Three-dimensional (3D) maps of urban green spaces (UGS) provide a significant dataset for estimating carbon sequestration and understanding urban ecosystem functions. However, existing 3D-UGS mapping methods are often designed for light laser detection and ranging (LiDAR) data with high cost and small coverage, limiting their scalability for city-level applications. In response to the above limitations, we developed a multi-view intelligent fusion network by using multi-view images to generate a 3D-UGS map. The proposed algorithm consists of three parts: (1) Height information estimation network; (2) UGS extraction network; and (3) 3D-UGS map generation module. A height information estimation network was used to retrieve detailed UGS height information. UGS extraction network-based multimodal feature fusion idea was proposed to extract 2D-UGS. 3D-UGS map generation module was developed to produce a fine-grained 3D-UGS map. The proposed algorithm yields a high-quality 3D-UGS map with a root mean square error (RMSE) ranging from 0.85 m to 1.06 m for the urban semantic 3D (US3D) dataset. The results showed that the proposed algorithm achieves remarkable 3D-UGS map performance with an average RMSE of 2.045 m at Beijing. Our research provides new insight into how multi-view images and artificial intelligence (AI) can be integrated to generate fine 3D-UGS maps at the city scale.

Deep neural networks provide innovative solutions for automatic modulation classification. However, the performance of deep neural networks (DNNs) is significantly affected by the incomplete signal pattern conditions, which are caused by the limited number of training samples and the diversity of modulation signals. Existing deep learning models typically utilize regularization methods to enhance the model’s generalization and robustness. However, these methods suffer from the issue of inconsistent training and testing output distributions, which can lead to a decline in model performance during testing. To address this issue, a consistency regularization training method is proposed for AMC under incomplete signal pattern conditions. The method utilizes the designed consistency regularization loss function to explicitly introduce regularization constraints between the submodel outputs and the full model outputs, thereby mitigating the inconsistency in output distributions caused by Dropout and ultimately enhancing the model’s generalization capability. Experimental results demonstrate that our method effectively improves the performance of DNNs in AMC tasks under incomplete signal pattern conditions.

Understanding how transportation infrastructure affects ecological connectivity at multiple spatial scales is critical for regional biodiversity conservation and sustainable urban development. However, how different road hierarchies interact with species dispersal capacities to reshape ecological connectivity across patches, corridors, and entire ecological networks remains unclear. Existing studies often oversimplify roads as homogeneous barriers, limiting our ability to capture scale-dependent and nonlinear fragmentation processes. To address this question, we develop a multi-scale, multi-level analytical framework that explicitly integrates road hierarchy as a continuous resistance gradient and species dispersal thresholds ranging from 10 to 30 km. Using the ChangshaZhuzhou-Xiangtan Urban Agglomeration (CZTUA) in China as a case study, we evaluate ecological connectivity under both single-grade and multi-grade road disturbance scenarios. Results show that (1) High-grade roads (e. g., expressways, railways): impose strong large-scale barriers to connectivity, particularly for species with limited mobility. Under the railway scenario, PC decreased by 30.66 % and IIC by 23.83 % at the 10 km dispersal threshold, with impacts weakening to 16.80 % and 17.32 % at 30 km. (2) Low-grade roads exhibit pronounced scale-dependent effects: provincial roads enhance local connectivity, while county roads shift from inhibitory to facilitative roles as dispersal distance increases. (3) Overlapping road hierarchies produce nonlinear, synergistic disruptions to ecological networks, with impacts magnified by species mobility traits. By explicitly linking road hierarchy, dispersal scale, and network organization, this study advances theoretical understanding of scaledependent landscape connectivity and provides an operational framework for integrating ecological integrity into regional transport planning and multi-jurisdictional environmental governance.

Although detail injection in high-frequency multispectral images (MSI) can improve the spatial resolution of hyperspectral image (HSI) reconstruction, existing methods generally focus on local feature enhancement and lack effective constraints on the detail injection process and global spatial-spectral relationship. This can easily lead to cross modal inconsistencies and boundary misalignments, and ultimately weaken spectral fidelity. To address this issue, we propose a double decomposition network (DD-Net), which integrates intrinsic decomposition, a dual-branch structure, and directional enhancement. Unlike existing methods that directly inject MSI details into a mixed feature space, this network first decouples the low-resolution HSI, separating reflectance-related information from illumination guidance, and then maps spatial-structure restoration and spectral refinement to two dedicated branches for collaborative optimization. Specifically, in the reflectance branch, complementary background priors and reflectance edge priors are constructed, and the reflectance complementary modulation is introduced to enhance structural compensation, boundary constraint, and detail restoration. On the illumination branch, the illumination-guided correction module progressively refines the spectrally guided representations by leveraging edge priors and MSI details, thereby mitigating spectral distortion caused by direct texture injection. Finally, the enhanced representations from the two branches are further refined by the spectral bucket attention module to reconstruct the high-resolution HSI. Experimental results on three datasets demonstrate that DD-Net outperforms existing state-of-the-art methods in both quantitative metrics and visual performance, exhibiting superior detail recovery and spectral fidelity in complex scenes.

The Impervious Surface (IS) is closely related to human activities and ecological environments, whose monitoring is vital for establishing safe, resilient, and sustainable human settlement environments. Synthetic Aperture Radar (SAR), with its timely, cloud-penetrating, and geometrically sensitive observational ability, provides a promising solution for large-scale, high-frequency IS extraction. However, accurately identifying the IS with clear boundaries in SAR amplitude images is a huge challenge due to speckle noise and the scattering heterogeneity of the IS. In response to the above challenge, we innovatively propose a Boundary Awareness Network (SARBA-Net) for IS extraction from SAR images. SARBA-Net employs a multi-scale Boundary Preserving Structure (BPS), which consists of multiple Boundary Awareness and Enhancement Modules (BAEMs) to progressively restore the boundary topology. Within BAEMs, a novel Boundary Contrast Loss (BCL) is designed to generate boundary class awareness, and a hard boundary anchor mining strategy further facilitates BCL to discriminate boundaries. Finally, a Boundary Consistency Auxiliary Loss (BCAL) is introduced in SARBA-Net to refine the boundary detail by the boundary duality between the ground truth and prediction. Extensive experiments on SAR images across diverse IS scenes, including dense urban, rural, road-dominated, and complex mountainous regions, show that SARBA-Net achieves remarkable IS segmentation performance in detail perception, shape preservation, and boundary refinement, with a 2.81% improvement in IoU and a reduction in boundary consistency error to 5.44%. These results demonstrate the effectiveness of SARBA-Net and highlight its potential for robust IS monitoring, particularly under cloudy and rainy conditions where optical imagery is limited.

Accurate characterization of global urban green space (UGS) changes is essential for understanding the urban climate changes and supporting sustainable development goal (SDG) 11 of the 2030 Agenda. However, there is still a lack of high-spatiotemporal resolution monitoring of global UGS changes over the past three decades. This study developed a crowdsourcing data engine driven deep adaption network for monitoring global UGS changes. First, a crowdsourcing data engine is developed to create UGS label samples. Second, a deep adaption UGSs extraction network is proposed to enhance global UGS dynamics mapping accuracy in the lack of time-series label samples. Our method yielded an average accuracy of 85.13% for annual global UGS mapping from 1993-2022. Analyzing the global UGS change results, we found that areas of global UGS during 1993-2022 increased by non-UGS convert to UGS, expanding by 76.92 thousand km(2). The proposed method has significant application value for SDG 11.7 indicator monitoring by leveraging geospatial artificial intelligence and big earth data.

Urban green space and water body play a significant role in urban ecosystems and urban planning tasks. In recent years, the great development of remote sensing technology has greatly assisted the segmentation of urban green space and water body, both in terms of data quantity and quality. However, due to significant inter-class differences and huge training costs, the segmentation of urban green space and water body remains a challenging task. In the past two years, the introduction of the Segment Anything Model (SAM) brought about innovation in the field of segmentation tasks. SAM exhibits good generalization and zero sample ability when processing various images and objects, and can segment any object in any image without the need for additional training. However, it is difficult to apply SAM to the field of remote sensing images. This is mainly because the complex situations such as occlusions and shadows in high-resolution remote sensing images, and retraining the large model requires significant computational and time costs. The proposal of Parameter Efficient Fine Tuning (PEFT) techniques for large models provides a solution. PEFT techniques can quickly adjust model parameters to suit specific tasks with minimal data and computing resources. In this paper, a LoRA-based SAM model is proposed for urban green space and water segmentation tasks. High-resolution satellite remote sensing images within Shenyang City, Liaoning Province, People's Republic of China are collected as samples for model fine-tuning training. To expand the training samples, a hybrid data augmentation strategy is used for the input images. The experimental results show that the fine-tuned SAM model based on the LoRA method achieves better results compared to other representative segmentation methods, which demonstrates the effectiveness of the proposed method.

Long-term ecological observations are invaluable for studying ecological processes, surprises, and effects. They are equally, if not more, crucial for understanding the science and implications of rapidly urbanizing cities, yet they remain surprisingly rare. Here we establish a long-term ecological observatory network consisting of six forest sites along the urban-rural gradient in Beijing, China (BEON), and present a decade-long analysis of nine environmental indicators from 2013 to 2023. Urbanization imprints a distinct, heterogeneous signature: multi-year means of key factors vary markedly over short distances, including CO₂ (405.3-536.5 ppm), O₃ (24.4-31.8 μg m-3), wind speed (0.05-0.19 m s-1), air temperature (13.1-15.9 °C), soil temperature (12.2-16.8 °C), air humidity (50-55 %), and soil moisture (20-28 %). While urban core sites generally exhibit higher CO₂, O₃, temperatures, and soil moisture (with lower humidity and wind speeds), site-specific traits (e.g., vegetation, topography, human activity) drive significant deviations from physical urban-rural gradients. Temporally, consistent diurnal and intra-annual cycles coexist with inconsistent inter-annual trends, and special periods (e.g., Chinese New Year, 2014 APEC Summit) reveal context-dependent shifts-CO₂/temperature declines and O₃ increases. These findings underscore urbanization's profound, heterogeneous impacts on forest microenvironments and validate the necessity of long-term urban ecological networks. Addressing challenges like site representativeness and resource sustainability, we advocate an integrated "Sky-Air-Ground-Network" platform and formulate a synthetic framework to comprehensively capture urban environmental complexity.

Maintaining universal exposure to green space is one of the crucial tasks in building livable cities. However, the prevalence of urban expansion in the past few decades worldwide has resulted in uneven exposure to green space as a spatial concomitant of disordered urban development. Previous literature has mainly focused on examining the impacts of biophysical conditions, socio-economic development, and institutional capacity in shaping green space exposure, with little attention given to the role of urban sprawl. To address the gap, we developed a conceptual framework to explore the relationships between urban sprawl and green space exposure. We then used Amap real-time accessibility model, landscape metrics, and regression models to examine whether urban sprawl can lessen green space exposure in Chinese cities. Statistically, it is clear that southern and southeastern cities, as well as those at higher administrative levels, have greater green space exposure, while eastern and smaller cities tend to exhibit urban sprawl. Furthermore, we partially confirmed that urban sprawl can lessen green space exposure, especially in dispersed and fragmented Chinese cities. This result can be attributed to planning strategies that prioritize city cores over suburban areas. However, this is not the case for complex urban forms. We speculate that this finding is linked to the universal characteristics of unsystematic growth and edge development in Chinese cities. This work may provide insights for planners and decision-makers in coordinating green space planning with urban development trajectories.

Urban green space and water body play important roles in urban planning and urban ecosystems. Accurate segmentation of urban green space and water body is one of the prerequisites for tasks such as land use/land cover (LULC) and ecological environment protection (EEP). Due to the high complexity, significant inter-class differences, and large temporal variations, challenges arise. With the rapid development of remote sensing technology in recent years, the high-resolution remote sensing image (HRRSI) from satellites provides tremendous assistance for the semantic segmentation of urban features. In this study, a dataset called NEU-RS1 is developed based on HRRSI. To our knowledge, this is the first dataset that includes various types of urban green space and water body from different periods, unified into two classes: urban green space and urban water body. The dataset aims to contribute to explore the interrelationships and combined impacts of green space and water body in urban environments. A data augmentation strategy is designed to generate more training samples and solve the problems in practical applications caused by map rasterization. To improve the accuracy of deep learning (DL) based semantic segmentation models, an improved adaptive threshold algorithm is proposed. The algorithm expands the number of thresholds and uses the Jaccard distance as the objective function. The thresholds with the best segmentation results for each class are solved based on the differential evolution algorithm. Ablation experiments are conducted on the NEU-RS1 dataset based on several widely used DL-based semantic segmentation models, namely FCN, UNet, PSPNet, UperNet, DeepLabv3+, HRNet, and SegFormer. The performance of the proposed data augmentation strategy and adaptive threshold algorithm is tested, and their effects on urban green space and water body segmentation are quantitatively and qualitatively analyzed. The results show that both proposed methods can effectively improve the accuracy of the DL-based models. The best mIoU and mAcc achieved by applying the proposed methods are 85.37

Open set automatic modulation classification (OAMC) can identify unexpected unknown modulation types in real-world scenarios, which is receiving increasing interests in electronic countermeasures and signal recognition. However, existing OAMC methods rely on a large number of accurate labeled data, which makes the generalization ability of models poor and undoubtedly brings tremendous difficulties and challenges, especially in the case of limited labeled information. In addition, they only implement simple unknown detection without fine-grained unknown classification, which is necessary especially in areas that require high adaptability and robustness to unknown classes. Therefore, this paper proposes a CNNbased learning framework via self-supervised pre-training with out-of-distribution (OOD) generation and quadratic discrimination for fine-grained OAMC, named FOSSP, to address related problems. First, we use self-supervised pre-training learning paradigm in which a reformed de-biasing weighted contrastive loss is introduced to handle sample imbalance. To effectively capture the semantic features of the signals, we employ a tailored blend of loss functions including reconstruction loss, center loss, and cross-entropy. These are complemented by a suitable distance metric to ensure that the minimal distance between semantic features exceeds the maximum intra-class distance. Despite in the absence of training data for specific signal classes, the proposed FOSSP can distinguish between known and unseen signals. Moreover, FOSSP can determine what kind of unknown class it belongs to when a new signal instance appears. Extensive experiments are conducted on four benchmarks and the results demonstrate the effectiveness of FOSSP.

Urban morphology is a key factor of land surface temperature (LST) variations. However, the heterogeneity of its impact mechanisms across spatial scales, seasons, and Local Climate Zones (LCZs) has not been fully explored. In this study, we first identify the optimal spatial scale at which urban morphological features exhibit the strongest explanatory power for LST. Based on this optimal scale, we systematically evaluate the relative contributions, marginal effects, and interaction mechanisms of impervious surfaces, vegetation, and water bodies on LST across LCZs and seasons. The results indicate that: the explanatory power of urban morphology on LST is highest at the 200-m scale; impervious surface metrics consistently contribute to higher LST across all LCZ types and seasons, whereas vegetation and water-related metrics generally exert cooling effects, especially during summer; and the interaction effects among morphological elements are significantly influenced by both seasonality and LCZ classification. This study uncovers the complex influence of multidimensional urban morphology on urban thermal environments under varying spatial and seasonal contexts. The findings can offer scientific insights for fine-grained and differentiated climate-adaptive urban planning and thermal environment management.

Real estate valuation course is a core major course in college that closely integrates theory and practice. The use of multi-source data for the evaluation of real estate price characteristic and correlation has pioneering application significance for expanding the application of macro real estate policy analysis. Three major urban agglomerations within the Yangtze River Economic Belt are taken as the object of the course practice research. Based on the comprehensive information such as massive commercial housing transaction data, social and economic statistical data and vector data, the practice process of quantitatively analyzing the mutual influence between the urban attraction and the commercial housing prices is completely showed using the improved urban gravity model and the coupling coordination degree model. The policy support points to promote the coordinated development of urban attraction and commercial housing price level are formed as well.

The evolution of a terminal lake at the end of a river not only reflects the climate change characteristics within the basin but also the impact of regional human activities, especially in arid areas. In the Hexi Interior of China, three terminal lakes (e.g., Halaqi Lake, East Juyanhai Lake, and Qingtu Lake) situated in the Shule River, Heihe River and Shiyang River, respectively, have been increasingly studied to support regional ecological protection and sustainable oasis development. In this study, Landsat TM/ETM+/OLI and Sentinel-2 MSI imagery were used to examine Halaqi Lake spanning from 2017 to 2022, East Juyanhai Lake from 1990 to 2022, and Qingtu Lake from 2009 to 2022. The focus of this investigation was to characterize changes in lake area and the impact of climate change and human activities. The results revealed a dramatic change in Halaqi Lake, which suddenly emerged in 2017, initially covering an area of 13.49 km2, gradually vanishing nearly in 2021, and reappearing in 2022 with a reduced area of 9.53 km2. The area of East Juyanhai Lake was 54.39 km2 in 1990 but reduced to 40.84 km2 by 2022. Throughout this period, it encountered episodes of drying up in 1992, 1995, 2001, and 2002. Qingtu Lake emerged in 2009, with an area of 0.09 km2, and subsequently expanded to 2.60 km2 by 2022. Climate change and human activities collectively influence the area fluctuations of these three terminal lakes. Among these factors, temperature changes have a greater impact on the lake area in East Juyanhai. Global warming has worsened glacier melting in the Qilian Mountains, resulting in increased inflow in certain years and substantial lake area expansion. Human activities are the primary drivers of changes in Halaqi Lake and Qingtu Lake. Industrial water consumption is the key factor influencing area changes in Halaqi Lake, whereas water usage in forestry, animal husbandry, and fisheries plays a dominant role in the area changes of Qingtu Lake. Furthermore, the introduction of ecological water conveyance projects has had an indispensable effect on rejuvenating and preserving the watershed areas of these three terminal lakes. It is important to emphasize that human-driven water resource management is the primary cause of sudden changes in the lake areas.

Hot extremes are among the most damaging climate extremes to human society, with urban areas being particularly vulnerable due to the surface urban heat island (SUHI). Despite the growing frequency and intensity of surface hot extreme (SHX), a comprehensive understanding of the patterns and drivers of SUHI responsiveness to SHX (difference in SUHI intensity [delta SUHII] between SHX and non-hot extreme conditions [NSHX]) remains obscure. This study aims to shed light on the spatial-temporal pattern of SHX and the distinct SUHI responses to SHX throughout climatically diverse Chinese urban clusters using the seamless daily land surface temperature (LST) dataset. Our results provided evidence that Chinese urban clusters have experienced increasingly frequent, prolonged, and intensive SHX events regardless of climatic context, with stronger occurrences in summer and fall. Spatially, urban clusters in the northern regions have been subjected to more intensified SHX episodes compared to their southern counterparts. Furthermore, our findings revealed distinct delta SUHII responses to SHX across China, demonstrating synergies between SUHI and SHX when SHX events constrained to urban areas whereas trade-offs when SHX isolated within rural settings. The magnitude of delta SUHII depended upon the relative intensity of SHX between urban and rural settings, and heightened delta SUHII generally associated with more intensified SHX events in urban than rural areas. Additionally, the explainable machine learning-based driver exploration showed that delta SUHII was largely controlled by disparities in evaporative cooling (delta ET) between SHX and NSHX during the daytime, whereas during nighttime, it was predominantly governed by changes in surface heat storage, including urban-rural disparities in surface albedo (delta ABD) and impervious surface fraction (ISF). Facing the intertwined challenges posed by climate change and urbanization, it is imperative for cities to develop effective cooling strategies that emphasize enhanced evaporative cooling and minimized heat storage. These strategies are essential for safeguarding urban residents from potential synergistic effects between SUHI and SHX towards sustainable cities and human settlements.

Satellite derived Urban green space (UGS) maps provide an efficient and effective tool for urban studies and contribute to targets and indicators of the sustainable development goal. However, remote sensing of mapping UGS is challenging due to the existence of mixed pixels and the cost and difficulty of collecting quality training data. This chapter presents a crowdsourced data driven GeoAI model training for mapping UGS from Sentinel-2A satellite images. The proposed GeoAI method consists of three parts: 1) a multi-scale feature extraction module; 2) a multi-modal information fuse module; and 3) and a boundary enhancement module. The results showed that the proposed GeoAI achieved a high overall classification accuracy of 94.6%, which presents a clear UGS structure of a large scale. This chapter provide a fresh insight into how remote sensing and crowdsourced geospatial big data can be integrated to improve urban mapping of green spaces.

Accurate urban traffic forecasting is essential for intelligent transportation systems (ITS). However, the majority of existing forecasting methodologies predominantly concentrate on point-based forecasts (e.g., traffic detector forecasts). A limited number of them pay attention to the urban bidirectional road segments and the complex road network topology. To advance accurate traffic forecasting in complex urban scenarios, this paper proposes a Graph Representation enhanced Fully Attentional Spatial-Temporal network (GR-FAST). First, we construct a refined bidirectional road network graph (BRG) to depict the urban road network topology more accurately, particularly focusing on the turning patterns at intersections. Then, we adopt the graph representation methodology and introduce spatial information encoding (SIE) to explicitly characterize the significance of roads and network structure from multiple perspectives. Enhanced by SIE, spatial attention can capture spatial dependencies from both road network topologies and traffic pattern similarities, thereby forming a unified urban spatial cognition. Finally, a multi-scale residual perception (MRP) module is designed to balance the interplay of short-term temporal variability and long-term periodicity. Experiments on a real-world urban dataset from Wuhan, China, demonstrate that GR-FAST outperforms the state-of-the-art deep learning methods, achieving an improvement of 9.19%. Furthermore, ablation studies suggest that the explicit incorporation of complex road spatial topologies can significantly enhance forecasting accuracy.