This PDF file contains the front matter associated with SPIE Proceedings Volume 13506, including the Title Page, Copyright information, Table of Contents, and Conference Committee information.

The appearance of many auxiliary devices in multi-story buildings enriches the movement mode of indoor pedestrians, making the object of indoor pedestrian pattern recognition not only limited to the common three modes of going up, down and walking, but also taking elevator, escalator and automatic sidewalk are new indoor pedestrian movement modes that need to be taken into consideration. In order to recognize the movement pattern of indoor pedestrians in multi-story buildings as precise comprehensively as possible, this paper proposes a new method of movement pattern recognition based on smart phone. Firstly, the relationship between the indoor pedestrian movement pattern and the changes of air pressure, acceleration data obtained by mobile phone was analyzed, and then nine movement patterns including walking, going up and down stairs, lift, escalator and automatic sidewalk were identified by the CNN-LSTM-BO after Bayesian optimization. The experimental results show that the accuracy can reach 0.96.

Accurate ocean tide load displacement (OTL) modeling plays a vital role in precision geodesy. This paper first compares the OTL displacement of the M2 constituent observed at seven GPS stations near French Polynesia with the modeled values. The analysis reveals that the root mean square error (RMS) of the residuals in the east component is 0.94mm, whereas the uncertainty of the GPS measurements is approximately 0.1mm. This indicates that the residuals are primarily influenced by the OTL modeling. Subsequently, by comparing the RMS between the OTL displacements modeled by different ocean tide models and the GPS observations, we found that the maximum difference in the east component was only 0.04mm. This suggests that the residuals may be influenced by the PREM model. Next, we consider the impact of anelastic dispersion in both the entire mantle and the asthenosphere on OTL modeling. The results show that considering the anelastic dispersion of the entire mantle can reduce the RMS of the east component residual to 0.75mm, indicating that its improvement effect in OTL modeling is more significant. Using the OTL displacement data of the M2 constituent observed by GPS, we verified the Q-value model in the French Polynesia region and found that the Q-value of the asthenosphere is more likely to be 71, while that of the lower mantle is more likely to be 104. After replacing the Q model of PREM and accounting for mantle anelastic dispersion, the RMS of the east component residual between the OTL model and the GPS observations dropped to 0.57mm, with the maximum residual reduced by 0.44mm. However, for the vertical residual, although it may be influenced by both the ocean tide model and anelastic dispersion, it is difficult to clearly distinguish between these factors. Therefore, further discussion and research are needed.

Ocean waves are irregular combinations of waves with multiple wave heights, periods, and directions of travel, and are an important element of the marine environment. Accurate wave forecasting plays an important role in reducing wave accidents and guiding offshore production activities. In response to the spatial dependence of wave forecasting data, this paper establishes an intelligent correction model for wave forecasting based on Transformer improved UNet (T-UNet). Introducing the Transformer structure into the encoder of the traditional UNet model and using a dual sampling module instead of traditional upsampling in the decoder to improve feature extraction capability, the improved model can better understand the dependency relationships between different positions in the sequence and achieve effective correlation between local and global features. This article uses wave height and wind speed forecast data and significant wave height reanalysis data provided by the European Centre for Medium Range Weather Forecasts (ECMWF) to conduct comparative experiments with traditional UNet models. The experimental results show that the root mean square error and average absolute error of the T-UNet model for the correction of future 24-hour forecasts are smaller than those of the UNet model. Compared with the root mean square error before correction, it has decreased by more than 72.4%, effectively reducing the error of numerical forecast data and improving the forecasting ability of wave height.

Electromagnetic interference is one of the most serious threats to satellite navigation receivers, and the most effective countermeasures in the existing research is the space-time array processing (STAP) anti-interference technology. However, the traditional STAP will cause the signal signal phase to deviate, resulting in a decrease in phase tracking accuracy, or even serious consequences such as receiver loop loss. There are three bottleneck problem for the STAP bias correction, including the delay uncertainty in anti-interference filters, the non-ideal nature of the analog devices, and the unclearness of signal direction. Here, we propose a method for modifying the GNSS antenna array receiver’s signal phase deviation. Compared with the traditional algorithms, the proposed algorithm is implemented in a simple way and requires no a priori information. Simulation and 2 types of real data test results show that the proposed algorithm can effectively suppress the signal phase deviation caused by the STAP, and the residual error after the algorithm processing is less than 8°, which improves the receiver's signal phase tracking performance.

In order to weaken the influence of multi-path error on positioning results in GNSS mine tower derrick deformation monitoring, this paper proposes a multi-path error processing algorithm based on CEEMDAN joint wavelet packet analysis. The original signal is processed by CEEMDAN decomposition and wavelet packet filtering to extract the multi-path error in the coordinate sequence, which is applied to the subsequent coordinate correction, significantly weakening the influence of multi-path error on the positioning accuracy. Through the experimental verification of simulation data and measured data, the results show that the algorithm effectively reduces the interference of multi-path error on GNSS positioning results, and the coordinate accuracy is significantly improved. During DOY199-200, the RMS improvement rate is better than 58.1%, 32.9% and 20.5% in ENU direction, respectively; during DOY200-201, the RMS improvement rate is better than 60.7%, 31.4% and 17.5% in ENU direction, respectively; and during DOY199-201, the RMS improvement rate is better than 56.9%, 20.4% and 15.4%. The results demonstrate that the algorithm performs well in complex mining environments, verifying its utility and reliability in GNSS mine tower derrick monitoring.

In deep space exploration missions targeting regions of high scientific significance, it is imperative to address the challenges posed by complex terrain conditions, which necessitate greater precision in hazard detection and avoidance. Recognizing the inherent surveying errors in terrain data acquired through sensor detection, this paper focuses on the safe landing site selection considering terrain hazards positional errors. A modelling approach for quantifying hazards boundary errors is proposed. The dual threshold segmentation and region growth techniques are exploited to expand the hazard boundaries. The hazard boundary errors and risks within their influence areas are estimated considering LiDAR sensor measurement inaccuracies. Leveraging the constructed risk map, the study conducts a search for safe landing sites by combining probability values with maneuver distance cost after normalization using equal weights. This procedure yields multiple candidate safe landing sites that are then ranked accordingly. The efficacy of this method is validated by comparing with a conventional safe area search method without considering terrain errors. The results demonstrate that the proposed approach can identify candidate safe landing sites with higher safety levels.

This study focuses on urban architecture as the research subject and conducts a technical investigation into the construction of three-dimensional (3D) models of cities using high-resolution satellite stereo imagery. By acquiring high-resolution satellite stereo images within the study area, collecting control data or conducting field measurements for image control points, performing regional network adjustment, constructing 3D models, applying universal textures, and attributing properties to building base surfaces, we ultimately establish a city 3D model that meets Level of Detail (LOD) 1.3 precision. Through field measurements of certain proportions of building base surface elevations, building heights, and planar locations—followed by comparative analysis with the constructed city 3D model—we demonstrate that the model achieves LOD 1.3 accuracy requirements. This has significant reference value for both the presentation and application of urban 3D models. The primary objective of this research is to explore new technological methods in the construction of urban 3D models in order to enhance work efficiency and reduce production costs.

This study introduces a cutting-edge zebra crossing warning system designed to enhance pedestrian safety and regulate crosswalk traffic through advanced sensing technologies and intelligent lighting control. The system employs LiDAR (Light Detection and Ranging) sensors to accurately detect pedestrians within the crosswalk area, enabling precise localization of individuals. By integrating with traffic light data, the system effectively determines whether a pedestrian is crossing during a red or green signal. In the event of a red light violation, all zebra stripe lights rapidly flash to alert the pedestrian against jaywalking. Conversely, during legitimate green light crossings, only the three zebra stripe lights nearest to the pedestrian are illuminated, creating a dynamic lighting pattern that follows the pedestrian's progress. Notably, when more than three individuals are crossing simultaneously under the green light, the system initiates a slow flashing sequence in the zebra stripe lights to caution motorists and ensure the safe passage of pedestrians. This intelligent approach to crosswalk illumination not only improves pedestrian safety but also fosters compliance with traffic regulations, contributing to an overall enhancement of roadway safety.

This paper studies error modeling and decoupling techniques in the fusion positioning of Inertial Navigation Systems (INS) and 5G systems. INS offers the advantage of high autonomy, but its errors accumulate over time. 5G, with features like high-frequency transmission and large-scale antenna arrays, provides high-precision correction information for INS. To improve the accuracy of the fusion positioning system, this paper proposes a method combining frequency domain decoupling and Kalman filtering. By separating INS's low-frequency errors and 5G's high-frequency errors in the frequency domain, and applying Kalman filtering for time-domain correction, effective decoupling and error correction in multi-source fusion positioning is achieved. Simulation results show that the proposed method significantly improves positioning accuracy.

Aiming at exploring the influence factors of orbit determination accuracy of satellites with ultra-low orbits, the influences of error sources related to Global Navigation Satellite System (GNSS) satellites, GNSS receivers and algorithms are systematically analyzed. Results indicate that m-level GNSS ephemeris error, GNSS clock bias error and carrier phase observation error significantly decrease the accuracy, while pseudo-range observation error has minor influence and cycle slips can be handled. GNSS observations with short duration (e.g., 1h) do not worsen the accuracy. Reduced-dynamic method leads to better accuracy than kinematic method. The intervals of pseudo-stochastic pulses in reduced-dynamic method should be short enough, while the a priori sigma shows minor influence. Version of planet ephemeris has insignificant influence on orbit determination accuracy. The above findings provide constructive guidance for improving orbit determination accuracy of satellites with ultra-low orbits.

Synthetic aperture radar interferometry (InSAR) is a key tool for surface monitoring, widely used in terrain change and land subsidence monitoring. However, InSAR interferograms are often affected by many kinds of noise, which seriously degrades the data quality. In this paper, we propose a denoising method, named BM filter, which combines bitonic filtering and morphological filtering, to effectively reduce noise and preserve important terrain features. Firstly, the random noise is removed by bitonic filter and the information of terrain edge is preserved. Then, morphological filtering is used to strengthen the image structure through corrosion and dilation operations to improve the sharpness. This method not only improves the denoising effect, but also enhances the interpretation of the image. The experimental results show that, compared with the traditional denoising method, the proposed method can recover and retain terrain features more effectively in complex environment, showing excellent denoising ability and edge preservation performance.

Wind turbine detection is essential for the power industry and regulatory agencies to efficiently and accurately determine their number and spatial distribution. Under complex underlying surface conditions, the detection accuracy of conventional methods is low, especially when dealing with multi-scale wind turbine targets. In addition, the underutilization of the near-infrared (NIR) band limits the capability of optical remote sensing sensors in wind turbine detection. To address the above problems, this paper proposes a wind turbine detection algorithm YOLOv8m-BD based on high spatial resolution four-band reflectance images using YOLOv8m as the baseline model. Firstly, the four-band image is input through channel adaptation modification. Next, the CSPDarknet53 to 2-stage FPN (C2f) module is enhanced by introducing deformable convolution (DCN) to expand the receptive field. Finally, the effectiveness of our design was validated through extensive experiments on a self-built wind turbine dataset. Compared to the YOLOv8m baseline model that only inputs RGB three bands, the improved YOLOv8m-BD achieves an accuracy of 96.4% and a false alarm rate of 2.0%, an increase of 3.4% in accuracy and a decrease of 2.5% in false alarm rate.

High precision observation of the sea surface wind field during typhoon sea states is of great significance for disaster prevention and mitigation. A novel method based on DenseNet, a variant of convolutional neural network, is proposed for retrieving sea surface wind field under typhoon sea state from Sentinel-1 synthetic aperture radar (SAR) images with a moderate resolution of 10km by 10km in this paper. In the case of no external wind directions input, the dual-polarized (VV+VH) backscatter spatial image feature and the radar incidence angle provided by SAR are used to achieve high-precision sea surface wind speed retrieval. In scenarios where wind fringe information in SAR images is limited, the spatial image feature of single-polarized (VV) backscatter is utilized to enable accurate retrieval of sea surface wind direction with high precision. Compared with the ERA-5 reanalysis data, the root mean square error (RMSE) of wind speed retrieved by the proposed method is 1.06m/s, and the RMSE of wind direction retrieved is 12.88° within 99% confidence interval. The accuracy of wind speed retrieval and wind direction retrieval is better than that of the traditional method. The approach presented in this paper offers novel insights and paves the way for exploring new avenues in ocean surface wind field retrieval.

Launched on December 16, 2022, the SWOT satellite employs a wide-swath radar altimeter to monitor small-scale ocean dynamics, capturing submesoscale features at a spatial resolution of 15km. SWOT’s KaRIn instrument is capable of effectively capturing sea surface features associated with internal solitary waves (ISWs). Based on KaRIn twodimensional SSHA images, the impact of ISWs on sea surface height changes was quantitatively analyzed. The SSHA changes caused by ISWs mostly ranged from 5cm to 25cm, with a maximum SSHA change by up to 40cm. Moreover, ISWs modulate the sea surface height at their initial generation, and the leading waves within the same wave packet cause the greatest variation in the SSHA.

The issue of oil pipeline leakage has consistently been a matter of significant concern for the country, as it pertains to the establishment of a dependable oil supply network and the advancement of national energy security, economic stability, and social development. Among the various damage detection methods, ground-penetrating radar offers the advantages of being non-destructive, efficient and accurate. This paper presents a practical application of ground-penetrating radar in oil pipeline leakage detection in South China. It determines the location of buried oil pipelines in different sections through ground-penetrating radar profile analysis and analyses the typical profile characteristics of buried pipelines under different geological conditions. The findings demonstrate the applicability of ground-penetrating radar in oil and gas pipeline leakage detection in oil and gas transmission companies.

The rapid development of cities and the continuous increase of population make the construction land increase continuously, so the type of land use is also constantly changing. Although these changes meet the needs of human production and life to a certain extent, the environmental effects caused by these changes have caused many adverse impacts on the long-term development of human beings. Therefore, the analysis of land use type has a very important guiding role and reference value for the future urban construction and development planning. This study takes Dadukou District of Chongqing as the research area, and with the support of GIS and RS technology, carries out the spatial analysis of regional urbanization construction and the change of humid and thermal environment, and preliminarily discusses the change of urbanization construction, so as to provide reference value for the next planning of urban construction. From the analysis of image inversion results, it can be seen that from 2001 to 2023, the construction land in Dadukou District has been showing a curve rising trend, and the cultivated land area is relatively stable in the range of 40%–50%. The area of unused land and grassland is decreasing year by year. The area of forest land shows a trend of fluctuating decline before 2020, and it will recover in 2023.

At present, single satellite target detection, extraction, and tracking have gradually matured, but multi satellite collaborative fusion is still in the development stage. With the increase in the number of satellites and the completeness of satellite types, higher requirements have been put forward for the fusion of multi-source data from multiple satellites on board. This article proposes a fusion algorithm for multisource trajectory data on spacecraft, which accumulates target point position information, calculates target speed and heading, predicts the target position at the next time based on the timestamp of the target point, sets target association thresholds based on different load positioning accuracies, and finally achieves target point trajectory batching through trajectory association based on Kalman filtering. This algorithm relies solely on point position information fusion and does not depend on specific satellites, making it suitable for constellation level multi-source data fusion. To meet the needs of localization, this algorithm is based on Ascend310 to achieve multisource data fusion on board the satellite. The external information flow interface design meets the requirements of the entire satellite information flow.

Land surface temperature is an important indicator of urban heat island effect, and studying the change of land surface temperature can guide the degree of urbanization construction and cope with urban heat island effect. In this paper, remote sensing images are used to retrieve the land surface temperature in Beibei District of Chongqing, and through analyzing the change of land surface temperature in Beibei District of Chongqing during the 8 years from 2001 to 2023, it can be concluded that the land surface temperature in Beibei District of Chongqing has been used for 20 years. The surface temperature in Beibei district has changed from medium temperature area to medium temperature area and low temperature area. This reflects that Beibei is focusing on green, energy saving and low carbon during the urbanization process. This can provide reference for the scientific development and management of the city, and can also provide reference for the land use planning and ecological environment management of mountainous towns in southwest China.

This study addresses the challenges encountered by traditional multibeam measurement systems when surveying in extremely shallow nearshore waters. It designs and implements three different mounting configurations for multibeam sensors with tilt angles of 30 degrees, 45 degrees, and 60 degrees. Comparative analysis reveals that the 45-degree tilted installation offers optimal performance in terms of both efficiency and accuracy of measurement. This method significantly enhances the capability of measuring specialized underwater topographies, such as wharves, bridge piers, and groins, particularly in areas where the water depth is less than 3 meters. It ensures the validity of the collected data and the safety of vessel operations. The paper elaborates on the design of the multibeam sensor mounting frame with tilt, experimental methodology, and analysis of results, offering an efficient and precise new technological solution for underwater engineering surveys.

The moisture and heat state of the surface, which contains rich geoscience principles, plays a vital role in the urban environment and the growth of vegetation and crops. As one of the important contents of remote sensing monitoring, surface humidity monitoring has gradually advanced from experimental research to practical application and industrialization. In this paper, land surface humidity data of Jiulongpo District, Chongqing City for 8 years from2001to2023 are retrieved by remote sensing. According to the inversion results, it can be seen that urban land surface humidity has been improved to some extent in recent years under the drive of green, low-carbon and energy-saving policies, which has a certain guiding role for urban construction and development.

This paper proposes a spatio-temporal fusion model based on ESRCNN-ScConv, which combines spatial and channel reconstruction techniques with the objective of optimising feature extraction. The enhanced hybrid image decomposition model comprises two key components: multi-exponential end-element extraction and the least squares residual method. The generation of high-resolution data images is achieved through the application of the classical and extended ESRCNN-ScConv spatiotemporal fusion algorithm models. The inversion of the vegetation cover of the area is conducted using the classical model and the improved hybrid image element decomposition model. The experiment demonstrates that the model exhibits superior performance compared to the traditional model in terms of specificity, R², and other indexes. This validates the model's efficacy and practicality, and provides a novel concept and effective technical foundation for advancing the methodological research of vegetation cover.

With the development of marine information in China, the application of spatial basic geographic information in the marine field is increasing. System integration and specialized applications based on marine spatial geographic information are becoming increasingly common. This article establishes a spatiotemporal database for marine geographic information applications. It produces and releases basic geographic base map data services for vector, terrain, and remote sensing images, as well as marine 3D sphere model services. Application of microservice architecture and container technology, platform portal customized integration technology, sea and land integrated data distribution integration technology to build marine map service platform and realize the base map sharing application, special data query and analysis, spatial processing analysis, online mapping and microservice API sharing, and other functions. The platform improves the utilization efficiency of marine spatial geographic information resources.

In the present work, an improved harmonic analysis method was proposed to examine the time series of vertical coordinates from three global navigation satellite system (GNSS) stations in Hong Kong, aiming to obtain the amplitudes of timevarying ocean tide loading displacement (OTLD). Singular spectrum analysis (SSA) was employed to extract the main components of the time-varying amplitudes, and the fast Fourier transform (FFT) was performed to detect periodic signals in the time-varying amplitude sequences. The results indicate that the mean errors in the vertical coordinates of each station range from 9mm to 13mm, demonstrating satisfactory solution accuracy. Filtering reduced white noise by an average of 60.4% and colored noise by 21.68% across all stations. The four major tidal constituents (M2, S2, O1, K1) all exhibited noticeable seasonal variations. Following SSA processing, these constituents showed reductions in differences between their time-varying amplitudes and static amplitudes by 52.22%, 63.89%, 50%, and 58.33%, respectively. This indicates a remarkable enhancement in the accuracy of time-varying OTLD amplitudes. Additionally, significant annual and semi-annual signals were observed in the four major tidal constituents across all stations. Particularly noteworthy was the M2 tide from the HKOH station, which exhibited a noticeable one-third-year periodic signal, with the annual signal being the most pronounced and the one-third-year signal the least. Overall, this study improves the accuracy of OTLD monitoring and introduces novel methods and data to advance the understanding of dynamic ocean tide processes.

Obtaining high-quality remote sensing images is crucial in generating a three-dimensional terrain for flight simulators. However, due to the presence of haze and other impact factors, collected remote sensing images usually suffer from blurred details, which limits its applicability real-world applications. To address this issue a deep learning-based dehazing model is proposed to advance hazy remote sensing images to build flight simulator visual systems. The encoder-decoder architecture is proposed to perform the image dehazing, in which an improved feature extraction module is designed to extract both global and local semantic features across multi-resolutions. In addition, the pixel-pixel and region-region mechanism are innovatively designed to achieve deep mapping relationships and further enhance image dehazing. Experimental results demonstrate that the proposed model outperforms other comparative baselines, achieving high-confidence dehazing in multi-scene remote sensing images, and with good generality and applicability in multi-resolution flight simulators and multi-terrain remote sensing.

Carbon emissions closely relate to the urban thermal environment, helping adjust emission reduction policies and develop a regional low-carbon economy through their impact on it. In this study, the annual mean surface temperature of Shanghai and Urumqi was inverted using the GEE platform and the atmospheric correction method, the fishing net was created and the values were extracted to the points through ArcGis, and the bivariate Moran's index correlation between the carbon emissions and the annual mean temperature of the study area was analysed using GEODA. The results show that: (1) The global Moran index of carbon emissions in Shanghai and Urumqi are both positively correlated, with significant clustering characteristics in space. The global Moran's index of annual mean surface temperature in Shanghai is not significant and Urumqi shows significant clustering but with a lower degree of clustering. (2) From the bivariate local Moran's index, The centre of Shanghai and the southern part of Baoshan District show high - high agglomeration characteristics, while the coastal areas show low - low agglomeration characteristics, which are not significant in the south. Urumqi presents high - high agglomeration in the south-eastern district of Misheng, low - low clustering in the south-western direction with the eastern part of Dabancheng district, The centre is the Basin Desert, so the correlation is not significant. (3) According to the local correlation analysis of the bivariate Moran index, both sites are coupled within the high - high clustering range, and the other ranges are less affected by it, with limitations.

With the development of global positioning technology, the GNSS/INS integrated navigation system is increasingly widely used in the fields of national defense and military, remote sensing surveying and mapping, transportation, and automatic driving. The demand for system performance testing and evaluation is increasing. How to effectively and comprehensively evaluate the performance of the system in practical applications is a major problem. This paper presents a performance evaluation method of GNSS/INS Integrated Navigation System Based on dynamic sports cars and gives a comprehensive evaluation index of integrated navigation system performance. Taking the fog-integrated navigation system, MEMS Integrated Navigation System, and reference station receiver as the test objects, the test scheme was designed, and the data were analyzed. The position, speed, and attitude performance of the integrated navigation system in open, elevated, and tunnel scenes. The stability of the pure inertial navigation system and the PPK dynamic position performance of the reference station receiver in open scenes were evaluated. The feasibility of the test scheme and evaluation method was verified. This method can effectively evaluate the performance of the GNSS/INS integrated navigation system. It can also provide some reference for hardware development, positioning algorithm optimization, product finalization, and standardized detection of positioning and navigation application terminals.

The time Difference of Arrival (TDOA) positioning method is commonly used in multi-station collaborative positioning. However, the traditional TDOA method has low positioning accuracy. Aiming at the problem, a four-station TDOA passive positioning method is proposed, and the Chan algorithm is used for positioning calculation. The factors affecting positioning accuracy are analyzed through the calculation method of geometric dilution of precision (GDOP), and simulation verification is carried out. The simulation shows that selecting a reasonable deployment of reconnaissance stations, increasing the baseline length appropriately, and setting the appropriate main reconnaissance station position can effectively improve positioning accuracy.

In this study, the relationships among swath width, overlap rate, and depth in single-beam and multi-beam sonar depth measurement techniques were derived based on geometric and trigonometric principles. The sea depth, swath width, and overlap rate were meticulously calculated under various parameters, leading to the construction of a mathematical model that evaluates the coverage width of depth sounding lines, with considerations for variations in the direction of sounding lines and seafloor slope. This model extensively analyzes the interplay among sea depth, swath width, and overlap rate, providing a theoretical basis for the optimization of marine depth measurement techniques. The findings not only broaden the applications in the fields of marine geology, ocean engineering, and marine resource exploration but also offer practical potential.

With the intensification of climate change, the frequency and intensity of global extreme weather and climate events are increasing. Extreme precipitation events are typical cases, which will cause huge economic losses and casualties to human society. However, there are few studies on the mechanism of aerosol influence on extreme precipitation. Therefore, in this paper, the correlation between aerosol optical depth (AOD) and extreme precipitation in China was studied. Through the study of AOD in 2000-2020, it was found that the AOD in eastern China showed a linear decrease trend in summer and autumn, and an increase trend in northeast China in spring. In winter, the AOD in North China and Sichuan Basin showed an increase trend. The spatial distribution of AOD was related to the influence of population and industry density and topography. By calculating the correlation between precipitation and AOD, it can be found that in spring, the increase of AOD promoted extreme precipitation in northeast China, while the decrease of AOD promoted extreme precipitation in South China. In summer, except for some coastal areas in northeast China and South China, AOD in eastern China had a negative correlation with summer precipitation, and AOD promoted the occurrence of summer extreme precipitation in China. In autumn, the decrease of autumn AOD inhibited extreme precipitation in North China and promoted extreme precipitation in South China. In winter, the correlation between AOD and precipitation showed a “positive - negative - positive” spatial distribution from northeast to southwest, indicating that aerosols promoted precipitation in northeast and southwest China, while inhibited precipitation in North China and the middle and lower reaches of the Yangtze River.

Multi-beam sonar data contains important information for determining the type of seabed sediment. The current research results on multi-beam data for sediment classification are mainly based on the echo intensity information that varies with the Angle. In this paper, based on the measured data of multi-beam sonar in Meizhou Bay, the feature extraction and substrate classification are studied by using multi-beam single-array wave echo waveform data. In the case of less measurement data of Marine data and greater environmental impact, feature extraction and classification of the same data are carried out from different angles. By comparing the results of each type of substrate and evaluating the final effect of different methods, it is conducive to fully obtaining the factors related to the type of substrate in the data and improving the classification accuracy.

Rendering has become essential in the construction of 3D scenes in the present day, but the traditional rendering technology cannot utilize multi-service multi-node for service, which makes its application very limited. This paper proposes a cloud rendering method based on multi-node resource allocation, which breaks through the limitations of single-node cloud rendering, integrates multiple servers into one, deploys multiple high-performance servers as rendering nodes, and supports concurrent rendering demand for multiple paths. These nodes will be connected through the network to form a unified rendering resource pool, and in the process of resource pool construction, hardware resources are finely optimized to ensure that rendering tasks can be efficiently and rapidly allocated and executed.

The performance of the QFW-6000 ground-based microwave radiometer (MWR) in Zengcheng, Guangzhou for temperature and relative humidity profiles was evaluated on the daily, monthly, and seasonal scales, which is supported by the data from the Zengcheng surface weather station and Qingyuan sounding measurements. Results showed that the MWR performed better in retrieving temperature profiles than relative humidity profiles and the detection performance of MWR varies with altitude. Notably, the retrieval of relative humidity by the MWR was more accurate at lower levels, and the MWR demonstrates optimal detection capability at altitudes below 2000 meters during autumn. These findings contribute to our understanding of the capabilities and limitations of MWRs in measuring temperature and humidity profiles.

In the face of massive spatial grid data from low-resolution encryption to high-resolution process, the traditional point-by-point interpolation algorithm needs to calculate each encryption point in turn, so there are many problems such as large amount of calculation, time-consuming and low search efficiency. Therefore, a KD-Tree (K-Dimensional Tree) algorithm based on multithreading is proposed. By dividing data scientifically and reasonably to construct KD-Tree subtree, the problem of low efficiency of encrypted point search is effectively solved. At the same time, multithreading technology is integrated to take advantage of parallel computing, and the number of threads is dynamically adjusted to realize the uniform distribution of KD-Tree search tasks, which greatly improves the ability to process large-scale data. Finally, the spatial grid is quickly encrypted by using the adjacent points obtained by the search combined with the inverse distance weight. Based on the 3′*3′ resolution 1401*1101 spatial grid of a certain area, the experimental results show that the encryption efficiency of the multi-threaded KD-Tree algorithm is 10~11 times higher than that of the traditional KD-Tree algorithm, and under the optimal number of threads, the encryption efficiency is 3~4 times higher than the ArcGIS inverse distance weighted encryption function.

In the era of big data, obtaining useful knowledge from massive data has become a big problem. With strong ability of data description and relationship recognition, knowledge graph can be applied to knowledge discovery in marine field. This paper reviews the research status of the marine domain knowledge graph, puts forward the technical framework and process of the construction of the domain knowledge graph, and constructs geographic knowledge graph based on the data resource system of sea area and island and the supervision requirements. Firstly, adopt a top-down method to design the schema layer, which is the abstract model of concepts, attributes and relationships. Then, automatically extract entities, relationships and attributes from the business database. Finally, use the graph database to store the above entity nodes and relationship edges, and build a knowledge graph, and support the comprehensive judgment and prediction analysis of marine resources management.

The Earth’s outgoing radiation is one of the key parameters in the global climate change. This study aims to analyze the feasibility of reconstructing global Earth’s outgoing radiation using disk-integrated measurements. We utilized Clouds and the Earth’s Radiant Energy System (CERES) Synoptic Radiative Fluxes and Clouds (SYN) datasets to simulate the disk-integrated measurements from sensors onboard a Moon-based platform, a satellite at the Sun-Earth Lagrange L1 point, the L2 point, and a polar orbit at the altitude of 380000 km. The reconstruction method is proposed based on the disk-integrated measurements. The results indicated that while the reconstructed radiation at the Sun-Earth L1 point is generally higher than the true value, the radiation reconstructed by the Moon-based and high polar orbit platforms is closer to the true value. Furthermore, we found that the sampling time interval significantly affects the reconstruction accuracy, with intervals of less than 4 hours introducing minimal errors. Our findings suggest that disk-integrated measurements are capable of effectively reconstructing Earth’s global radiation, offering significant guidance for the planning of future missions.

Glacier is one of the key objects of investigation and monitoring of natural resources in Qinghai-Tibet Plateau. In this paper, the Qinghai-Tibet Plateau is taken as the research area, and glaciers are taken as research objects. The glacier data of the Qinghai-Tibet Plateau in 2013 and 2017 are taken as samples. A model based on the random forest algorithm (R²=0.72) was constructed by machine learning, and the dataset of 1km-scale glaciers in Qinghai-Tibet Plateau from 2000 to 2020 were obtained. The results show that the area of glacier in the Qinghai-Tibet Plateau has decreased from 53377km² to 49615km² since 2000, with a decrease of 7.05%. Temperature and precipitation are the main factors affecting the trend and rate of glaciers change, and the local topographic differences are the main factors causing the abnormal glaciers change.

In this study, an innovative approach is proposed to use two machine learning algorithms, Gradient Boosted Decision Tree (GBDT) and Adaptive Boosting (AdaBoost), to constrain the GNSS geodetic height time series data by combining key geophysical parameters, such as hydrology, atmospheric pressure, and temperature, for accurate modeling and Prediction. Through in-depth analysis of data from six continuously operating reference stations (CORS) from 2008 to 2018, these machine learning models are found to significantly outperform the traditional least squares (LS) method in fitting accuracy. In addition, we evaluated the prediction performance of these models for the next four years, and the MAE values of the prediction accuracy of the AdaBoost model ranged from 3.485-9.783mm, which was superior to the prediction of the GBDT machine learning model.

In order to study the determination method of reasonable accuracy in monitoring measurement, in this paper, DGM(1,1), DGM(2,1) and polynomial models of surface subsidence and arch crown settlement are calculated. The calculation results show that the grey model has some advantages under the condition of few data and information. In order to obtain high precision, it is necessary to select suitable model aiming at the characteristics of data. When the data of monitoring measurement are close to the approximation of homogeneous exponent sequence, it is necessary to consider using DGM(1,1) model firstly. The DGM(2,1) model should be preferred, when the data are under the condition of accelerating change. The polynomial simulation has some advantages in the case of sufficient data. The research conclusions have reference significance for the informationization design and construction of highway tunnels.

This study analyzes the Dongting Lake's spatiotemporal changes using long-term remote sensing data. The key findings include: (1) From 2005 to 2015, the lake's water area decreased by 61.12% during the wet season and increased by 50.38% in the arid season. (2) The centroid migration was most significant from 2005 to 2010 (8565.82 meters) due to water area fluctuations but decreased to 0.95 meters from 2010 to 2015. (3) Climate factors, particularly precipitation, had a moderate correlation with area changes (0.20 in the wet season, 0.31 in the dry season). The study also explores the impact of human activities on water area changes, providing valuable insights for sustainable development strategies.

This paper proposes an algorithm that integrates image salient features and dense templates for the correspondence extraction of multi-modal geospatial data (e.g., optical images, infrared images, synthetic aperture radar images, and digital maps). In the proposed method, the local structural information of each pixel is first extracted using the local self-similarity (LSS) model, which serves as the basis for subsequent feature and template construction. Then, a rotation invariant feature descriptor is constructed and combined with scale space to realize coarse matching with geometric invariance. Finally, a dense image template is designed to achieve fine matching with the support of feature matching results. Experiments on four types of multi-modal geospatial data demonstrate that the performance of our algorithm outperforms several state-of-the-art matching algorithms in terms of the number of correct matches, precision, recall, and root mean square error.

In order to solve the problem of the applicability of the interpolation accuracy of different kernel functions under different terrain conditions, the radial basis multi-kernel function algorithm (RBF) is taken as the research object of four different geomorphological terrains: mountains, plains, urban areas and mountains Spatial interpolation and evaluation of the interpolation accuracy of different kernel functions in a variety of terrains. The experimental results show that: 1. In the plain terrain, the interpolation of the cubic kernel function is the best; mountains, urban areas and mountains performed better in the quintic kernel function. 2. When the kernel function is fixed, the more average the elevation point data (the value range is small and the dispersion is small), the higher the interpolation accuracy. vice versa; 3. Generate DEM comprehensive analysis, the quintic kernel function has the best interpolation effect in urban areas, mountains, and mountains, while the cubic kernel function is more suitable for plains.

In recent years, extreme precipitation weather occurs frequently, which greatly affects the production and life of the society, and the traditional water vapor detection method is difficult to obtain the atmospheric Precipitable Water Vapor (PWV) in real time with high accuracy, so the inversion of PWV by using the BDS real-time Precision Point Positioning (PPP) technique has become a hot research topic. Therefore, this paper compares the accuracy of ZTD and PWV based on real-time PPP by using the BDS observations from 17 continuous operation reference stations in Hong Kong during September 1 to 11, 2023, uses the ZTD and PWV obtained by post PPP as the reference values, and further analyzes the change trend of PWV during the rainstorm in Hong Kong during this period. Results show that, the ZTD and PWV root mean square error (RMSE) obtained based on real-time PPP are 4.9mm and 1.6mm, respectively, which satisfy the needs of meteorological research, compared with the post PPP; and the PWV value began to rise sharply before the rainstorm, and the actual surface precipitation reached its maximum value after the PWV value reached its peak for 1~2h. During heavy rains, the maximum PWV value generally exceeds 60mm or more, and after the precipitation ends, the PWV value begins to drop rapidly. The real-time PPP inversion PWV based on BDS has the characteristics of all-weather, high precision, and high spatial and temporal resolution, which can be used for rainstorm early warning research and has important practical significance for production and life.

The Qilian Mountain National Nature Reserve is an important ecological barrier in western China, and the aboveground carbon stock of grassland within this area plays an important role in the carbon cycle of terrestrial ecosystems. In this study, the aboveground carbon stock of grassland from 1990 to 2023 in Qilian Mountain National Nature Reserve was estimated using XGBoost model and aboveground biomass conversion method, based on land use data, Landsat images, meteorological and elevation data. The results showed that the aboveground carbon stock of grassland in the area increased from 5.91×10⁵t to 9.11×10⁵t in the past 33 years, with an average annual increase of 9.7×10³t, which showed a significant spatial growth trend. The XGBoost model can accurately estimate the aboveground carbon stock of grassland. Also, we found that the average monthly precipitation in the growing season can significantly improve the estimation accuracy of the model.

This paper takes the dense matched point cloud as the research object, utilizes the progressive encrypted triangular mesh filtering algorithm to filter the dense matched point cloud, combines the texture information of the dense matched point cloud and the method of generating the contour lines by DEM, rejects the false matched points, carries out the accurate classification to obtain the ground points in the reservoir area, and then obtains the key points of the terrain features by data thinning, and finally carries out the reservoir capacity calculation by generating the irregular triangular mesh. Taking the highest peak reservoir as an example, different ground distances and elevation threshold parameters are selected to extract topographic feature points for the ground points, and the reservoir capacity calculation method based on the dense matching of model key points in the point cloud of the tilt model is investigated, and compared and analyzed in terms of the data compression rate and the calculation of the reservoir capacity. It is found that the calculation method not only ensures the accuracy of TIN elevation model, but also improves the working efficiency, and also improves the accuracy of reservoir capacity calculation, which provides a new calculation method for future reservoir capacity calculation.

Obtaining high-quality remote sensing images is crucial for ecological environment monitoring in the Yangtze River basin and for the protection and development of the regions along the river. Currently, due to the challenges faced by satellite remote sensing instruments in simultaneously acquiring high spatial and temporal resolution data, spatio-temporal fusion technology is one of the main methods to address data gaps and the difficulty of obtaining high-quality data. This study uses a portion of the Sanjiang Confluence area in Yibin City as the experimental area, selecting Landsat and MODIS data as data sources. Four spatiotemporal fusion algorithms (STARFM, FSDAF, ISTRUM, FSDAF2.0) are employed to achieve the spatiotemporal fusion of Landsat and MODIS data, analyzing the applicability of these four algorithms in the Yibin area. The results indicate that the ISTRUM algorithm has the smallest overall error compared to the actual images, with a high degree of restoration for spatial and spectral information, yielding the best predictive fusion results and demonstrating good applicability in areas with high heterogeneity.

To address the challenges of distribution discrepancies in remote sensing data across different regions and temporal contexts, this study introduces a novel cross-domain crop recognition method. The proposed approach aims to bridge the distribution gap between source and target domain remote sensing images by optimizing the feature distributions of the source domain, aligning them more closely with the visual characteristics of the target domain. This process enhances the model’s generalization capabilities across domains. The training process involves pre-training the model on source domain data, followed by fine-tuning with augmented source data to accommodate the distribution variations in the target domain. Experimental results show that this method significantly improves recognition accuracy and robustness in cross-domain crop identification tasks, offering an effective solution for advancing the application of remote sensing data in agricultural monitoring.

Machine learning is a leading data-driven approach currently being explored for landslide prediction. To evaluate the characteristics and predictive performance of various machine learning models in karst regions, Jianshi County—a typical karst area in Northwest Hubei—was selected for landslide susceptibility assessment. This study examines four machine learning methods: logistic regression (LR), random forests (RF), support vector machines (SVM), and extreme gradient boosting (XGB). The models were compared using statistical indices and receiver operating characteristic curves (ROC). Results indicated that the RF, XGB, LR, and SVM models demonstrated robust predictive capabilities for landslide susceptibility, with area under the curve (AUC) values of 0.7669, 0.7612, 0.7594, and 0.7444, respectively. Among these, the RF model exhibited the highest predictive ability, followed by the XGB, SVM, and LR models. Furthermore, the practical susceptibility assessment revealed frequency ratio (FR) accuracies of 0.6529 for LR, 0.8803 for RF, 0.6387 for SVM, and 0.6118 for XGB. Notably, the RF model significantly outperformed the other methods in practical applications. This finding suggests that ensemble algorithms offer superior predictive power compared to single classification methods in addressing karst area landslide classification challenges.

In recent years, large-scale algal blooms, commonly referred to as "green tides," have frequently occurred in the Yellow Sea, significantly impacting the nearshore ecosystem, coastal environment, and local economy. This phenomenon has become one of the region's most pressing ecological challenges. In response, this paper proposes an innovative algal removal decision-making method using raw image data, ocean dynamics data and intelligent algorithms is proposed. To estimate the distribution probability of scattered algal patches, we employ the DBSCAN density clustering algorithm, while a Bezier curve optimization technique smooths the results to better reflect actual distribution patterns. Through algorithm optimization and implementation, the coverage area and distribution area of enteromorpha were obtained more accurately. Through algorithmic optimization, we enhance overall efficiency, ensuring timely and effective operational responses.

Air pollution has garnered widespread concern for its detrimental effects on human health. This study aims to investigate the spatiotemporal dynamics of air pollution in Northeast China from 2019 to 2023 from a health perspective. Focusing on 36 cities, this study analyzed the pollution characteristics and health hazards of PM_2.5, PM₁₀, SO₂, CO, NO₂, and O₃. The findings indicate that the annual average concentrations of air pollutants did not exceed the limit values. Except for O₃, the concentrations of pollutants showed a downward trend year by year, while O₃ exhibited an opposite trend. The overall spatial pollution level was Liaoning>Jilin>Heilongjiang. The air quality in various cities was predominantly categorized as 'Good' and ' Moderate'. Based on the USEtox model, it is estimated that PM contributes to an average of 3174.66 premature deaths annually, with a mortality rate of 0.11%. This translates to an estimated 1103.60 premature deaths per million people each year. The calculation of the Hazard Quotient (HQ) for pollutants suggests that they do not pose a health risk to the population.

Aiming at the timely return of Enteromorpha salvage ships from the Enteromorpha salvage area, this paper proposes a path planning algorithm based on the improved A* algorithm. In order to enhance the search efficiency and accuracy of the traditional A* algorithm, firstly, in response to the problems of the traditional 8-neighborhood search method, such as the existence of fewer search directions and sharp corners, we propose a hollow 16-neighborhood search strategy, which searches in different environments using different neighborhood sizes; secondly, regarding the issue that the paths planned by the traditional A* algorithm are tightly attached to the obstacles at the corners, by establishing buffer zones for obstacles, the planned paths are made to conform more closely to the reality of the ship navigation. Finally, to address the path jitter problem that arises when the traditional A* algorithm fails to accurately describe the optimal direction, dynamic direction is added on the foundation of the fixed direction to carry out a secondary search for the jittery path, thus enhancing the path smoothness and avoiding the path from falling into the local optimum. Experimental results show that compared with the traditional A* algorithm , the improved A* algorithm significantly reduces the search time, the number of search nodes, the number of inflection points, and the final planning path is smoother.

The estimation of building height using optical and synthetic aperture radar (SAR) remote sensing imagery is of critical importance for advancing our understanding of urban morphology and facilitating the spatial optimization of urban resources. However, current building height datasets subject to several restrictions: the small sample sizes hinder the effective extraction of remote sensing information required for robust, data-driven deep learning models; the limited spatial coverage of these datasets restricts the geographic diversity and representativeness necessary for capturing spatial heterogeneity; and the lack of open access to these datasets constrains their applicability and validation in broader research contexts. To address these challenges, this study constructs a large-scale building height regression dataset tailored for deep learning applications, encompassing the central urban areas of 62 cities in China and comprising 5606 samples, including Sentinel-1 and Sentinel-2 imagery alongside ground truth height values. Using this dataset, we evaluate the performance of multiple deep learning models for building height estimation, providing insights and references for future research in this domain. The dataset can be downloaded at (https://github.com/CSPON2035/SSBH).

Considering the demand for digital protection of linear cultural heritage, the method based on modern spatial information technology such as oblique photogrammetry, three-dimensional (3D) laser scanning and 3D modeling is proposed in this paper. Firstly, the multi-source data acquisition of air-ground collaboration is carried out by using oblique photogrammetry and three-dimensional laser scanning. Secondly, 3D reconstruction is carried out based on multi-source data. Finally, information is extracted from 3D model and organized. The method was applied to the Xianghe section of the North Canal in China to generate digitized products in various forms of points, lines and planes. And a management platform was established to realize the exhibition, information query, spatial analysis, etc of linear cultural heritage, which provides support for the protection, virtual restoration and flood control command of linear cultural heritage.

Multiple surveying techniques, including GNSS-RTK and Unmanned Aerial Vehicle (UAV), were employed in the exploration of two separate coastal ecological restoration projects in Shi Dao Bay and Chaoyang Port. By comparing and analyzing the performance of various methods in terms of accuracy, terrain characterizing, cost-effectiveness, efficiency, and convenience, the applicability of multiple survey methods in coastal ecological restoration projects was assessed. The research indicated that the elevation accuracy of UAV photogrammetry with photo-control points and UAV LiDAR could be controlled within centimeters. However, without photo-control points, the height error derived by UAV photogrammetry would increase by orders of magnitude at flight heights from 100 meters to 150 meters. Both UAV photogrammetry and UAV LiDAR could achieve plane accuracy within centimeters. Furthermore, UAV LiDAR was capable of mapping the various types of coastal terrains, including units like ecological seawalls and the coastal boundary. In contrast, UAV photogrammetry tent to be only suitable for capturing relatively uniform and flat terrains, such as beach faces. For the measurement of underwater terrains, neither of these two methods can be used, only conventional techniques of GNSS-RTK or single-beam echo sounders are applicable. Moreover, UAV-related survey methods were found to be more cost-effective. In summary, GNSS-RTK surpassed other methods in high-accuracy surveys and offered a wider range of applications. By comparison, UAV-related survey methods, particularly UAV LiDAR, were more efficient and convenient for conducting large-scale survey.

With the advancement of remote sensing technology and the enhancement of image spatial resolution and spectral resolution, the evaluation of hyperspectral image quality has emerged as a highly significant concern. Therefore, this paper selected the radiation state parameter, sharpness, signal-to-noise ratio (SNR), information entropy and other quantitative evaluation indicators to compare and evaluate the quality of two typical domestic spaceborne hyperspectral data, GF-5 and ZY-1 02D. Meanwhile the spectral characteristics of the bare soil and vegetation in the same area of Qinshui Basin were compared in the two data. The results show that the mean and sharpness of ZY-1 02D in the near infrared band are slightly higher, and the SNR and information entropy of the two data are basically the same. In general, the two data both have rich image information, and can also reflect the spectral characteristics of ground objects. This study quantitatively evaluates the data quality of hyperspectral data in China, and supports the subsequent hyperspectral remote sensing applications.

A major concern in modern society is the equilibrium of society development, resource management, ecological health, and economic prosperity. This research focuses on Zaozhuang City, leverages remote sensing data from the Landsat series of satellites, and relies on the computational power of the Google Earth Engine platform to measure the remote sensing ecological index (RSEI), quantitative assessment of the ecological environment quality of Zaozhuang City from 2000 to 2024. The results show that: (1) The ecological environment quality of the research region is generally average, with the mean value of the RSEI dropping from 0.4969 in 2000 to 0.4444 in 2024. (2) Between 2000 and 2024, the percentage of regions identified with excellent ecological conditions fell by 23.33%, while the combined proportion of areas with poor and relatively poor ecological environment quality increased by 21.06%. This research offers datasets and technical assistance for assessing regional ecological conditions.

To improve the scientific approach of precipitation enhancement operations during low vortex weather processes, this study utilizes the actual data from eight spring low vortex precipitation processes in Liaoning Province, China, FY2E satellite data, and ERA5 data to analyze the relationship between cloud characteristic parameters and rainfall intensity of precipitation cloud systems. The results show that the optical thickness, cloud top height, subcooling layer thickness, effective particle radius, and liquid water path increase monotonically with the increase of precipitation intensity, while the cloud top temperature decreases monotonically, indicating that the deeper the cloud layer, the greater the precipitation intensity, the higher the cloud water content, and the larger the particle size. Using the TS and HSS scoring methods, only the scores of optical thickness and cloud top height showed a trend of increasing first and then decreasing. The values corresponding to the highest scores were selected as the recognition thresholds for precipitation clouds, which were 25 and 4.5 km, respectively. The direction and location of cloud parameter high-value areas are relatively close to the direction and location of heavy rain bands, and weak precipitation areas with abundant supercooled water and lack of ice crystals have the potential for cold cloud catalyzed precipitation enhancement. When Liaoning Province, China is affected by a low vortex weather system in spring, comprehensive analysis of cloud identification indicators, cloud moisture content, ice crystal number and other factors can identify target clouds for rain enhancement operations and delineate potential areas for precipitation enhancement, which can effectively guide artificial precipitation enhancement operations.

Aiming at the problems that traditional measurement means are difficult to implement in gully terrain, and it is difficult to obtain its full-view information and do long-term continuous deformation monitoring, a new measurement method using airborne LiDAR is proposed. Firstly, the point cloud data of the gully terrain is rapidly acquired by the UAV-mounted LiDAR. Then the characteristics of three commonly used point cloud filtering algorithms are combined, and the optimal filtering method suitable for the characteristics of the gully terrain is compared and analysed. Finally, the two sets of data before and after the cave-in in the experimental area are interpolated and differentiated, and the digital elevation model of the cave-in area is derived. The results show that: the measurement method using airborne LiDAR proposed in this paper can obtain gully data quickly and efficiently, and can monitor deformation in real time and dynamically, with high feasibility and applicability, and has a good application prospect compared with the traditional measurement means.

The icing process is primarily influenced by climatic and micro-topographical factors. However, there is currently limited research on the relationship between icing meteorology and geographic information conditions. This paper takes the second cold wave event of the winter of 2024 in Bijie City as a case study. The city is divided into 12 blocks based on topographic features and administrative regions, and a large number of monitoring terminals were installed to conduct meteorological monitoring across various altitudes and terrain areas within Bijie. On this basis, an in-depth analysis of the variations in temperature, humidity, wind direction, and wind speed associated with icing meteorology across different altitudes and terrain areas was conducted, revealing the variation patterns of icing meteorology under different geographic information conditions. The results indicate that temperature is negatively correlated with altitude, and both the total duration and maximum continuous duration of temperatures below 0°C are negatively correlated with altitude. Humidity is not a major factor contributing to the differences in conductor icing in Bijie City. Wind direction is significantly influenced by topographical factors—the more open the terrain and the fewer the surrounding obstructions, the greater the range of variation in wind direction, leading to more frequent changes in wind direction. Conversely, the more enclosed the terrain and the more obstructions present, the more concentrated the wind direction becomes.

Using the amplitude information of the air gun excitation signal recorded by the active source observation station in Qilian Mountains, Gansu Province, the attenuation parameter t* with time was calculated using spectral ratio method through filtering, stacking, and cross-correlation processing. The attenuation changes of different seismic phases at each station of the observation system were obtained. The results showed a clear correlation between attenuation changes and travel time changes. The calculation of attenuation changes based on the active source of an air gun is an effective method for detecting changes in underground media, providing another way for the continuous detection of changes in underground media by the active source of an air gun.

Small-diameter drilling machines are important equipment for geological explorations. With the development of artificial intelligence and Internet of Things (IoT), future small-diameter drilling machines will have a higher level of intelligence to improve work efficiency and safety. Lithology (mineralization) identification technology is one of the key challenges in implementing the intelligent system for small diameter drilling machines. Through the investigation on the international patent layout and R&D trends, domestic R&D progress and application status, this paper analyzes the domestic R&D foundation, examines the status quo of the applicable core (mineralization) identification technologies and the development trend of intelligent core identification technology. This paper also compares and selects applicable lithology (mineralization) identification technologies, comprehensively analyzes the difficulties and bottlenecks in realizing the intelligent lithology (mineralization) identification technology for small diameter drilling machines, and puts forward three working suggestions with the hope of providing theoretical support and technical references.

Polarization Time-Series InSAR (Pol-TS-InSAR) method is an effective strategy for enhancing both the quantity and reliability of persistent scatterer (PS) pixels. However, existing researches have not fully exploited the phase information from different polarization channels nor considered the scattering mechanisms of pixels during the later stages of Pol-TSInSAR process. This study, investigates three kernels for deriving model parameter using multi-polarization data during the model parameter estimation stage during time-series InSAR analysis, namely the Temporal Coherence Optimization approach (TCO), peak temporal coherence approach (PEAK), and weight average approach (WS-MPMP). Experiment were conducted experiments using 45 sets of Sentinel-1A VV and VH polarization data to analyze the deformation of Jieyang Chaoshan International Airport. The performance of the three kernels was evaluated based on reliability, number of measurement points obtained, and computational efficiency. The experimental results indicate that the TCO method achieves marginally better performance in increasing the number of pixels and enhancing accuracy, but exhibits significantly lower computational efficiency compared to the PEAK and WS-MPMP methods.

The occurrence of El Niño-Southern Oscillation (ENSO) events affects global climate change and leads to the occurrence of extreme weather events. Precipitable water vapor (PWV) is an important parameter to characterize climate change, and its response to the occurrence of ENSO events exists. In this paper, based on the data fusion between CMONOC and the European Centre for Medium-Range Weather Forecasts (ECMWF), we form a PWV time series from 2000 to 2021 in mainland China, and analyze the nonlinear trend of the PWV time series and its response to the ENSO events by using singular spectrum analysis. It is found that more than 82% of the stations in mainland China show an overall upward trend from 2000 to 2021, and the occurrence of inflection points during the PWV trend changes corresponds well with the occurrence of ENSO events of moderate intensity or above; the PWV nonlinear trends of all stations show an inflection point from 2007 to 2012, and ENSO events with different phases occur consecutively in this time period; More than 80% of the sites showed an inflection point between 2015 and 2016, with the occurrence of super-intense ENSO events during that time period.

Liverpool has the highest proportion of highly deprived neighbourhoods in England. To explore the spatial pattern of employment deprivation, data from the 2011 Census, and IMD (Index of Multiple Deprivation) from the government are used. Spatial autocorrelation analysis (Moran's I and LISA methods) is applied to determine if there is a systematic pattern in the spatial distribution of the variable and to identify clusters. Time series analysis is conducted to illustrate unemployment levels in Liverpool. The results indicate a positive spatial autocorrelation for employment deprivation in Liverpool, suggesting that neighbouring areas are similar, potentially due to a lack of schools, job opportunities, and labour force demand in certain locations. This pattern of employment deprivation is concentrated, with clusters found in the northern part, southern part, and inner urban cores. Despite improvements in the employment situation, it still falls behind the overall level of the United Kingdom.

The visualization of ocean scalar field is of great importance to scientists engaged in the analysis of marine extreme weather, the delineation of ship routes and the prediction of fish populations. This paper presents a text format transformation method for gridded marine environment data, and completed visualization with the assistance of ArcGIS API for JavaScript. This visualization method realizes the lightweight architecture of the ocean scalar field display and addresses the issues previously encountered, such as the occupation of numerous server resources and the slow response speed of the system. The visualization architecture proposed in this paper enables the effective display of diverse marine environment scalar elements, including sea temperature, salinity, sea surface height, and others, within the ODINWESTPAC system. This system provides information services for marine managers and researchers.

Recently, 3D Gaussian Splatting (3DGS) has made remarkable progress in real-time rendering. It has significantly advanced realistic 3D scene rendering in various academic and industrial applications. However, 3DGS requires a substantial number of consecutive views for scene modeling and faces challenges in synthesis quality due to inputs such as sparse view input. Overfitting of the training views by 3DGS can result in artifacts and floaters in the rendered images, particularly when the number of training views is limited. To address this challenge, we introduce a novel method for training 3D models from limited data. Our approach involves continuously tracking Gaussian variations to guide the synthesis of novel view synthesis, thereby reducing local geometric overfitting and improving rendering quality without sacrificing speed. Additionally, we propose an annealing strategy to prevent the emergence of ambiguous scene representations while considering both global structure and fine texture details. Experimental results demonstrate that our method can achieve superior results in novel view synthesis with limited data. In the single-view training results compared to the baseline on the Nerf-LLFF dataset, our method exhibited an improvement of 5.40% in PSNR, 9.36% in SSIM, 8.54%p in LPIPS without additional inference costs.

Satellites such as Sentinel-1A/B, the newly planned NASA-ISRO SAR (NISAR), and the German Tandem-L with reentry intervals as short as 6-12 days have caused numerous SAR data. These data has provided a rich database for InSAR time-series analyses but posed a challenge to efficiently and consistently invert surface deformation. The traditional Small Baseline Subset Interferometric Synthetic Aperture Radar technique (SBAS InSAR) can be applied for surface deformation inversion but with limitation of processing efficiency. SBAS InSAR is challenging to meet the demand for dynamic, efficient and continuous surface deformation monitoring. In this study, we propose a high-efficient deformation monitoring method based on the Kalman filter (KF) and InSAR data. First, we use SBAS technology to obtain the surface deformation information and construct the state equations according to the time series characteristics of deformation. Second, the new interferograms are generated with the new and original images, and the observation equations are constructed with the unwrapped phase of the interferometric phase as the observation value. Finally, the surface deformation at the new moment is quickly estimated based on our KF dynamic processing. Compared with the traditional SBAS method, our method effectively reduces redundant calculations and improves computational efficiency. Actual experimental results from a mine in Xinjiang show that our method not only has high storage performance, but also demonstrates obvious advantages in dynamic estimation computational efficiency. Our method improves the timeliness and accuracy of deformation monitoring and adapts to the real-time monitoring needs brought by the growth of SAR data.

Haitang Town in Qufu is located in the southern part of the central Shandong Tai-Meng Mountain area. This region has distinct fault structures and is geologically favorable for geothermal resources. This paper focuses on the investigation of Haitang Town and its surrounding areas using the CSAMT and microtremor exploration method. Through the analysis and interpretation of the data obtained from these methods, the main fault structures and the abnormal burial depths of the hydrothermal reservoir aquifer layers in the study area have been identified. The results show that the geothermal resources in Haitang Town are primarily located in the weathered fracture zone and the fracture karst development zone of the upper Ordovician Majiagou Formation's carbonate rocks, with a burial depth of approximately 1500–1900 meters. The paper also recommends the location and depths for geothermal wells, providing a scientific basis for subsequent exploration and development of geothermal resources in this area.

The complex and unique hydrogeological environment in Jinan serves as a natural barrier to the construction of urban rail transit. To achieve the coordinated development of spring water protection and subway construction, this study focused on the M3 subway line and applied 3D geological models to the karst area subway planning. Based on comprehensive collection and fusion of geological data from multi-source in Jinan, a 3D geological model of Jinan spring area was constructed using the MapGIS K10-TED 3D platform, enabling a 3D visualization of the complex terrain and geological structures in the spring area. By analysing the 3D model, an evaluation of the conditions for subway construction in the spring protection core area between the Qianfoshan fault and the Wenhuaqiao fault was conducted. The results indicated that the thickness of the volcanic rock layer in the northern part of Jinan urban area is significant, creating favourable conditions for spring water gushing. Some areas between the Qianfoshan fault and the Wenhuaqiao fault are identified as the spring water protection core area, with the initial water level ranging from 5 - 15 m, which can be considered as the roof of the lower confined water, and the stable water level of the confined water is at 32.0 m, with an aquifer permeability coefficient of 0.62-1.61 m/d. Under the assumption of neglecting the influence of shallow karst water runoff, the risk of the M3 subway line crossing the spring water area is considered to be basically controllable.

Aiming at the prominent problem of low efficiency in the process of elevation datum transformation of DEM or DSM images, a fast calculation method based on MPI ( Message Passing Interface ) parallel technology is proposed. The purpose is to use multi-core technology to effectively accelerate the working speed and comprehensively improve the efficiency and quality of elevation datum transformation. The experiment uses two sets of DEM images with different image resolutions (DEM images are used in this paper due to the same processing method of DSM images and DEM images), and conducts batch image elevation datum conversion experiments on the Ubutun 20.4.3LTS operating system. The experiment uses the MPI parallel algorithm to make full use of the performance advantages of multi-core computers to achieve the purpose of processing multiple images at the same time. The results show that the use of MPI parallel technology can greatly improve the efficiency of image elevation datum conversion. The efficiency of using this method to calculate two sets of images is increased by 10.9 and 10.859 times, respectively. MPI parallel technology can give full play to the advantages of multi-core computers, and the parallel effect is best when the number of processes is about the total number of tasks.

This paper provides a rapid registration method for realistic 3D models in water conservancy and hydropower projects to provide a realistic 3D model dataset of underground chamber groups for the construction of engineering digital twins. Firstly, the SLAM algorithm is used to realize the construction of the 3D model of the cavern based on images. And then, we adopt BIM overturning technology to obtain the real spatial position of the three-dimensional model of the construction cavern via CAD design drawings. And the 3D model of the real scene is quickly registered to the correct spatial position. Through the calculation of the occurrence of the catalogued elements on the realistic 3D model and the actual engineering cavern, less than 5° of difference between the occurrence of the geological structure elements obtained on the rapid registration model and the real value. The proposed method can provide well aligned 3D models which meet the requirements of the geological cataloging operation in the engineering construction process.

At present, there are fewer studies on the fusion of ground 3D laser scanning technology and optical image for 3D modeling. To address this issue, this paper takes the rapid construction of the 3D model of the reservoir hub as the research object, combines the high-precision features of the ground 3D laser point cloud and the advantages of the rich texture information of the optical image, fuses the two kinds of heterogeneous data, and builds a 3D model of the reservoir hub quickly and efficiently, and develops a number of 3D model application functions by using the software development tools according to the actual application requirements. A number of 3D modeling application functions are developed using software development tools according to practical application requirements. Meanwhile, an improved target point cloud splicing method is proposed and successfully utilized in the case study. This study provides an important reference for the application of 3D laser scanning technology in the rapid construction of 3D models.

In the field of land cover classification, the multimodal fusion of hyperspectral imagery (HSI) and light detection and ranging (LiDAR) data has become a key research direction in remote sensing. Hyperspectral imagery provides rich spectral information, enabling the identification of subtle material differences, while LiDAR data offers high-precision spatial geometric structures. However, the complementarity between these two data modalities faces challenges related to spatial resolution discrepancies and feature inconsistency. To address these challenges, this paper proposes a Multiscale Spatial Adaptive Calibration Network (MSACNet) for joint classification of hyperspectral and LiDAR data. The network extracts multiscale convolutional features and introduces a Spatial Adaptive Calibration Module (SACM) at each scale to adaptively calibrate the fused multimodal features, adjusting for discrepancies between spectral and spatial geometric feature distributions. Experimental results on the Houston2013 dataset demonstrate that the proposed method achieves superior classification performance compared to existing classification models.

This paper presents a method for removing airwaves in Marine Controlled-Source Electromagnetic (MCSEM) surveys. Airwave interference poses a major challenge in shallow water electromagnetic exploration, as it diminishes the accuracy of subsurface resistivity detection. Based on the electric field theory of a vertical electric dipole in a vertically transverse isotropic (VTI) half-space, frequency-wavenumber expressions for the electric field are provided, illustrating the propagation characteristics and mechanisms of airwave interference. Numerical simulations are conducted to compare detection capabilities before and after airwave removal. The results show that removing airwaves significantly improves detection depth for low-resistivity and deeply buried targets. These findings demonstrate the positive impact of airwave removal on enhancing the effectiveness of MCSEM surveys, providing a foundation for more accurate marine resource exploration.

Due to the complex and variable nature of the seabed environment, differences in the properties of various units lead to issues with the accuracy of acoustic model predictions and poor adaptability to the environment. These disadvantages significantly hinder their practical application. To enhance the identification performance of acoustic models, we employ a knowledge-guided learning method and in-situ acoustic reflection measurement approach, integrating the spatial distribution characteristics of actual seabed acoustic parameters with underwater sound propagation strategies. It conducts a study on the acoustic characterization method for layered seabed using a knowledge and data fusion model. Additionally, through numerical simulations and laboratory experiments, the research investigates the influence of seabed parameter uncertainties in shallow water environments on model performance based on Fisher information theory and refines and improves the model.

Terrestrial laser scanning (TLS) technology has become an important tool in forest survey, and multi-station scanning registration is a key prerequisite for subsequent data processing. At present, the TLS registration of multi-station scanning in forests is mainly based on the registration of target spheres, which is very time-consuming and cumbersome. In this paper, an automatic registration method based on improved two-dimensional feature descriptors is proposed. The method mainly includes five steps: height normalization, point cloud slicing, trunk position extraction, feature descriptor construction, and trunk matching. The algorithm can efficiently and accurately extract tree trunk position information from the TLS data of multiple stations, and construct and match two-dimensional feature descriptors for point cloud registration based on the relative spatial position relationship of tree trunks. The algorithm was tested under the point cloud data of six sample plots and the experimental results show that the algorithm has high accuracy, with the average accuracy, recall rate and F1 score of trunk extraction reaching 94.6%, 93.9% and 94.3%, respectively, and the average RMSE of point cloud registration is 0.035.

Skylights, as a unique and scarce type of pits on Mars, are significant for scientific research and engineering applications. Compared with visual interpretation to detect skylights, deep learning methods are more automatic and objective. However, existing deep-learning methods face low accuracy due to very few skylight samples and the model's inability to capture the subtle features that differentiate skylights from typical pits. To solve the problem, skylight samples generated with StyleGAN2-ADA network are used to expand the original dataset and not-skylight pits are used as negative samples for adversarial training. These improvements reduce the number of misidentified skylights to approximately 1.3% of the original. Furthermore, to enhance the feature extraction capability, we propose the YOLOv9-PSA model, leading to a 14.1% increase in precision. The results indicate that expanded dataset and the YOLOv9-PSA model significantly improve detection accuracy, with final metrics achieving a precision of 94.1%, a recall rate of 88.9%, and an F1 score of 91.4%. The method proposed in this paper provides a foundation for skylight detection of other celestial bodies.

The Surface Wave Investigation and Monitoring (SWIM) aboard the China-French Oceanic Satellite (CFOSAT) can measure ocean wave spectrum (70-500 m wavelength range). SWIM can effectively mitigate the limitations of synthetic aperture radar (SAR); however, it still encounters certain challenges, particularly in terms of mitigating the impact of speckle noise. To enhance speckle noise removal and improve the accuracy of SWIM wave spectrum inversion, this paper extracts the observed speckle noise spectrum from CFOSAT SWIM L1B data and NDBC buoy wave height spectrum data. A nonlinear least squares fitting method is employed to effectively capture the complex relationships in the data, leading to the construction of a new speckle noise model. Based on whether the speckle noise is influenced by the Doppler bandwidth, the model is divided into two distinct fan-shaped parts: the background speckle noise part and the along-flight speckle noise part. The findings demonstrate that the newly proposed speckle noise model exhibits superior performance in simulating observed speckle noise compared to both the Jackson model and the Hauser model, thereby enhancing the accuracy of removing speckle noise.

Launched by NASA in 2018, ICESat-2 is a next-generation laser altimetry satellite equipped with the Advanced Topographic Laser Altimeter System (ATLAS) capable of penetrating water bodies. Nonetheless, the data collected is heavily affected by noise due to interference from sunlight and water quality, which compromises its application efficiency. Consequently, the effective removal of noise from photon point clouds is a critical challenge and a current focus of research. This paper reviews the denoising algorithms applied to ICESat-2 photon point cloud data, evaluates the effectiveness and limitations of various methods, and discusses potential research directions for future denoising technologies.

Remote sensing image radiometric resolution transformation plays a fundamental role in data storage, transmission efficiency, processing speed, image visualization, data simplification, and analysis. As a crucial preprocessing step, radiometric resolution transformation is commonly based on gray-level stretching methods. However, these methods are often time-consuming due to the complexities involved in method selection and parameter adjustment. This paper proposes a method based on Generative Adversarial Networks (GANs), inspired by histogram matching, to achieve radiometric resolution transformation of remote sensing images. Given the significant dynamic range changes in radiometric resolution transformation, establishing a direct mapping relationship is challenging. Therefore, a Discrete Histogram Approximation Loss is designed to compute the similarity between High Radiometric Resolution Images(HRRI) and Standard-Resolution Radiometric Images (SRRI), strengthening the mapping correlation and optimizing the visual features of the generated images. Remote sensing images contain a variety of ground objects and complex detail textures; hence, Structure Correction Loss is used to constrain the structural features, ensuring the clarity and accuracy of the resultant images. This study was validated using a self-made dataset, and comparative experiments with other methods were conducted. The results demonstrate that the proposed method achieves a peak signal-to-noise ratio of 29.20dB and a structural similarity index of 86.8%, significantly outperforming other methods. These findings confirm the feasibility of the proposed approach, with the generated images exhibiting high color and structural accuracy and realism.

Self-calibration of three-dimensional multi-beam laser scanning (MBLS) is essential for precise data collection. However, the use of extensive point clouds for self-calibration results in significant computational demands, which may adversely affect processing efficiency. To address this issue, in this paper, a point cloud data simplification strategy is introduced that involves hierarchical selection from point clouds of multiple self-calibration planes. A self-calibration method by exploiting a two-steps hierarchical data selection is presented, with the aim to select representative points. The verification experiment demonstrates a significant improvement in computational efficiency, achieving a 97.0% reduction in processing time.

Road extraction from remote sensing images has important application value in urban planning, traffic management, and disaster response fields. However, the road features in remote sensing images are complex, with variable scales and uneven distributions, making the road extraction task extremely challenging. We propose the Spatial Attention Stacking Network (SASNet). Through alternating training and continuous optimization between the base model and the confidence model, the confidence evaluation model gradually learns to recognise the prediction errors of the base model in specific regions. Thus, prior information based on spatial confidence analysis is introduced into the stacking algorithm, strengthening the extraction of key information such as road edges and corners. The experimental results on the Massachusetts Road Dataset show that compared with the currently popular road extraction methods, this method has significantly improved performance indicators.

Building extraction methods based on deep learning have technical characteristics such as high extraction accuracy and fast processing speed. However, when using high-resolution remote sensing images, traditional methods often struggle to extract occluded buildings. To address this problem, this paper introduces the Multi-Feature Iterative Model (MFIM), a multi-scale feature iterative model that extracts features at different scales and fuses them in a compressed manner to obtain a structure that preserves global features without losing local details. In order to evaluate the performance of the method, we conducted a comparative analysis on the WHU dataset, using U-Net, DeepLab V3+, and HR-Net as benchmark models. The experimental results indicate that the MFIM method achieves IoU, F1, Recall, and Precision scores of 89.85%, 94.74%, 94.86%, and 94.81%, respectively. This method significantly improves the accuracy of building extraction, and its effectiveness in complex scenes makes it valuable for building extraction tasks.

Remote sensing image data is a method of collecting ground target data through remote sensors and forming images using the interaction between electromagnetic radiation and the ground surface. However, due to various practical factors, the resolution of remote sensing image data is often low, and software processing methods are more cost-effective. With the rapid development of deep learning theory and applications, image super-resolution technology has become a key technology in the field of image processing. This article constructs a dataset using remote sensing images of natural scenes, and improves the original SRGAN algorithm to make it applicable to remote sensing images and enhance visual effects and evaluation metrics.

To enhance the accuracy and robustness of traffic sign extraction in complex road scenarios, this paper proposes a method for traffic sign extraction in highway environments that integrates road linear topology relationship constraints. Firstly, the Segment Anything Model (SAM) is employed to process road surface images, extracting road edges and constructing the road's topological structure. Then, by using the edges as spatial constraints within the topological relationship, the detection range for traffic signs is narrowed. Additionally, the Intersection over Union (IoU) confidence is dynamically adjusted based on the topological relationship between signs of different scales, preventing missed detections of small-scale signs. Finally, the latest YOLOv10 object detection network is utilized for precise recognition and localization of traffic signs. Experimental results from tests on actual highway sections demonstrate that the proposed method significantly improves the accuracy and stability of traffic sign detection in highway scenarios while notably reducing the false detection rate, showcasing excellent performance.

Wetland protection and restoration are important components of the healthy development of the ecological environment. Based on the Google Earth Engine (GEE) cloud platform and the Landsat long-term remote sensing dataset, this study extracted remote sensing information of wetlands in Zaozhuang city and analyzed its spatiotemporal evolution characteristics using Random Forest Algorithm (RFA). The results show that: (1) The Overall Accuracy (OA) and Kappa coefficient were no less than 94.44% and 0.93, respectively. The precision evaluation indicated that the method has good accuracy and stability. (2) The wetland area in Zaozhuang city showed a significant upward trend, increasing from 52.52km² in 2000 to 69.98km² in 2024, with a cumulative increase of 17.46km². The area increased by 3.99km², 3.31km², and 10.17km² in different stages. (3) Spatially, the wetlands migrated from the southwestern region to the northern and central regions, with a decrease in spatial fragmentation and an increase in aggregation. The study could supply technical assistance for wetland resource inventory and spatial resource management.

A new Non-Subsampled Shearlet Transform (NSST) domain-based optical and SAR image fusion framework is proposed to address the lack of target fusion rules in the image fusion process, which leads to inconsistent regional edge information. The method combines the WLE and WSEML algorithms to fuse the low-frequency sub-band information. It introduces the multi-scale morphological gradient (MSMG) into the adaptive dual-channel Pulse Coupled Neural Network (PCNN) in the high-frequency sub-band for edge detection and enhanced edge features. The results show that the method of this paper has superior performance in both qualitative and quantitative evaluation compared to the other five fusion algorithms. It can better preserve the spectral information, resulting in a clear and appropriate visual perception of the fused image.

The importance of optical remote sensing images in Earth observation is increasing, but the presence of clouds significantly affects the clarity and information integrity of the images and becomes a major challenge in remote sensing image processing. Although the traditional cloud detection methods based on physical features are simple and efficient, they often have the problems of omission and misjudgment when facing complex surface features. For this reason, this paper proposes a high-precision cloud detection model CADC-Net based on improved DeeplabV3+, which improves the accuracy and stability of the model by introducing the coordinate attention mechanism and the DenseASPP multi-scale feature extraction module. The coordinate attention mechanism helps the model to recognize the spatial location of cloud targets more accurately and enhances the ability to process the details of cloud boundaries. The DenseASPP module captures the cloud structure information at different scales through multilevel feature sharing, which effectively improves the accuracy of recognizing cloudy regions. Compared with traditional methods, CADC-Net is not only more robust in feature extraction, but also significantly reduces the phenomenon of mis-segmentation. Experiments on the L8-Biom dataset show that the overall accuracy of the CADC-Net model is improved to 97.47%, which is 3.95% higher than that of the traditional DeeplabV3+. In addition, the model shows better performance in several metrics such as intersection-to-union ratio, precision and recall, especially in cloud boundary and complex feature scenarios. This indicates that the model possesses higher adaptability and practical value in cloud detection tasks of remote sensing images.

The internal environment of vertical well shafts is complex, and the traditional methods of monitoring the deformation of the rigid guide are cumbersome work processes, waste of manpower and material resources, Measurement accuracy is low, and 3D laser scanner technology is used to obtain point cloud information of the rigid guide to judge the verticality of the rigid guide. When extracting the point cloud data of the rigid guide, the traditional DBSCAN algorithm needs to manually select the EPS and Minpts, and needs to repeatedly change the two parameters to ensure the accuracy of the clustering. Using an improved density clustering algorithm, the radius EPS and the number Minpts can be automatically derived from the point cloud information. The point cloud data of the clustered rigid guide is sliced to fit the intersection line of the rigid guide using the least squares algorithm to obtain the coordinates of the intersection point of the rigid guide, and the difference of the intersection point coordinates is processed to obtain the deformation information of the rigid guide. Compared with traditional methods, the use of 3D laser scanners can obtain massive point cloud information, improve monitoring accuracy and efficiency, and achieve real-time and continuous monitoring of rigid guide deformation.

With the development of sensor technology, new technologies such as drone oblique photogrammetry and lidar are becoming increasingly popular in survey and design, but the data do not completely correspond in geometric coordinates. There are differences in imaging characteristics, geometry, scale, perspective, and dimension between multimodal data, as well as the influence of different environments, weather, and seasonal conditions, which leads to the sparseness and low accuracy of the homonymous points in existing image matching. To solve this problem, this paper proposes a feature description network, which extracts matching feature vectors through a lightweight transformer feature extractor and a progressive feature upsampling module and optimizes the log-likelihood loss to train the network. The experimental results of this method on the MegaDepth dataset and real dataset show that it can accurately match optical images and depth images, providing an effective solution for improving multimodal data processing capabilities.

To achieve continuous monitoring and tracking of ships over a long time series, a method based on multi-temporal remote sensing images from GF-4 satellite for continuous monitoring and tracking of ship wakes is provided. This method first constructs a GF-4 satellite remote sensing dataset for ship wake recognition, performs texture enhancement processing on the remote sensing images, and finally trains the model based on the YOLOv5x network. Experimental results show that image texture enhancement can significantly improve the recognition effect, and the highest detection rate can reach 91% in cloudless and fragmented cloud environments. Through testing, the single-band imaging mode can be used for continuous tracking and monitoring of multiple targets in the same area, obtaining motion information such as target position, heading, speed, and trajectory.

Surface emission spectra reflects the ability of a target to emit radiation of different wavelengths, and plays an important role in multiple fields as an inherent property of the target object. Currently, common methods require input of multiple data, whose acquisition will directly affect the accuracy and the timeliness of the results obtained. This article proposed a new physics-informed deep neural network for estimation of surface emission spectra, which not only utilizes the ability of deep learning models to handle nonlinear relationships and has been successfully applied to the inversion of various environmental parameters, but also combines physical knowledge of the maximum and minimum emissivity differences in multi band emission spectra to guide deep neural network training. The proposed model achieved accurate results in the simulated dataset. Compare the results of this article with ASTER surface emissivity products, which have similar spatial distributions and numerical ranges. However, the input of the new model only requires remotely sensed data, which has stronger robustness and timeliness.

High-resolution remote sensing images provide detailed surface information and are extensively utilized in urban planning, geographic monitoring, and smart city applications. Semantic segmentation, a crucial image analysis method, assigns pixel-level dense labels for scene understanding. However, this task faces challenges such as numerous small objects, complex shapes, detailed edge handling, and the effective fusion of multi-scale features. Existing methods often struggle to balance global information and local details, resulting in limited segmentation accuracy and efficiency. To address these challenges, we propose a novel network architecture named FAttFormer, which leverages the strengths of both Convolutional Neural Networks (CNNs) and Transformers. Within this architecture, we introduce two key modules: the Feature Attention Refinement Module (FARM) and the Multi-Scale Hybrid Attention (MSHA) module. The FAttFormer network is designed to significantly improve semantic segmentation accuracy and efficiency through multi-level and multiscale feature fusion and refinement. Specifically, the FARM module focuses on effectively integrating encoder and decoder features to handle small objects and complex edges. The MSHA module enhances the ability to capture and integrate global and local information at different scales, further refining segmentation results. Extensive experiments on the ISPRS Vaihingen and Potsdam datasets demonstrate that FAttFormer outperforms existing models in segmenting impervious surfaces and buildings, showing significant improvements in accuracy and detail preservation.

Aiming at the problem of insufficient research on rapid assessment methods for traditional villages to cope with flood impacts, this paper proposes a method to optimize the flood assessment model using UAV images and deep learning, and uses the YOLOv7 target detection model as well as the dataset for the training of the flood assessment model for traditional villages, and the results show that the target detection model using deep learning can be trained to produce a highly accurate traditional village flood impact assessment model, and it is easy to operate with a high degree of automation, and a high recognition speed can be achieved according to the equipment hardware.

Water body extraction is the preliminary step to define the range of remote sensing for water quality, which is of great significance to define the monitoring boundary of conventional water pollutants in Nansi Lake Basin. The original Sentinel-2 L2A multispectral image was processed with super resolution, and its spatial resolution of 13 bands was unified to 10m. According to the difference of spectral features of surfaces objects, characteristics of NDWI, NDWI3, MNDWI, NDPI and other water indexes were analyzed. By comparing effects of water body extraction with four water body indexes, it was found that the applicability of different water indexes in different surface object types was different. The remote sensing image of Nansi Lake Basin is mainly divided into five surface object types, including vegetation, towns, small rivers, mountains and shadows. The EM algorithm was used to classify surface objects of remote sensing images and complete the definition of five surface object types. In different areas of surface object types, the water index with better applicability was respectively used to extract water body. The results showed that the method of water body extraction integrating surface object types and spectral characteristics is highly efficient, and can obtain better accuracy of water body extraction than the one using only a single water index, and effect of water body extraction for large-scale remote sensing images is effectively improved.

Image matching is a fundamental task in image processing, particularly in photogrammetry and remote sensing. Conventional image matching methods often struggle with multimodal images due to significant geometric distortions and nonlinear radiometric effects. To address these challenges, researchers have developed a variety of algorithms specifically tailored for multi-modal data. Despite this, there is a notable lack of comprehensive theoretical summaries of multi-modal image matching. This paper aims to bridge this gap by discussing the complexities of matching multimodal images and providing a succinct overview of existing methods. We evaluate three Conventional algorithms—OS-SIFT, RIFT, and POS-GIFT—and two deep learning methods—SemLA and SuperPoint+SuperGlue. These methods are assessed using six types of multimodal remote sensing datasets: Optical-Optical, Optical-Infrared, Optical-Depth, Optical-Map, Optical-Synthetic Aperture Radar (SAR), and Day-Night. The results indicate that the POS-GIFT algorithm outperforms others, achieving an average of 1217 correct matches with a success rate of 96.70%. This paper concludes by discussing the challenges of multimodal image matching and suggesting future research directions, serving as a valuable resource for researchers and practitioners in remote sensing.

The traditional method of remeasurement of operational railroad has many processes, complicated operations and low efficiency, and the time of railroad maintenance days is decreasing, so new demands are made on the remeasurement method of operational railroad. In response to this problem, this paper researches the rapid measurement technology of operational railroad to achieve the rapid and high-precision acquisition of spatial measurement data in a wide range of operational railroad and the rapid and high-precision measurement of railroad track line information. This paper mainly studies the railroad line 3D coordinate acquisition trolley and its acquisition system, mainly including the key technology research of data synchronization technology, data fusion Kalman filter technology, combined navigation solution technology and track flat and longitudinal section fitting optimization research. Through this research, the three-dimensional coordinate acquisition trolley of railroad line can obtain high-precision railroad track alignment information without relying on CPⅢ control network, reduce the number of times on the track, reduce the operation safety hazards, save cost and human resources.

High-precision calibration and processing of optical remote sensing satellites is a necessary condition for obtaining all kinds of remote sensing observation products, and it is the basis for all kinds of remote sensing observation products to be fully and effectively applied. Radiometric calibration is to establish the quantitative relationship between the output value of each detector of the remote sensor and the corresponding actual ground radiation brightness. In this paper, an on-orbit yaw calibration method based on satellite yaw of 90 degrees is proposed, including the processing of yaw data and the acquisition of relative radiation calibration coefficient. Based on the data of hyperspectral detector of terrestrial ecological carbon monitoring satellite, the experimental results show that this method has a good effect on the relative radiation correction of hyperspectral detector, and the stripe noise of the original image has been well removed, which lays a foundation for the subsequent quantitative research and application of hyperspectral data.

Aiming at the problem that key objects with cultural features cannot be prominently expressed in the construction of traditional village scenes, this paper proposes a We-map symbolic traditional village expression method based on object importance. The approach utilizes the object importance assessment model as its theoretical foundation, determining the semantic significance of the object by considering the cultural characteristics of geographical objects in the village scene. Additionally, it integrates spatial three elements to establish the importance assessment model of the object. The evaluation results are then symbolized based on the We-map concept, with objects of high importance expressed preferentially through abstract symbols and those of secondary importance through simple symbols. Compared with the traditional loading method, the number of data transmissions per second is improved by 29 times, and the scene loading speed is improved by 10-15 seconds. The findings indicate that the approach has a significant positive impact on enhancing the productivity of constructing traditional village scenes, while also emphasizing the cultural attributes of important elements within the depicted scene.

[Objective] This paper aims to study the spatial and temporal distribution of rural residential areas in Lvchun County, and provide scientific reference for the optimal layout adjustment of residential areas. [Method] Voronoi diagram, kernel density analysis, geographic detector and analytic hierarchy process were used to study the spatial distribution of rural residential areas and the influencing factors of residential areas in Lvchun County. [Result] The distribution of rural residents in Lvchun County generally showed the distribution characteristics of “tight in the northeast and loose in the southwest”, with the trend of expanding from point-like aggregation to belt-like aggregation. The spatial layout of residential areas is mainly influenced by the distance from towns and villages, schools, roads, health centers and other spatial settlement factors, and the constraints of natural environment factors are decreasing. [Conclusion] The highly livable areas in Lvchun County are all along the road and close to towns, and about 11% of the villages are demolished and merged, which can provide reference value for optimizing the layout of residential areas and merging villages for poverty alleviation and demolition in different places.

In this paper, an automatic extraction method of fuzzy kernel for remote sensing images is proposed for the core evaluation index of remote sensing imaging clarity. In this paper, the deep learning based two-stage VIT classification network TIMCT (Two-stage Improved Multi-Scale Classified Transformer) is firstly trained, by using the dataset with known remote sensing image feature types to train the network and classify the images, these output scores obtained through training with high options, i.e., the network selects the region results. Then the remote sensing image is screened in two stages using edge intensity map and classification network to select the region suitable for fuzzy kernel degradation extraction. After that, ISD (iterative support detection) algorithm is used to iteratively optimize the estimation to get the final fuzzy kernel of each slice, and the result is converted into Energy concerntration (EC) to reflect the clarity of the image. The experimental results show that the method proposed in this paper can realize good region selection effect and effectively improve the ability of extracting image fuzzy kernel, and the method in this paper is ahead of the latest technology.

Soil organic carbon (SOC) serves as a critical indicator of soil health, influencing both fertility and carbon sequestration potential. This study examines future soil organic carbon (SOC) stocks in Mexico under climate change scenarios, employing an innovative framework integrating high-resolution spatial data, machine learning, and climate projections. Utilizing a dataset of 24,000 samples and NASA's Carbon Monitoring System Program, we developed climate data-driven models to generate SOC maps for Mexico from 2021 to 2100 under SSP245 and SSP370 scenarios. Our optimized Random Forest model (R-squared = 0.4457) identified temperature seasonality, annual temperature range, and precipitation as key SOC drivers. Projections reveal complex SOC density changes, with both scenarios showing a 16.17% increase by 2040. However, by 2100, SSP245 maintains a 9.93% increase, while SSP370 preserves only a 2.97% increase. The southeastern region, particularly states like Chiapas and Tabasco, showed the highest SOC densities (8-12 kg/m²), while northern and northwestern regions displayed the lowest (<2 kg/m²). This research enhances our understanding of spatiotemporal SOC variations under future climate scenarios, providing critical insights for targeted interventions to preserve and enhance soil carbon stocks. Our findings have significant implications for sustainable land management and climate change mitigation strategies in Mexico, supporting policymakers in pursuing carbon neutrality goals.

Building change detection is an important research field in remote sensing imagery. This paper proposes an improved Pyramid Scene Parsing Network (PSPNet) to address the problem of missing small ground objects and insufficient building change samples caused by interference factors such as bare land and roads in complex scenes. The method is based on PSPNet and introduces an attention feature transfer mechanism in the feature extraction stage to effectively combine high-level semantic information with low-level building details, thereby enhancing the capture of salient features, suppressing invalid features, and improving the accuracy of change detection. In response to the high similarity between changes and backgrounds, a DySample lightweight dynamic upsampling module is introduced to concentrate sampling points in the target area while ignoring the background, reducing the number of parameters and achieving effective recognition of changes; In addition, in response to the shortage of change samples, the idea of multi task learning is combined to enable semantic segmentation and change detection tasks to be processed in the same network, in order to alleviate the problem of insufficient samples. The improved PSPNet model exhibits good adaptability in building change detection tasks and significantly improves accuracy, recall, and F1 score while maintaining low computational resource consumption.

Time-series SAR interferometry (InSAR) combining persistent scatterers (PS) and distributed scatterers (DS), has made significant progress recently. DS corresponding to natural target, different from PS, is affected by temporal decorrelation. Due to the low signal-to-noise ratio of DS interferometric phase, phase optimization is needed. This paper made use of the coherence matrix decomposition to filter interferograms relevant to DS in small baseline processing. The optimized phase was estimated by eigenvalue decomposition of the coherence matrix constructed from accurately identified homogeneous pixels. Moreover, temporal coherence of DS was estimated by selecting low-noise and clear fringes interferometric pairs in the process of DS pixel extraction. The results show that this strategy can provide high deformation measurement spatial density while ensuring measurement accuracy, which is an effective way to monitor land subsidence.

In order to solve the problem that the existing zenith tropospheric delay (ZTD) models in China oftenuse the exponential function with low accuracy at low elevation for vertical correction and ignore the residuals of the fitting function, the ERA5 data from 2015~2017 were used to evaluate the accuracy of the fitting function and the ZTD vertical correction model based on the cubic polynomial function and the exponential function, respectively, and a new model based on the residual compensation of the cubic polynomial function (CZTD_cHi) was established. The accuracy of the three models was compared with the GPT3 model, and the results showed that the root mean square error (RMSE) of the CZTD_cHi model was 2.11 cm using GNSS-ZTD as the reference value in 2018, and the accuracy of the CZTD_cHi was about 40.3%, 16.0% and 5.3% higher than that of the GPT3 model, the CZTD_eH model based on exponential function and the CZTD_cH model based on cubic polynomial function, respectively. The CZTD_cHi model established in this study can provide high-precision ZTD vertical correction services for GNSS users in China, and can provide an important reference for real-time GNSS water vapor detection and navigation and positioning in China.

From the current well network distribution in the XX oilfield, it can be seen that the well pattern effectively controls control major faults, structural trends, and distribution morphology of sand body. However, due to the truth that many fault points are located within strata, well placement often leads to missing well points, which makes it difficult to control the spatial distribution of faults in oil layers. Moreover, for some smaller faults, their orientation cannot be accurately determined by using fault point data. Based on the existing drilling data, it is challenging to identify micro-amplitude structures. Therefore, through fine structural and fault interpretation of three-dimensional seismic data, and with the aid of a combination of well and seismic methods to study structural development characteristics and fault development features, we have clarified the structural development characteristics of the top surface of the P oil layer group and identified structures with an amplitude greater than 5 meters. The study shows that many local structures exist within the block, such as fault blocks, fault anticlines, fault noses, and domes. These areas are potential regions in the expansion zone and there are zones of remaining oil accumulation in the developed area, making them suitable for the deployment of high-efficiency wells. It is builds the foundation for the development adjustment and evaluation of remaining potential in the block.

On September 3, 2017, the People's Republic of Korea conducted its largest underground nuclear test to date. The analysis of interferometric synthetic aperture radar (InSAR) data can explore surface displacements associated with nuclear explosions. This study calculated the actual deformation of the mountain after the explosion using D-InSAR and photogrammetric methods. Due to the large-scale surface deformation caused by the nuclear explosion, the experimental site experienced a total volume change of 0.058 cubic kilometers. In addition, this study has made some innovations in the method of adjusting the SAR image coordinate system to a spatial Cartesian coordinate system. Compared with previous scholars' research, this project has established a model to study the volume changes of different experimental fields in less involved areas. Ultimately, it is possible to compute the deformation of the mountain caused by the internal explosion, laying the foundation for the early warning department to formulate protective measures.

Utilizing daily and hourly observation data from national meteorological stations in Shenyang spanning from 1981 to 2023, along with encrypted automatic station data at an hourly resolution, this study examines the temporal trends of precipitation variations during the warm season (May-September) within the region, as well as investigates the spatial distribution and characteristics of short-term heavy rainfall events. The findings indicate a significant rise in warm season precipitation during the initial phase (May) of Shenyang urban area over the past 40 years. However, during the intermediate stage from June to July, there is a transitional shift from an increasing trend to a decreasing trend. In the later stage of the warm season (August and September), there is a notable downward trajectory with an escalating decline. Due to the substantial proportion of precipitation in July and August, there is a persistent decline in annual precipitation. There are significant differences in the distribution of extreme hourly precipitation and short-term heavy precipitation stations in Shenyang urban area in the past 10 years, with the high-value areas mainly distributed in the urban area and the eastern part of northern Shenyang; eastern Shenbei is located in the downwind area 30-50km from the center of Shenyang urban area, which is conducive to the increase of precipitation and the generation of extreme heavy rainfall.

Ghana is located on the northern coast of the Gulf of Guinea in West Africa, where mineral resources are widely distributed. Through the detailed study of Ghanaian magnetite breccia, it is found that Ghanaian magnetite breccia is mainly distributed in magmatic rocks, and the breccia shapes are diverse, including round, sub-angular and angular, and the proportion of round to sub-angular and angular content reaches about one-half, which has obvious characteristics of fragmentation. Most of the metal minerals in the magnet breccia are wrapped with other minerals, and the secondary increase phenomenon can be seen on the edge of the breccia, and the individual breccia develops pores, because of the different iron content. The cement of magnet breccia is mainly fine-grained clastic material or clay minerals, with particle size less than 0.03mm. It is believed that the breccia in Ghana magnet breccia can be divided into two types, namely rock breccia and metal mineral breccia. The formation process of breccia is mainly divided into two stages, namely magma formation stage and magma solidification diagenesis stage. The analysis of mineralogy and diagenetic process of Ghana magnetite breccia is of great geological significance for understanding the magma fluid interaction and metallogenesis in this area.

Subsurface salt bodies provide good confined storage space for oil and natural gas, and effective identification of various types of geologic bodies is the core content of oil and gas exploration. However, current data-driven salt body interpretation methods generally suffer from poor generalization. In the face of the inherent differences between synthetic seismic data and field seismic data, under the premise that field salt body markers are difficult to obtain, it becomes crucial to rationally apply field seismic data and enable the model to deeply learn the correlation features between the two types of data. We proposed an unsupervised embedding-based method for dense prediction of 3D salt bodies (3D CLP-gb), presented contrastive loss for field-oriented seismic data, and demonstrated that unsupervised learning of diverse salt body augmentations with high-frequency amplitude features of other geologic bodies as positive and negative samples of each other can help to solve the challenge of the generalizability of the salt body to field data segmentation. The experimental results show that with only 1.56% of the tagged data used, 3D CLP-gb as an upstream task in combination with the SOTA model is able to extract a more complete distribution of salt bodies from the field seismic body, achieving an IOU value of 87.5% for the SEAM synthetic validation set, and a generalization performance of 85.8% for the F3 field dataset.

The concentration of suspended matter in coastal seawater has a significant impact on seawater quality, marine organisms, and regional marine ecosystems. Traditional remote sensing monitoring models have long faced the challenge of determining the optimal inversion bands. Based on singular value theory, this paper proposes a novel method for band selection. Singular Value Decomposition (SVD) reduces the complexity of data, making it easier to capture data features. Taking the coastal waters of Hong Kong as the study area, this paper employs SVD to conduct random forest algorithms and compares them with inversion methods based on band combinations. After verifying the accuracy with measured data from stations, it is found that the random forest algorithm based on SVD achieves the highest accuracy with an R² of 0.75 and an RMSE of 1.05, demonstrating the best model universality.

In the face of the gradually increasing damage to the ecological environment caused by mining, the construction of a high-precision digital elevation model (DEM) of the mining area is of vital significance for the in-depth study of surface deformation and rock movement parameters. Aiming at the problem that the traditional point cloud filter algorithm is difficult to obtain a high-precision DEM when processing airborne LiDAR (light detection and ranging) point cloud data under the complex terrain conditions in the coal mine area, an improved progressive TIN densification filter (IPTDF) algorithm is proposed, which is able to obtain point cloud data closer to the real ground surface to construct a high-precision DEM. In order to verify the effectiveness of the proposed method, firstly, four point cloud filter methods are used to carry out point cloud filter experiments on two complex terrain study areas, and then the confusion matrix is used to evaluate the point cloud filter error; secondly, the DEM is constructed by using the inverse distance weighted (IDW) interpolation algorithm, and the effect of the DEM is judged by visual interpretation. The final results prove that the IPTDF algorithm proposed in this paper can obtain higher point cloud filter accuracy under complex terrain conditions, has good applicability and reliability, and provides data support for the construction of high-precision DEM.

To improve the accuracy of dam safety monitoring, the study introduces precision single-point positioning on the basis of BeiDou positioning and proposes precision single-point positioning BeiDou positioning system. Selected monitoring data from a dam monitoring station in Sichuan, the precision single-point positioning BeiDou positioning system and the traditional satellite positioning system for comparison experiments, the results show that, in the dam displacement monitoring comparison experiments, the precision single-point positioning BeiDou positioning system in the different test points of the accuracy of 87.33%, 94.83% and 94.44%, are much higher than the accuracy of the traditional satellite positioning system monitoring. The results show that the dam safety monitoring system based on the precision single-point positioning Beidou positioning system can obtain higher positioning accuracy and can be used as a long-term effective means for dam safety monitoring.

Water leakage in wellbores significantly impacts the structural safety and longevity of the wellbore, making timely and accurate detection of leakage areas crucial. This paper proposes a method for detecting wellbore water leakage based on point cloud reflection intensity, aimed at enhancing detection accuracy and efficiency. The approach begins by using a histogram of point cloud reflection intensity to determine the threshold for segmenting leakage regions, allowing for an initial rough extraction. This is followed by precise clustering of the leakage areas using an improved DBSCAN algorithm, enhanced with a K-D tree for efficiency. Finally, boundary extraction and convex hull computation are employed to accurately calculate the area of the leakage region. The proposed method is validated against experimental results, demonstrating its ability to quickly and accurately detect wellbore water leakage, showing promising potential for practical applications.

With the acceleration of urbanization, the urban heat island effect has become more and more serious, which has seriously affected the quality of life of urban residents and the sustainable development of cities. In this paper, Kunming City of Yunnan Province is taken as the research area. Using Landsat series satellite image data from 2000 to 2020, the land surface temperature (LST) is retrieved by single-window algorithm, and the local Molan’I, Markov transfer matrix, and geographical detector are combined to quantitatively analyze the spatio-temporal evolution characteristics and its influencing mechanism of LST in Kunming City. The results show that: (1) From 2000 to 2020, the LST in Kunming showed a significant rising trend, the average annual warming rate was 0.14°C/a, the heat island effect was gradually intensified, and the area of high temperature zone (>30deg;C) increased from 17.3% in 2000 to 29.6% in 2020, and the spatial agglomeration of each temperature zone is continuously enhanced; (2) Fractional Vegetation Cover(FVC) is the dominant factor affecting LST, and has a significant negative correlation with LST. Every 10% decrease in LST will increase by 1.3deg;C; (3) The natural land loss caused by urban expansion is another important cause of surface warming. From 2000 to 2020, the area of construction land increased by 157.64%, and its average temperature was 4.3deg;C higher than that of forest land and 10.72deg;C higher than that of water; This study reveals the evolution law of the surface thermal environment in Kunming City during the rapid urbanization process from the aspects of spatio-temporal dynamic change, landscape pattern characteristics and influencing mechanism, which can provide important scientific basis for regional urban planning and ecological civilization construction.

The wheat yellow rust disease was selected as the object in this paper, and the vegetation index (VI) and texture feature (TF) were extracted by UAV multi-spectral images, and the monitoring model of the wheat yellow rust was constructed by machine learning methods, such as KNN, BP, SVM and RF. The results show that: The RF model training set R² of VI and TF is the highest compared with the other three models, which are 0.691 and 0.529, respectively. When the two remote sensing information sources are fused, the RF model training set R² model value is 0.783, which improves the model accuracy compared with single data source. The VI was more beneficial to for estimating wheat yellow rust than TF, and the model accuracy of the model was better than that of a single data source under the synergistic effect of the two data sources. In addition, the monitoring accuracy of RF model is better than the other three algorithm models, and it is more suitable for the UAV remote sensing monitoring of wheat yellow rust.

Jiangqiao is located in the western slope area of the northern part of the Songliao Basin, which is currently the most favorable area for heavy oil exploration. The main target layers are S and G oil layers, and the oil-water relationship is complex. How to effectively identify oil bearing reservoirs is one of the key factors for achieving exploration breakthroughs in this area. This article takes the S1 oil reservoir group in J75 work area as an example, and based on data analysis, uses seismic waveform indication simulation to predict the oil bearing reservoir of the S1 oil reservoir group. The prediction results indicate that the reservoir prediction results simulated by waveform indication on the profile are basically consistent with those obtained from logging. By extracting the plan view of the S1 oil reservoir and conducting well analysis, the predicted accuracy rate was 94.8%.

The complex geological environment in karst areas often leads to unexpected geological disasters during tunnel excavation due to adverse conditions such as caves and aquifers, posing significant safety hazards to construction projects. Single geophysical exploration methods have limitations and are often inadequate for meeting the demands of actual engineering projects. This paper employs an integrated approach using high-density resistivity, semi-airborne transient electromagnetic (SATEM), and ground-penetrating radar (GPR) methods to conduct geological surveys in a proposed highway tunnel traversing complex karst terrain. The timely acquisition of detailed and comprehensive karst geological data enables accurate identification of the spatial location and distribution of adverse geological features within the tunnel. This approach facilitates the determination of potential karst geological disaster risks and provides a precise design basis for related engineering projects. Verification through actual excavation revealed a medium-sized cave at the geophysical anomaly locations, proving the effectiveness of the integrated geophysical exploration methods and offering a valuable reference for similar tunnel exploration endeavors.

Aiming to investigate the evolution in vegetation water use efficiency (WUE) in the Yun-Gui Plateau over the past 22 years. This study estimated the WUE in this region from 2001 to 2022 using the Google Earth Engine. The Theil-Sen, Mann-Kendall, and coefficient of variation methods were employed to explore the spatiotemporal evolution of WUE in the Yun-Gui Plateau. The results showed that: (1) WUE in the Yun-Gui Plateau has been fluctuating but showing an increasing trend, with an average value of 3.01 g C kg^-1 H2O. (2) Regions with significant and slight decreases in WUE are mainly distributed in the southwest, while regions with significant increases are mainly found in the central part, and regions with slight increases are predominantly located in the northeast. Most areas exhibit a stable trend in annual average WUE. (3) WUE in shrublands and grasslands is decreasing, while other vegetation types show an increasing trend. The research findings have scientific reference value for the improvement of the ecosystem structure and function in the region.

In this paper, based on examples of micro-motion and geomagnetic bathymetric surveys, and on the basis of the establishment of a reasonable geologic model, data such as the type of one-dimensional inversion curves of the geomagnetic method, frequency-visual resistivity, impedance-phase contours, two-dimensional inversion resistivity cross-section maps, and velocity cross-section maps of the micro-motion surface-wave method are comprehensively applied by means of the study of geologic data in the study area. Adopting multi-parameter and multi-data processing methods can overcome the interference of large-thickness low-resistance and low-velocity layers, and obtain real and effective geological information. Adopting the research idea of "surface - line - point", focusing on the geothermal anomalies step by step, analyzing and interpreting the structure of the survey area, analyzing the development of fracture gectonic, constructing vertical gectonic structure model, analyzing the depth of thermal storage, providing the basis for the development of geothermal resources, and ultimately obtaining a good exploration effect.

To address the issue of insufficient monitoring due to the complex terrain of the Sulfur Gully mining area in Changji City, where single-track SAR data monitoring is affected by topographical features, a monitoring method integrating ascending and descending radar data using the Small Baseline Subset (SBAS-InSAR) technique is proposed. Additionally, a two-dimensional calculation method is employed to process the line-of-sight (LOS) results derived from the ascending and descending SBAS monitoring. These results are compared and validated against leveling data. The findings indicate that the two-dimensional calculation method can significantly reduce positional discrepancies, enhance the precision of single-track SBAS monitoring, and achieve more accurate two-dimensional deformation results.

Explore the spatial distribution and matching status of agricultural water and water resources under different changes of land use in the future, and provide support for the planning and decision-making of agricultural sustainable development at watershed scale. FLUS model was used to predict the change of land use spatial pattern in Shi yang River basin in 2035 under different scenarios, and InVEST model was used to simulate water yield and evaluate the spatial-temporal matching relationship of agricultural soil and water resources in Shi yang River basin. The results show that the matching pattern of water and soil resources in Shi yang River Basin is better in the west than in the east, and the Gini coefficient is between 0.2 and 0.3, which shows a slight upward trend in the future, but is still in a relatively balanced state. In 2035, the average matching coefficient of water and water resources in Shiyang River Basin was 797m³/hm², and the matching coefficient of water and water resources in 2020 was 640m³/hm², and the overall trend was positive. The research results can provide a reference for the balanced management of agricultural soil and water resources in Shi yang River Basin.

Mining-induced large gradient subsidence and vegetation growth are the main problems in surface deformation monitoring. The L-band radar wavelength offers strong vegetation penetration and relatively large deformation monitoring range, making it suitable for monitoring mining-induced deformation. However, challenges such as limited available data, long revisit periods, and high costs persist. This study employs Sentinel-1A as the main data and ALOS-2 serving as auxiliary data. The SIP method is applied to perform three-dimensional decomposition of ALOS-2 data, using the Sentinel-1A coordinate system as the reference and its imaging geometry for synthesizing line-of-sight (LOS) deformation. Time-series InSAR analysis is conducted to obtain high temporal resolution deformation time-series in mining areas.

In order to accurately evaluate the surface subsidence caused by mining in the loess plateau mining area, solve the production safety and environmental protection problems of the mining area, and study the use of airborne LiDAR point cloud data to construct a high-precision loess plateau mining area working face subsidence basin model. Firstly, preprocess the collected raw monitoring data of the mining area to obtain the true color point cloud data of the subsidence area of the working face. Then, by analyzing the terrain factors of the subsidence basin using models of different resolutions, the optimal resolution of the surface subsidence basin model suitable for the mining area working face was determined. Finally, the sliding window filtering algorithm was used to optimize the preliminary constructed subsidence basin model, and the accuracy of the optimized model was verified using three quantitative evaluation indicators. The results show that the subsidence basin model established and optimized in this article has good accuracy, can comprehensively and reliably reflect the subsidence situation of the working face surface in the loess plateau mining area, and has high application value.

Mangrove forests are one of the most productive marine ecosystems in the tropical and subtropical coastal zone, playing an extremely important role in purifying seawater, preventing wind and waves, maintaining biodiversity, and carbon sequestration and storage. Monitoring mangrove resources has important ecological service functions and economic and social value. Conducting long-term monitoring of mangrove resources is of great significance for the research and protection of mangrove resources. This study utilized long-term remote sensing image data from 1980 to 2024, dividing the research area from 1978 to 2024 into five time periods: 1980–1990, 1990–2000, 2000–2013, 2013–2018, and 2018–2024. Six remote sensing image base maps of Pearl Bay in Beilun River estuary, Guangxi were produced from 1978 to 2024, and the area and changes of mangrove resources in the research area were monitored. The main reasons for the growth and decrease of mangrove resources in Pearl Bay area were analyzed, providing data reference for the protection of mangrove resources in Beilun River estuary, Guangxi. The results show that the overall trend of mangrove resources in the Pearl Bay area of Beilun River estuary in Guangxi has been good in the past 46 years, with the number of mangroves recovering and increasing. At the same time, there are also factors such as natural reduction, a small amount of aquaculture occupation, and pollution discharge. This study fully utilizes new technologies such as satellite remote sensing to provide important data support for strengthening the dynamic supervision of mangroves, continuously restoring the ecological environment of mangroves, and promoting the natural ecological restoration of mangroves. It also provides important decision-making support for formulating scientific and reasonable protection and restoration policies for mangrove management.

The drought in farmland, caused by moisture deficit, is reflected by soil moisture content, a key parameter for assessing surface drought. Traditional models fail to account for varying vegetation cover during wheat's reproductive period. This study introduces a remote sensing-integrated model for drought monitoring throughout wheat's life cycle. Three indices, PDI (low coverage), MPDI (medium coverage), and TLDI (high coverage), were selected to reflect soil moisture under different vegetation covers using MODIS data. These indices were validated with soil moisture data from 16 agrometeorological stations in the Guanzhong region (2010-2012). The model provided spatial and temporal distributions of soil moisture, enabling effective on-farm drought monitoring. Results showed correlation coefficients of 0.68, 0.64, and 0.62 for PDI, MPDI, and TLDI, respectively. The temporal drought distribution revealed extensive drought in late October, diminishing in mid-November, and light to moderate drought by early April. Spatially, the east experienced severe drought, central parts moderate, and the west mild. The model's outputs closely matched actual drought conditions, demonstrating its value for agricultural drought monitoring.

Located between the main body of the Qinghai-Tibet Plateau and the Sichuan Basin, the Western Sichuan Plateau is a part of the Hengduan Mountains and the southeast edge of the Qinghai-Tibet Plateau. The Western Sichuan Plateau has complex geological structure and strong fault activity. It is a high incidence area of geological disasters, and is one of the regions with the most serious geological disasters such as collapse, landslide and debris flow in China. In this paper, Yajiang County in the Western Sichuan Plateau is taken as the research area, and GIS is used to carry out geological hazard risk assessment research. The quantitative analysis of the relationship between the geological hazard potential points and the landform factors, geological structure factors, meteorological and hydrological factors, human engineering activities factors in Yajiang County is carried out. The spatial relationship between the geological hazard potential points and elevation, stratum lithology, annual precipitation, NDVI, etc. is revealed, and the geological hazard risk in Yajiang County is systematically evaluated. The geological hazard vulnerability evaluation index system of Yajiang County is established. Finally, risk assessment of geological hazards in Yajiang County is obtained through superposition and calculation of evaluation results of geological hazard risk and vulnerability of Yajiang County, and the results of assessment are divided into five grades. Based on geological hazard risk division, the suggestions on prevention and control of geological disasters are offered.

In this paper, an innovative "dual terminal fusion" remote sensing dynamic monitoring system for mine is proposed. Through multi-source data fusion and "1+1+N" architecture, the system realizes the comprehensive integration of data collection, storage and analysis. Although the system has significant advantages in terms of data processing speed and user interface friendliness, its high cost and complexity may limit its application in small mines. In addition, the adaptability of the system under specific geological conditions still needs to be further validated and optimized. The system performs well in terms of processing speed, accuracy, stability and scalability, but the environmental impact and sustainability still need further in-depth analysis.

As an important target of lithologic and unconventional oil and gas reservoir exploration, the widely distributed shallow delta sand bodies provide good conditions for large-area and high-abundance conventional and unconventional oil and gas accumulation. In order to clarify the basic sedimentary characteristics of shallow braided river delta front and its control on reservoir quality, this study takes He 8 in Jingbian gas field as an example, based on core description, combined with analysis and testing and logging data, discusses the microfacies division of braided river delta front and the environmental constraints of reservoir quality. The results show that underwater distributary channel, distributary bay, natural levee and mouth bar are mainly developed in the front of braided river delta. The underwater distributary channel is mainly composed of medium-fine sandstone with good connectivity, with various cross bedding or parallel bedding, which is the skeleton sand body of the reservoir. The microfacies of natural levees are mainly siltstone and mudstone, which develop along the rivers on both sides and can only be used as reservoirs locally. Reservoir quality is controlled by microfacies and sand body thickness.

The resilience of urban network structure is an important measure of urban development stability. Taking the Yellow River Basin as an example, the resilience development of the urban network structure in the Yellow River Basin is analyzed by evaluating the transmission and agglomeration of the urban network structure. The results show that: (1) The intensity of urban linkages in the Yellow River Basin is significantly different, and the urban development in the region is unbalanced. (2) The agglomeration of downstream cities is significant, and the dependence on the network structure is strong. (3) The core node cities lack the central driving effect on the non-core node cities. Finally, some suggestions are put forward for the optimization of urban network structure in the Yellow River Basin.

Using remote sensing technology to monitor crop growth is one of the very important research topics in precision agricultural remote sensing. Crop growth monitoring refers to monitoring crop growth status and predicting its development trend through satellite remote sensing. Real-time knowledge of crop growth status can provide timely and reliable information for field management, lay the foundation for estimating crop yields, and provide the National Agricultural Department with a basis for food management and decision-making. This paper takes the Redstar Farm of the Beian Branch of the Heilongjiang General Administration of Agricultural Reclamation as the research area, selects the Sentinel-2 image data and vector data files in the area in 2023, and uses basic software such as ENVI5.3 and ArcMap10.4 to obtain the crop phenological phase growth information map of the study area. By comparing and analyzing monthly growth charts, we can obtain the characteristics of changes in time and space of crop growth in Redstar Farm in 2023 and analyze the main reasons for these changes.

This study focuses on the central urban area of Harbin, analyzing the spatial similarity, interdependence of its buildings, and their influencing factors using correlation and spatial autocorrelation analysis. Furthermore, a multi-scale geographically weighted regression (MGWR) model is employed to explore the spatial heterogeneity of three-dimensional buildings in greater depth. Compared to traditional multiple linear regression and geographically weighted regression (GWR) models, the MGWR approach not only provides a more nuanced analysis of spatial heterogeneity but also dynamically adjusts the regression bandwidths for different influencing factors. This allows for a more precise understanding of how these factors impact the three-dimensional spatial pattern. The findings reveal that the architectural landscape pattern in Harbin's central area is strongly influenced by spatial autocorrelation. The distinct characteristics of urban buildings shape the landscape pattern, contributing to notable spatial differentiation across the city.

The anomalous zonal wind moves southward during ENSO mature phase in boreal winter. Previous studies suggest that it may be caused by the nonlinear interaction of annual cycles or the influence of background mean state changes. In this research, the numeric experiments using ECHAM4.6 is conducted to confirm the mechanism of the anomalous zonal wind southward shifting. The results show that the north-south asymmetry mode of ENSO anomalous wind field is not the result of the nonlinear interaction between ENSO and the annual cycle. The annual cycle changes the mean state field, and the cross-equatorial north winds in winter affect the southward shifting of ENSO anomalous wind field through advection.

A field study employing high spatiotemporal-resolution measurements of velocity, salinity, and suspended sediment concentrations was conducted to analyse the characteristic of Tiaozhoumen channel fronts. A significant difference in frontal structure was observed between flood and ebb tides. At ebb tide, the current velocity was slightly lower on the shallow side of the front than at the deep side. Furthermore, the salinity of water in the frontal zone and its shallow side was higher than that on the deep side of the front. However, flood tide trips to low-velocity zones occurred near the frontal zone, and the salinity and flow velocity of the frontal zone were lower than those of either side of the front. Additionally, the field data support the hypothesis that the presence of fronts causes the improvement of lateral flow and an increase in the suspended sediment concentration of the middle and deep layers in the front area.

The sound speed in seawater is primarily influenced by temperature, salinity, and pressure. Given known temperature and salinity profiles, empirical formulas can be used to calculate acoustic travel times across these profiles. Conversely, by employing an Inverted Echo Sounder (IES) device to measure sound propagation time data along a profile, the corresponding temperature and salinity profile information can be inferred using the Gravest Empirical Mode (GEM) method. However, the structure of water masses in the ocean is not constant, and in regions with complex water mass properties, such as the KESS (Kuroshio Extension System Study) area, the effectiveness of the GEM method for inversion may be significantly reduced. This study developed a multidimensional Deep Learning Gravest Empirical Mode (DL-GEM) method based on feedforward neural networks, incorporating sea surface temperature and latitude-longitude coordinates into the standard GEM approach. When latitude coordinates and sea surface temperature are taken as input parameters, the MSE in the training process drops from 0.045 to 0.025, which effectively improves the performance of the model. The addition of longitude coordinates has little effect on the training effect of the model. It can be concluded that the division of water mass in KESS area is mainly reflected in the north-south direction. The DL-GEM method showed an increase of 5%-25% in the variance capture rate for temperature and salinity profiles over the traditional GEM method. The DL-GEM model also produced more accurate thermohaline results when applied to Inverted Echo Sounder (IES) observational data. This research enhances the inversion precision of IES observational data in regions with complex water mass properties, offering more accurate depictions of mesoscale oceanic dynamics.

Surveying and mapping data is the foundation of basic mapping and the basic basis for all spatial location information. In the context of the transition from traditional basic mapping to new basic mapping, exploring the construction content of provincial modern surveying and mapping data has significant practical significance. This paper aims at the massive space information acquisition of “real scene three-dimensional” (3D Real Scene), which introduces the construction situation of Shandong province’s modern surveying and mapping datum system with land-sea coordination. It also summarizes the experience of building this system and discusses the construction direction between modern surveying and mapping datum systems and new basic mapping.