Abstract. Ground surface deformation in permafrost terrain provides critical information on heat and mass transfer during soil freezing and thawing, and serves as a key indicator of permafrost dynamics. However, continuous observations at minute-scale resolution remain extremely rare, sub-daily deformation processes are still poorly documented, and the consistency among remote sensing, contact, and non-contact in situ measurements has not been systematically evaluated. To address these gaps, we established an intensively instrumented permafrost deformation monitoring supersite in an alpine meadow on the central Tibetan Plateau. Ground surface deformation was measured using collocated linear variable differential transformer (LVDT) sensors, ultrasonic ranging, GNSS interferometric reflectometry (GNSS-IR), and Sentinel-1 SBAS-InSAR, together with multi-layer soil temperature and moisture observations. Automated LVDT-based observations provided continuous 5 min deformation records with sub-millimetre precision from 2022 to 2026, thereby resolving, for the first time, sub-daily deformation processes across different freeze–thaw stages. Results show that active layer thaw settlement commonly develops in a stepwise manner within a day, whereas late-thaw-season subsidence associated with excess ground ice melt is more continuous, reflecting sustained drainage and compression. This process also gives rise to a breakpoint-style acceleration in subsidence during the late thaw season. The sub-daily record further clarifies the origin of short-lived late-winter heave events, which are most consistent with infiltration and rapid refreezing of liquid water in shallow soil. The four deformation datasets show broadly consistent temporal patterns. LVDT and InSAR agree closely (r = 0.91), whereas GNSS-IR and ultrasonic ranging show even stronger agreement (r = 0.96). InSAR maintains strong agreement with the other observation methods in both the freezing and thawing seasons, with correlation coefficients consistently exceeding 0.86. GNSS-IR and ultrasonic ranging are more affected by snow and vegetation and require substantial filtering, making them more suitable for seasonal- to multi-year monitoring than for resolving subtle sub-daily signals. At our site, settlement during the active layer thaw stage is strongly correlated with the square root of thawing degree days across all four thaw seasons (R2 > 0.96), and ice–water phase change within the active layer can explain more than 78 % of the observed seasonal deformation. These results provide a valuable observational benchmark for permafrost deformation from sub-daily to multi-year timescales and support improved process-based modelling and interpretation of GNSS-IR and InSAR observations in permafrost regions.

The spatiotemporal dynamics of the Air Quality Index (AQI) and its response to vegetation regulation require further investigation. Using multi-source data from 2020 in Shandong Province, China, this study analyzed the effects of vegetation greenness (NDVI, LAI, EVI), ecological efficiency (Net Primary Productivity, NPP), and landscape structure on AQI within 3 km grids. Monthly correlation analyses revealed that AQI peaks in January (125.09), June (99.01), and December (105.81). PM2.5, O3, and PM10 were the primary pollutants in winter, summer, and spring/autumn, respectively. Vegetation showed a significant purifying effect from June to September. NPP (r = -0.83) was more effective in mitigating air pollution than greenness-related indices (r = -0.48). Pollution mitigation was enhanced by vegetation patches with complex shapes and dispersed configurations. During the non-growing season, the vegetation alleviating effect weakened considerably, and a decoupling between greenness and ecological efficiency occurred. This decoupling was associated with a stronger positive correlation between population density and AQI. The findings highlight the importance of seasonal vegetation dynamics and landscape optimization for regional air quality management.

Maize production is highly vulnerable to rising vapor pressure deficit (VPD), yet its agricultural-scale impacts remain insufficiently characterized due to the lack of integrated assessments linking spatiotemporal VPD dynamics with crop-level water productivity (WP), particularly the relative contributions of yield and evapotranspiration (ET) under varying atmospheric drought conditions. Using the Coupled Model Intercomparison Project Phase 6 (CMIP6) data, we analyzed the spatiotemporal evolution of VPD across five maize ecoregions in the Yellow River Basin (YRB) under three Shared Socioeconomic Pathway (SSP) scenarios (baseline: 1985-2014; future: 2021-2100) and quantified its effects on maize WP. Mean VPD across the YRB maize ecoregions is projected to increase by 21% from 2021 to 2100 (vs. 1985-2014), peaking in the 2080 s under SSP585. Eastern ecoregions (II > I) experience greater VPD escalation than western counterparts (V > III > IV), with post-2030s VPD growth accelerating to 0.036 kPa/yr under SSP585, while SSP126 stabilizes. Projected VPD increases may hinder WP sustainability in regions II-V across all climate scenarios. ET primarily governs WP under mild atmospheric drought (VPD <= 4 kPa), whereas yield constraints dominate under severe drought. Thus, a VPD-dependent shift in the WP demands a process-based optimization of irrigation and yield management under varying atmospheric drought conditions and offers actionable insights for climate-resilient agricultural strategies in (semi-)arid maize-growing regions.

More judicious application of phosphorus (P) fertilizer can moderate the impacts of global warming on crop water productivity (WP) in (semi-)arid regions, but its long-term feedback on future climate and maize ecoregions remains unclear. This study quantified future changes in maize WP under multiple climate scenarios in the Yellow River Basin (YRB) of China and evaluated the potential of optimized P fertilization to mitigate WP decline. Based on the process-maize-model and combined with 5 global climate models (GCMs) & times; 3 shared socioeconomic pathway scenarios (SSPs; SSP126, SSP370, and SSP585), our results showed that WP of maize will decrease by an average of 5% in future relative to the baseline period in the whole YRB, with a greater decline under the SSP585 scenario. WP in different ecoregions also declined, particularly in the Huang-Huai region (II), which experienced significant decreases (p < 0.05) under SSP585 in the 2021-2040 (2030s), 2041-2070 (2050s), and 2071-2100 (2080s). Rising temperatures were the primary driver, with a threshold effect observed: WP turned point at above 20 degrees C in eastern YRB (regions I and II), while in the rest of other ecoregions (III-V), it peaked around 12 degrees C then declining. After crossing temperature thresholds under an ensemble of projected climate scenarios, optimizing P fertilization increased maize WP by an average of 10.5% compared with the no-fertilization treatment. A combined nitrogen (N) and P application (180 kg ha(-1) and 90 kg ha(-1) respectively) maximized WP across all SSPs. These findings provide critical temperature thresholds and effective fertilization strategies for sustaining future maize production and water consumption in the YRB and similar regions.

Wetlands, as one of the most critical natural ecosystems, face increasingly prominent threats of degradation, with the continuous decline of hydrological connectivity being considered a key cause. However, the impact of submerged aquatic vegetation (SAV) on wetland hydrological connectivity has long been overlooked. This study constructs the EFDC hydrodynamic model that takes into account the spatiotemporal changes of SAV and analyzes the evolution of hydrological connectivity in the Baiyangdian Wetland, the largest wetland in North China. The results show that under the growth conditions of SAV, the hydrological connectivity of wetlands exhibits seasonal variation characteristics. The connectivity-domain area, accounting for the effects of SAV density and plant height, decreased on average by 3.5%, 39.2%, and 25.1% across the different growth stages, which are more pronounced in areas with dense vegetation distribution. The water-impeding effect of SAV is the main reason for the decrease in hydrological connectivity, especially during the peak growth period when the impact of SAV on flow velocity exceeds the influence of water depth changes by 14 times. The spatiotemporal differences in SAV distribution density and plant height are the main drivers of the spatiotemporal heterogeneity of hydrological connectivity. In addition, SAV may reduce the flood conveyance capacity of wetlands, leading to increased flood risk. We suggest that remote sensing technology should be combined to determine the spatial distribution of wetland plants, and artificial intervention should be made during the peak growth period to control its growth at key nodes to improve the hydrological connectivity of wetlands. This study enriches the theoretical framework of wetland hydrological connectivity research and provides a scientific basis for wetland SAV management.

Automatic classification of remote sensing images for land use and land cover (LULC) applications encounter numerous challenges because of multi-resolution data, heterogeneous appearance and multi-spectral complexity. Existing deep learning models, especially conventional CNNs, struggle to include inherent features of objects because of their limited ability in modeling the long-range dependency and global contextual information. To address these limitations, this paper proposes a novel Heterogeneous Dual-Path Wide Residual Network (HDP-WRN) model for enhanced LULC classification. The initial feature extraction is performed using a post-activation residual block and two parallel paths using different convolutional operators with different attention mechanisms. Branch A integrates standard convolutions and Efficient Channel Attention (ECA) to extract fine-grained local textures, and Branch B merges Ghost convolutions and Convolutional Block Attention Module (CBAM) to boost global spatial channel feature extraction effectively and their complementary representations are fused for obtaining discriminative joint feature space effectively. Extensive experiments show that HDP-WRN achieves competitive performance on multiple benchmark datasets, with accuracy of 98.67% on EuroSAT, 95.11% on RSSCN, 96.13% on SIRI-WHU, and 97.24% on UC-Merced. Under our experimental setup, the proposed model outperforms the reported results of several existing methods, though direct comparisons should be interpreted with caution due to differences in training protocols across studies. Results verify the model for accurate and robust LULC classification for diverse spatial resolutions and remote sensing images.

The mid-high latitudes of the Northern Hemisphere are one of the major gathering areas of global vegetation, playing a key role in regulating the climate and carbon cycle. Studying the interactive effects of vegetation phenological dynamics and climate change will help predict future vegetation dynamics and ecological protection. However, the response of vegetation phenology to multi-dimensional climatic factors is still unclear, and the spatiotemporal heterogeneity of their contributions to phenological changes and the complex interactions between them are not yet determined. This study used the solar-induced chlorophyll fluorescence (SIF) to invert accurate spring phenological parameters, and combined ERA5-Land datasets to obtain "water-light-heat" multidimensional climatic factors. We quantified the responses of the start of the growing season (SOS) to climate factors and the contributions and spatiotemporal heterogeneity of climate impacts, and discovered the complex relationships and influences. The results show that SOS predominantly occurred between 140-160 days during 2001-2020. SOS was slowly delayed until 2009, and then advanced before that. The climate sensitivity of SOS has obvious spatial differences, and different gradient levels of climate factors significantly affect the sensitivity of SOS. Furthermore, temperature, precipitation, and solar radiation are the dominant factors affecting SOS. Their contributions to SOS changes show obvious latitude patterns, with relative contributions and dominant areas varying across different sub-periods, but temperature consistently controlled the most area of SOS changes. This study reveals the complex interactions and mechanisms between climate change and spring phenology of vegetation on a large scale, aiding in predicting and addressing potential future ecological changes.

This study presents a detailed on-orbit performance analysis of the Hard X-ray Imager (HXI) aboard the Advanced Space-based Solar Observatory (ASO-S), confirming its overall stability within design specifications for key parameters like detection efficiency and energy resolution. However, the analysis focuses primarily on characterizing temporal variations, including distinct periodic fluctuations linked to orbital (∼ 99 minutes) and annual cycles, as well as non-periodic events. Temperature variations highlight orbital/seasonal effects and suggest potential long-term thermal leakage around the Solar Aspect System (SAS) via a gradual rise on the front plate. High-voltage (HV) remains stable during nominal operations, but its management during South Atlantic Anomaly (SAA) passages is critical. Gain analysis identifies a generally stable trend punctuated by five significant abrupt decreases attributed to operational parameter reset errors and radiation exposure during SAA passages (influenced by geomagnetic storms or operational choices). Detection efficiency and energy resolution remained largely stable, with notable deviations primarily linked to the parameter reset error. These findings demonstrate the instrument’s general robustness while highlighting specific anomalies and underscoring the need for ongoing monitoring, optimized operational protocols (especially HV management), and time-dependent calibration to ensure the highest data quality for solar flare science.

Land Use and Land Cover (LULC) classification is critical for environmental monitoring and sustainable resource management, but faces challenges in accurately capturing complex spatial-spectral features and long-range dependencies in remote sensing imagery. To address this, we introduce a Hybrid Wide Residual Network-Dual Spatial Positional Embedding Vision Transformer (WRN-DSPViT) framework enhanced with Squeeze-and-Excitation (SE) blocks and dual spatial positional embeddings. This model integrates a Wide Residual Network (WRN) for local spatial feature extraction and a Vision Transformer (ViT) with novel dual spatial encoding to capture global context via multi-head self-attention, where SE blocks dynamically recalibrate channel-wise features. Attention pooling is employed to fuse spatial features, allowing for adaptive weighting of important regions in the image, further enhancing classification accuracy. For multimodal hyperspectral-LiDAR data (Houston 2013), we extend this framework with parallel WRN-SE streams and cross-modal transformers, preserving spatial relationships through dual encodings. Evaluated on three benchmarks—EuroSAT (Multispectral), Houston 2013 (hyperspectral-LiDAR), and DeepGlobe (road extraction) —the hybrid WRN-DSPViT achieves state-of-the-art performance: 98.80% accuracy on EuroSAT (3.26 M parameters, 201.68 MFLOPS), 91.24% overall accuracy and 92.18% average accuracy on Houston 2013, and 0.802 F1-score/0.660 IoU on DeepGlobe.

Clarifying the spatio-temporal characteristics of urban vegetation phenology (UVP) is essential for understanding the potential effects of future climate change on vegetation growth. However, current studies lack quantitative analysis of urbanization impacts on spatio-temporal characteristics in UVP. This study measured UVP (including start, end and length of the growing season: SOS, EOS and GSL) changes in both long-term (from 2003 to 2020) and inner-outer city (between inner and outer city areas) in 365 Chinese cities using multi-source data. SHapley Additive exPlanations (SHAP) and accumulated local effect models were used to identify the key drivers and their changing patterns. The results indicated that UVP changed more consistently in inner-outer changes than in long-term changes. SOS in 75 % of cities advanced by -15.96+10.30 days in long-term changes, while GSL and EOS in 74 % and 72 % of cities extended by 23.89+13.81 days and delayed by 11.93+8.99 days, respectively. SOS in 80 % of cities advanced in inner-outer changes (-12.81+9.33 days), and GSL and EOS in 85 % and 73 % of cities extended and delayed (18.28+13.17 and 10.91+7.59 days), respectively. Long-term and inner-outer changes in UVP showed significant differences across city sizes. In the climate zones from south to north, there was a clear gradient in UVP in long-term changes, but little change in UVP in inner-outer changes. SHAP analyses revealed that temperature is the most important factor affecting UVP, followed by night lighting, radiation, and these drivers showed complex non-linear relationships with UVP. This study clarified the spatiotemporal characteristics of UVP, which contributed to a better understanding of the carbon sink potential of urban vegetation.

Forests, being the largest and most intricate terrestrial ecosystems, play an indispensable role in sustaining ecological balance. To effectively monitor forest productivity, it is imperative to accurately extract structural parameters such as the tree height and diameter at breast height (DBH). Airborne LiDAR technology, which possesses the capability to penetrate canopies, has demonstrated remarkable efficacy in extracting these forest structural parameters. However, current research rarely models different tree species separately, particularly lacking comparative evaluations of tree height-DBH models for diverse tree species. In this study, we chose sample plots within the Bila River basin, nestled in the Greater Hinggan Mountains of the Inner Mongolia Autonomous Region, as the research area. Utilizing both airborne LiDAR and field survey data, individual tree positions and heights were extracted based on the canopy height model (CHM) and normalized point cloud (NPC). Six tree height-DBH models were selected for fitting and validation, tailored to the dominant tree species within the sample plots. The results revealed that the CHM-based method achieved a lower RMSE of 1.97 m, compared to 2.27 m with the NPC-based method. Both methods exhibited a commendable performance in plots with lower average tree heights. However, the NPC-based method showed a more pronounced deficiency in capturing individual tree information. The precision of grid interpolation and the point cloud density emerged as pivotal factors influencing the accuracy of both methods. Among the six tree height-DBH models, a multiexponential model demonstrated a superior performance for both oak and ”birch–poplar” trees, with R2 values of 0.479 and 0.341, respectively. This study furnishes a scientific foundation for extracting forest structural parameters in boreal forest ecosystems.

Vegetation changes and human activities in both natural and urban environments have played a crucial role in carbon cycling and sustainable development globally. However, there is an insufficient comparison in national vegetation changes across regions with varying intensities of human activities to those natural areas. Based on urban boundary and night-time light datasets, we have identified and extracted rural, urban-low activity, and urban-high activity areas within China. Geodetector model was applied and conducted to assess the vegetation impacts of seven distinct natural and human factors on vegetation. Results show that overall vegetation change trend was characterized by Significant greening from 2000 to 2020. Areas with less than 1% Significant degradation are predominantly located in the southeastern China. Despite the dominance of forest growth, cropland in urban areas exhibits more stability under human control. Human involvement has significant restraint on vegetation growth from rural to urban, while there was little difference in vegetation growth between areas with strong human activity and areas with weak human activity. Moreover, human factors and land use/land cover have gradually become the dominant impact combinations in the eastern China over the past 20 years. The influence of temperature is increasing annually in southern China but decreasing in northern China. Meanwhile, factors related to water (e. g. precipitation and soil moisture) had more pronounced influences on western China. Our results provide a comprehensive insight in vegetation dynamic change in areas under different degrees of human activities, as well as the alterations in main driving factors on spatial heterogeneity of vegetation.

Tree species classification is crucial for forest resource monitoring and biodiversity conservation, particularly in ecologically fragile boreal forests like the Greater Hinggan Mountains region in China. In this study, multiple features, including spectral features, texture features, and topographic features derived from multi-temporal Sentinel-2 imagery and the ASTER GDEM V3 dataset, were incorporated into machine learning models. Six algorithms of Support Vector Machine (SVM), Random Forest (RF), Extreme Gradient Boosting (XG-Boost), Decision Tree (DT), Multi-Layer Perceptron (MLP) and k-Nearest Neighbors (KNN) were then fitted and evaluated. The results revealed that the SVM algorithm achieved the highest overall accuracy of 93.79%, with spectral features from May and vegetation indices from February contributing most significantly to classification performance. This study can provide a reference for feature computation and machine learning algorithm selection for tree species classification in boreal forests.

Phenology refers to the natural responses of vegetation to environmental factors and disturbances throughout its life cycle. Focusing on the mid-high latitudes vegetation in the Northern Hemisphere, this study reported the differences in phenology extracted using “photosynthesis” and “structure” indices, specifically solar-induced chlorophyll fluorescence (SIF) and leaf area index (LAI), respectively. We analyzed the dynamics of phenology inverted by different indices and validated the inversion accuracy using data from FLUXNET2015 gross primary productivity (GPP). The results revealed that: (1) Both the start of the growing seasons (SOS) and the end of the growing seasons (EOS) showed significant advancing trend. On average, the SOS derived from SIF (143 days) was later than that derived from LAI (136 days), whereas the EOS derived from SIF (236 days) occurred earlier than that from LAI (245 days); (2) Across all four climatic zones, phenology was dominated by an advancing trend, with the rate of phenological change derived from SIF generally faster than that from LAI. Distinct differences in phenological trends were observed among various vegetation types; (3) Phenological parameters extracted using SIF were closer to actual observations than those derived from LAI, with the accuracy of SOS inversion generally higher than that of EOS for both indices. The study demonstrates that “photosynthesis” indices can produce more accurate photosynthetic phenology than “structure” indices, which provided a reference for the subsequent vegetation phenology studies.

Cultural ecosystem services (CESs) play a critical role in urban residents' well-being, yet conventional evaluations rely heavily on green-space area and overlook how facility quality and basic services influence the delivery of actual cultural benefits. To address this methodological gap, this study develops a three-tier evaluation framework-service potential, actual supply capacity, and actual service utility-to quantify multistage attenuation in CES provision across 95 parks in seven central districts of Shenyang, China. The framework integrates 114 quantitative and qualitative indicators from field surveys, national facility standards, and perception-based assessments, enabling a scientifically robust and replicable assessment of how cultural benefits are transformed from ecological structure to human experience. Results reveal that single-index, area-based assessments substantially overestimate CES supply: district-level supply-demand ratios drop from 66 to 195% to only 11-55% once quality and basic services are incorporated. Comprehensive and special parks retain the highest CES potential, whereas community and linear parks undergo significant losses due to aging facilities, insufficient maintenance, and inadequate infrastructure. Education and cultural services exhibit the most severe shortages, with deficits reaching 59-84%, underscoring structural limitations in learning-oriented spaces. By distinguishing structural (quantity), functional (quality), and experiential (basic service) constraints, the framework provides clear diagnostic guidance for targeted planning and management. Its multistage structure also reflects broader principles of sustainable urban development: improving CES requires not only expanding ecological elements but also enhancing service quality, strengthening infrastructure, and promoting equitable access to cultural benefits. The framework's generalizability makes it applicable to high-density cities worldwide facing land scarcity and green-space inequality, supporting efforts aligned with SDG 11 to build inclusive, resilient, and culturally vibrant urban environments.

Climate variability, including global warming, spatiotemporal changes in precipitation, etc., has implications for the growth and development of cereal crops, posing a concealed threat to future food security. The Yellow River Basin (YRB) holds significant importance in China's agricultural landscape, serving as a vital hub for the cultivation of maize and the production of grains. This study employed meteorological grid data simulated by the Climate–Weather Research and Forecasting model (CWRF), in conjunction with the Decision Support System for Agrotechnology Transfer (DSSAT) model, to forecast the prospective ramifications of climate change on maize phenology, yield, and grain nitrogen content (GNC) within the YRB from 2020 to 2050. Our results indicated that the YRB is expected to experience increasing trends in temperature, radiation, and precipitation from 2020 to 2050. Significant spatial variations were observed across the YRB in maize phenology, yield, and GNC. The overall growth period of maize in the entire basin is projected to be reduced by 13 days, comprising a 5-day decrease from sowing to flowering and an 8-day decrease from flowering to maturity. The future maize yield in the YRB exhibits a general declining pattern, with an average reduction of 10.5

Urban green spaces (UGSs) are critical for landscape, ecological, and climate studies. However, the generation of long-term annual UGSs maps is often constrained by the lack of sufficient, high-quality training samples for training classifiers. In this study, we introduce an automatic training sample migration method based on visually interpreted reference data and long-term Landsat imagery, implemented on the Google Earth Engine (GEE) platform, to produce annual UGSs maps for Tianjin from 1984 to 2022. Migrating training samples to each year significantly improved classification performance, especially for UGSs and water bodies. UGSs coverage in sample areas increased from 5% to 38%, resulting in more reliable trend detection. Our spatiotemporal analysis revealed that green coverage in the study area reached up to 40%, dominated by tree cover that is significantly underestimated in existing global and regional land cover products. Distinct temporal patterns emerged between the old built-up area (OBUA) and new built-up area (NBUA). Early UGS decline was largely driven by NBUAs, while post-2007 greening involved both OBUAs and NBUAs, as captured by classification maps and vegetation indices. Our study proposes a scalable and practical framework for long-term land cover mapping in rapidly urbanizing regions, with enhanced potential as higher-resolution data becomes increasingly accessible.