Abstract. Flash floods pose a significant threat in mountainous regions, where rapid-onset inundation requires timely and reliable forecasting for effective emergency response. Physically based hydrodynamic models provide strong interpretability but require detailed terrain, roughness, rainfall, and initial-condition information, and their computational cost limits real-time forecasting and repeated scenario analysis. Data-driven models offer higher efficiency, but grid-based architectures may distort irregular hydrodynamic meshes, and purely data-driven graph models can generate physically implausible predictions during long-horizon autoregressive simulation. To address these limitations, this study proposes a physics-informed spatiotemporal graph convolutional network, termed PI-STGCN, for rapid flash flood inundation simulation over irregular triangular control-volume meshes. The model represents the hydrodynamic domain as a directed dual graph, where triangular cells are treated as nodes and shared interfaces are treated as directed edges with geometric and hydraulic attributes. PI-STGCN integrates edge-conditioned spatial graph convolution, rainfall encoding, causal temporal convolution, multi-step residual prediction, and rollout-aware training through scheduled sampling and pushforward rollout. Shallow-water-equation residuals, water-volume consistency, and near-channel boundary constraints are used as soft physical regularization terms to improve local hydrodynamic plausibility. The model was trained using hydrodynamic simulation outputs from design storm events and evaluated on an unseen 100a24h extreme design storm and an independent real rainfall-driven flash flood event. For the 100a24h event, PI-STGCN achieved R² = 0.9142, NSE = 0.9088, KGE = 0.8565, MAE = 0.3390, and RMSE = 0.7892. Ablation and residual diagnostics show that spatial graph convolution and residual prediction are critical for accuracy and rollout stability, while physics-informed regularization improves momentum-equation consistency, although its effect on conventional predictive metrics is event- and metric-dependent. The trained model reduced inference time from 63–105 s for hydrodynamic simulations to 1.06–1.08 s. These results indicate that PI-STGCN provides an efficient physics-regularized surrogate for rapid inundation prediction on irregular meshes in mountainous catchments.

Understanding the evolution and mechanisms of livestock industry agglomeration provides valuable policy insights for reconciling growing meat demand with constrained resource endowments. This study analyzes the spatial agglomeration of livestock industry at county level across China from 2000 to 2022 using the localization quotient and Moran’s I. An interpretable machine learning approach is employed to test hypotheses concerning the driving mechanisms underlying the spatial distribution of livestock industry. The results show that the agglomeration of China’s livestock industry is intensifying, with the agro-pastoral transitional zone (APTZ) emerging as a prominent agglomeration area and distinct agglomeration patterns observed within the zone as well as in its eastern and western regions. Proximity to markets has become an increasingly important determinant of livestock industry agglomeration in China. This market-driven shift has heightened the demand for agricultural feed, prompting the livestock industry to relax its dependence on local natural resource endowments and gradually relocate eastward. Regionally, the agglomeration within the APTZ is shaped by the joint effects of natural and social factors. Natural factors dominate agglomeration dynamics in the western regions of the zone, whereas social factors are more influential in its eastern regions.

Accurate estimation of grazing intensity is essential for understanding grassland dynamics and supporting sustainable grazing management. However, traditional estimation methods rely heavily on field surveys or statistical yearbooks and are constrained by low spatial resolution, infrequent updates, and limited adaptability to large-scale dynamic management. Moreover, many large-scale assessments fail to consider the spatial heterogeneity of livestock foraging behavior, leading to misrepresentation of localized grazing impacts. In this study, we used MODIS LAI data to construct a full-growth curve via a sliding window and locally estimated scatterplot smoothing. Dynamic Time Warping (DTW) was further applied to correct for climatic disturbances, enabling a more accurate differentiation between LAI variations induced by grazing and those driven by climate fluctuations. Deviations from this curve allowed us to identify grazed pixel and estimate actual grazing intensity on the northern Tianshan Mountains. Key findings include: (1) Grazing intensity on the northern Tianshan Mountains was estimated at 0-41.07, 0-32.80, 0-35.54, and 0-35.00 SU & centerdot;ha(-1) in 2005, 2010, 2015, and 2020, respectively. (2) The estimated grazing intensity showed a significant positive correlation with county-level livestock census data (R-2 = 0.65-0.69, p < 0.01). (3) Compared with the LHGI dataset, estimates achieved high consistency (R-2 = 0.77-0.86; RMSE = 0.60-0.68). This study provides a scientific basis for the management of grassland ecosystems and the optimization of grazing on the northern Tianshan Mountains, contributing to the improved sustainable utilization of grassland resources.

Achieving Sustainable Development Goal 11 (SDG 11) under rapid urbanization, particularly in regions governed by stringent policy constraints such as China, requires robust and adaptive urban growth modeling tools. Conventional models often rely on static rules, limiting their ability to capture complex, nonlinear urban dynamics and policy-driven changes. To address this, we proposed a novel Dynamic Cellular Automata (DCA) model that innovatively couples the Maximum Entropy model for suitability analysis with a dynamically adaptive cellular automata system. The DCA model integrates agent preference mechanisms with real-time rule adjustment, enabling more accurate simulation of diverse urban growth patterns and policy scenarios. Validated using multisource spatial and statistical data across China, the DCA model demonstrated higher simulation accuracy, stability, and versatility compared with traditional simulation models across diverse geographical environments. It effectively simulated infill development, edge expansion, and leapfrog development modes. Case study results showed higher urban growth suitability in eastern and central China, particularly in major urban clusters such as Yangtze River Delta, Beijing-Tianjin-Hebei region, and Pearl River Delta. Meanwhile, ecological constraints and limited infrastructure continue to restrict development in the Qinghai-Xizang Plateau and Inner Mongolia. Scenario analyses further reveal tendencies toward consolidation in core cities, balanced expansion along the central and southeastern coasts, and infrastructure-driven growth in western regions. Overall, the DCA framework provides strong support for strategic land use planning in China and offers a transferable tool for policy-driven urban growth modeling in other heterogeneous regions.

Pluvial flooding disrupts road transport and emergency rescue operations, amplifying cascading risks across urban traffic systems and constraining disaster response effectiveness. Although traffic signal control can sustain partial flow and accelerate post-flood recovery, existing systems are insufficiently adaptive to rapidly changing inundation and traffic conditions, limiting their capacity to maintain critical traffic functions during extreme rainfall events. To address this issue, this study developed an adaptive signal control framework based on a coupled hydrodynamic-traffic-rescue multi-agent model to quantify how real-time signal strategies mitigate compound and cascading flood-traffic-rescue interactions and improve emergency response capacity under pluvial flood risk. Taking the Liangshui River Basin in Beijing as a case study, this study integrated empirical peak and off-peak traffic data with designed rainfall scenarios with return periods of 20, 50, and 100 years to evaluate two adaptive mechanisms: yellow-light phase control (YLPC), triggered by critical water depth, and green-light ratio optimization (GLRO), driven by real-time traffic feedback. The results reveal a strong spatial coupling between inundation corridors and congestion-prone traffic arteries, forming cascading flood-traffic risk pathways that amplify congestion propagation and emergency response delays. During flooding, maintaining baseline fixed-cycle signal operation under low-demand conditions shows stronger system resistance than premature YLPC activation, shortening emergency response time by up to 49.2 min, equivalent to a 12.6

Increasingly frequent compound dry-hot extremes (CDHEs) are influencing atmospheric chemistry through land-atmosphere feedback. Here we show that summertime surface ozone (O3) concentrations across natural background regions of North China (>35°N) unexpectedly exceed those in urban areas, accompanied by an increasing trend in tropospheric NO2 vertical column densities (VCDs). This compelling spatial anomaly confirms that natural nitrogen oxides (NOx) sources, particularly soil NOx emissions (SNOx), may play a pivotal role in driving O3 formation. To test this hypothesis, numerical experiments with the Unified Inputs for WRF-Chem (UI-WRF-Chem) are conducted, in which the SNOx scheme as a function of the soil temperature and moisture is developed with the constraint of satellite-retrieved SNOx fluxes in 2018-2024. During a severe CDHE in summer 2022, extreme dry-hot forcings triggered potent SNOx in natural backgrounds. Although natural soils have substantially lower nitrogen availability than croplands, their SNOx could contribute to 43% of O3 production. Our study highlights the urgent need to incorporate NOx emissions from natural lands in air quality management strategies as CDHEs are projected to intensify under climate change.

Accurate and efficient surface water quality monitoring is essential for ecological protection and sustainable development. However, conventional monitoring methods, such as fixed-site observations, often suffer from spatial limitations and overlook crucial auxiliary variables. This study proposes an innovative modeling framework for large-scale river water quality inversion that integrates multi-source data—including Sentinel-2 imagery, meteorological conditions, land use classification, and landscape pattern indices. To improve predictive accuracy, three tree-based machine learning models (Random Forest, XGBoost, and LightGBM) were constructed and further optimized using the Whale Optimization Algorithm (WOA), a nature-inspired metaheuristic technique. Additionally, model interpretability was enhanced using SHAP (Shapley Additive Explanations), enabling a transparent understanding of each variable’s contribution. The framework was applied to the Red River Basin (RRB) to predict six key water quality parameters: dissolved oxygen (DO), ammonia nitrogen (NH3-N), total phosphorus (TP), total nitrogen (TN), pH, and permanganate index (CODMn). Results demonstrate that integrating landscape and meteorological variables significantly improves model performance compared to remote sensing alone. The best-performing models achieved R2 values exceeding 0.45 for all parameters (DO: 0.70, NH3-N: 0.46, TP: 0.59, TN: 0.71, pH: 0.83, CODMn: 0.57). Among them, WOA-optimized LightGBM consistently delivered superior performance. The study also confirms the feasibility of applying the models across the entire basin, offering a transferable and interpretable approach to spatiotemporal water quality prediction in other large-scale or data-scarce regions.

Cropland quantity, quality, and ecological condition may improve in a composite assessment without becoming better aligned with zone-specific management targets. Conventional aggregation can offset deterioration in one dimension with gains in another, obscuring the resulting risk profile. We therefore propose a Target-Oriented Vector Coordination Model (TO-VCM) to evaluate whether county-level changes in cropland quantity (Q), quality (U), and ecology (E) align with policy-referenced management mandates. The model distinguishes structural alignment (Cstru) from effective advancement (Ldev) and integrates these indices with pathway classification and obstacle diagnosis. Applied to Qinghai Province, China, for 2020–2024, the framework showed that 22 of 35 counties improved in overall performance, while 9 of these counties also deviated further from zone-specific targets. Thus, higher composite scores did not necessarily indicate closer convergence with the intended management pathway. The TO-VCM provides a diagnostic basis for identifying such mismatches and informing differentiated cropland governance.

Abstract. Flash floods in mountainous regions are becoming more frequent and destructive under climate warming, yet cross-regional understanding of their triggering mechanisms, cascading impacts, and governance remains fragmented. This review synthesises 1,967 studies published during 2000–2025 to establish a globally comparable baseline of mountain flash-flood research. By integrating bibliometric and topic analyses with qualitative synthesis, we reveal pronounced geographical and thematic imbalances, with research concentrated in Europe and Asia. At the same time, many high-risk mountain regions in Africa and South America remain overlooked. Across regions, flash-flood initiation and impacts are shown to be strongly state-dependent and coupled, emerging from interactions between storm intensity, duration and spatial concentration, antecedent hydrological conditions, and hillslope-channel connectivity. This coupling helps explain why fixed rainfall thresholds are difficult to generalise and highlights the need for dynamic, multi-source early-warning approaches. Comparing evidence on early warning, structural protection, and Nature-Based Solutions, the review shows that cascading processes dominate risk management challenges. We therefore propose an adaptive governance framework that links monitoring and forecasting, spatial planning, grey-green integration, and basin-scale risk sharing under non-stationary climate conditions. Overall, this synthesis consolidates fragmented evidence into a cross-regional knowledge base to support flash-flood risk reduction in mountainous regions, where data and capacity are limited.

China faces a critical spatial imbalance between its agricultural water/land resources and food production, threatening national food security. To address this, this study analyzes available agricultural water-land resources and food production data from 335 prefecture-level Chinese cities (2000-2020) to measure spatial imbalance and explore its underlying drivers. Panel regression models are employed to examine the determinants of water-food and land-food imbalance. Results show that over 20 years, available agricultural water and land declined by 9.96% and 19.39%, respectively. Notably, 46.76% of cities improved their water-food imbalance, while 35% improved their land-food imbalance. Regression results indicate that higher grain yield, yield per unit area, and agricultural employment exacerbate imbalance, whereas mechanization and urbanization alleviate them. The positive association between water-food imbalance and land-food imbalance suggests a mutually reinforcing effect that may accelerate systemic spatial imbalance within the water-land-food nexus. This study provides a novel prefecture-level agricultural water-land dataset and offers evidence for region-specific policies to optimize resource allocation and strengthen China's food security.

Mountainous and Hilly Areas (MHAs) are critical for ecological security, water resource conservation, and agricultural production. However, Agricultural Non-Point Source (ANPS) pollution in MHAs remains poorly understood due to the failure of existing models to capture the pronounced spatiotemporal heterogeneity of smallholder farming systems. This study focuses on the Yuanjiang River Basin, a representative subtropical, agriculture-intensive watershed in MHAs, to address both methodological gaps in simulating ANPS pollution and knowledge gaps in understanding its spatiotemporal characteristics. A customized version of the Soil and Water Assessment Tool Plus (SWAT+), referred to as SWAT+-Smallholder Farming Systems (SWAT+-SFS), was developed to simulate variability within Hydrological Response Units (HRUs) in fertilization timing and dynamic postponement behavior in response to real-time rainfall and soil moisture. Three fertilization representation methods were compared: a conventional discrete-event scheme, a spatially split HRU approach, and the proposed SWAT+-SFS method. While the split-HRU approach disrupted the hydrological unit structure, SWAT+-SFS balanced computational efficiency and accuracy. Compared to the conventional discrete-event scheme, the SWAT+-SFS model reduced overestimation of total nitrogen (TN) during the rainy season and substantially improved total phosphorus (TP) simulation, with average validation Kling-Gupta Efficiency (KGE) values increasing by 67.38 % for TN and 57.03 % for TP across monitoring stations. Simulations revealed that surface runoff-driven losses are significantly amplified by fragmented, prolonged fertilization schedules. This study underscores the critical need to account for complex human behaviors when understanding contemporary watershed hydrological and water quality processes, and highlights the importance of timely, science-based management guidance tailored to smallholder farming systems to support sustainable agriculture in MHAs.

Surface water quality is critically shaped by diverse natural and anthropogenic drivers operating at different spatial scales. However, few studies have systematically quantified how these drivers vary in strength and importance across scales. This study develops an innovative multi-level modeling framework to disentangle and quantify the scale-dependent effects of natural and socioeconomic factors on surface water quality. By integrating sub-watershed-level natural indicators with county-level socioeconomic metrics, our hierarchical model partitions variance across spatial levels. Taking China’s Red River Basin as case study area, the model results show that natural factors at the sub-watershed level explain 51.03% of water quality variation, while county-level socioeconomic factors account for 22.58%. Scale-specific effects are evident: meteorological variables such as precipitation and humidity dominate at the sub-watershed scale, while primary industry intensity and livestock density have stronger impacts at the county scale, both contributing to water quality deterioration. In contrast, modernized facility agriculture contributes positively, with effects concentrated at the county level. These findings underscore the importance of scale-matched indicators in policy design. The proposed framework offers a robust method for multiscale environmental diagnosis and supports adaptive, spatially differentiated water quality governance.

Water security and its sustainable management are critical to human survival and livelihoods. Under the pressures of climate change and population growth, an increasing number of natural watersheds are being regulated by dams and reservoirs. However, the operation of these man-made infrastructures, particularly small-scale reservoirs managed by local governments, is often characterized by flexible and irregular management practices, significantly complicating streamflow modeling-a critical aspect of water management. Remote sensing provides valuable reservoir storage data for data-scarce basins, but its coarse temporal resolution requires integration with ground-based observations or simulations. Unlike traditional physical models that rely on explicit hydrological processes and predefined reservoir operation rules, or data-driven methods that struggle with multi-timescale data and their dependencies, the Multi-TimeScale Long Short-Term Memory (MTS-LSTM) model is a deep learning framework designed to integrate multi-timescale data. This study evaluates the MTSLSTM in integrating monthly remote sensing-derived reservoir storage variation data to simulate daily reservoir-regulated streamflow. The case study on the Yuanjiang River Basin demonstrated that the MTS-LSTM effectively bridges the gap between SWAT-simulated natural streamflow and observed regulated streamflow, a gap primarily caused by reservoir storage variations. The model achieved strong performance at two hydrological stations. For monthly simulations, the mean Correlation Coefficient (CC) was 0.92, Nash-Sutcliffe Efficiency (NSE) was 0.81, and Kling-Gupta Efficiency (KGE) was 0.80. For daily simulations, the mean CC was 0.79, NSE was 0.58, and KGE was 0.71. Integrating remote sensing data significantly enhances simulation accuracy, outperforming naive LSTM models. This study presents a systematic methodology for incorporating multi-source and multi-timescale data to enhance the accuracy of reservoir-regulated streamflow simulations, with a particular focus on regions with limited data and hybrid cascade reservoir systems.

Agriculture green development faces the challenge of balancing the application of both organic and chemical fertilizers. Formulating effective policies to harmonize these applications is essential for ensuring food security and environmental sustainability. However, there is a lack of studies investigating the economic and environmental benefits of policy implementation at the grid scale, and insufficient research has been conducted on the subsequent impacts these policies have on diverse crops. Therefore, this paper develops the SIMPLE-G-Heihe model to simulate the impacts of green agricultural policies on multiple crops planting in arid regions. This model addresses limitations found in the SIMPLE-G model, which supported a single crop and lacked organic fertilizer substitution simulation. The results show that: (1) Merely increasing subsidies for organic fertilizers is insufficient to significantly enhance yields, and improving the efficiency of fertilizer utilization should also be emphasized. (2) Combining improvements in fertilizer technology with subsidies for organic fertilizers can concurrently support economic advancement and environmental mitigation. Research indicates that integrating a 3% organic fertilizer subsidy with an 8% improvement in chemical fertilizer technology increases vegetable output value by 0.27% and reduces N2O emissions by 1.09%. (3) The upstream areas of the Heihe River Basin show a significant reduction in fertilizer use. Fertilizer input is mainly reduced for vegetables in the upstream areas, while for wheat and maize, the reduction is primarily in the midstream and upstream areas. The findings discuss the spatial heterogeneity and flexibility of policy impacts, providing valuable insights for the implementation of green agriculture strategies in China.

Ecological degradation, environmental pollution, and lack of resources have brought challenges to sustainable urban development, whereas the concept of resilience provides a new normative and analytical tool to investigate urban responses to changes. To explore the mechanisms of resilience in cities at different development phases, this study employs a system dynamics approach to simulate and compare the evolutionary state of different resilient subsystems, including economic, social, ecological and infrastructure modules within cities. Then taking Beijing, Fuzhou, and Urumqi in China, three cities at different development stages as examples, the evolution of urban resilience systems from 2000 to 2035 is simulated under the socio-economic development scenario. The results show urban resilience in the three case study cities evolves in different trajectories with distinctive features showcasing different development phase. This study constructs a resilient city system dynamics model, which provides references for future quantitative resilience simulations and can be adapted and applied to resilience simulations of cities in different contexts. In addition, the case study provides distinct policy recommendations for resilience building in cities at various stages of development and points out key areas for resilience building in different cities.

Accurately identifying the optimal spatial scales of effect analysis of influencing factors on water quality is crucial for effective water environment management. To address this, we propose a framework that consists of mix scale division (watersheds, riparian buffers, and circular buffers), and conduct a case study in the Yuanjiang River Basin (YJRB). Scale effects and non-linear impacts of various influencing factors on water quality were identified. The case study results revealed the variations of these water quality indicators were predominated by different influencing factors at various sales, such as by slope (SL) at the scale of riparian buffer (w) = 500 m and circular buffer (r) = 15 km (with a contribution percentage of 16.6 %), mean annual temperature (TE) at the watershed scale (23.4 %), annual precipitation (PR) at the scale of w = watershed and r = 15 km etc. The percentage of cultivated land (CL) > 28 % at the scale (w = 500 m, r = 50 km) will lead to increase in total phosphorus (TP) and SL > 26° at the scale (w = 500 m, r = 15 km) will lead to increase in pondus hydrogenii (pH). While, POP > 30 person per unit area at the scale (w = 100 m, r = 1 km), SL > 24° at the scale (w = 500 m, r = 15 km), NDVI > 0.48 at the scale (w = 500 m, r = 5 km) and PR > 1300 mm at the scale (w = watershed, r = 15 km) will lead to decrease in ammonia nitrogen (NH3-N), total nitrogen (TN), electrical conductivity (EC) and turbidity (NTU) respectively. Results indicated that the precision can be improved by regulating influencing factors within scale effects and considering the non-linear effects of factors on water quality.

The impact of climate change on rice yield varies among different rice varieties. Designing effective agronomic adaptation strategies is crucial for global rice provision. However, considerable uncertainty remains as to which approaches/strategies should be used in different regions. To this end, we conducted a meta-analysis aimed at quantifying firstly the marginal effects of climate change (i.e., temperature, precipitation, and CO2) and four adaptation strategies (i.e., changing varieties, adjusting fertilization, adjusting irrigation, and altering planting dates) on rice yield in Indica, Japonica, and Hybrid rice. We further assessed climate risks to rice yield and identified optimum adaptation strategies under three shared socio-economic pathway (SSP) scenarios. The results of the meta-analysis showed that temperature has the greatest negative marginal effect of −3.11

Context Land use/cover change (LUCC) can directly and indirectly affect surface urban heat island intensity (SUHII) and the effects need to be decomposed. Objectives To perform long-term trend analyses of contribution indexes (CIs) of land use types to urban heat environment in cities and to deconstruct direct and indirect effects of LUCC on SUHII within geographical regions. Methods Mann–Kendall test and Sen’s slope were used to examine the trends of CIs and SUHII in 365 cities during summer of 2005–2019. Structural equation models were established to quantify direct and indirect effects of land use types’ CIs on SUHII in six geographical regions of China. Results First, SUHII in 78.08% and 73.70% of the Chinese cities increased during summer daytime and nighttime, respectively. Second, the CI of built-up land significantly increased across more than half of the cities in all the six regions. Third, not all land use types exerted both direct and indirect effects on SUHII. At daytime, the CI of cropland (direct) was the dominant factor in East China (1.386), South-central (− 0.637), and Northwest (− 0.399) regions. At nighttime, the CI of water bodies (both direct and indirect) was the dominant factor in Northwest (0.506) and Northeast (0.697) regions while CI of built-up land (both direct and indirect) determined in North China (0.476). Conclusions Separation of direct and indirect effects of land use types on SUHII had practical implications for cities to optimize the structures and functions of ecosystems and to take regionally based actions improving the urban heat environment.