[Purposes]To address the issue of the lack of long-term continuous positioning performance evaluation of China's BDS-3 new frequencies in the Antarctic region.[Methods]This paper uses the dual-frequency ionospheric-free combination model technology.Based on the continuous one year observation data from multiple tracking stations in the Antarctic region,the static and dynamic PPP performance of BDS-3 is evaluated,and a comparative analysis is conducted with the compatible frequencies of other navigation and positioning systems.[Findings]The experimental results show that in the Antarctic region,the satellite constellation status of the compatible frequencies of BDS-3 and Galileo is good and comparable,while that of the GPS satellite constellation is poor.The compatible frequencies of BDS-3 and Galileo can achieve stable and reliable PPP positioning,with the positioning accuracy reaching about 1 cm per epoch.However,the PPP accuracy of the GPS system is poor,and it cannot achieve stable and continuous positioning.The satellite constellation status and PPP accuracy of the BDS+Galileo+GPS combined constellation are significantly improved compared with any single system,and the positioning accuracy can reach the millimeter level at most.[Conclusions]The new frequencies of BDS-3 have the ability to provide high precision and high reliability positioning services in the Antarctic region,and the combined application of multiple systems is an effective way to improve the navigation and positioning performance in the polar regions.

[Purposes]Satellite clock anomalies are among the critical factors affecting the accuracy and reliability of GNSS precise point positioning (PPP).Conventional detection approaches often rely on prediction models,which suffer from low sensitivity and limited ability to directly mitigate their impact on PPP performance.[Methods]To address this issue,this study proposes a real-time detection method for BDS satellite clock anomalies using multi-station observations.The method incorporates satellite clock bias parameters into an extended PPP model and establishes an anomaly identification and elimination mechanism through combined equation adjustment and residual analysis.[Findings]Experimental results demonstrate that the proposed method can effectively detect anomalies in both simulated and real datasets.The station residual RMS is reduced by 37%and 51.4%,respectively,and the 3D positioning accuracy is improved by up to 42.7% and 62.5%,respectively.Both schemes can achieve a residual RMS of approximately 0.017 m and a 3D positioning accuracy of 0.09 m.[Conclusions]The method significantly strengthens anomaly detection capability and improves PPP accuracy,providing robust support for high-precision applications of the BDS.

[Purposes] The quality of raw observations at reference stations is a critical factor in delivering high-precision services via the BeiDou ground-based augmentation system (BDS-GBAS),and a systemic quality analysis is of great significance for improving service accuracy and reliability.[Methods] This study analyzes observational data from five consecutive days at 110 domestic reference stations in the BDS-GBAS network.Using the Anubis software,we conduct data quality assessments in terms of completeness rate,cycle slips,multipath error,and carrier-to-noise ratio (C/No).[Findings]The results indicate that,91.8% of stations achieve a completeness rate above 85%; 99% of stations have a cycle slip rate (CSR)below 1 (per 1000); 99% of stations exhibit multipath error below 50 cm; 99% of stations have C/No above 40 dB·Hz,with the BDS system showing superior performance.We also identify a small number of anomalous stations,and propose possible causes and optimization suggestions based on station geographic location,antenna environment,etc.[Conclusions]The results confirm that the national BDS-GBAS network exhibits high overall observational quality.These findings provide important references for the layout,operational strategies,and data quality assurance of augmentation system reference stations.

[Purposes]To address the issues of low accuracy and adaptability with current remote sensing image coastline extraction methods.[Methods]This paper proposes a new coastline extraction method that combines adaptive simple linear iterative clustering (SLIC) and fractal net evolution approach (FNEA),providing dependable data support for precise coastal zone monitoring and protection.Firstly,according to the complexity of the image,the optimal number of superpixel segmentation blocks was calculated,and the SLIC theory algorithm was used to realize the adaptive superpixel segmentation of remote sensing images.Then,the regional similarity between superpixels is defined according to multiple features,and FNEA is used to merge superpixels to achieve coastline extraction.[Findings]In order to verify the superiority of the proposed algorithm,the proposed method and four comparison methods are used to conduct experiments on coastline remote sensing images with different complexities.The average recall rates are 96.46%,85.34%,89.54%,92.29%and 92.15%,respectively,and the average Kappa coefficients are 0.93,0.63,0.80,0.81 and 0.86,respectively.[Conclusions]The proposed method achieves both adaptive superpixel segmentation and precise coastline extraction.

[Purposes]Landslide recognition in remote sensing images is crucial in geological disaster detection and emergency response.However,existing methods often face challenges such as limited detection accuracy and poor recognition of small-scale landslide targets.Based on the YOLO11s model,this paper proposes an improved landslide detection algorithm named YOLO11s-RDS,which balances detection accuracy,efficiency,and multi-scale feature extraction capabilities.[Methods]The algorithm enhances the model's detection ability for complex landslide features and small-scale landslide targets by integrating the structural reparameterization module RepVGG into the YOLO11s backbone network,introducing the dynamic upsampler DySample into the neck network,and inserting a multi-scale sequence feature fusion module between the neck network and the head network.[Findings]The experimental results on the remote sensing landslide dataset show that the improved YOLO11s-RDS model achieves improvements of 1.3,8.8,5.5,5.5,and 5.7 percentage points in precision,recall,mAP_0.5,mAP_0.5:0.95,and F1-score,respectively,compared to the original model.[Condusions]Compared to other models in the YOLO series,the improved model YOLO11s-RDS demonstrates strong overall performance in landslide identification in remote sensing imagery.

[Purposes]To systematically evaluate the operational monitoring potential of LT-1 L-band SAR in the dense vegetated areas of northwest Yunnan.[Methods]This study quantitatively compared the coherence point density,peak coherence,and integrity of typical subsidence zones using SBAS-InSAR deformation inversion based on 16 scenes of LT-1 and 21 scenes of Sentinel-1 data from Lanping county,Yunnan.[Findings]Results indicate: ①LT-1 exhibits significantly superior coherence performance than Sentinel-1 in high-vegetation areas,with a correlation point density of 6685 points per km² and a peak coherence of 0.85,compared to Sentinel-1's 5790 points per km² and 0.25,respectively.②LT-1 demonstrated higher deformation monitoring accuracy,with average RMSE₁ of 1.60 mm and MAE₁ of 1.24 mm,both lower than Sentinel-1's 4.69 mm and 3.47 mm.③LT-1 detected a maximum annual average subsidence rate of 362 mm/a in mining areas,far exceeding Sentinel-1's 65 mm/a,while revealing the coupling between photovoltaic power station subsidence and river flow direction.[Conclusions]Leveraging its L-band penetration capability,LT-1 enables millimeter-level monitoring,providing operational support for mine safety,photovoltaic foundation stability,and geological hazard early warning.

[Purposes]Surface subsidence induced by mining activities in mining areas is characterized by rapid deformation rates,significant damage severity,and extensive spatial coverage.To address the challenge that interferometric synthetic aperture radar (InSAR) struggles to capture complete basin information during large-scale subsidence monitoring due to spatiotemporal decoherence.[Methods]This paper proposes an InSAR-based large-scale subsidence monitoring method that integrates unmanned aerial vehicle laser radar (UAV-LiDAR)data.This method first acquires cumulative subsidence basins from both time-series InSAR and UAV-LiDAR data within the mining area.Reliable regions for both datasets are delineated based on deformation threshold segmentation.Kriging interpolation is then applied to spatially fuse transitional zones,enabling comprehensive reconstruction of the mining subsidence basin.[Findings]Verification conducted at the S1306 working face of Shanxi Lu'an Gucheng Coal Mine demonstrates that InSAR effectively monitors non-central subsidence zones,while UAV-LiDAR compensates for its accuracy limitations in large-scale subsidence areas.The fused results yield continuous and complete subsidence basins with well-preserved fine-scale features.[Conclusions]Compared with leveling survey data,the root mean square error (RMSE),mean absolute error (MAE),and mean squared error (MSE)of the fusion results were 81.6,63.0,and 6.7 mm,respectively.This validates the effectiveness of this method in obtaining complete and accurate subsidence information in mining areas under complex mining conditions.

[Purposes]Aiming at the problems of noise interference and insufficient signal continuity in the process of seabed photon extraction from ICESat-2 laser altimetry data.[Methods]This paper proposes an improved denoising model based on the density-based spatial clustering of applications with noise (DBSCAN)algorithm.The model introduces an elliptical filtering kernel to adapt to the spatial distribution characteristics of seabed photons,incorporates a small cluster removal mechanism to eliminate discrete noise,and applies a continuity constraint strategy to enhance the spatial coherence of the signal photons.[Findings]Experimental results in the Oahu area of Hawaii show that the proposed method can effectively extract seabed photon signals,with denoising accuracy indices reaching R²=99.45%,RMSE=0.48 m,MAE=0.38 m,and MRE=14.82%.[Conclusions]The proposed method demonstrates significant advantages in suppressing noise and maintaining signal integrity,accurately reflecting the spatial structural characteristics of seabed topography,and providing a reliable technical approach for underwater topographic mapping in coastal zones.

[Purposes]Incorporating building facade 3D data from the ground-view perspective can effectively mitigate geometric distortion and texture blurring from the 3D building models using aerial imagery alone.However,existing studies rarely address automated methods for acquiring ground data.With recent advancements in multi-sensor fusion and simultaneous localization and mapping (SLAM)technologies,unmanned systems have achieved significant improvements in environmental perception and autonomous navigation.[Methods]Based on these new developments,this paper proposes a method for autonomous acquisition of 3D building facade data using an unmanned ground vehicle (UGV).The method automatically generates key acquisition viewpoints from facade geometry information and extracts feasible vehicle pathway from point-cloud data,thereby enabling autonomous path planning and navigation between viewpoints.[Findings]Experimental results demonstrate that:①the point cloud-based passable region classification method for UGV reaches an average accuracy of 92.81%; ②the trajectory error of locally planned paths is less than 0.2 m;③integration of ground data significantly enhances the visual quality and realism of reconstructed 3D models.[Conclusions]Overall,the proposed method exhibits strong accuracy and robustness,providing a highly automated solution for real scene 3D modeling.

[Purposes]Since its launch in 2013,GF-1 has been widely applied in fields such as agriculture,forestry,urban environment,and land use change monitoring.As fundamental quantitative product,surface reflectance has long been a focus of attention.To further strengthen the application and promotion of the GF-1 products,this paper carries out extensive verification of the accuracy of the products in overseas regions.[Methods]Using the ACFrC software based on atmospheric lookup tables to perform atmospheric correction on images from GF-1 PMS and WFV sensors in Southeast Asian countries including Thailand,Bangladesh,and Laos.Indicators such as accuracy (A),precision (P) and uncertainty (U) are adopted to evaluate the cross-validation accuracy between GF-1 surface reflectance products and Landsat 8/9 OLI data.[Findings]The results show that for the three regionsthe minimum value of the A,P and U indicators is 6.79×10^-5,and the maximum is 7.82×10^-2,both of which are lower than the error constraint expectation of 0.015+0.15R (R is the surface reflectance),which are 1.90×10^-2 and 8.10×10^-2 respectively.Thus,the products meet the accuracy requirements.So there is a high degree of consistency between the surface reflectance products generated by the ACFrC and Landsat 8/9 data,and the accuracy (A) of cross-validation between the blue band of GF-1 PMS and Landsat 8/9 surface reflectance in Bangladesh and Laos is close to 0,indicating a high level of consistency between the two.[Conclusions]The GF-1 PMS and WFV surface reflectance products generated by the ACFrC system provide preliminary validation for the operational application of such data and offer references for the quantitative application research of global medium-to-high resolution satellite data.

[Purposes]In response to the limited depth information of distant objects in stereo-LiDAR fusion features and the issue of error accumulation and amplification in the disparity-to-depth conversion process of existing depth completion models,this paper proposes a depth completion method—RefDC,focused on long-range depth feature refinement.[Methods]Firstly,this paper optimizes the depth voxel representation by constructing a volumetric representation using true metric depth as the voxel dimension,and enhance depth precision through a soft encoding strategy,improving depth detail features.Then,this paper introduces an iterative hypothesis-guided depth refinement module to extract global and local semantic and geometric features.Using a lightweight ConvGRU,the module iteratively refines the depth results and gradually corrects accumulated errors.[Findings]On both real and synthetic datasets,RefDC achieves state-of-the-art accuracy,with overall performance improvements of 4.2%and 11.9%,and a 6.0%enhancement in long-range depth completion.[Conclusions]Compared to traditional methods,RefDC can capture long-range depth features more accurately while significantly reducing computational burden.

[Purposes]High-precision digital elevation models are crucial for ecological assessment and disaster risk management in plateau mountainous regions.This paper proposes an accuracy correction method based on ASTER GDEM to generate higher-precision and more timely DEM products.[Methods]Firstly,this study integrates ICESat-2,GEDI LiDAR data,and environmental covariates to perform error modeling and correction using an XGBoost model.Then, high-quality terrain control points are selected to developed land-use-specific correction models.Finally, the corrected elevation estimates are spatially interpolated to generate the refined DEM product.[Findings]Experimental results demonstrate that the corrected ASTER GDEM reduces ME from 39.94 to -0.021 m,MAE from 40.09 to 10.49 m,and RMSE from 41.44 to 14.59 m.Accuracy in terrain elevation and slope significantly improves across diverse land cover types.[Conclusions]The proposed method effectively enhances the accuracy of ASTER GDEM in the highland mountainous regions of Yunnan province.It provides a scalable solution for improving DEM accuracy in complex topography and ecologically diverse areas,offering reliable support for terrain analysis and disaster risk assessment.

[Purposes]A nutrient enrichment inversion model for aquaculture sea areas based on ZY-1 hyperspectral images is constructed to provide scientific reference for marine fishery management and marine environmental protection.[Methods]Using Zhao'an Bay and Dongshan Bay in Fujian province as research areas,we select 4 spectral bands with the strongest correlation to the logarithm of COD,DIP,and DIN based on measured spectral reflectance,water quality data,and ZY-1 hyperspectral imagery.Three models (CatBoost,random forest,and multiple linear regression) are developed.After precision evaluation,the optimal model is identified to assess eutrophication status and conduct spatiotemporal feature analysis.[Findings]Results show that the CatBoost model demonstrated higher inversion accuracy.The R² values for COD,DIP,and DIN are 0.80,0.71,and 0.88,respectively.[Conclusions]The inversion results reveal significant seasonal variations in Zhaoan Bay's eutrophication index,showing higher levels in some areas during autumn and lower levels in spring and winter.Dongshan Bay exhibit better overall water quality than Zha'an Bay,with most regions maintaining low eutrophication indices across different periods,indicating relatively good water conditions.

[Purposes]To address the low unwrapping accuracy in regions with complex noise and phase discontinuities, as well as the difficulty of traditional phase unwrapping algorithms in balancing precision, robustness and computational efficiency,the study proposes a deep learning method called quality-guided residual U-Net (QG-ResUNet).[Methods]This approach innovatively incorporates pixel-level phase quality maps as prior information through an attention mechanism into a residual-connected U-Net architecture,guiding the network to focus on high-reliability phase regions while intelligently handling low-quality areas.[Findings]Using normalized wrapped phase and quality maps as dual-channel inputs,the network end-to-end predicts detwisted phase.On a simulated dataset featuring diverse complex terrains and noise levels,QG-ResUNet significantly outperforms traditional methods like SNAPHU,baseline U-Net,and PhaseNet in RMSE,MAE,and SSIM metrics,reducing RMSE to 0.51 rad and improving SSIM to 0.95.Ablation studies demonstrate that the quality map-guided mechanism reduces RMSE by approximately 34.6%,validating its core role.[Conclusions]Testing on real Sentinel-1 data further shows that this method effectively handles complex noise and discontinuities,generating high-quality disentanglement results.

[Purposes]In complex application scenarios such as the industrial internet of things (IIoT),factors including wall penetration,multi-source signal interference,and human obstruction can affect electromagnetic wave signals,resulting in indoor positioning errors.[Methods]To address these issues,this paper proposes an indoor positioning algorithm based on CNN-LSTM occlusion detection model,which improves positioning accuracy by rapidly eliminating abnormal signal fluctuations caused by occlusion.Using a signal loss model as a reference,the algorithm inputs RSSI feature vectors into CNN-LSTM model for training to construct the occlusion detection model,and introduces the crested Porcupine optimizer (CPO) to search for optimal hyperparameters and enhance model performance.[Findings]To verify the adaptability of the model,comparative experiments are conducted by incorporating the occlusion detection model into the WKNN and IWKNN positioning algorithms.Experimental results show that the positioning accuracy of the CNN-LSTM-IWKNN algorithm is improved by 19.4%,11.0%,and 5.6% compared with WKNN,IWKNN,and CNN-LSTM-WKNN respectively.[Conclusions]Therefore,the CNN-LSTM occlusion detection model effectively mitigates positioning errors caused by occlusion and significantly improves indoor positioning accuracy.

[Purposes]To improve the detection of low-amplitude gross errors inglobal navigation satellite system real time kinematic (GNSS RTK) signals,which is limited by the traditional 3σ method,and to address the low sensitivity of empirical mode decomposition (EMD) to regional gross errors,an improved 3σ-EMD algorithm is proposed.[Methods]The algorithm combines experimental and simulated data,utilizing statistical and signal processing techniques to detect and reduce noise in GNSS RTK monitoring signals.It follows a multi-stage strategy: rapid identification and correction of significant gross errors using 3σ,adaptive separation of high-frequency noise via EMD,and residual error optimization through a secondary 3σ process.Applied to dynamic deformation monitoring of Tianjin Haihe Bridge,GNSS-RTK was adopted,and signal denoising was conducted by combining the improved 3σ-EMD method.[Findings]Results show that the improved method achieves a gross error detection rate of 66.7%,outperforming standalone 3σ (59.0%) and EMD (25.6%).The signal to noise ratio is increased to 10.44 dB,an 81.9% improvement over the noisy signal.The algorithm reduced the signal amplitude from 69.70 cm to 34.36 cm,optimized the standard deviation to 6.78 cm,and accurately extracted the bridge's fundamental vibration frequency (0.37 Hz).[Conclusions]The study confirms the robustness of the improved 3σ-EMD algorithm in complex environments,enhancing GNSS RTK signal quality for reliable dynamic deformation monitoring of bridges.

[Purposes] To address the inadequacies in expressing urban road semantics and the disconnect between spatial and semantic information in online maps,we propose a map-graph interaction model to enhance semantic exploration and cognitive reasoning capabilities.[Methods]By establishing a bidirectional mapping mechanism between geometric space and semantic space,we construct a map-graph interaction model.Integrating large language models,we build a natural language question-answering interface to achieve closed-loop “spatial-semantic” interaction.[Findings] The prototype system effectively implements bidirectional interaction between spatial and semantic elements,enabling natural language-based road knowledge queries and visualization.[Conclusions] This model effectively bridges the gap between road spatial representation and road semantic knowledge.It not only enhances the efficiency of road information transmission but also lowers the cognitive threshold for understanding complex road semantics through natural language interaction,providing robust support for future urban planning and spatial intelligence decision-making.

[Purposes]With the growing demand for 3D modeling of dense building areas in architectural planning,there is an urgent need for efficient and high-precision geographic information acquisition and modeling technologies.Oblique photogrammetry technology has blind spots in the 3D modeling of dense building areas,and is particularly prone to model voids and model distortions in areas with vegetation occlusion and weak textures.[Methods]To address the limitations of single-technology modeling,this study adopts unmanned aerial vehicle oblique photogrammetry and LiDAR technology,and conducts 3D real-scene reconstruction through coarse-to-fine point cloud fusion method.[Findings]The results demonstrate that the point cloud fusion technology effectively overcomes the shortcomings of single-data-source modeling: specifically,it improves geometric integrity in complex areas such as those with vegetation occlusion and weak textures,and resolves the issues of model voids and distortions.The generated 3D model has clear textures and accurate geometry,with both its planar accuracy and elevation accuracy stably maintained at the centimeter level.[Conclusions]This method can provide a technical pathway for the fine modeling of dense building areas,and the generated model as well as its derived digital line graphs can also demonstrate good application potential in building cluster planning and spatial analysis.

[Purposes]The Chengdu-pengzhou aviation oil pipeline passes through plains,hills,low mountains and other landforms,with a large terrain span.Geological disasters such as landslides and water damage occur frequently.Therefore,early identification of geological hazards along the pipeline is of great significance for ensuring the safe operation of the pipeline.[Methods]Aiming at the complex terrain along the pipeline route,this paper proposes a multi-factor-based method for hazard identification along pipelines.By integrating slope,aspect,and the side-looking imaging characteristics of SAR satellites,visibility models for both ascending and descending orbit images are constructed.This approach helps to delineate optimal zones for geohazard interpretation,thereby enhancing both the accuracy and efficiency of the identification process.[Findings]This paper conducts experiments based on Chengdu-pengzhou aviation oil pipeline in Longquanshan as the research area.The experimental results show that the combined lifting orbits can effectively extract deformation information along the pipeline,and carry out geological hazard identification along the pipeline through comprehensive multi-factor analysis of InSAR visibility of the lifting orbit.Compared with single-track disaster identification,the effective visible area of the combined lifting rail can be increased to 92.367%.More conducive to the effective identification of disaster hazards in the study area.According to the time series characteristics of surface deformation,landform and optical image consensus,26 hidden points are identified.[Conclusions]The research shows that this method can effectively identify the potential geological hazards along the pipeline,and can provide reference for the prevention and control of geological hazards along the pipeline.

[Purposes]This study intends to solve the problem of conducting effective monitoring of landslides.[Methods] Taking the “Tongshan—Pizhou” coal mining belt in Xuzhou,Jiangsu province as an example,based on the LT-1 and digital elevation model from December 2022 to July 2023,the D-InSAR algorithm is used to extract surface deformation information of the mountains within the coal mining belt.High resolution optical remote sensing satellites are combined to determine landslides,and typical landslide areas are analyzed based on soil type data.[Findings]The results show that: ①Compared with field investigation data,the method proposed in this paper can effectively identify landslide phenomena in coal mining mineralization zones,identifying a total of 114 landslide areas with an accuracy rate of 91.2%; ②More than 54.2% of surface deformation points are located at slopes greater than 30% and towards the southwest slope.As the elevation increases,the number of surface deformation points sharply decreases;③A total of 80.7% of landslide events occurred in three types of soil areas: saturated alluvial soil,saturated subsoil,and humus low activity strong acid soil,with 52,21,and 19 incidents respectively,accounting for 45.6%,18.4%,and 16.7% of the total landslide events.[Conclusions]The research results provide reference basis for landslide warning and comprehensive land improvement work in the coal mine belt of northern Jiangsu.

[Purposes]Large-scale underwater topographic surveys using unmanned surface vehicles (USVs)often suffer from sparse and unevenly distributed survey lines.The underwater digital elevation model constructed by conventional spatial interpolation methods often has severe terrain expression distortion.Optimizing the underwater terrain modeling method is very important for engineering construction.[Methods]This paper proposes a modeling approach that integrates the fitting of valley-bottom feature lines with the simulation of terrain skeleton structures.Valley-bottom topographic features are extracted through profile projection analysis,followed by the use of rubber-sheeting-based spatial elastic deformation to fit valley-bottom characteristic lines.Contour-based equidistant interpolation is applied to simulate the terrain skeleton,and bathymetric data are used for 3D elevation adsorption.Finally,a progressive interpolation strategy is adopted for modeling.[Findings]Experimental results demonstrate that the proposed model accurately reflects underwater terrain trends and improve the quality of local feature representation.The local fitting accuracy of the model is better than 0.5 m,and the elevation mean square error is less than 5 m.[Conclusions]This method meets the accuracy requirements for national-level 3D real-scene topographic modeling,has been well applied and verified in water resources survey work.

[Purposes]Addressing the insufficient detection accuracy caused by the “low contrast,strong texture,and cross-scale” characteristics of defects in real-world scenarios,this paper proposes a lightweight road defect detection method combining frequency domain decomposition and contextual multi-scale convolutional attention.[Methods]This method introduces Haar wavelet downsampling into the YOLO11 framework to suppress aliasing and preserve high-frequency details such as cracks in the frequency domain.Simultaneously,it constructs multi-scale convolutional attention to stably respond to slender and small-scale targets.Furthermore,a context-guided module fuses local details and void context to suppress complex background interference and improve target boundary continuity,thereby enhancing detection robustness without significantly increasing computational overhead.[Findings]Training and testing were conducted using the SVRDD street view road damage dataset.The results show that,under lightweight conditions with approximately 3.26×10⁶ parameters and approximately 5.6 ms per inference layer,the model achieves mAP₅₀ and F1 scores of 0.756 and 0.752 7,respectively,which are significant improvements compared to the baseline model.[Conclusions]Furthermore,it achieves a better balance between accuracy and efficiency when compared with typical object detection methods such as Faster R-CNN and YOLOv5,thus verifying the effectiveness of the proposed method.

[Purposes]To reduce the influence of coal shed environment and mechanical equipment on 3D modeling of coal yard and improve the accuracy of point cloud data of coal yard.This paper proposes a combined denoising algorithm for coal yard point clouds based on regional growth algorithm and multi-scale elevation difference.[Methods]The algorithm mainly includes two steps: coarse sampling and fine screening.Firstly,based on the similarity of geometric features,the regional growth algorithm is used to find the abnormal points that are different from the main structure of the coal yard point cloud,and the rough extraction of the noise point set is completed.Secondly,according to the uniqueness of the neighborhood spatial distribution of abnormal points,the multi-scale elevation difference algorithm is used to eliminate non noise points,and complete the fine screening of noise point set.[Findings]The experimental results show that the similarity error of the proposed algorithm based on 1.23% noise removal rate is 0.162 343 and 0.038 870 6 respectively.[Conclusions]It is about 6.54% and 46.8% higher than the traditional area growth algorithm.The proposed algorithm can effectively remove noise points while preserving the complete subject information of the coal yard point cloud,providing data support for the subsequent volume measurement of the coal yard.

[Purposes]To address the challenges of noise-susceptible backgrounds and small,irregular discharge regions in ultraviolet (UV) imagery—factors that undermine the accuracy and robustness of traditional approaches—and to improve both detection/segmentation performance and engineering applicability for corona discharge.[Methods]We augment the U-Net framework with atrous spatial pyramid pooling to enhance multi-scale feature representation and embed a convolutional attention module into skip connections to highlight discharge regions while suppressing background clutter.A composite loss combining binary cross-entropy and Dice loss balances pixel-level accuracy and region overlap.The model is trained and validated on a UAV-acquired UV dataset of 220 kV substation equipment.[Findings]Experimental results show that the proposed method achieves an mIoU of approximately 0.83 and a Dice score of approximately 0.89 on real data, with precision, recall, and inference speed of 0.875, 0.935, and 1.99 frames/s, respectively, demonstrating a favorable balance among segmentation accuracy, target coverage, and real-time performance.[Conclusions]The improved U-Net strikes a favorable balance between accuracy and efficiency,offering a reliable technical solution for automated and intelligent inspection of corona discharge in power equipment.

[Purposes] With the comprehensive promotion and continuous deepening of smart city construction,there is an increasing demand for relevant scientific and technological talents,and the required comprehensive scientific and technological quality is also getting higher.This brings new professional development opportunities for surveying and mapping talents who undertake the construction of urban “digital base”,while also posing new challenges to professional talent cultivation.Faced with new demands and developments,the reform of talent cultivation in surveying and mapping is urgent,but there are certain gaps and deficiencies in the existing surveying and mapping talent cultivation system.[Methods] Starting from the demand for surveying and mapping talents in smart cities,this article analyzes the core problems of lagging curriculum system,insufficient practical teaching links,and limited teaching staff.The system proposes a path and approach for adjustment and reform:by integrating cutting-edge technologies related to smart city construction such as urban management,urban engineering,intelligent perception,big data analysis,and artificial intelligence,the curriculum system can be optimized and reconstructed; Deepen project driven teaching to enhance students' ability to solve complex engineering problems; Build a deep integration mechanism of industry,academia,research and application,and smooth the channels of industry,academia and research.[Findings] Research has shown that through dynamic optimization of courses,deep integration of industry and education,and interdisciplinary ability building,the talent supply-demand gap can be effectively bridged.Taking Shandong University of Science and Technology,Chongqing Jiaotong University,and Shandong University of Architecture as examples for practical application and exploration,certain achievements have been made.[Conclusions]The research results of this article provide reference and experience for the training of surveying and mapping professionals for the construction of smart cities,which will help surveying and mapping professionals play a greater role in the construction of new smart cities and achieve a deep connection between professional education and national strategic needs.

[Purposes]Aiming at the complex working conditions of narrow and airtight space,high occlusion and significant electromagnetic interference in tunnel construction,and to meet the practical needs of high-precision positioning and safety monitoring of construction personnel,engineering vehicles and mechanical equipment,a ultra-wide band (UWB) tunnel construction positioning system integrated with wireless AP function was designed by integrating the technical advantages of internet of things (IoT) and UWB positioning technology.[Methods]By optimizing the layout scheme of UWB base stations in tunnel scenarios,a three-layer architecture system of “positioning perception-network transmission-information application” is built to realize the functions of real-time positioning,trajectory tracing and one-key emergency alarm of monitoring targets in the tunnel.[Findings] An improved ranging algorithm based on time difference of arrival (TDOA) is adopted to effectively reduce the positioning deviation in the tunnel non-line-of-sight (NLOS) environment,and the positioning accuracy of the system was verified through field tests.The experimental results show that the positioning error of the system is stably within 0.4 m in the tunnel construction environment,which can meet the high-precision positioning and monitoring requirements of personnel and mechanical equipment in tunnel construction.At the same time,the integrated design of wireless AP simplifies the wiring process in the tunnel and improves the stability of data transmission.[Conclusions]The application of this system can provide technical support for the information management and safety emergency disposal of tunnel construction,and effectively improve the intelligent level of tunnel construction safety supervision.

[Purposes] In order to monitor and analyze the characteristics and responses of surface deformation during CO₂ injection at a carbon storage area in Tangshan,Hebei,and to ensure the safe operation of the CCS project.[Methods] This study utilizes 113 descending orbit Sentinel-1A time-series SAR images to extract surface deformation information based on the SBAS-InSAR technique,and analyzes the spatiotemporal characteristics of the deformation in conjunction with engineering background data.Quantification of the surface deformation impact induced by CO₂ injection is conducted using LOWESS trend separation and the Mann-Whitney U test.[Findings] The results show that subsidence was the dominant deformation pattern in the study area,with both its rate and cumulative magnitude within safe limits.CO₂ injection induced a diffuse spreading deformation pattern without significant impact on overall stability,although a marked negative step-change occurred around the injection well surrounding the CO₂ leakage.[Conclusions] This methodology provides a long-term and continuous deformation monitoring solution for carbon storage areas,addressing the spatiotemporal limitations of traditional measurement techniques,and offers technical support for engineering safety assessments.

[Purposes]To achieve dynamic monitoring and stability assessment of the coastline in Weifang,Shandong province,we established a monthly-quarterly remote-sensing monitoring system covering the city's 158 km shoreline.The primary dataset comprised seven dates of 2 m resolution domestic satellite images (GF-1/6,ZY-1/E/F,etc.) acquired from December 2023 to June 2024.[Methods]After ortho-rectification in PIE-Ortho,block adjustment,NND fusion and projection to CGCS2000,ortho-images with a planimetric RMSE ≤ 5.09 m were generated.Shoreline-change patches were extracted by integrating visual interpretation and deep learning,and 15 suspicious sites were field-checked.[Findings]Based on shoreline type,land-use status and seaward land-cover classification,the results show that: ①Artificial shoreline accounts for 52.0% of the total length,with reclamation expanding at a rate of 0.85 km per quarter; ②Areas of high human-activity intensity are concentrated in Hanting District and Shouguang Port; and③The coastline is generally stable,with erosion/encroachment detected only on the east bank of the Mihe diversion gate.[Conclusions]This study provides timely and high-precision quantitative remote-sensing information for the refined management of the coastal zone.