Search

To enhance the accuracy of land use classification in mining areas, the Object-based Convolutional Neural Network (OCNN) method has been widely used. However, existing researches tend to neglect the importance of decision-level fusion, focusing only on feature-level fusion. This study proposes a new classification framework with Multi-level Fusion of Object-based analysis and CNN (MFOCNN) to achieves high-accuracy land use classification in mining areas. First, Simple Linear Iterative Cluster (SLIC) is employed to generate image objects, which serve as the basic unit for classification. Second, an improved DenseNet is proposed to extract deep features from image patches, which represent image objects, and provide the classification result. Third, handcrafted features including spectral, textural and geometric of the image objects are extracted and fused with the deep features to obtain the classification result with random forest classifier. Finally, the Dempster-Shafer (DS) evidence theory is applied to fuse the two previously described classification results at the decision-level to obtain the final result. Experiments conducted in the mining areas of Erdos using Gaofen-6 images demonstrate that the proposed MFOCNN achieves the best visual performance and accuracy among all tested methods. The MFOCNN, with its feature-level fusion and decision-level fusion, significantly improves the accuracy of land use classification in mining areas. The results suggest that the proposed MFOCNN is a promising method for achieving high-accuracy land use classification in mining areas.

Urban functional zones (UFZs) serve as the spatial carriers embodying urban economic and social activities, thus making the accurate mapping of UFZs imperative for urban planning, management, and sustainable development. Traditional remote sensing-based methods for mapping UFZs primarily capture the physical attributes of ground objects (such as land cover and spatial patterns) while overlooking the inherent social and economic characteristics, as well as the comprehensiveness, heterogeneity, and scale-dependency. With the rapid development of intelligent sensors, the available geospatial big data, reflecting individual human activities, have greatly increased and enable users to analyze UFZs from both physical and socioeconomic aspects. In this study, we provide a comprehensive review of the existing literature on UFZ mapping using remote sensing and geospatial big data. Specifically, this study summarizes the state of the art from three perspectives: spatial analysis units, representation features derived from multi-source data, and the function classification methods of UFZs. Spatial analysis units encompass regular grids, road blocks, image segmentation units, traffic analysis zones, and buildings. Data features consist of the remote sensing image-derived features (such as visual, spatial pattern, and abstract features) and the geospatial big data-derived features (such as spatial, attribute, and temporal features). For function classification, kernel density estimation, cluster analysis, supervised machine learning, probabilistic topic models, and deep learning methods have been applied. Finally, this study discusses the challenges and limitations of UFZ mapping units, the bias issues of geospatial big data, and the integration of remote sensing and geospatial big data. Meanwhile, future opportunities to these issues and the expansion of functions from 2D to 3D are discussed, in order to formulate an enhanced UFZ mapping framework.

ABSTRACTUrban green space (UGS) is important to the urban ecological environment. It has physical characteristics and social function characteristics and plays an important role in urban climate change, sustainable development goals (SDG) and residents’ health. However, existing researches mostly focus on the extraction of UGS physical features, neglecting the importance of UGS social functions, resulting in the unresolved problem of multi-type and fine-grained functional mapping of UGS. Therefore, based on natural language processing (NLP) and multi-source data fusion, this paper proposes a multi-type and fine-grained UGS function mapping method. First, the social functional standards of UGS have been re-established, with a total of 19 categories. Second, the semantic information in the POI data name text is extracted using the deep learning model, and the reclassification of the UGS type of POI data is realized. Then, combined with multi-source data, 18 types of UGS are extracted. Finally, combining multi-source data to extract urban road green spaces (GS), a fine-grained UGS functional map of Shanghai is created. The results show that the overall accuracy rate of the method is 93.6%, and the Kappa coefficient is 0.93, which proves that the method has good performance in large-scale spatial UGS classification.

Automatic urban informal settlement (UIS) extraction based on high-resolution remote sensing image (HRI) is of great significance for urban planning and management. This study proposes a scene classification-assisted and texture feature-enhanced multiscale network (STMNet) for UIS extraction. First, STMNet takes HRI and computed handcrafted texture feature (HTF) as input. Second, it employs a pseudo-siamese network to extract multidimensional deep features from HRI and HTF, respectively. In addition, a feature attention fusion module is constructed to fuse the aforementioned features. Finally, skip connection and feature decoder are utilized to obtain UIS extraction results. In detail, considering the sparse and dispersed distribution of UIS, a scene information aggregation and classification module is constructed to determine whether the input image patch contains UIS. For the characteristics of high spatial heterogeneity and various shapes and scales of UIS, an improved atrous spatial pyramid pooling is presented to extract multiscale and multireceptive field features. An edge loss function is applied during network training to minimize errors in the edge regions of UIS. The effectiveness of STMNet is tested on a self-produced UIS extraction dataset and the publicly available UIS-Shenzhen dataset. Quantitative results demonstrate that STMNet achieved the best performance in terms of $\text{fa}$, $F1$, and $\text{IoU}$. The $\text{mr}$ is slightly higher than that of MAResU-Net and UisNet. In addition, STMNet achieved the best visual interpretation results and the fastest inference speed on the self-produced UIS extraction dataset.

Automatic extraction of tailing ponds from Very High-Resolution (VHR) remotely sensed images is vital for mineral resource management. This study proposes a Pseudo-Siamese Visual Geometry Group Encoder-Decoder network (PSVED) to achieve high accuracy tailing ponds extraction from VHR images. First, handcrafted feature (HCF) images are calculated from VHR images based on the index calculation algorithm, highlighting the tailing ponds’ signals. Second, considering the information gap between VHR images and HCF images, the Pseudo-Siamese Visual Geometry Group (Pseudo-Siamese VGG) is utilized to extract independent and representative deep semantic features from VHR images and HCF images, respectively. Third, the deep supervision mechanism is attached to handle the optimization problem of gradients vanishing or exploding. A self-made tailing ponds extraction dataset (TPSet) produced with the Gaofen-6 images of part of Hebei province, China, was employed to conduct experiments. The results show that the proposed method achieves the best visual performance and accuracy for tailing ponds extraction in all the tested methods, whereas the running time of the proposed method maintains at the same level as other methods. This study has practical significance in automatically extracting tailing ponds from VHR images which is beneficial to tailing ponds management and monitoring.

The building height holds significant importance for comprehensively understanding urban morphology, enhancing urban planning, and fostering sustainable development. Although many methods using optical and SAR images have been presented for building height estimation, these methods fall short in capturing the influences of economic and social attributes on building height. In this study, we introduced a Nature-Economy-Society (NES) feature model to comprehensively represent building height information, and established a multi-scale One-Dimensional (1-D) Convolutional Neural Network for predicting building heights, referred to as NES-CNN. First, we derived the natural attributes of urban buildings from time-series Sentinel-1 SAR images and Sentinel-2 multispectral images, as well as World Settlement Footprint (WSF) data and Digital Elevation Model (DEM), economic attributes from nighttime light and Gross Domestic Product (GDP) data, and social function attributes from Points of Interest (POI) data. Second, an autoencoder is employed to reduce the dimensionality of the high-dimensional natural attribute features, minimizing data redundancy. Finally, the multi-scale 1-D CNN model is presented to explore the correlations between the multi-source and heterogeneous NES features and building height information, facilitating the prediction of building height. In experiments, we applied the proposed method to estimate building heights in Beijing and Shanghai at a spatial resolution of 10 m. The results indicated that for Beijing, the RMSE, MAE, and R values are 6.93 m, 4.41 m, and 0.84, respectively, while for Shanghai, these values are 7.57 m, 5.38 m, and 0.80, respectively. The addition of social and economic attribute information decreases the RMSE by 6 % in both Beijing and Shanghai compared with using only natural attributes. In comparison to existing studies at the same mapping resolution, RMSE decreases by 39 % for Beijing and 51 % for Shanghai. The innovative and inspiring nature of this study lies in its application to large-scale building height estimation.

Assessing the urban residential environment quality (REQ) is essential for advancing sustainable urban development and enhancing urban residents’ living standards. Traditional REQ assessments rely on statistical data, prone to delays and lacking holistic insight. This study takes residential blocks as the analysis units and is conducted within the area of the Sixth Ring Road in Beijing. It synthesizes multi-source geospatial data to devise a comprehensive framework for assessing urban REQ, incorporating facets of environmental health and comfort, housing comfort, transportation convenience, city security, and life convenience. Utilizing the principle of minimal relative informational entropy, this study integrates the Analytic Hierarchy Process (AHP) with the entropy method to determine the weight of each evaluative criterion. Subsequently, a linear weighting technique is employed to ascertain the scores for each evaluative criterion, thus facilitating a detailed examination of the REQ. Finally, the research probes into the complex interrelation between the assessed REQ and the city’s Gross Domestic Product (GDP) and carbon emissions across varying scales. Findings reveal that (1) the overall REQ within Beijing’s Sixth Ring Road is superior at the center and diminishes towards the periphery. (2) The dispersion of environmental health and comfort and city security metrics is relatively uniform, showing minor variations; however, a marked disparity is observed in the distribution of housing comfort metrics. (3) Regions characterized by higher GDP tend to demonstrate relatively higher levels of the REQ. Conversely, areas boasting higher-quality urban REQ are more inclined to exhibit increased levels of carbon emissions.

Studying driving factors of the urban heat island phenomenon is vital for enhancing urban ecological environments. Urban functional zones (UFZs), key for planning and management, have a substantial impact on the urban thermal environment through their two-dimensional (2D)/three-dimensional (3D) morphology. Despite prior research on land use and landscape patterns, understanding the effects of 2D/3D urban morphology in different UFZs is lacking. This study employs Landsat-8 remote sensing data to retrieve the land surface temperature (LST). A method combining supervised and unsupervised classification is proposed for UFZ mapping, utilizing multi-source geospatial data. Subsequently, parameters defining the 2D/3D urban morphology of UFZs are established. Finally, the Pearson correlation analysis and GeoDetector are used to analyze the driving factors. The results indicate the following: (1) In the Fifth Ring Road area of Beijing, the residential zones exhibit the highest LST, followed by the industrial zones. (2) In 2D urban morphology, the percentage of built-up landscape (built-PLAND) and Shannon’s diversity index (SHDI) are the main factors influencing LST. In 3D urban morphology, building density, the sky view factor (SVF), and the area-weighted mean shape index (shape index) are the main factors influencing LST. Therefore, low-density buildings with simple and dispersed shapes contribute to mitigating LST, while fragmented distributions of trees, grasslands, and water bodies also play important roles in alleviating LST. (3) In the interactive detection results, all UFZs show the highest interaction detection results with the built-PLAND. (4) Spatial variations are observed in the impact of different UFZs on LST. For instance, in the residential zones, industrial zones, green space zones, and public service zones, the SVF is negatively correlated with LST, while in the commercial zones, the SVF exhibits a positive correlation with LST.

As basic units of urban areas, urban functional zones (UFZs) are fundamental to urban planning, management, and renewal. UFZs are mainly determined by human activities, economic behaviors, and geographical factors, but existing methods 1) do not fully use multimodal geographic data owing to a lack of semantic modeling and feature fusion of geographic objects and 2) are composed of multiple stages, which lead to the accumulation of errors through multiple stages and increase the mapping complexity. Accordingly, this study designs a multimodal data fusion framework (MDFF) to map fine-grained UFZs end-to-end, which effectively integrates very-high-resolution remote sensing images and social sensing data. The MDFF extracts physical attributes from remote sensing images and models socioeconomic semantics of geographic objects from social sensing data, and then fuses multimodal information to classify UFZs where object semantics guide the fine-grained classification. Experimental results in Beijing and Shanghai, two major cities of China, show that the MDFF greatly improves the quality of UFZ mapping with the accuracy about 5% higher than state-of-the-art methods. The proposed method significantly reduces the complexity of UFZ mapping to complete the urban structure analysis conveniently.

Notable achievements have been made in remote sensing images change detection with sample-driven supervised deep learning methods. However, the requirement of the number of labeled samples is impractical for many practical applications, which is a major constraint to the development of supervised deep learning methods. Self-supervised learning using unlabeled data to construct pretext tasks for model pretraining can largely alleviate the sample dilemma faced by deep learning. And the construction of pretext task is the key to the performance of downstream task. In this work, an improved contrastive self-supervised pretext task that is more suitable for the downstream change detection is proposed. Specifically, an improved Siamese network, which is a change detection-like architecture, is trained to extract multilevel fusion features from different image pairs, both globally and locally. And on this basis, the contrastive loss between feature pairs is minimized to extract more valuable feature representation for downstream change detection. In addition, to further alleviate the problem of little priori information and much image noise in the downstream few-sample change detection, we propose to use variational information bottleneck theory to provide explicit regularization constraint for the model. Compared with other methods, our method shows better performance with stronger robustness and finer detection results in both quantitative and qualitative results of two publicly available datasets.

Abstract Global population growth and land development are highly imbalanced, marked by 43% of population increase but 150% of built-up area expansion from 1990 to 2018. This results in the widely concerned ghost city phenomenon and runs against the sustainable development goals. Existing studies identify ghost cities by population densities, but ignore the spatial heterogeneity of land carrying capacities (LCC). Accordingly, this study proposes a general concept termed underload city to define cities carrying fewer people and lower economic strength than their LCC. The underload city essentially describes imbalanced human-land relationship and is understood in a broader context than the usually applied ghost city. In this study, very high-resolution satellite images are analyzed to obtain land functional structures, and further combined with population and GDP data to derive LCC. We empirically identify eight underload cities among 81 major Chinese cities, differing from previous findings of ghost cities. Accordingly, the proposed underload city considers heterogeneous human-land relationships when assessing city loads and contributes to sustainable city developments.

The quality of urban living environments has become a focal point for local governments and citizens. By conducting a thorough analysis of the human settlement environment, the study can not only gain an intuitive insight into the quality of life of residents but also propose forward-thinking and sustainable suggestions for areas of improvement. This study optimizes and analyzes open platform data closely related to residents and assesses the suitability of different areas for living from diverse perspectives and methodologies. This study has chosen Beijing and Xi’an as the primary case studies. The local living environment is categorized into residential, living, recreational environment, transportation convenience, and safety. Our evaluation combines subjective and objective analysis methods and considers hotspot and cold spot analyses. This study employs the Analytic Hierarchy Process (AHP) as a subjective analysis method and the entropy method for objective analysis. By integrating both methods, it assesses the living environment conditions of Beijing and Xi’an. Furthermore, using GIS software, hotspot analysis is conducted for both cities, identifying areas of high and low quality. Detailed analysis is subsequently carried out for the low-quality clusters. Ultimately, this study, grounded in the theory of Transit-Oriented Development (TOD), presents recommendations for sustainable development aimed at representative rural towns and streets. City centers in Beijing and Xi’an have high-quality environments, while the outskirts show declining quality. Xi’an has uneven resource distribution, while Beijing is more balanced, with hotspot analyses indicating specific high- and low-quality cluster locations in both cities. These disparities and characteristics of the low-quality clusters offer insights for future urban development.

Land covers provide essential information for understanding and detecting ecosystem, resources, and environmental dynamics. However, they are generally mapped at coarser temporal scales to study the inter-annual changes, while scant attention has been paid to map intra-annual land cover dynamics at finer temporal scales. Moreover, existing studies are still limited in intra-annual land cover mapping with dense satellite image time series (SITS). Accordingly, this study proposed a novel approach to accurately classify dense SITS for mapping intra-annual land cover dynamics. First, dense SITS is segmented at multiple spatiotemporal scales to generate optimal spatiotemporal cubes (ST-cubes), which are chosen as classification units. Second, the ST-cubes based on spectral, textural, spatial, and temporal features are integratively defined and employed in SITS classification. Third, the spatiotemporal context is modeled by a spatiotemporally extended conditional random field model that measures both spatiotemporal features and semantic correlation between geographic objects. Finally, the proposed method is applied to map the intra-annual land cover dynamics. Comparative experiments of SITS classification are carried out between our method and three existing competitors in a suburban area in Beijing, China, with a dense Sentinel-2 SITS. Moreover, based on the classification results, we analyzed the quantitative intra-annual dynamics of land cover. The result shows that our approach achieves significant improvements in classification accuracy over existing methods, indicating the effectiveness and superiority of the proposed method in mapping intra-annual land cover dynamics with dense SITS.

Automatic change detection based on remote sensing is playing an increasingly important role in the national economy construction. To address the problem of limited change detection accuracy in existing single-level difference networks, this study proposes the Multi-level Difference Network (MDNet) for automatic change detection of ground targets from very high-resolution (VHR) remote sensing images. An early-difference network and a late-difference network are combined by MDNet to extract multi-level change features. The early-difference network can focus on change information throughout to reduce the spurious changes in the change detection results, and the late-difference network can provide deep features of a single image for reducing rough boundaries and scattered holes in the change detection results, thus improving the accuracy. However, not all high-level features extracted by MDNet contribute to the recognition of image differences, and the multi-level change features suffer from cross-channel heterogeneity. Stacking them directly on channels does not make effective use of change information, thus limiting the performance of MDNet. Therefore, the Multi-level Change Features Fusion Module (MCFFM) is proposed in this study for the effective fusion of multi-level change features. In the experiments, the publicly available open-pit mine change detection (OMCD) dataset was used first to achieve a change detection of open-pit mines over a large area, with an F1-score of 89.2%, increasing by 1.3% to 5.9% compared to the benchmark methods. Then, a self-made OMCD dataset was used to achieve an F1-score of 92.8% for the localized and fine-scale change detection in open-pit mines, which is an improvement of 0.7% to 5.4% compared to the benchmark methods. Finally, the Season-varying Change Detection Dataset is used to verify that the MDNet proposed can detect changes in other scenarios very well. The experimental results show that the proposed MDNet has significantly improved the performance of change detection on the three datasets compared with six advanced deep learning models, which will contribute to the development of change detection with VHR remote sensing images.

Urban parks have important impacts on urban ecosystems and in disaster prevention. They also have diverse social functions that are important to the living conditions and spatial structures of cities. Identifying and classifying the different types of urban parks are important for analyzing the sustainable development and the greening progress in cities. Existing studies have predominantly focused on the data extraction of urban green spaces as a whole, while there have been relatively few studies that have considered different categories of urban parks and their impact, which makes it difficult to characterize or predict the spatial distribution and structures of urban parks and limits further refinement of urban research. At present, the classification of urban parks relies on the physical features observed in remote sensing images, but these methods are limited when mapping the diverse functions and attributes of urban parks. Crowd-sourced geographic data may more accurately express the social functions of points of interest (POIs) in cities, and, therefore, employing open data sources may assist in data extraction and the classification of different types of urban parks. This paper proposed a multi-source data fusion approach for urban park classification including POI and urban functional zone (UFZ) data. First, the POI data were automatically reclassified using improved natural language processing (NLP) (i.e., text similarity measurements and topic modeling) to establish the links between urban park green-space types and POIs. The reclassified POI data as well as the UFZ data were then subjected to scene-based data fusion, and various types of urban parks were extracted using data attribute analysis and social attribute recognition for urban park mapping. Experimental analysis was conducted across Beijing and Hangzhou to verify the effectiveness of the proposed method, which had an overall classification accuracy of 82.8%. Finally, the urban park types of the two cities were compared and analyzed to obtain the characteristics of urban park types and structures in the two cities, which have different climates and urban structures.