In response to the problems of low accuracy, long time consumption, and complex data processing existing in traditional bridge alignment measurement technologies, this paper utilizes vehicle-borne 3D laser scanning technology to achieve rapid measurement of bridge alignments.Firstly, the Alpha3D vehicle-borne 3D laser measurement system is used to conduct multi-round dynamic data collection for a large cross-river bridge.Then, methods such as KNN filtering and cloth filtering are employed to eliminate the interference of dynamic vehicles on the bridge deck with the ground point cloud data, so as to extract the ground point cloud.Finally, the measurement accuracy is verified based on the measured data.The experimental results show that the internal consistency mean error of the vertical alignment measurement results in the stable approach bridge area can reach within 5 mm, and the mean error in the cross-river section is approximately 9 mm.Moreover, the multiple measurement results of the cross-river section can reflect obvious vertical dynamic characteristics of the bridge.It can be seen that the method in this paper has the advantages of high efficiency, high accuracy, and simple operation, showing good popularization and application prospects in the daily operation and maintenance detection of large bridges.

In view of the poor performance of optical remote sensing images in road extraction at night or under insufficient lighting conditions, this paper introduces SAR images road extraction methods based on semi-automatic and fully automatic approaches, as well as deep learning methods based on discriminative models and generative models.It summarizes their technical principles and applicable scenarios, analyzes the advantages and limitations of various methods in terms of extraction accuracy, computational complexity, and generalization, which provides technical references for road extraction based on SAR in complex environments.Meanwhile, in order to improve the effect of road extraction based on SAR, it is pointed out that the technologies of knowledge distillation, diffusion model-assisted annotation and large model architectures will have great potential.

To address the issues of low automation level and limited applicability in verticality detection of pier structures using laser point clouds, this paper proposes an automatic extraction method for pier central axis based on point cloud slicing and rotating calipers algorithm, which assists in calculating verticality for piers with diverse cross-sectional types.Firstly, a stationary scanner is properly positioned, and point clouds from multiple stations are registered via identical targets to obtain the complete point cloud of the pier.Subsequently, the integrated point cloud is sliced along the pier height direction, with outliers removed through statistical filtering.Then, principal component analysis (PCA)is employed to perform projection transformation on sliced point clouds, followed by the calculation of 2D convex hulls for each projected slice.The minimum bounding rectangle is determined using the rotating calipers algorithm, with its geometric center identified as the cross-sectional centroid.Finally, the central axis of the pier is derived through least-squares linear fitting of all slice centroids.Experimental results demonstrate that this method effectively extracts central axes of non-cylindrical piers with automated processing, showing a 3.2 mm deviation compared to the on-site verticality measurement result of 3.8 mm.The proposed approach not only applies to bridge pier central axis extraction, but also provides reference value for central axis extraction in tunnels, buildings, and other structures.

Multi-granularity spatio-temporal objects, as the basic data model of the pan-spatial information system, have special advantages in performing spatial data materialization modeling.Based on the cognitive and behavioral ability characteristics of the spatio-temporal object model, the construction and application of Geo-AI has become a research direction that urgently needs to be broken through in the field.On the basis of completing the basic theoretical analysis of Geo-AI and multi-granularity spatio-temporal object model, combining with the basic principle of ant colony algorithm, a Geo-AI group intelligence model based on spatio-temporal object data and applied to bus route planning is constructed, and experiments and applications are carried out in conjunction with practical cases.The results show that combining spatio-temporal object models with artificial intelligence algorithms enables Geo-AI modeling and applications directly based on GIS, and provides advantages in computation, simulation and real-time interaction for spatial planning problems.

In view of the problems of cross-regional commuting, such as the difficulty in traditional data integration and the high requirement for data timeliness, first of all, this paper combines the characteristics and trends of cross-regional transportation at home and abroad, and proposes the emerging connotations of regional commuting in the new era, such as functional connection and network structure, as well as the core value of people-oriented transportation travel efficiency and quality improvement.Secondly, based on the big data of Internet location, this paper proposes the concept definition and identification method of commuting in the new era at the regional scale, and further verifies the feasibility of the technical methods from multiple dimensions such as cities, districts and counties, and cross-city.Finally, taking Shanghai and its surrounding cities as examples, this paper analyzes the characteristics of cross-city commuting at different scales such as cities, districts and counties.Generally, it presents features such as “networked convection” and “near-area flowing”, further optimizing and empirically verifying the universality and feasibility of cross-regional commuting methods at different scales.

This paper addresses the issues of complex ground penetrating radar (GPR)image features and low accuracy in defect detection by proposing a multi-attribute fusion method for tunnel lining defect detection.By extracting instantaneous amplitude, instantaneous phase, and instantaneous frequency attributes of radar signals, combined with wave-particle duality theory to design a Wave module as the backbone network, a multi-modal feature fusion framework is constructed.The method employs a pyramid structure to extract low-level and high-level semantic features in layers, and introduces a lightweight MLP architecture to optimize network dynamics and computational efficiency.Experimental results demonstrate that the model fusing instantaneous attributes with the Wave module achieves a mean average precision (mAP)of 91.7%in cavity, loose, and steel detection tasks, an improvement of 3.8%compared to the baseline YOLOv8 model.Ablation experiments and comparative analysis verify the effectiveness of the multi-attribute fusion strategy and the Wave module in enhancing feature expression capability, providing a new approach for precise non-destructive detection of tunnel lining defects.

This study addresses the challenges in high-resolution remote sensing image semantic segmentation, such as large variations in object scales, blurred boundaries, and spectral similarity.A lightweight segmentation model is proposed, which integrates multi-scale features and dual attention mechanisms.The model is based on SegNeXt, incorporating a convolutional block attention module (CBAM)into its multi-scale convolutional attention network to refine feature representations through channel and spatial dual attention mechanisms.During the decoding stage, the Hamburger structure is used to integrate mid-to-high-level semantic information.Experiments on the GF-2 remote sensing image dataset show that the model achieves noticeable improvements over the original SegNeXt across various metrics, with particularly superior performance in handling fuzzy boundaries and linear feature categories.The results demonstrate that this method achieves a balance between accuracy and efficiency while maintaining a lightweight design, offering a feasible solution for real-time semantic interpretation of remote sensing images in resource-constrained environments.

This study focuses on tidal creek extraction in the central Jiangsu coastal zone using a semi-automatic method based on multi-source time-series remote sensing imagery, including Sentinel-1 and Sentinel-2A data, within the Google Earth Engine (GEE)platform.The method integrates spectral, index, texture, and polarization features, employing a Random Forest supervised classification algorithm for extraction.Time-series analysis enhances multispectral data, and various vegetation and water indices, along with texture features, are calculated for classification.The results show an overall accuracy of 0.95, outperforming single-data-source methods and providing accurate tidal creek boundary information.This approach offers technical support for ecological conservation and sustainable management in estuarine tidal creek regions.

In the safe operation and maintenance of large structures such as gate piers, the information on surface cracks is of great significance for the overall health assessment of the structure.In response to the problem of rapid and accurate measurement of surface cracks on large structures such as piers that cannot be targeted for safety maintenance, this paper proposes a quantitative identification method for surface cracks of gate piers by integrating multi-source image data to improve the accuracy and reliability of image-based measurement methods.Firstly, based on traditional visible light and thermal infrared images, the image data of the distance matrix is fused to form an image registration method that combines local feature regions with mutual information to achieve fused image acquisition containing multiple texture information.Subsequently, based on the construction of a surface crack recognition method that can effectively distinguish between crack areas and non crack areas, the distance matrix image information is overlaid to achieve the quantification of pixel spatial positions in the fused image of crack surfaces; Finally, by combining the morphological information of the identified crack area, a feature parameter measurement method for the surface cracks of the pier was formed.The accuracy and reliability of this method were verified through case analysis.The results show that this paper provides a new method for measuring the surface cracks of the pier based on multi-source image fusion, which can provide data support for the health diagnosis of the pier structure and has broad application prospects.

With the widespread application of wide-angle vision sensors in various fields, traditional perspective cameras have gradually revealed their limitations, especially in complex visual perception tasks.Although wide-angle lenses provide a broader field of view, the introduced image distortion severely affects the accuracy and reliability of feature matching.To address this issue, this study proposes an innovative panoramic image matching technique based on spherical isometric mapping.The technique firstly projects the panoramic image into a spherical isometric mapping image using the scaramuzza camera model, eliminating the inconsistency distortion inherent in panoramic images in a planar structure.On this basis, a corresponding spherical data structure is constructed, and the FAST corner detection method is adopted.Furthermore, an improved rBRIEF algorithm tailored for spherical pixel structures is proposed, along with the construction of corresponding feature descriptors, to achieve high-precision and highly robust panoramic image matching.

In view of the current difficulties in early detection of coal fires in open-pit coal mines and the late discovery time, this paper proposes an improved UAV infrared image target detection and recognition method for early coal seam spontaneous combustion in open-pit coal mines based on YOLOv12n.By replacing the backbone network with PP-LCNet, introducing a secondary improved MCADSA attention mechanism module, and adding an improved NWD loss function, the detection accuracy of the infrared image coal fire dataset has been improved, providing assistance for the intelligent inspection of early coal seam spontaneous combustion.

This paper proposes a monocular vision-driven real-time high-precision dense scene reconstruction algorithm for robots, based on deep dense monocular visual SLAM and rapid uncertainty propagation techniques to reconstruct 3D scenes from images.The algorithm achieves dense, accurate, and real-time 3D scene reconstruction while demonstrating robustness against extreme noise in depth estimation from monocular visual SLAM.Unlike traditional methods that rely on specialized depth filters or estimate depth uncertainty from RGB-D sensor models, this approach directly utilizes the information matrix from the underlying bundle adjustment problem in SLAM to generate probabilistic depth uncertainty.This depth uncertainty provides a critical signal for weighting depth maps during volumetric fusion.Our method produces more precise 3D meshes with significantly reduced artifacts.Experimental validation on the challenging Euroc dataset shows that compared to methods that directly fuse depths from monocular visual SLAM improves mapping accuracy by 85%.

In response to the challenges faced by autonomous vehicles in achieving accurate localization and dense mapping in complex environments, this paper proposes a real-time monocular dense SLAM algorithm based on 3D reconstruction priors.By incorporating robust geometric priors, the algorithm demonstrates exceptional robustness in unstructured environments and does not rely on predefined camera models, making it adaptable to various general time-varying camera models.The algorithm's architecture consists of four core modules:point-to-map matching, tracking and local fusion, map construction and loop closure detection, a second-order global optimization mechanism.Through adaptive parameter calibration, the algorithm achieves leading performance in multiple benchmark tests under complex scenarios such as dynamic lighting and weak textures.Additionally, the algorithm is capable of operating in real-time.

To mitigate the multi-path effects in GPS triple-frequency PPP time transfer and improve the accuracy and stability of time transfer.Three baselines are selected for the experiment: USN7-USN8 (0 km), IENG-WAB2 (212 km)and PTBB-WAB2 (3406 km).Using the residuals from seven days of GPS triple-frequency PPP observations (from solar day 061 to 068 in 2024)at these six stations, a GPS triple-frequency observation multi-path hemispherical map model is constructed.The performance of GPS dual-frequency and triple-frequency PPP time transfer before and after multipath correction on solar day 069 in 2024 is compared.The Allan variance of receiver clock differences for the three baselines of GPS dual-frequency and triple-frequency PPP solutions improved by 36.1%, 25.3%, and 12.4%, respectively, and by 36.8%, 25.6%, and 12.8%, respectively.The multipath hemispherical map model constructed based on observation residuals can effectively suppress multi-path errors and enhance the stability and accuracy of GPS triple-frequency PPP time transfer.

To address the limitations of the Saastamoinen model in calculating zenith hydrostatic delay(ZHD)due to its reliance on ground-measured atmospheric pressure data and the lack of meteorological instruments at most GNSS stations, this study proposes an improved method based on a general regression neural network(GRNN).By integrating radio occultation data and radiosonde station observations to construct a training dataset, a GRNN-ZHD prediction model was developed.Combined with ZTD derived from GNSS observations of the Crustal Movement Observation Network of China(CMONOC), a novel model for retrieving precipitable water vapor(PWV)was established.The results demonstrate that the GRNN model achieves an average RMSE of 15.23 mm for ZHD retrieval, showing a 46.8% improvement compared to the GPT3 model(28.64 mm).For PWV retrieval, the GRNN model achieves an average RMSE of 5.17 mm, outperforming the GPT3 model's 10.76 mm (51.9% accuracy improvement).Among the 20 validation stations, the GRNN model maintains PWV retrieval deviations below 7 mm at 15 stations, whereas the GPT3 model achieves this threshold at only 3 stations.

To address the problems of degraded ultra-wideband (UWB)positioning accuracy in non-line of sight (NLOS)environments and the divergence of long-term Inertial Navigation System (INS)positioning results, a UWB/INS integrated positioning method based on adaptive unscented Kalman filter (AUKF)and graph optimization is proposed.First, based on an improved IGGIII function, M-estimation is performed by assigning different weights to the deviations between UWB-measured pseudoranges and INS-positioning pseudoranges, achieving NLOS identification and reconstruction.Second, AUKF is employed to fuse observations for UWB/INS integrated positioning estimation.By introducing an adaptive factor that adjusts the Kalman gain according to the innovation variation, the accuracy of the integrated positioning estimate is enhanced.Finally, a graph optimization method constrained by INS increments and line-of-sight (LOS)UWB pseudoranges is adopted, further suppressing the NLOS error in the integrated positioning and improving the accuracy of the positioning estimate.Positioning experiments demonstrate that the proposed algorithm achieves an average positioning accuracy of 0.14 m, representing an improvement of approximately 22%compared to traditional integrated positioning methods, and effectively ensures positioning performance in complex indoor scenarios.

The global navigation satellite occultation sounder Ⅱ (GNOS Ⅱ), carried by the Fengyun-3 satellite, has achieved operational products such as sea surface wind speed, soil moisture, sea ice coverage and sea ice thickness based on the global navigation satellite system reflectometry (GNSS-R)data acquired in orbit.This study employs neural network (NN)technology to develop an SWH inversion model using Beidou Navigation Satellite System reflectometry (BDS-R)and Global Positioning System reflectometry (GPS-R)data provided by GNOS Ⅱ on the FY-3E satellite.A triplet comparison analysis method is adopted to compare and analyze the inherent errors in SWH inversion based on BDS-R and GPS-R data.The research results indicate that the inversion accuracy of BDS-R for significant wave height, assessed solely with data from the European Centre for Medium-Range Weather Forecasts (ECMWF)is 0.43 meters, compared to 0.46 meters for GPS-R.When validated independently using buoy data from the National Data Buoy Center (NDBC), the inversion accuracies for BDS-R and GPS-R are 0.45 and 0.50 meters, respectively.Using the triplet comparison analysis method, the inherent errors in SWH inversion for BDS-R and GPS-R are estimated to be 0.40 and 0.43 meters.This method effectively reduces the impact of inherent errors in the comparison data on the assessment results.Overall, the inversion accuracy of BDS-R for SWH is approximately 7%better than that of GPS-R.The findings of this study provide a reference for the operational application of SWH retrievals from FY-3 GNOS Ⅱ.

Next point-of-interest (POI)recommendation is one of the key applications in geolocation-based social networks.To address the issues of inadequate representation of users' cyclic temporal behavior and insufficient mining of social relationships in existing methods, this paper proposes a POI recommendation method that integrates both cyclic temporal behavior and social relationships.We analyze users' behavioral patterns from their check-in sequences across three time granularities: short-term, periodic, and long-term, and extract cyclical time-sequential behavioral features.Additionally, we mine social relationships between users by examining the overlap in their check-in records and the similarity of their friends, extracting dual-layer social features.The method introduces feature fusion with an adaptive weight allocation strategy and calculates users' preference scores for POIs.Based on these scores, the next POI is recommended to the user.Experimental results on the Sina Weibo (Shanghai)and Foursquare (New York)datasets demonstrate that the proposed method significantly improves hit rate (HR)and normalized discounted cumulative gain (NDCG).

Traditional geographic information systems (GIS)often encounter multiple challenges in the human-computer interaction process, such as cumbersome operation procedures and limited intelligence.With the rapid development of general artificial intelligence technology, new engines centered on generative AI are driving the geographic information industry to accelerate its evolution from digitalization to intelligence.Typical practices include innovative research such as Autonomous GIS, MapGPT, and LLM-Find.Existing studies have confirmed the huge potential of large language models (LLMs)in tasks such as GIS knowledge Q&A and map-making.However, current research still has the following limitations: on the one hand, the models lack the ability to autonomously understand geographic information data and perform complex spatial task analysis; on the other hand, they highly rely on the task parsing and code generation capabilities of the large models themselves.In addition, the API calling mode may lead to the risk of privacy and sensitive geographic data leakage.To address these challenges, this paper innovatively proposes a geographic information intelligent agent, Geo-Agent, based on an open-source architecture.This framework proposes a multi-level instruction parsing strategy based on spatial thinking chains and a data retrieval strategy oriented to graph structures, effectively solving the problems of geographic semantic understanding deviation and spatial logic disconnection.Experimental verification shows that Geo-Agent can understand, manage, and deeply analyze geographic information data, and can complete complex spatial analysis tasks through natural language interaction, providing an innovative path for realizing fully autonomous and intelligent next-generation geographic information systems.

Sluice play an important role in disaster prevention and mitigation, shipping and transportation, as well as agricultural irrigation.It is significant to strengthen the settlement monitoring of sluice to ensure their safe and stable operation.The settlement monitoring of Dongxingang sluice and Dongjiajiang sluice was carried out using PS-InSAR technology to extract the time-series settlement results of the sluices from 2016 to 2020.The results were then compared with those obtained using the SBAS-InSAR method.Finally, the causes of sluice settlement were discussed.The results indicated that Dongxingang sluice and Dongjiajiang sluice settled at a rate of -5.00 and -3.37 mm/a, respectively; By comparing the monitoring results of PS-InSAR and SBAS-InSAR, it can be seen that the annual deformation rate differences for both sluice gates are within 2 mm, and the coefficient of determination for the time-series settlement monitoring results of both methods is above 0.77, indicating a high degree of consistency and reliability between the two methods.The soil texture type dominated by powdery loam and the special topographic features of Hissing Horse Bend may affected the stability of the sluice.Decreases in groundwater levels and sluice water levels may aggravate the settlement of the sluice.Additionally, high-density urban development impacted soil stress, potentially causing settlement in the sluice area.

Low-altitude route planning and infrastructure constitute important research directions in the low-altitude economy.To further enhance the application of real-scene 3D modeling in urban digital transformation and provide a sustainable 3D digital foundation for low-altitude economic growth, this paper proposes a low-altitude route planning algorithm based on RRT+Floyd, supported by real-scene 3D data, tailored for complex urban environments.Experiments are conducted in the eastern high-tech zone and northern built-up area of Jinan city, evaluating route planning at four altitudes (30, 80, 120, and 300 m).Five metrics(route length, node count, smoothness, average turning angle, and maximum turning angle)are compared with the traditional RRT algorithm.Results show that the proposed algorithm significantly improves planning efficiency and speed in low-altitude airspace, particularly in class W airspace.

Research on the identification and classification of urban underlying surface elements based on optical remote sensing images has attracted significant attention.However, the currently available optical remote sensing image datasets for underlying surface classification suffer from issues such as low data source accuracy, limited classification categories, and lack of standardization.These problems make it difficult to meet the research needs for land and sea use classification and sponge city construction.To address this, this paper aims to meet the research needs for fine classification of underlying surface elements for urban hydrological simulation.We construct a high-resolution remote sensing image urban underlying surface fine classification dataset (DUSC-7)based on aerial images with a resolution of 0.1 meters.We extract the underlying surface elements from the images to create sample slices, and perform semi-automatic annotation with reference to the results of the third national land survey and topographic maps.This results in a classification dataset of urban underlying surface elements containing 7 categories and 8859 instances.The images for each category in the dataset are randomly divided into test and training sets in a 3∶7 ratio, and validation experiments are conducted.The experimental results show that, in the verification of the effectiveness of general classification models, the overall test accuracy of the current mIoU models achieve more than 0.648 8.The constructed DUSC-7 dataset can effectively meet the verification requirements for urban underlying surface element classification algorithms.

As a large-scale (Category 2)water conservancy hub within the Longjiang River basin of Fuqing city, Fujian province, the Dongzhang reservoir has long faced the combined risks of seepage erosion and deformation due to the influence of its gently dipping, weathered clayey interbedded layers in the dam foundation.This study integrates 2024 full-cycle monitoring data to establish a multi-source monitoring system combining Sentinel-1 satellite InSAR, 3D laser scanning, and BOTDA distributed fibre optic sensing.It analyses the evolution patterns of dam horizontal displacement, circumferential seepage, and foundation uplift pressure.Results indicate, reservoir volume fluctuations predominantly drive displacement, with an extreme displacement range of 5.05 mm at monitoring points YZ10—YZ13, and a mere 2.8%deviation between GNSS and tension-line data.The primary seepage channel RS-1 exhibits an extreme water level variation of 3.318 m, correlating with the 8～15m depth dielectric constant anomaly zone identified by ground-penetrating radar.The uplift pressure variation at the new Y3-2 monitoring point in the dam foundation reached 3.714 m, coinciding with the downstream thermal anomaly zone and fissure belt, where a 10% increase in seepage corresponded to a 0.3℃ temperature decrease.The study innovatively establishes a dynamic coupled model of seepage, displacement, and temperature.It proposes prevention and control paradigms including BeiDou positioning-optimised monitoring grids, digital twin extreme condition simulations, and curtain grouting reinforcement, providing support for the safety management of reservoirs in complex geological settings.

With the in-depth promotion of smart city construction, the city information model (CIM), as a core component of digital twin cities, has become an important means of fine governance and intelligent management in cities.This article focuses on the construction of city information models based on the three-dimensional full-space digital base of cities, and explores its supporting methods and technical system.Firstly, the concept of city 3D full-space digital base, its constituent elements and its importance in the construction of city information model are elaborated.Then, a framework of the supporting methods for the construction of city information model is proposed, including air-ground integrated data acquisition and processing, intelligent multi-source data fusion, and intelligent analysis and visualization.The feasibility and effectiveness of the proposed method are verified through application analysis.Finally, the research results are summarized and the future development direction is prospected, providing theoretical support and technical references for the construction and application of city information model and the construction of smart cities.

The current urban surveying and mapping results are abundant and the time series is good.The value and role of historical surveying and mapping results should be fully explored.This article uses spatial statistical analysis, standard deviation ellipse, and kernel density analysis methods to study and explore the spatio-temporal evolution of surveying and mapping results on urban construction land, deeply analyze the development status of urban construction, and quantitatively evaluate the classification of urban construction land and the spatio-temporal evolution analysis of urban spatial expansion, explore the characteristics and trends of urban changes and development, and provide scientific theoretical and technical support for urban development, utilization and protection, national spatial planning, use control, and implementation supervision.

Aiming at the problem that the existing research on the changes and influencing factors of glaciers in the Qilian Mountains of Gansu province is insufficient, this paper uses the high-resolution remote sensing images in 2013, 2019 and 2023 to monitor glacier changes based on the second glacier inventory data, and uses the geographic detector to analyze driving factors.The results indicate:①From 2006 to 2023, the number of glaciers increased by 301, the area decreased by 105.07 km².The average annual retreat rate was 0.83 %, and it continued to accelerate.②In the area with small scale, low altitude and steep terrain, the glacier retreats quickly, and the southwestward glacier retreats the fastest, reaching 21.45%.③The increase of temperature changes the thermal balance, which is the key factor for glacier retreat.Slope affects glacier movement and solar radiation promotes ablation, which is the main driving factor.In addition, pollutants such as PM₁₀ produced by increased human activities in recent years have changed the physical properties of glacier surface and enhanced the ablation process.In order to slow down the glacier retreat, it is recommended to control the discharge of cooling, reduce human pollution emissions, and strengthen the protection of glaciers in complex terrain areas.

Currently, there are various methods for identifying ecological sources in ecological network construction research.In-depth research on different ecological sources identification methods is of great significance for constructing comparable ecological network results, as well as understanding their impact and applicability.This study takes the Weihe River basin as an example, utilizing four ecological sources identification methods—remote sensing ecological index, ecosystem services, morphological spatial pattern analysis and habitat quality—along with their various combination approaches (totaling 15 methods)to extract ecological sources.The rank sum ratio method is employed to evaluate the ecological networks constructed using different ecological source identification methods.The results indicate that when constructing ecological networks in the Weihe River basin using different ecological sources identification methods:①The method based on habitat quality identifies the largest number of ecological sources, while the method combining all four methods identifies the largest ecological sources area;②The method based on ecosystem services+habitat quality identifies the largest number of ecological corridors, while the method based on habitat quality generates the longest total length of ecological corridors;③The ecological network constructed using the ecosystem services+habitat quality method performs the best.This study provides a methodological foundation for constructing a unified and comparable ecological network.

To address the challenges of diverse and difficult-to-recognize merging and diverging locations in lane-level maps, this paper proposes a prior query-based detection network for merging and diverging locations.Firstly, candidate locations are recalled using geometric intersections of lane lines as anchors, and the surrounding lane information is rendered as images.Secondly, multi-scale image features are extracted using the ResNet network, and an encoder is employed to build long-range dependencies within the image features, thereby expanding the receptive field.Thirdly, a prior query encoding module is designed to extract the prior positional information of the candidate merging and diverging lane lines.Finally, the decoder is adapted to incorporate the prior positional information, and a binary classification task is designed to identify merging and diverging locations.Experimental results demonstrate that the proposed model exhibits strong noise robustness on maps with incomplete lane lines, achieving an identification accuracy of 96.27%.The proposed method demonstrates high accuracy and robustness in the identification of merging and diverging locations.

HJ-2F is a new-generation small low earth orbit (LEO)synthetic aperture radar (SAR)remote sensing satellite developed by China, primarily designed for all-weather, all-day operational monitoring of ecological and environmental conditions.High-precision orbit determination is essential to ensure the quality and reliability of its remote sensing observations.This study analyzes onboard dual-frequency GPS observation data from HJ-2F acquired in January 2025.The observation quality is systematically evaluated in terms of satellite visibility, cycle slip rate, and measurement noise.A simplified dynamic model is then employed to conduct precise orbit determination (POD).The results demonstrate that the domestically developed onboard dual-frequency GPS receiver enables a radial orbit accuracy better than 1cm and a 3D orbit accuracy better than 3cm, meeting the requirements of high-precision remote sensing applications.

With the new round of scientific and technological revolution and the rise of emerging industries, the core competence of GIS presents three characteristics: “emerging”, “innovative” and “new”.In this context, the engineering thinking of GIS undergraduates needs to be focused on particularly, and the practical ability needs to be improved to serve the development of the new quality productive forces.Taking the course GIS Engineering and Development as an example, this paper analyzes the connotation and importance of GIS engineering thinking, and puts forward the course structure and lesson design of “three-level class”, aiming at improving undergraduates' GIS engineering thinking and ability to solve practical problems and supporting the cultivation of innovative talents for GIS under the background of education reform.