adjacent land covers. Third, we iteratively trained and refined random forest models using tide-level and phenological features extracted from composite optical and radar imagery, producing a global coastal dataset centered on 2020 with 11 distinct land cover types. The mapping result was cross-compared with multiple global and regional coastal datasets and validated using the temporally cleaned external dataset, achieving an overall accuracy of 92.7%. By optimizing the selection of Sentinel scenes to approximately 10-50 in time-series composite mapping, this approach balances computational efficiency with intertidal classification accuracy. The dataset delineates the global distribution of tidal wetlands and their adjacent environments at a 10-m resolution; specifically, tidal wetlands-including mangroves, tidal marshes, and tidal flats, amount to 425,509 +/- 932 km2. This fine-resolution and multi-category mapping framework enables a more precise identification of potential trade-offs between conservation and development, providing valuable references for sustainable coastal management.

China's New-Type Urbanization Plan since 2015-the world's largest urbanization endeavor-reshapes the nation's socioeconomic landscape but lacks high-precision, fine-scale progress monitoring. Urban construction sites (UCSs)-barometers of urban spatial expansion and renewal-offer a detailed observational window. A sub-meter-resolution deep-learning framework for nationwide UCS mapping is proposed. Using a Segment Anything Model-enabled weakly supervised method for pixel-level UCS annotation, a spectral-texture dual-branch segmentation network with 94.3% overall accuracy identifies 541 177 UCSs (including 10-m² micro-sites) across 372 cities. K-means clustering partitions cities into four typologies, uncovering a dual-track parallel pattern (incremental expansion + stock optimization) vs the classical 'growth-decline-renewal' trajectory. Spatial analysis shows that UCS construction correlates with annual PM₂.₅ concentrations; green dust-proof net coverage (<10%) fails to curb pollution. The framework serves as a 'microscope' for evaluating new-type urbanization and supports sustainable planning.

Deep convection plays a vital role in transporting Asian pollutants from the planetary boundary layer (PBL) into the Asian summer monsoon anticyclone (ASMA). However, the efficiency and effectiveness of transporting pollutants with various chemical and physical properties to the ASMA remain unclear. In this study, we use the global atmospheric chemistry and climate model EMAC to investigate the deep convective transport of trace gases such as CO, NH3 and SO2 from the PBL to the ASMA over the years 2010-2020. We quantify the deep convective transport efficiency of different trace gases into the ASMA. We show that the strongest convective transport tendency occurs over northern India and the southern edge of the Tibetan Plateau for CO (0.2-0.5 ppbv h-1), over the south and eastern parts of the Tibetan Plateau for NH3 (0.02-0.05 ppbv h-1), and over central India and eastern China for SO2 (0.002-0.005 ppbv h-1). We find that, in contrast to CO and NH3, the SO2 enhancements within the ASMA are very weak, and there can even be a decrease in SO2 over the southern Tibetan Plateau relative to the surroundings. Our analysis indicates that gas-liquid partitioning in clouds and subsequent wet deposition over South Asia are more effective at reducing SO2 than NH3 reaching the Tibetan Plateau and the ASMA. In view of ongoing changes in regional emissions, the effects of deep convective transport of various pollutants and associated gas-aerosol-cloud interactions on the chemical features of the ASMA require continued investigation.

Cities are changing faster than the tools used to understand them. A new generation of AI can now synthesize images, model aspects of human behavior and reason across heterogeneous data—capabilities that are beginning to reshape how researchers and practitioners observe, model and support decisions about urban life. Here, in this Review, we explore what these tools can genuinely offer, where the evidence is strong and where it remains thin. The answer depends less on technical novelty than on whether generative AI can be evaluated honestly, validated in context and deployed in ways that keep human judgment and accountability at the center. Understanding cities has never been more important. This Review considers the capabilities of generative AI, synthesizing insights into the importance of honestly evaluating AI, validating it in context and deploying it to center human judgment and accountability.

Over the past few decades, the Tibetan Plateau (TP) has witnessed changes in summer precipitation that follow a "south drying-north wetting" diploe pattern. Although the role of adjacent oceanic moisture export (OME) in modulating the TP precipitation has been highlighted, there is no consensus on the relationship between OME and summer TP rainfall by far. Here, we quantify the contributions of the OME from three seas, namely, the Arabian Sea (AS), the Bay of Bengal (BOB), and the South China Sea (SCS), to the TP summer precipitation and particularly examine the relationship between them from the standpoint of long-term trend based on the 35-year Lagrangian modeling and OME-based precipitation diagnostic. Climatologically, the observed and the diagnosed OME-based precipitation are compatible, both in terms of spatial pattern and magnitude. However, the preferred regions contributed by the three sea sources vary from one another. The AS with fraction contribution 62.6 % rank first with influenced regions almost covering the whole TP, followed by the BOB (21 %) and the SCS (16.4 %) with impacted regions mainly distributed over the southeastern and eastern TP, respectively. Although this order of contribution to the trends of whole TP precipitation remains unchanged, the AS comparatively acts as a main contributor to the "south drying" over the TP, while the BOB makes more contribution to the "north wetting", implying the distinct mechanisms on the precipitation changes among the TP subsectors.

Air pollution remains a major global environmental challenge, posing serious threats to ecosystems and human health. PM2.5 pollution is particularly severe in the Beijing-Tianjin-Hebei (BTH) region and is characterized by strong regional associations. In this study, a time-lagged PM2.5 association network was constructed for the BTH region during 2015-2023 based on complex network theory. Seasonal and interannual variations in network structure were examined, key nodes, major association pathways, and dynamic community structures were identified, and the influence of meteorological conditions on network topology under different winter weather types was further investigated by combining network analysis with a self-organizing map (SOM) approach. The results show that the PM2.5 association network in the BTH region exhibits pronounced seasonal and interannual variability. Network density and clustering coefficient are highest in winter, and overall connectivity increases over time. Langfang, Baoding, Tianjin, Beijing, and Tangshan are identified as key nodes in the network, and the major association pathways linked to these cities remain structurally robust across seasons and winter weather types. The PageRank hierarchy of the network remains broadly stable during the study period, although the distribution of node influence shows a tendency to evolve from concentration toward relative homogenization. The community structure exhibits a clear south-north contrast, with denser internal connections in southern communities and relatively weaker connections in northern communities, while some cities show high stability in community affiliation. Multi-threshold sensitivity analyses further confirm the robustness of key-node identification and community partitioning. Network topology also differs substantially among winter weather types: high-stagnation conditions correspond to stronger connectivity and clustering, suggesting enhanced regional pollutant retention and mixing, whereas strong-wind conditions are associated with sparser networks and greater community separation. Overall, within the winter SOM analysis, wind speed and the stagnation index are identified as the key meteorological factors associated with network connectivity and the strength of regional PM2.5 associations. These findings provide a transferable analytical framework for investigating regional PM2.5 association networks in other areas and offer a scientific basis for season-specific, region-specific, and weatherspecific air pollution prevention and control in the BTH urban agglomeration.

Accurately elucidating the mechanisms and pathways through which grey-green spaces mitigate land surface temperature (LST) is crucial for alleviating urban heat island effects under rapid urbanization. However, the seasonal and diurnal variations in the impacts of 2D/3D grey-green spaces on LST across different local climate zones (LCZs) remain insufficiently explored. The Categorical Boosting (CatBoost) model was employed to quantify the contributions and marginal effects of 2D/3D grey-green spatial morphology on multidimensional LST, and partial least squares path model (PLS-PM) was applied to uncover their direct and indirect impact pathways. Results revealed that LST exhibited significant diurnal and seasonal variations across the seven LCZs, with greater daytime differences than at night. Compact LCZs exhibited higher LST than open LCZs. Daytime LST was primarily influenced by 2D/3D grey spaces and 3D green spaces, whereas 2D green spaces dominated at night. Among all factors, building density, canopy height, and Core_Green were identified as the most influential variables. Pathway analysis further revealed that 3D_Green serves as a key indicator mitigating both daytime and nighttime LST. It not only directly reduces LST but also indirectly offsets the LST increase induced by 2D_Grey and reinforces the cooling effects of 2D_Green. This study advances the understanding of how 2D/3D grey-green space influence LST, suggesting that the adverse thermal effects of built-up grey spaces can be indirectly alleviated through their synergistic integration with green spaces. These findings provide novel insights for optimizing urban grey-green spaces within existing urban fabric to improve the urban thermal environment.

Nighttime light (NTL), a crucial indicator of human activity intensity, has not been systematically analyzed for its interactive mechanisms with air pollution and climate change. This study first investigates the spatiotemporal evolution of China's total nighttime light (TNTL) and average nighttime light (ANTL), alongside key indicators of meteorological parameters and air pollution, at the grid scale from 2000 to 2023. We then employ prefecture-level city data and a geographically and temporally weighted regression (GTWR) model to quantify the spatiotemporally heterogeneous associations of temperature (TMP), precipitation (PRE), fine particulate matter (PM2.5), ozone (O-3), land use (LUL), topography, and socioeconomic factors with NTL. The results indicate that (1) China's NTL exhibits a significant overall upward trend, with areas of increase or significant increase comprising 92.04% of the total study area. TNTL growth demonstrates regional heterogeneity, expanding by a factor of 4.91 in East China and 2.65 in Northeast China; (2) meteorological and air pollution indicators display spatiotemporal non-stationarity, with the synergistic effect between O-3 and PRE being the strongest; (3) among NTL drivers, LUL contributes most significantly (0.44), followed by TMP (0.14) > PM2.5 (-0.33 & times; 10(-1)) > O-3 (0.17 & times; 10(-1)) > PRE (-0.33 & times; 10(-6)); (4) TMP and PRE may primarily influence NTL by altering ecological conditions and nighttime activity patterns. TMP shows a strong positive correlation with NTL in the junction zone of South, East, and Central China, whereas PRE predominantly exerts a negative influence; (5) air pollution exhibits distinct spatiotemporal effects: high PM2.5 and O-3 generally correspond to lower NTL, though positive correlations persist in some areas due to industrial structures, highlighting the need for integrated policies that balance air quality management with sustainable urban planning; (6) the 2013 "Air Pollution Prevention and Control Action Plan" significantly strengthened the negative correlation between PM2.5 and NTL in North China. However, O-3 concentrations increased by 28.9% after 2017, underscoring the challenge of coordinating VOC and NOx controls for long-term atmospheric sustainability.

Urban parks are essential public service resources that enhance living environments and promote physical and mental well-being. The opportunity cost of green land in cities is massive, as its importance to city productivity, warranting more serious attention to systematic monitoring and evaluation of park systems. Here, we propose a data-driven framework grounded in the supply-demand dynamics of the park-people nexus, to inform strategic greening solutions. Using the Greater Bay Area (GBA) of China as an empirical case study, we reveal that only 55.11 % of residents can access parks within a 15-minute walking distance, while 64 % live in high-density (>5000 persons per km(2)) areas with a per-capita park area of just 4.73 m(2)-nearly one-third of China's national standard of 12 m(2) per person. By evaluating the potential for urban park enhancement and the feasibility of construction costs, our framework presents various greening options, with the most cost-effective scenario involving the development of potential urban parks covering approximately 33.59 km(2). This option could achieve the full park coverage goal, requiring an estimated investment of 16.1 billion CNY. Our method provides a scalable framework for better optimizing urban green spaces in rapidly urbanizing regions.

Rare earth (RE) materials are critical to modern technologies, yet current research has largely overlooked the indirect exports of RE embedded in intermediate and final products. Leveraging the detailed inventories of China’s RE flows and downstream usage constructed herein, this study addresses that gap by systematically tracing and decomposing China’s indirect RE exports from 1990 to 2020 through the integration of environmentally extended multi-regional input-output (EEMRIO) analysis along with global value chain (GVC) and structural path analysis (SPA) methods. We find that over one-third of China’s RE exports were indirectly embedded in traded goods with other regions during this period. This substantial outflow of embedded REs underscores China’s important—yet often underrecognized—role in sustaining the global RE supply chain. These findings further reveal the complexity of RE trade and emphasize the need for comprehensive tracking of metal flows.

The rapid evolution of satellite-borne Earth Observation (EO) systems has revolutionized terrestrial monitoring, yielding petabyte-scale archives. However, the immense computational and storage requirements for global-scale analysis often preclude widespread use, hindering planetary-scale studies. To address these barriers, we present Embedded Seamless Data (ESD), an ultra-lightweight, 30-m global Earth embedding database spanning the 25-year period from 2000 to 2024. By transforming high-dimensional, multi-sensor observations from the Landsat series (5, 7, 8, and 9) and MODIS Terra into information-dense, quantized latent vectors, ESD distills essential geophysical and semantic features into a unified latent space. Utilizing the ESDNet architecture and Finite Scalar Quantization (FSQ), the dataset achieves a transformative ~340-fold reduction in data volume compared to raw archives. This compression allows the entire global land surface for a single year to be encapsulated within approximately 2.4 TB, enabling decadal-scale global analysis on standard local workstations. Rigorous validation demonstrates high reconstructive fidelity (MAE: 0.0130; RMSE: 0.0179; CC: 0.8543). By condensing the annual phenological cycle into 12 temporal steps, the embeddings provide inherent denoising and a semantically organized space that outperforms raw reflectance in land-cover classification, achieving 79.74% accuracy (vs. 76.92% for raw fusion). With robust few-shot learning capabilities and longitudinal consistency, ESD provides a versatile foundation for democratizing planetary-scale research and advancing next-generation geospatial artificial intelligence.

While atmospheric rivers (ARs) are major drivers of extreme precipitation (EP) in East Asia and hold promise for improving forecasts, this predictive potential is limited because not all ARs result in EP. This raises a key question: what distinguishes the environments of ARs that lead to EP? This study addresses this issue by comparing the environments of ARs associated with EP (AR&EP) with those without EP (AR&NONEP) during summer seasons spanning 32 years, focusing on the lower Yangtze River Basin (LYRB). Results show that while moisture sources for both AR types are spatially similar, AR&EP events exhibit enhanced moisture originating from regions west and north of the LYRB. This indicates that AR&EP events depend more heavily on excess moisture supplied from mid- to high-latitudes. Key large-scale circulation anomalies were also identified during AR&EP events, including a strengthened and westward-extended western North Pacific subtropical high, an enhanced westerly jet, an intensified and eastward-extended South Asian high, and a deepened upper-level trough. Among these, the deep EAT and strengthened westerly jets emerge as the primary factors differentiating AR&EP from AR&NONEP events. Locally, significant differences in convective activity were observed, underscoring the essential role of the Mei-yu front in triggering AR&EP. In this context, the synergy among the deepened upper-level trough, moisture availability, and localized uplift provided by the Mei-yu front is crucial for accurately predicting AR&EP in the LYRB region.

Extreme PM2.5 pollution events pose substantial threats to public health and environmental sustainability. However, under climate change, the relationship between global fire activity and extreme PM2.5 pollution remains insufficiently understood. Using 0.25° × 0.25° gridded data from 2004 to 2023, this study employs quantile regression models to assess the influence of global fire activity on extreme PM2.5 pollution. Results show that fire activity has a significant positive relationship with PM2.5 concentrations across all quantiles, with this relationship becoming particularly pronounced under extreme pollution conditions. At the 95th percentile, the fire-related regression coefficient reaches 0.899 (p < 0.05), which is 2.9 and 4.5 times higher than the coefficients at the 50th (0.310) and 10th (0.197) percentiles, respectively. Spatial autocorrelation analysis further reveals that regions where extreme PM2.5 pollution is strongly associated with fire activity exhibit significant spatial clustering (Moran’s I = 0.036, p < 0.01). Notably, Canada in North America, Siberia in Asia, Brazil in South America, and Indonesia in Southeast Asia are identified as the most strongly affected regions. These findings improve understanding of the role of fire activity in extreme PM2.5 pollution and provide important evidence for strengthening global air quality management and emergency response strategies.

Artificial intelligence (AI) is rapidly transforming Earth science, offering unprecedented capabilities to tackle the most pressing challenges in the field. This work explores significant advances and emerging challenges across the AI for atmosphere-ocean sciences, while outlining critical ways forward. We review deep-learning methods and their application in weather and climate forecasting, which outperforms dynamical models in accuracy and computational efficiency. The role of AI in detecting complex phenomena, enhancing data assimilation and reconstruction, bias correction and downscaling coarse model outputs is also examined. However, the 'black-box' nature of complex AI models necessitates a focus on explainable AI to build trust and extract mechanistic insight. The most promising path forward is identified as the development of hybrid physics-AI modeling, which integrates the data-driven power of AI with the foundational constraints of physical laws to ensure generalizability and causal consistency. A new framework for AI-based model intercomparison is essential for rigorous benchmark performance. Finally, we contextualize these technical developments by discussing the usefulness and applicability of AI to society, including the improvement of multi-hazard early-warning systems and green energy production. We conclude by envisioning the future of AI agents for Earth science-autonomous, goal-oriented systems capable of designing and running experiments, generating and testing hypotheses, and learning dynamics from multisource data. This synthesis underscores that AI is not merely a tool, but a paradigm shift, which will significantly improve how we understand and adapt to a changing climate.

The critical root-zone soil moisture threshold (θ_r^c) is a fundamental parameter that marks the transition from energy-limited to soil-moisture-limited evapotranspiration (ET) regimes, yet regional and global studies often rely on near-surface soil moisture and its associated threshold as a proxy. Here, we synthesize a global network, measurement-based evaluation of θ_r^c by analyzing 879 dry-down events across 33 eddy covariance flux tower sites equipped with multi-layered soil moisture sensors reaching depths of at least 1 meter. The results demonstrate that θ_s^c and θ_r^c are significantly inconsistent, with an overall root mean square error (RMSE) of 0.06 m³ m⁻³. This discrepancy is primarily driven by the vertical soil moisture gradient and decoupling of the surface and root zone layers during dry periods, which leads to substantial errors in identifying the onset and duration of plant water stress. For instance, at some forest sites, using θ_s^c delayed the detected onset of moisture stress by 27 days and underestimated the duration of the moisture-limited regime by 36 days compared to θ_r^c. Across the diverse biomes and climate types studied, the multi-site mean θ_r^c was 0.15 ± 0.11 m³ m⁻³. These findings provide a critical observational benchmark for the evaporative fraction-root zone soil moisture (EF-θ_r) relationship, highlighting that transitioning from surface to root-zone-based assessments is essential for accurate land-surface model evaluation and the quantification of ecosystem vulnerability to drought.

Protected areas (PAs) are vital for biodiversity conservation and ecosystem services, but their benefits are unevenly distributed. However, the extent of this inequality remains unclear. A quantitative accessibility-availability framework is developed to evaluate the spatial equality of PAs in China. Using data on 3,710 PAs, a county-level spatial distribution index is constructed by combining road-network accessibility (distance to the nearest PA) and availability (PA quantity and coverage) across 2,859 county-level divisions. Distributional equality is summarized with the Gini coefficient, supplemented by spatial clustering and Geographically Weighted Regression (GWR). Results reveal four main patterns: (i) approximately one-third of county-level divisions lack any PA sites, and PA quantity is more unevenly distributed than coverage rates; (ii) accessibility analysis shows an average road distance of 56.57 km to the nearest PA (maximum 503.45 km), with half of all county-level divisions located within 40 km; (iii) pronounced regional asymmetries emerge-eastern, northeastern, and central China host 62.9% of PA sites but only 8.3% of total PA area, whereas western China contains 37.1% of sites and 91.7% of area-indicating fragmented, small PAs in the east versus large, sparse PAs in the west; and (iv) GWR identifies population, road length, GDP per capita, and biodiversity value as the dominant correlates of spatial equality, reflecting the combined demographic, infrastructural, economic, and ecological determinants shaping inequality. Policy implications for PA system planning: (i) integrating spatial equality into PA planning objectives; (ii) prioritizing new PAs near urban and densely populated regions; (iii) enhancing connectivity and gateway infrastructure in remote regions; (iv) supporting small and community-based PAs; and (v) integrating public health into PA planning. Embedding the Gini-based accessibility-availability metric in national and provincial planning can advance an inclusive, just, and ecologically effective PA system.