This PDF file contains the front matter associated with SPIE Proceedings Volume 12551, including the Title Page, Copyright information, Table of Contents, and Conference Committee Page.

Based on the three-dimensional acceleration recordings from the strong motion network nearby the Luding earthquake, we calculate the ground motion related parameters, discuss the distribution of PGA and analyze attenuation relation.

During the earthquake interval in northeast China, it is of great significance to monitor the ground deformation in Songyuan City with an earthquake in 2018. In this paper, the Sentinel-1 images in two orbits during 2017 and 2020 were used to monitor the long time-series and large area ground deformation in Songyuan City. The deformation was obtained by setting the spatial baseline, coherence threshold and other parameters reasonably. And the results mainly showed that there was ground deformation phenomenon and differences spatially, and coherent points P1 and P3 showed relatively significant linear subsidence characteristics from July 2018 to July 2019, which indicated the ground stable was mainly influenced by the earthquake.

The characteristics of cloud top physical quantities of severe convective clouds were analyzed by using Himawari-8 satellite data for two short-time heavy precipitation processes over Liaoning Province on September 9, 2021 (referred to as the "9.9" process) and October 3, 2021 (referred to as "10.3" process ). The results demonstrate that the range of convective clouds is wider and the value of the water vapor brightness temperature difference and infrared cloud top height of the "10.3" process are greater than those of the "9.9" process. In the "10.3" process, the top of the convective cloud is always near the troposphere or above the top of the troposphere, whereas in the "9.9" process, it is only in the initial and mature stages of convection. The physical features of cloud tops can be used to determine the life cycle of cloud masses. Strong convective clouds are already visible in the early stages of convection, and both the rate of cloud growth and the rate of decline in infrared brightness temperature are faster in the beginning stage than in the mature stage. The early stage of cloud production, on the side of the low-level inflow area, has the most pronounced cooling area. As a result, the cloud body's optical thickness grows at this point, and the freezing process for water clouds starts to take place. In the stage of cloud maturity, the height of the cloud top achieves its highest point, and in comparison to the early stage, the optical thickness, density of water or ice cloud particles, and cloud height all diminish.

Based on GIS technology, this paper comprehensively combs and summarizes the economic value evaluation of the national owned forest resources assets at the county level, analyzes the basic principles of the assets evaluation, summarizes and analyzes the evaluation ideas, selects the appropriate evaluation ideas according to the characteristics of forest resources, selects the specific evaluation formula in combination with the existing data to evaluate the national owned forest resources assets in Jinjiang City, and analyzes the charts. Finally, it expounds the support and application of this achievement, The details are as follows: (1) Based on the existing basic data of Jinjiang City, the scope of the appraisal work (map spot) is determined through GIS superposition analysis and processing, and the economic value appraisal data of Jinjiang forest resources assets are sorted out. (2) Combined with consulting relevant literature, this paper analyzes the basic principles of asset appraisal, selects appropriate appraisal ideas according to the characteristics of forest species, production period and age, and obtains the value of forest resources assets owned by the whole people in Jinjiang City. (3) The ArcGIS software is used to generate the economic value heat map of forest resources assets owned by the whole people in Jinjiang City, the economic value heat map of forest resources assets owned by the whole people in Jinjiang City, the economic value heat map of forest land assets owned by the whole people in Jinjiang City, and the utilization efficiency heat map of forest land owned by the whole people in Jinjiang City. The economic value of forest land assets can be obtained through visual analysis. (4) The support and application of this achievement are described.

Ship trajectory prediction with high accuracy plays a significant role in maritime traffic management. The collision can be effectively decreased with the help of real-time prediction to plan a navigation course and monitor the ship's travel status. We propose a short-term trajectory prediction method based on bidirectional long short-term memory (Bi-LSTM) network by using AIS data from the Hainan-1 satellite. The main steps include (1) eliminating the abnormal data by filtering the historical data, smoothing the trace by linear interpolation, and normalizing into uniformly distributed time-series data; (2) creating the Bi-LSTM model; (3) predicting the next position of the ship. The experimental results show that the model has a relatively low root-mean-square error, which demonstrates its efficiency for trajectory prediction and can be utilized to avoid collisions and improve the safety of maritime traffic.

In order to improve the reliability of large deformation prediction of layered soft rock tunnel under complex geological conditions, a tunnel large deformation prediction method based on convolution neural network is proposed, which solves the problems of complicated calculation of multiple evaluation indexes' weights and diverse limit values in large deformation prediction. In order to fully consider the influence of layered weak surrounding rock strength, surrounding rock structure type, in-situ stress and groundwater on large deformation of tunnel, six sub-indexes, namely, compressive strength of rock mass, bedding dip angle, initial in-situ stress state, buried depth, corrected quality index of rock mass and groundwater development, are selected to predict the large deformation grade. According to the classification standard of large deformation, a large deformation prediction model based on in-situ stress inversion and on-site large deformation monitoring information is constructed. By using the large deformation information of the tunnel, a convolution neural network large deformation prediction model which accords with the actual law of the target tunnel site is constructed. Convolutional neural network model has high accuracy in predicting sample test sets.

With the development of remote sensing technology and artificial intelligence, land use data emerge in endlessly. In order to explore the characteristics, suitability and advantages and disadvantages of different land use datasets, this paper takes Shanghai as an example to analyse the land use change trend of ESACCI, GlobeLand 30 and MCD12Q1 datasets from 2000 to 2020 by using land use dynamic degree. Based on CA-Markov model, the development trend of land use change in 2020 and 2030 is simulated and predicted, respectively. Results: (1) The CA-Markov model was verified, and the Kappa coefficient of accuracy test was greater than 0.75, indicating that the model was reliable. (2) Comparing and analysing the existing data of the three datasets, the results show that there are great differences in different land use datasets; (3) Comparing the simulation results of three datasets, the results show that there is no necessary connection between data resolution and simulation results.

"Extremely low frequency ground penetrating (WEM) project" is one of the major national scientific and technological infrastructure construction projects approved by the National Development and Reform Commission in the "11th Five-Year Plan," which is used for resource exploration, earthquake prediction and other frontier scientific research. Extremely low frequency (0.1 ~ 300Hz) is an innovative exploration technology that artificially emits extremely low frequency electromagnetic waves to effectively cover a larger area and depth. It will play an irreplaceable role in earthquake prediction. The article introduces the background of the project and the construction of the subsystem in detail. [1]

The special locations and serious weather conditions of plateau hydropower stations make it particularly important to conduct condition monitoring on these plants. Through analyzing the features of condition monitoring of high plateau hydropower stations, the present paper summarizes the application of GPS in this area and the challenges standing in the way. Based on the comparison, this study concludes the impacts of satellite elevation angles on condition monitoring of plateau hydropower stations to back the application of GPS in this field, which is beneficial to the promotion of satellite remote sensing technology in condition monitoring.

Aiming at the application of the geographical conditions monitoring data in the results of the third national land resource survey, this paper proposes a comparison and fusion model of the third national land resource survey and the geographical conditions monitoring based on regression analysis. This study analyses the reasons for the differences between the two groups of data, and formulates the fusion rules, and takes Jieshou city as the study area to carry out experiments. We randomly selected sample areas for fusion experiments to obtain statistical data, consider the relationship between the fused cultivated land area (Y) and the cultivated land overlapping area ( x₁ ), and the cultivated land difference area ( x₂ ), and established a regression equation. The model can solve the problem of the difference between the data of the third national land resource survey and the geographical conditions monitoring, and realize the effective integration of cultivated land types. Based on the regression analysis method, the regression equation of cultivated land area after the integration of Jieshou city is obtained. The model proposed in this paper can not only effectively connect the third national land resource survey data with the geographical conditions monitoring data, but also scientifically and accurately predict the area of cultivated land in Jieshou city. The research results can provide technical support for the distribution of cultivated land quality monitoring points, land environmental protection, land management, etc.

As extreme weather events, severe typhoons have a great impact on people's daily production and social stability. Therefore, many scholars have used different methods to predict the formation of severe typhoons in order to mitigate the impact of severe typhoons. However, the severe typhoon in reality is a small probability event and the number of severe typhoon samples used for prediction is much smaller than the number of ordinary typhoon samples. Existing models rarely consider the impact of unbalanced data on model training and prediction, which makes it difficult to apply the model to the actual severe typhoon prediction. Therefore, we propose a severe typhoon formation prediction model based on unbalanced data. The model uses a convolutional neural network to obtain features from the severe typhoon environmental field and an LSTM model to implement the prediction of severe typhoon formation. A customized loss function designed in this paper is used to add different classification weights for normal typhoon samples and severe typhoon samples, so that the model improves the prediction of unbalanced severe typhoon formation. The experiments show that the severe typhoon formation prediction model based on unbalanced data outperforms the traditional machine learning model for unbalanced severe typhoon formation prediction.

It has become apparent in recent years that climate change is becoming one of our most significant environmental issues. This is because of the greenhouse effect, melting glaciers, and other factors. Sea level rise and extreme rise have been receiving more and more attention of late. Furthermore, numerous flooding events have become increasingly frequent over the past few decades, especially for coastal cities. Therefore, it becomes increasingly critical to evaluate the impact of storms on coastal cities in order to comprehend the specific impact of storms on coastal cities. Prediction of overtopping rates considering future SLR (Sea Level Rise) is also an essential part of the process. The purpose of this report is to assess flood impact in terms of overtopping rate with a numerical model, XBeach. ArcGIS and MATLAB were utilized for data processing and visualization.

To discuss the space-time variation features of land use in Hebei Province, the dynamic optimization model of land use, transfer matrix of land use, and comprehensive index of land use were utilized to analyze this. The results showed that: (1) cultivated field, woodland, and grassland constitute the important land use types in Hebei Province; (2) Arable land, forest land, grassland, and building land are the principal land types that change. The land use change of the water area, unused land, and construction land with a small area proportion is relatively fast, and the land use change of cultivated land, forest land, and grassland with a relatively large area proportion tends to be stable; (3) In Hebei Province, the land use degree in the southeast is always higher than that in the northwest.

The bedrock of Wumishan formation in Yanling buried hill belt of Jizhong Sag has high temperature, high heat, crack cave development, and great geothermal exploitation potential. Due to the control of multi-stage structure and karst, the reservoir space has various types and complex structures. Fractures are the main factors affecting heat carbonate reservoir development. Combining with the characteristics of regional tectonic evolution and stress, through the field outcrop, core, thin section, 3D seismic data analysis, fracture characteristics for the development of reservoir in the studied area is described, and that the region is mainly the large angle of the development structure cracks and dissolved fracture, along the seam with solution pores. Since fracture development is mainly affected by tectonic stress factors, the multi-angle fusion technology of seismic attributes such as coherence, curvature, texture and ant body is preferred to identify fracture network system in this study, and the distribution characteristics of three types of structural fracture development zones are summarized. As a result of the buried hill fracture development degree of karst has control effect, combined with Jizhong depression exposure period unconformity karst development characteristics of karst, recognize the distribution of karst development focused on the tectonic fault zone and buried hill near the top surface weathering crust, and validated by karst properties prediction profile and production well, drilling mud leak rate alignment is higher. It is further shown that the mud loss site is mainly the development location of dissolution pores. Finally, the eastern part of Yanling tectonic belt and the western part of Chuan tectonic belt are the most favorable reservoir development areas.

For the microseismic events induced by water inrush in underground coal mine, this paper carried out the numerical calculation and simulation study of water-induced microseismic signals, and on this basis carried out the development of a monitoring system suitable for underground coal mine. In the research process, the numerical calculation of the characteristic parameters such as energy, amplitude and frequency of the vibration signal generated by the induced water disaster was realized. Combined with the internal structure of the moving coil detector, the electrical signal response calculation of the vibration signal was completed, and the electrical characteristic waveforms of different vibration signals were obtained theoretically. Based on the theoretical research results, the whole microseismic monitoring system is designed and developed. In this paper, the self-developed monitoring system is used to realize the effective capture of underground microseismic signals. The detection results show that the system can monitor the vibration signals generated by the disaster inducing water in real time, and provide technical support for the mine water disaster warning.

The construction of marine space-time frame networks is an important part of ocean exploitation and utilization. Accurate measurement of the coordinates of seafloor datum points is the basis of the network. The temporal and spatial variation of sound speed in seawater will cause refraction of sound waves. Effectively eliminating the refraction effect of sound waves is crucial to improving the accuracy of underwater acoustic positioning. However, sound ray correction requires a trade-off between ranging accuracy and computational efficiency. In this paper, we proposed a method to improve the positioning accuracy of the equivalent sound speed profile method, and the simulation experiment results show that the algorithm can accurately calculate the coordinates of the seafloor datum point with high computational efficiency.

Achieving a smooth and high-fidelity visualization of 3D city scenes in virtual geographic environment is a key technology for digital twins. Geographical information system (GIS) researchers have proposed many methods for loading and visualizing such city-scale scenes. However, if the scale of the scene is too large, the current methods cannot achieve a smooth and high-fidelity visualization of 3D city scenes. In this study, we propose an efficient visualization method for 3D city scenes through calculating potential visible sets (PVS) and hierarchical levels of detail (HLOD). We use adaptive quadtree partition to divide the 3D city scenes. Then, we calculate the potential visible set of each leaf node using an improved ray-casting algorithm, and visual perception parameters model were evaluated by employing the screen error, which was used to constrain the dynamic scheduling of HLOD. We implemented our method on Unreal Engine (UE4) and designed multiple sets of controlled experiments to illustrate the advantages of our method in terms of visualization efficiency and rendering effect. The results show that our method can significantly improve the visualization fluency, maintain good global and local detail.

Aiming at the problems of long integration time, data redundancy, and the inability to share across systems of land use data in villages and towns, through analysis of land use status classification and survey data characteristics, a multi-level semantic constraint village land resource survey data is proposed using integration and service publishing methods. Firstly, in terms of time, space, attributes and other characteristics, the land resource survey data is described uniformly, and the relationship between the land use situation of the village and town and the data type is constructed. Secondly, starting from the three-level constraints of task semantic constraints, data semantic constraints and temporal semantic constraints, a multi-level semantic constraint land resource survey data integration framework is built. Finally, the virtual database method is used to calculate and integrate matched publishing rules through matching rules to form a publishing rule instance set, which realizes the integration and service publishing of rural land resource survey data under multi-level semantic constraints. Practical analysis shows that this method can effectively improve the efficiency of the integration and release of rural land resource survey data, and provide a real-time and effective basis for accurate analysis of the current situation of rural land use.

Based on the observation daily record and data of cavern strain in Xibozi seismic station of Yanqing, the interference factors and its characteristics of cavern strain are analyzed. It provides the basis for precursor abnormality identification of this cavern strain in the future.

Aiming at the problems of complex scene objects, lack of correlation between them, close coupling between modelling knowledge and modelling process, and low flexibility of scene configuration in dam safety monitoring scene construction, this paper proposes a multi-level semantic constraint method for dam safety monitoring scene construction. First, based on the analysis of the characteristics of dam safety monitoring scene modelling operations and modelling knowledge expression, modelling knowledge is extracted from spatial pose semantic constraints, spatial layout semantic constraints and component combination semantic constraints, and modelling knowledge and modelling operations are stored in a parametric way. Then, after the multi-level semantic constraint rules constrain and guide the way the scenario objects are combined, the dam safety monitoring scenario is generated. A prototype dam safety monitoring system is developed based on this approach to validate the effectiveness of the method.

The gas reservoir, which is located in the Anda-Wangjiatun area of the Xujiaweizi fault depression in north of Songliao basin, is one of the huge volcanic reservoirs in deep layer. The reservoir displays a series of characteristics, such as complicated lithology and lithofacies, strong heterogeneity, complicated seismic response and so on. The favorable gas-bearing reservoir prediction is much more difficult. With use of time-frequency analysis technique, which is based on wavelet change, combined with the high and low frequency information in seismic data, the gas-bearing property prediction has been carried out in the volcanic reservoirs of Yingcheng formation, which developed in Anda-Wangjiatun area. Through comparison and analysis of all kinds of geophysical parameters’ response to volcanic reservoir which are rich in natural gas, it indicates that these parameters, such as attenuation gradient, energy percentage, specified the ratio of energy to the corresponding frequency and show a better result. The gas-bearing prediction result in this area is consistent with that of wells. The coincidence rate even reaches to 75%. Three favorable gas-bearing areas have been pointed out, which provides a basis for volcanic gas reservoir exploration.

3D geological visualization has a broad application prospect and value in the study of Geological Structure (GS). Therefore, this paper studies the visualization of GS based on UAV and GIS. The 3D GS model simplification and multi-resolution model are analyzed, and the GS model refinement and data dynamic scheduling are discussed. This paper briefly introduces UAV and GIS, and applies them to GS visualization for research and analysis. Based on seismic data multi attribute fusion technology, the hydrogeological database structure and 3D visualization system are processed, and then the GS visualization is effectively analyzed in detail.

In order to realise the transmission of underwater data, this paper proposes an improved 2FSK modulation method based on field programmable gate array (FPGA) and PWM modulation technique implementation. The method generates a sinusoidal signal by writing a Verilog language on vivado software, and performs conventional 2FSK modulation on this signal, and finally the signal after 2FSK modulation is PWM modulated, which enhances the immunity of the signal to interference.

To enhance the interferogram filtering effect in Interferometric Synthetic Aperture Radar (InSAR), an improved Goldstein filter is proposed, which improves the filter block size, smoothing operator, and filtering parameters. In addition, the filter window size can be adjusted adaptively according to the noise standard deviation. Meanwhile, the new smoothing operator is designed to achieve maximized noise suppression while ensuring resolution. Finally, new filtering parameter is constructed by considering the effects of interferogram amplitude and noise standard deviation. To evaluate the effectiveness and practicality of the proposed algorithm, simulated and real data are used for verification, respectively. The results show that the proposed algorithm significantly reduces the interferometric phase noise, enhances the filtering effect, and has obvious advantages in maintaining phase continuity.

At 2:00 p.m. on August 6, 2019, the Leidashi landslide was reactivated near Danjing town, Gaoxin east zone, Chengdu, Sichuan province, China. From the data of geological exploration, multi-temporal satellite images, and unmanned aerial vehicle images, it is found that the Leidashi landslide, the slow-tilting thrust-type rock landslide on the impact of rainfall and human engineering activities, is the revival of the 2010 landslide. In this study, the discrete element method software MatDEM was used to simulate the sliding process of the Leidashi landslide in 2019. The results of starting and sliding process and local deformation characteristics of the landslide are obtained. Meanwhile, the displacement field, heat field, and energy change in the sliding process of the landslide are analyzed. Comparing the landslide morphology before and after changing the friction coefficient of the soft layer and slope elevation and adding an "anti-slide pile group" in the process of numerical simulation, it is believed that the deformation and failure of the Leidashi landslide are mainly caused by two factors: artificial excavation and torrential rainfall.

The lower ordovician Yingying third member in Gucheng area developed the carbonate distal steepening gentle slope platform depositional system. There are four types of facies: inner gentle slope, inner shallow gentle slope, outer shallow gentle slope and deep gentle slope. On the basis of sequence stratigraphic correlation between single well and consecutive well, the third member of ordovician in the ancient city area is divided into three quaternary sequences from bottom to top, The interior of each quaternary sequence is composed of high-level and transgression system domains, respectively. The lithofacies paleogeographic layout of the three quaternary sequence periods of the ordovician third member in the Gucheng area is compiled. This paper describes the features of lithofacies paleogeography of the quaternary sequence of Yingsan and provides a foundation for the further exploration of oil and gas of the study area.

In order to clarify the seismic sedimentology interpretation method under coal seam interference, time-frequency analysis technology was used to analyze the seismic reflection frequency of the coal seam. Combined with seismic filtering, the seismic interference of the coal seam is suppressed through the sedimentological interpretation of stratigraphic slices with different dominant frequencies. The spatial distribution characteristics of the sedimentary system are determined by comprehensive mixed display technology. The results show that the delta plain subfacies are developed, which are composed of four distributary channels, crevasse splays, and swamps in the near north-south direction, and a large-scale channel migration zone is formed in the southwest.

Automatic classification of point clouds in urban scenes has great application requirements. Different feature dimensions and feature combinations have different effects on the result of point cloud classification. This paper summarizes a series of point cloud feature description methods from multiple perspectives, extracts 22-dimensional point cloud feature vectors, then constructs different combinations of geometric features, point color features and neighborhood color features, and discusses the classification effect of different feature combinations. In order to verify the effectiveness of the feature combination methods, Random Forest (RF) and Extreme Gradient Boosting (XGBoost) machine learning classification models are used for experimental verification and comparative analysis. The results show that the two classification models have good robustness. When only geometric features are used for classification, the F1 score of the two methods is only about 52%, while the overall classification precision of the two methods is improved by more than 20% after combining color features.

The study and analysis of dynamic evolution in paleoclimate research links is an important element in the study of paleoclimate dynamics. The use of temporal GIS can simulate the natural movements and human activities in the paleolandscapes, so as to analyze the frequency and amplitude of changes in different seasons, different time periods and the same moment in each time point under different weather conditions. The analysis of dynamic evolution in the paleoclimate research link can reveal the variation patterns of frequency and amplitude in different seasons and at different time points. The study of historical evolution analysis can help us understand the dynamic evolution of paleoclimate, and this can also play a guiding role in paleoclimate research, so as to provide some reference for future historical development and human activities.

Under the Dual Carbon Goal, geothermal energy is a renewable energy source that can effectively reduce carbon emissions. However, the construction process of geothermal wells also causes carbon emissions. Therefore, geothermal energy projects need a mathematical model based on carbon emission reduction to determine the depth of a geothermal well. This model can quantify the carbon emission reductions caused by different depths of geothermal wells from the perspective of drilling rig oil consumption, drilling well water consumption, geothermal power generation, associated natural gas, etc., and can compare the most suitable depth of geothermal wells by calculation. This model reveals that geothermal gradients are a crucial parameter in determining the depth of geothermal wells based on carbon emissions. For a given scenario, a 3000 m geothermal well can achieve positive net carbon emission reduction in a shorter period (4436.6 hours) than a 2000 m geothermal well.

Because of the strong random noise in surface microseismic data, the utility of the data may be decreased. Therefore, it is necessary to use some efficient techniques for denoising the microseismic data. Shearlet transform is a new multiscale transform which can adaptively capture the geometrical characteristic of multidimensional signals and represent signals containing edges optimally. Most traditional nonlinear thresholding methods based on Shearlet transform denoising presume that shearlet coefficients are independent. However, the shearlet coefficients of surface microseismic signals have significant dependencies. For this reason, these denoising schemes suppress too many coefficients that might contain useful microseismic signals information. We proposed a new multivariate model based on shearlet transform to improve this problem. This new multivariate model can not only adaptively capture the inter-scale dependency according to the anisotropic property of variances of shearlet coefficients in different sub-bands, but also take the inter-direction dependency into account. We perform tests on synthetic and field desert seismic data and the denoising results show that the proposed method can effectively preserve effective signals and remove random noise.

For the evaluation of the implementation effect of the experimental project that uses Eichhornia crassipes for removal of pollutants in Caohai Lake since 2011, the scale and growth states of Eichhornia crassipes planting in the Caohai Lake were monitored using high-frequency remote sensing data. By analyzing Sentinel-2 data in three annual periods from 2017 to 2019, the distribution area and NDVI value of time series of the Eichhornia crassipes were extracted. The results showed that: on the general trend, the NDVI value in the Eichhornia crassipes region was basically consistent with the area change, and after the growth reached its peak, artificial salvage was started to prevent secondary pollution; however, the Eichhornia crassipes was not planted during the period from late February to late July when it had growth potential. In late November to late February, the Eichhornia crassipes was in recession, and the Eichhornia crassipes that could not be salvaged in time must cause secondary pollution to the water body in this area. According to the growth cycle of the Eichhornia crassipes, the Eichhornia crassipes should be planted and salvaged in a timely manner, which is expected to achieve more ideal pollution control effect.

Drought affects food security and social production and life seriously, so it’s crucial to obtain drought information timely. This paper conducted a comparative study of drought indices in Xinjiang. We used MOD13A2 and MOD11A2 to calculate AVI and TVDI, used daily precipitation and temperature data from 41 meteorological stations to calculate PDSI, compared their ability and analyzed their consistency, finally. The results showed that: (1) The average value of AVI and TVDI was consistent on inter-annual drought response and the correlation reached -0.54. (2) AVI responded poorly at the early stage of vegetation growth; TVDI responded poorly to drought events in the northern Xinjiang; and remote sensing indices possessed lag; PDSI responded to drought events not only in terms of magnitude of values, but also in terms of dynamic decline of values. (3) AVI-TVDI spatial correlation was poor; TVDI lag I / II -PDSI correlation was -0.28 and -0.31 in drought years, the consistency in the eastern part was larger than the western part, the northern part was larger than the southern part. The consistency of AVI-PDSI was poorer than TVDI-PDSI. The results aimed to provide help for drought assessment and new ideas to enrich the means of drought monitoring in Xinjiang.

In InSAR interferogram processing, phase filtering is an extremely critical step, and traditional filtering methods have defects such as difficulty in identifying edge information, and easy confusion of noise and mutation signals. In order to solve the above problems, we propose a filtering method based on the Steerable Pyramid. The results of applying this method to the interferogram are compared with five traditional filtering methods, and we found that it not only performs well in the visual result of the filtered interferogram, but also shows superiority in various filtering quantitative index evaluation.

Bridge engineering construction is a kind of engineering construction with high risk. Only by building scientific and reasonable safety assessment and monitoring measures can the construction safety be improved and the risk be controlled. Based on the construction of a large cantilever steel truss traffic bridge project, this paper introduces the application of three-dimensional laser scanning technology in the monitoring of bridge transverse construction from the aspects of measuring point layout, monitoring scheme, scanning measurement and data analysis and processing. This paper introduces the bridge construction safety monitoring technology, as well as the content and steps of construction safety monitoring, and summarizes the measures of bridge construction safety assessment monitoring.

The reason for the existence of small targets and sub-pixel targets in hyperspectral images is that the low spatial resolution leads to the phenomenon of mixed pixels, which makes some targets exist in the form of small targets and sub-pixel targets in the image. Moreover, the traditional anomaly detection algorithm introduces the regularization parameter into hyperspectral anomaly detection, but the regularization parameter needs to be manually set and adjusted. However, since the anomaly detection is carried out without prior information, in real situations, it is difficult to obtain the prior information of the target, so the regularization parameter may be difficult to determine. In addition, considering that traditional algorithms are not sensitive to small target pixel targets, a hyperspectral anomaly detection network based on self encoder is designed for hyperspectral images with small targets and sub-pixel targets.

Hyperspectral images contain large amounts of information and have high spectral resolution. The performance of hyperspectral images in describing and distinguishing target categories has been greatly improved. With the development of unmanned aerial vehicles (UAVs), a lightweight, adaptable and low-cost way has greatly expanded the application field of hyperspectral images. This paper proposes a spatial spectrum attention mechanism based on Deep Deterministic Policy Gradient (DDPG) for hyperspectral classification. This attention mechanism is combined with 3DCNN to assign different weights to different channels in the classification process. The classification accuracy is improved by activating the useful spatial spectrum information and suppressing the useless spatial spectrum information in the hyperspectral image. A large number of experiments have been carried out to prove the effectiveness of the structure.

Environment monitoring, traffic management, and autonomous driving systems are numerous industries that use semantic segmentation of remote observation sensing data. High-resolution remote sensing satellite images have a large application area, include a lot of object information, and are challenging to extract information features due to the rapid growth of remote sensing technology. Motivated by the recent success of Convolutional Neural Network (CNN) for image semantic segmentation, this paper explores the power of CNN feature learning ability for remote sensing images, and serviceable image segmentation results are provided. A conventional Encoder-Decoder segmentation network is proposed for remotely sensed data. To further boost the performance of semantic segmentation, a variety of segmentation losses are explained and utilized. Evaluation metrics are introduced, and a series of experiments are conducted. The results demonstrate our proposed loss-modified CNN model's competitive performance against other methods.

The identification and characterization of the thermocline play an important role in the fishery economy and military strategic planning. Because the thermocline is close to the sea surface, it is most easily influenced by wind, rain and solar heat, etc. So, the range of the thermocline characterizations are widely distributed. This paper adopts a kind of combinatorial algorithm for thermocline identification. So, the identification of thermocline is not limited by the lowest critical value and gets rid of the artificial subjective factors. At the same time, the discontinuity caused by the traditional vertical method in the shallow and deep-water area of the shelf can be eliminated. The result shows that the combination method of these two methods can get thermoclines that are thicker and more reasonable.

In order to improve the spatial resolution of hyperspectral images, this paper improves the network structure of SSRNET algorithm and analyzes the functions of various parts of the network model. A comparative experimental analysis was carried out on the urban dataset. Experimental results show that the improved method in this paper has improved to a certain extent in both subjective and objective evaluation.

In view of the fact that the source and target parts are contained in the reference target, the existing model expression is inconsistent with the actual cognition. A quantitative detail direction relationship expression model is proposed to establish a two-layer grid direction relationship matrix for spatial objects. The expression ability of the two-layer grid direction relation matrix is tested and analysed. Through the comparison and analysis with the calculation results of the direction relation binary in literature [7] and the grid direction relation matrix in literature [8], the recognition ability of the two-layer grid direction relation matrix to displacement and rotation is verified. Finally, it is proved that the method in this paper is effective in the calculation of spatial direction similarity, extended cartographic comprehensive evaluation, data matching and fusion applicability of work tasks such as change detection and update.

The rich band information of hyperspectral images provides data support for accurate classification, but the information redundancy also brings bad effects to image classification. In order to improve the classification accuracy, this paper takes Liutang Town of Guilin City, Guangxi Province as the research area, and "OVS-1A/B" hyperspectral image as the data source. Based on two different dimensionality reduction methods, Principal Components Analysis (PCA) and CfsSubsetEval GreedyStepwise (CG), combined with Support Vector Machine (SVM), classifiers are used to classify hyperspectral images based on four multi-feature fusion schemes, and their classification accuracy is compared. The results show that: (1) In the two classification schemes with different dimensionality reduction methods, the classification accuracy of ground objects increases continuously with the increase of the types of fused features. (2) In the same combination scheme, the classification accuracy obtained by CG feature selection is higher than that obtained by PCA feature extraction. (3) The classification accuracy of scheme 4 using CG method for feature selection was the highest, the overall classification accuracy was 97.12%, and the Kappa coefficient was 0.9556.

Vehicle-mounted measurement system can obtain massive point cloud and panoramic image through 3D laser scanner and panoramic camera. How to organize these data effectively, fast matching and fusion visualization have become urgent problems to be solved at present. Aiming at the above problems, according to the characteristics of road point cloud, this paper uses octree construction algorithm based on spatial partition to carry out the distributed construction of multi-resolution point cloud and designs and obtains an irregular octree data organization structure of point cloud. ThreeJS is used to design point cloud scheduling mechanism to achieve rapid visualization of massive point cloud and panorama. A visual process of point cloud panoramic matching and fusion based on geometric features of spatial objects is proposed. Experiments show that the technical process in this paper can realize the stable and rapid fusion visualization of vehicle-mounted point cloud and panoramic data at the Web end, and the fusion process is simple, the fusion precision is high, and it has high practical value.

In order to meet the needs of landslide stability evaluation, we introduced the 3D laser scanning technology into the landslide physical model test slope surface deformation monitoring to obtain the deformation of the physical model of landslide under the action of external rain factor using the DoD method, the C2C method and the M3C2 method. The results show that: 1) All three methods can obtain the overall deformation and displacement of the landslide surface, avoiding the limitations of traditional point-based monitoring methods; 2) the DoD method requires the prior construction of the DEM using point clouds, and the process is more complicated than the C2C and M3C2 methods. The DoD method can only obtain the deformation in the vertical direction, which does not reflect the true 3D deformation of the landslide surface; 3) The C2C and M3C2 methods directly compare the point clouds without constructing the DEM, which has the advantages of fast and efficient. But the distances calculated by the C2C method do not necessarily represent the true deformations occurring on the surface. The M3C2 method computes the 3D distance between the reference and comparison point clouds using the point cloud normal vectors of the best planes constructed at multiple scales, which further optimizes the assertion criterion of the C2C method for extracting approximate homonymous points and can distinguish the collapse and accumulation zones of landslides based on the difference in color values, making the monitoring results more reliable. The deformation information of the landslide surface extracted from the laser point cloud provides a reliable information and scientific basis for the correct assessment of landslide stability.

Side-scan sonar is widely used in maritime search and rescue, submarine geological exploration, target detection and identification, etc. with its long range of action, comprehensive coverage, and large amount of collected data. Image denoising is the premise of the correct interpretation of side-scan sonar images. In this paper, a new weighted kernel function is proposed on the basis of the original Non-Local Means (NLM) denoising algorithm, which leads to an improved non-local mean denoising algorithm. By experimenting with Gaussian white noise images with different noise levels, it appears that the efficiency of this algorithm is superior to that of the NLM algorithm, and the denoising performance is significantly enhanced compared with the NLM and Fast Non-Local Means (FNLM) algorithm, especially for images under solid noise.

LiDAR point clouds and optical images are two widely used geospatial data. The fusion of LiDAR point clouds and optical images can take full advantage of these two types of data. Since LiDAR point clouds and optical images vary in dimension (3D vs. 2D), spectral (near-infrared vs. visible) and data acquisition principles ( time of flight vs. perspective projection), the fusion of LiDAR point clouds and optical images is challenging. This paper deals with the registration of LiDAR point clouds and optical images. Feature point-based matching methods with different feature detector and descriptor combinations are evaluated, and find that different combinations affect the matching performance greatly. Among the evaluated 112 combinations, FAST-SIFT and AGAST-SIFT combinations have the best matching performance. Besides, to remove the large amount mismatches in the matching results, the paper proposed a template and RANSAC based mismatch removal algorithm. The experimental results show that the proposed mismatch removal algorithm greatly improved the matching success rate and the correct matching rate.

The exploration and development of P oilfield in Melut basin South Sudan is early. The water-flooded layers are caused by bottom water coning, edge water propulsion and injected water. At present, the oilfield has entered the middle and high water cut stage. In order to stabilize oil and control water and enhance oil recovery, it is necessary to carry out the evaluation of water flooded layer. Based on the flow unit index theory, the reservoir is divided into three types by cumulative frequency method using experimental data of mercury injection and physical properties. Then the water flooding level is qualitatively and quantitatively evaluated by comprehensively using logging and masterlog data. According to the sensitivity optimization of logging and masterlog, the spontaneous potential curve, resistivity curve and total gas curve are selected to identify the water-flooded layers. Based on the result of petrophysical facies, the original water saturation, remaining oil saturation and irreducible water saturation of different reservoir types are finely evaluated. Then the water cut content is determined accurately. The water-flooded layers are quantitatively evaluated. The method is used in the production practice, the result shows that the accurate identification rate of water-flooded layer is 90.9%. The result of calculation match well with the core analysis data. The quantitative evaluation of water-flooded layers match well with the production. The comprehensive evaluation method of water-flooded layer meet the needs of oilfield development and production. This method organically combines the petrophysical facies with the comprehensive evaluation of water flooded layer. The water flooded layer is finely evaluated; this result provides an important basis for stabilizing oil, controlling water and improving oil recovery in old oilfields.

The extraction of roads from satellite images is a necessary step for urban planning, intelligent transportation, etc. We propose Lite-HRNet-OCR, a lightweight and efficient CNN structure for road segmentation. The network of Lite-HRNet-OCR begins with a lightweight Lite-HRNet backbone that learns the weights of all channels and resolutions. The weights serve as the channel for information exchange across channels and resolutions. The multi-resolution output of the lightweight backbone is input to OCRNet, which organizes contextual pixels into object regions and exploits the relationships between pixels and object regions to augment the representation of their pixels. Two loss functions, cross-entropy loss and Tversky loss, are used to solve the problem of sample imbalance. Experimental results show that our method achieves competitive performance on the public CHN6-CUG road dataset. Specifically, the Lite-HRNet-OCR achieves 64.39% Mean IOU and 96.52% F1 with 2.9 MParams and 29.4 GFLOPs.

Synthetic aperture radar (SAR) image processing using different filters can produce different results for different parts of the image. Thus, specific applications require specific filters. In this study, we evaluated and compared the performance of several widely used filters, including the mean, median, Lee, Frost, Kuan, enhanced Lee, enhanced Frost, and gamma maximum a posteriori (MAP) filters, in suppressing speckle noise in GF-3 FSII images. We applied these filters to different images to determine the optimal filter and processing window size for different image parts, including strongly textured, homogeneous, and fringe areas. We found that the optimal filter for the strongly textured area was the Kuan filter (window size, 7 × 7). The optimal filters for homogeneous areas were the Kuan (7 × 7, 5 × 5, and 3 × 3), Lee (7 × 7), and mean filters (7 × 7). The optimal filters for fringe areas were the enhanced Lee (7 × 7 and 5 × 5) and enhanced Frost filters (7 × 7).

Monomerization is one of the important aspects of tilt photography models in practical applications, but the monomerized models lack rich semantic information. In this paper, a multi-scale coupling mechanism of semantic information and monomerization model is designed. Firstly, semantic consistency processing is performed on the crowd source data to obtain rich semantic information. Then, a semantic information classification table is constructed, and two-dimensional semantic information based on CityGML level of detail fusion rules are designed to join the multi-scale functional areas based on CityGML. Finally, multi-scale semantic information is assigned to the monolithic model based on the inclusion relationship between functional areas and buildings. The results of this paper show that the proposed method can integrate the query of the structure and location of 3D monolithic models as well as their rich semantic information, thus providing data support for the application of 2D and 3D systems for data analysis.

This paper summarizes the normal dynamic characteristics of water temperature in well Lu32 and the morphological characteristics of natural environmental disturbance, and explores the abnormal change characteristics of water temperature in well Lu32. Based on the water temperature observation and hydrochemical observation data carried out in the same well in recent years, combined with the comparison of the occurrence time and space of moderate earthquakes within the near field of 200 km, and the abnormal period of water temperature and water radon, it is found that there is a good correspondence between the two, and the comprehensive analysis of the existing data indicates that the water temperature in well Lu32 has the tectonic conditions of additional geothermal field anomalies, which can be used as one of the short-term and medium discriminative indicators of near-field moderate earthquake precursive anomalies and far-field large earthquake precursive anomalies.

It is important to access the disaster distribution for earthquake disaster assessment and emergency command. Quick access to post-seismic disaster information is necessary for emergency command. Based on the post-earthquake landslide image dataset, a residual network model is used to identify disaster information on landslide image data using migration learning techniques. The research results show that the use of deep learning methods can better analyse landslide images, with recognition accuracy reaching over 93%, and can effectively extract disaster information from the images, providing technical support for the automatic analysis of emergency disaster data.

For the complex channel system with multi seismic facies, it is difficult to accurately identify the channel by conventional seismic attributes. In this paper, the channel identification method was applied based on spectrum decomposition and growing neural gas network (GNG). Firstly, the most sensitive frequency components with channel sand were selected through spectrum decomposition and as the input of the growth neural gas network. Then, GNG training was carried out to obtain the target neuron set. Finally, the GNG trained neural network can be used to estimate probabilities for each facies and a facies map is generated with related probability. The method was used in in Jiangqiao area of Songliao Basin, China. The results show that there are three types of channel facies (red, green and blue) in Saertu oil layer. This method can be used as an effective tool for identification of complex channels.

A flight test of airborne gravimetry was conducted in Antarctic based on a strapdown inertial airborne gravimeter called SGA02 during the 36th Chinese National Antarctic Research Expedition. This test was aimed at evaluating accuracy of the system, collecting gravity data, and measuring the system’s practicability. Airborne gravimetry is one of the most significant methods in gathering gravity data with mGal-level precision over several kilometers of spatial resolution, which is a kind of strategic resource in the field of geophysics and many other important areas. A lack of southern hemisphere research and geophysical data exists due to the difficult accessibility of the Antarctic region. Since geophysical data of this region can hardly be acquired, the test performed in Antarctic can, in some degree, make up the gap between the gravity data and other geographic features. SGA02 is independently developed by NUDT (National University of Defense Technology), based on integrated SINS/DGPS system (strapdown inertial navigation system and differential global positioning system). In the flight test of the airborne gravimetry in the Antarctic region, SGA02 performed with 1.5 mGal precision (160s of filtering). The constitution of the whole system is introduced and the theory of the strapdown inertial airborne gravimeter is briefly discussed. The data gained from this test is displayed and analyzed, followed by the discussion and evaluation of the results.

The eight-sensor planar cross magnetic gradient tensor detection system can detect both first-order and second-order magnetic gradient tensors. Compared with the traditional four-sensor and five-sensor planar cross-shaped magnetic gradient tensor system, the measured tensor data is more stable. However, the alignment error correction has not been studied in the eight-sensor planar cross magnetic gradient tensor detection system. Aiming at this problem, alignment error correction is carried out in this paper. The simulation results show that the proposed method can effectively correct the alignment error.

Grain for Green Program (GGP) is an ecological restoration and compensation project carried out by China to protect and improve the environment, which can not only ensure the economic income of peasant households in areas where farmland is returned to forests, but also effectively reduce natural disasters. Taking Chaoyang City as an example, the land cover information of the area is extracted based on the Landsat images from 1985 to 2020, and the changes of the barren hills, cultivated land, forest and grassland during the 36-year GGP are analyzed. The results show that in the past 36 years, the afforestation area of barren hills in Chaoyang has reached 2745.5 km², and the afforestation area of cultivated land was about 933 km², GGP has achieved remarkable results. The abandoned lands and reforestation areas are mainly distributed in areas with slope >25° and altitude area over 500 m, and the changes of the two are mainly manifested in the internal structural conversion of the proportion of cultivated land with an altitude of less than 500 m and slope < 25°, which is basically in line with the requirements of the policy of GGP.

The identification of rock slices is the basis of geological research. However, due to the complex mineral composition and structure of volcanic rocks, microscopic analysis is difficult and requires a lot of time for professionals to complete. This paper uses deep learning technology, which has emerged in recent years, to explore a new method of intelligent rock slice identification to achieve automatic identification of volcanic rock slices. The deep residual shrinkage neural network model is used to study the intelligent recognition of volcanic rock slice images. The study investigated 11 basic types of volcanic rock and collected 12,000 high-definition images of rock slices using an electron polarized light microscope. Each image was processed via histogram equalization and image sharpening; data were expanded by random cropping; and the expanded image data were used as a training set to train the network. When the number of model layers reached 50 layers, the best accuracy rate was achieved. After a series of optimizations and improvements of the network model type, the accuracy rate of the test set classification results exceeded 92%.

Northwest Pacific Ocean is an important area for resource exploration and abyssal exploration. In order to ensure the safety of manned submersible operations in this region, this paper adopts synoptic analysis method and selects ERA_interim data and optimal path set data of tropical cyclone to study the spatial distribution and temporal variation of wind and wave elements in this region. According to the operation requirements of level-four sea state distribution and level-five sea state recovery for Jiaolong manned submersible, the sea state from June to August is relatively good, which is more suitable for submersible operation. However, attention should be paid to the generation and movement path of tropical cyclone, and favorable sea state conditions should be selected for operation at other times.

At present, when the ultra-long linear 3D model is loaded in the 3D scene, it is difficult to adapt the ultra-long line model to the terrain model due to the influence of projection distortion. At present, some solutions mainly focus on the model rendering level, without considering the necessity of linear engineering construction, which is hard to encounter the requirements of 3D visualization in the meadow of linear engineering management. Therefore, this paper conducts research from two aspects of projection distortion calculation and analysis, linear engineering independent coordinate system establishment and so on, and puts forward the method of line model loading based on linear independent coordinate system piecewise technology. Firstly, the linear engineering projection distortion is calculated and analyzed. Then, based on the requirements of linear engineering construction accuracy, multiple independent coordinate systems that meet the accuracy requirements are constructed according to the characteristics of linear engineering. Finally, the 3D model of the ultra-long line is piecewise loaded, and the influence caused by the distortion of the projection length is apportioned through each independent coordinate system. The practical consequence indicate that this method can virtually optimize the difficult problem of model and terrain acclimatization caused by projection distortion, and meet the rendering requirements of linear engineering 3D models.

Detailed sedimentary microfacies research is important to determine favorable target areas for oil and gas exploration. In order to clarify the sedimentary microfacies characteristics of Chang 7 in the Ordos Basin, the types and distribution of Chang 7 sedimentary microfacies were studied using logging, core, analysis and testing data. The results show that delta facies is developed in the Chang 7, mainly in the delta front subfacies, including distributary channel, natural levee, interdistributary bay and mouth bar. The underwater distributary channel sand body, as the skeleton sand body, is relatively well developed and is the main place for oil and gas accumulation. The thick fine sediments in Chang 7 sedimentary period are mainly mudstone and silty mudstone. Multi channel sand bodies and extremely thick fine-grained rocks form a high-quality source reservoir cap rock association.

This paper introduces the basic composition of the Global Ocean Stereoscopic Observation Network and expounds the necessity and importance of observation data processing and quality control. Taking data of ocean stations and shore-based radar stations as examples, the paper elaborates the data processing and quality control technology, including processing principles, objects, contents, methods and parameters, and focuses on 16 quality control methods including range test, missing test and statistical characteristic test. The quality control for the sea-level data of the ocean station was carried out, and the error detection rate was 1.8%. The data after quality control was more authentic and reliable. Then, it analyses the quality control software and sharing service platform of the observation data. Lastly, it points out that further studies on homogenization test and correction of ocean station data would make the results more reasonable and scientific.

Traditional villages are valuable material, wealth left behind in urbanization construction and have rich cultural resources. Traditional villages are both symbols of excellent traditional history and culture, and cultural carriers of inherited folk customs. In the context of rural revitalization in the new era, it is of great practical significance to analyze the spatial distribution characteristics of traditional villages and their influencing factors to provide a theoretical basis for the protection of traditional villages and the development of traditional villages in Chongqing. Using geometric analysis and ArcGIS10.7 spatial analysis tools, the spatial distribution characteristics and influencing factors of 110 national traditional villages in Chongqing were explored. The results show that the spatial distribution type of traditional villages in Chongqing is cohesive, with the nearest neighbor ratio R=0.66, which is less than 1. From the municipal perspective, the geographic concentration index of traditional villages is G=37.59, with Youyang, Xiushan, Pengshui, and Shizhu counties as the key concentration areas. From the overall spatial perspective, the imbalance index of traditional villages S=0.79, and the spatial Gini coefficient G=0.617, the traditional villages are unevenly distributed in the four major geographic divisions of Chongqing, with more than 70% of traditional villages concentrated in southeastern Chongqing. In addition, the occupancy rates of western Chongqing, eastern Chongqing and the main urban area are 18.18%, 9.09% and 1.82%, respectively, with serious polarization. Among the many factors affecting the distribution of traditional villages in Chongqing, natural topography, socio-economic, historical and cultural and documentary policy factors dominate. Among the many factors affecting the distribution of traditional villages in Chongqing, natural topography, socio-economic, historical and cultural factors and documentary policies dominate and constitute the unique spatial distribution of traditional villages in Chongqing.

Mineral resources are an important pillar of China's social and economic development. However, abandoned mines have brought serious problems such as occupying cultivated land, destroying ecology, polluting the environment and causing hidden dangers of geological safety. This paper takes an abandoned mine in Yantai City as an example. In the historical mining process of the study area, the geomorphic landscape was destroyed, soil erosion was serious, the function of land resources was lost, and the slope of mining pit was steep, and local rock mass was broken. This paper introduces the problems of geological environment in this area and analyzes the reasons for the destruction of the geological environment of the mine in this region, to restore farmland, grassland, woodland for the purpose, analyzes the status quo of the mining pit, residual mound and material yard in the region, and puts forward the corresponding restoration scheme for each failure unit: blasting the residual mound, backfilling the mining pit, and backfilling the slope and site leveling.

In the research of 3D linear scene construction, the scene construction method based on time or data is difficult to be event-centered, and there is the problem of data redundancy in the construction process. The event driver drives the change in the impact factor of an event during transaction dispatch by calling the corresponding event execution function to finally solve the problem and prevent the transaction from piling up. Based on linear reference model and event-driven, this paper proposes an event-driven model under linear reference system, and then realizes the construction of 3D linear scene driven by event model. Firstly, the basic event model is established by combining the characteristics of linear scene data band distribution. According to the types and logical relations of events in linear scene, the events are classified and the event coding rules under linear reference system are designed to make the event change drive the model attribute change in linear scene. The unified mapping relationship between events and objects is established, and the linear scene construction rules based on event-driven model are designed. Finally, experiments on the construction of dynamic highway scenes were conducted, and the construction of 3D linear scenes based on the event-driven model was realized, and the smoothness of scene rendering was significantly improved compared with the traditional method. The feasibility of efficient construction of 3D linear scenes based on event-driven model is verified. It is of great significance for highway emergency security.

Research on carbon cycle and carbon balance has always been one of the hot spots of global attention, and forest ecosystems, as the largest carbon reservoir in terrestrial ecosystems, are of great importance to the global carbon cycle. As an important component of forest ecosystems, the study and analysis of carbon fluxes in coniferous forest ecosystems is key to understanding the impact of global climate change on the carbon cycle. However, traditional carbon flux data management suffers from lack of data and untimely sharing. Therefore, based on the observation of ecosystem carbon fluxes, this paper designs and studies a carbon flux data management information system to unify the data management, improve the processing and analysis of carbon fluxes, provide data support and services for the carbon cycle of coniferous forest ecosystems, and also help predict future climate change trends, which is the key to cope with global changes.

This paper investigates the deposit geology, ore body characteristics, and ore characteristics of the ornamental granite mine located in Xinyi, Guangdong Province. According to the research, ornamental granite is mined in the Silurian period at an elevation of +943m~+423m. The ore is composed primarily of gneissic granite, and the principal constituent minerals are potassium feldspar, plagioclase, quartz, and biotite; potassium feldspar accounts for 25%~30%, plagioclase for 30%~35%, quartz for 30%~35%, and biotite for 7%~9%; the principal chemical components are SiO₂ and Al₂O₃, accounting for 68.15% and 13.29%, respectively. The density is 2.61~2.69g/cm³, the water absorption rate is 0.09~0.19%, the abrasion resistance is 66~119cm^-3, the bending resistance is 13.5~18.3MPa, the gloss is 87.0%~87.7%, the dry compressive strength is 82.5~141.0MPa, the saturated compressive strength is 71.0118.0MPa, the internal irradiation index is I_Ra=0.1, and the external irradiation index I_r=0.3～0.4，which can meet the decoration and construction purposes.

The ecological impact of urban development and population increase is an area of increasing relevance, and urban ecological environment's quality can have a direct impact on people's experience of life. The assessment of urban ecological quality is an important reference for urban ecological protection. Therefore, based on the Google earth engine (GEE) platform, this paper applied a remote sensing ecological index (RSEI) method to study the ecological environment quality (EEQ) changes in Xiamen City in the past 20 years. The four remote sensing indices, Normalized Difference Vegetation Index (NDVI), the heat index (land surface temperature, LST), the wetness index (WET), and the dryness index (normalized difference impervious surface, NDBSI) were first composed. Furthermore, a principal component analysis (PCA) can be implemented to obtain RSEI image. Finally, the EEQ of Xiamen in 2000, 2005, 2010, 2015, and 2020 was evaluated. The results showed that the mean RSEI value of Xiamen in four corresponding years were 0.562, 0.586, 0.564, 0.575, and 0.569, respectively. The area with excellent ecological grade in Xiamen has significantly grown from 310.75 km² in 2000 to 509.69 km² in 2020, with an increase rate of 12.21%. The area with excellent ecological grade can be mainly transferred from the good ecological grade. It indicated that the area with the improved EEQ was much higher than the area with the deteriorated EEQ in Xiamen in the past 20 years. The best EEQ among the six districts in Xiamen invariably appears in Tong’an District, which is covered the highest forest area and has the smallest population. On the contrary, the EEQ of Huli District is the worst, mainly due to the high building density, and the large impervious surface areas in Huli District. Despite all this, the average RSEI value in Huli District also shows a steady increasing trend since various construction of urban ecology has been promoted, such the great increase of green space and the continuous reconstruction of the old town in Huli District.

Affected by Emei rifting, Northeast Sichuan was in a tensile environment in the late Maokou Formation, and the Kaijiang Liangping trough gradually began to take shape. The Changxing period was the peak period of trough development. The platform margin of Changxing Formation developed a reef reservoir, which is also the main exploration layer series of LG gas field. However, no in-depth study was carried out on the development of pore type fault solution reservoir in the platform. In Yanshan Himalayan period, LG area developed NW-SE fault zone under the joint action of Daba Mountain and Huaying Mountain, and the development position of fault zone controlled the distribution law of fault solution. Through the fine interpretation of 3D seismic data, the distribution law of NW trending fault trains in the area is implemented; according to the drilling conditions of lg11 well, combined with the practical experience and theoretical research foundation of fault solution in Tahe block, the forward modeling model is established. Through the analysis of forward modeling results, the seismic reflection characteristics of fault solution are determined, and the development model of fault solution in Changxing Formation in LG area is established; on the basis of tracing and depicting the characteristics of the fault solution in the profile, the distribution law and distribution characteristics of the fault solution in the Changxing Formation in the study area are defined by using seismic means such as multi-attribute analysis, which provides a new direction for oil and gas exploration in the Changxing Formation.

The Mao-2 member reservoir in the northeast of Sichuan Basin consists of a set of bright bioclastic, algal clastic limestone and silty fine dolomite. Through the analysis of the porosity and permeability of two wells in Yuanba area, the samples with porosity of 2-5% account for 23.8% of the total, and the permeability is mainly distributed in 0.001-0.01 × 10^-3 μ m², accounting for 66.7%. In general, Mao-2 reservoir is a low porosity and low permeability reservoir, and there are medium and high permeability reservoirs locally. Affected by karstification and basement fault activity, Mao-2 member is easy to form karst fracture cave type reservoir, and industrial gas flow is obtained in many wells in the adjacent area, which proves that Mao-2 member has good exploration potential.

Water yield is one of the important ecosystem services. It is vital to study the spatio-temporal distribution characteristics and driving mechanism of water yield for water resources protection, improving water resources planning and maintaining ecological security. In this study, we used the InVEST model to quantitatively evaluate the water yield of Qingdao in 2000, 2010 and 2020, and analyzed the spatio-temporal changes and driving factors of water yield. The result shows that the water yield of Qingdao has an upward trend from 2000 to 2020, with an average annual increase of 8.77%. The water yield is low in the north and high in the south. The main impact factors that result in difference spatial patterns of water yield between the north and south include the change of construction land, economic development, variation of precipitation and evapotranspiration. The results can provide a scientific reference for promoting the construction of ecological civilization in Qingdao and maintaining the coordinated development of the economy and ecology.

The Yellow Sea Green Tide, which occurs repeatedly every year, has a significant impact on China's ecological environment and economy. At present, resource scheduling in the process of emergency response to green tide disasters depends more on disposal experience. In order to improve the reliability and efficiency of decision-making, we propose a resource scheduling model, which is used to optimize the allocation of resources. First, determine the main influencing factors, including the details of salvage objects and available resources. Next, with the maximum salvage efficiency as the goal, the resource scheduling model of marine green tide disaster is established. An improved particle swarm optimization (SAPSO) is used to solve the model. Finally, a test case is taken as an example for experimental verification. The results show that, compared with PSO algorithms, the resource scheduling scheme obtained by SAPSO algorithm is better. The model proposed in this paper can help decision-makers to make effective resource scheduling schemes.

Correlation analysis of monthly precipitation and monthly mean maximum and minimum temperatures for 2001-2020 in Jinan, Qingdao, Wuhan and Shanghai is conducted in this paper. Programming tools such as python and matlab are used in statistical analysis of one-dimensional linear regression in order to compare the differences between coastal and inland climate elements and analyse their causes. Experiment indicates that inland cities usually have more extreme precipitation and higher maximum temperatures in all seasons, while coastal cities always have more stable precipitation and higher minimum temperatures, except for spring. By establishing a unary linear regression model, so as to better judge the temperature change in inland coastal cities, the experiment is in line with the expected results.

The Shahezi formation in Xujiaweizi fault depression is one of the key strata for exploration in Songliao Basin, which is characterized by various sedimentary types, strong reservoir heterogeneity, complex geological conditions and difficult benefit development, the conventional two-dimensional method is difficult to realize the fine and efficient utilization of this kind of reservoir. In order to fully tap the deep natural gas resources, based on Petrel software, on the basis of structural modeling, select the main factors that affect the reservoir in the area and use sequential Gauss simulation method to build a 3D geological model of the high-quality reservoir with shale content and reservoir convergence, in this paper, the authors describe the geological characteristics of frequent interbeds and complex lithology of sand and mud in the target zone and establish a multi-parameter mechanical model by using the GPA software and the data of geology, earthquake, logging and laboratory rock mechanics and physics experiments, based on the above geological model and stress model, the fracture propagation pattern is simulated, and the fracture parameters of each layer are optimized to realize the optimization of reconstruction scale and cost control. The well A in Xujiaweizi fault depression is designed by using 3D geological modeling method of complex geological body. The field application shows that the 3D geological modeling method of complex geological body has important guidance and reference significance for reservoir fine characterization, fracturing optimization and post-reconstruction of complex geological body.

When using the random forest algorithm to classify remote sensing images of each target year in the study area, the number of decision trees and the maximum number of features for constructing the optimal model of decision trees have a great influence on the accuracy of the random forest classification results. Based on this, this paper proposes an adaptive parameter tuning strategy based on GridSearchCV to improve the random forest algorithm. The method can select the best parameters according to different sample data and study area conditions. By comparing with unoptimized random forest, decision tree, and support vector machine algorithms, the results suggest that: the optimized random forest algorithm has good classification accuracy, and the overall accuracy and Kappa coefficient of classification results are above 0.90.

By exploring the spatial-temporal pattern changes of ecological land in Yangzonghai watershed of Yunnan Province, this paper obtained the quantity and spatial distribution of ecological land, which promoted the effective protection, rational utilization and sustainable development of ecological environment in the watershed. Using the data of 2005, 2010, 2015 and 2020, ArcGIS and Fragstats4.2 software were used to analyze the spatial and temporal patterns of ecological land use in the study area by using dynamic variation degree, transfer matrix and landscape pattern index. The ecological land in Yangzonghai watershed was mainly basic ecological land and auxiliary ecological land, which accounted for about 77% of the total study area. From 2005 to 2020, the ecological land showed a decreasing trend, with a total decrease of 798.61 hm². From the perspective of ecological land transfer matrix, forest land, cultivated land, bare land and construction land are converted frequently. The increase of forest land was mainly from cultivated land and bare land. The sources of cultivated land increase were forest land and bare land. The increasing sources of construction land were cultivated land, bare land and forest land. The decreasing sources of forest land became bare land and cultivated land. The decreasing sources of cultivated land became forestland and construction land. According to the landscape pattern index, woodland, water area and farmland were the main landscape types in the study area. From 2005 to 2020, woodland, water area and cultivated land all showed a trend of fragmentation, and the patch geometry changed from simple to complex. However, on the whole, the connectivity of dominant patch types in the study area was good, and the landscape diversity and dominance were relatively stable although there were certain fluctuations. The quantity, transfer and landscape pattern of ecological land in Yangzonghai basin from 2005 to 2020 were analyzed. It is expected to provide data support and decision-making basis for ecological land protection and ecological security system construction in the watershed.

Reservoir characteristics of Jurassic Daanzhai Segment in Yilong-Pingchang area are one of the important factors that affect the oiliness of the study area. Therefore, it is necessary to analyze and study the reservoir characteristics to find out the reasons for the difference of oiliness in this area. On the basis of collecting and sorting out a large number of previous research results, this paper studies the petrology, physical properties and characteristics of reservoir space in the first and second sub-segments of Daanzhai Member of Ziliujing Formation through laboratory test data, such as relevant rock slices, casting slices and scanning electron microscopy (SEM), and analyzes the control factors of reservoir development from three aspects: sedimentation, diagenesis and tectonism. The results show that the reservoir rock types of Daanzhai member in Yilong Pingchang area mainly include shell limestone and argillaceous shell limestone, with average porosity of 1.79% and average permeability of 0.036 × 10^-3μM²ⁱ in Daanzhai-1 member (hereinafter referred to as J₁dn¹), and average porosity of 1.9% and the average permeability of 0.056×10^-3μm² of Daanzhai-2 member (hereinafter referred to as J₁dn²). The reservoir space of J₁dn¹ member is mainly composed of micro crack intercrustal pores, while that of J₁dn² member is mainly composed of inter-crustal pores and intra-crustal pores.

In this paper, the extreme rainstorm in Zhengzhou, Henan is analysed. First of all, a brief overview of the flood in Zhengzhou, Henan is introduced. Secondly, the definition of rainstorm and flood is discussed. Thirdly, the factors affecting the frequency of floods in Henan province, and specifically the uniqueness of this flood in Zhengzhou, is investigated. Finally, some suggestions are provided to reduce the harm of extreme rainstorm to a minimum.

Based on the land use classification information from 1990 to 2020, considering the spatial heterogeneity and the characteristics of the land-sea gradient, this paper adopts the strip segmentation method and constructs the landscape ecological risk index to analyze the temporal and spatial changes of the ecological environment of Manila Bay, and explores the ecological risk changes of Manila Bay in the past 30 years. The results show that the ecological environment changes in Manila Bay have strong differentiation characteristics in spatial distribution and are greatly affected by the sea-land gradient; in recent years, the overall land use intensity has been relatively high, and the ecological status has been declining.

Xiaopinggou, the northeast area of the Dabancheng, outcrop a set of pillow lava assemblage which has exposed this area completely. In this paper, we have taken this place as an example and explained the geological meaning of the U-Pb age of the single zircon grain which was based on magmatic crystallization and statistical theory. It provided an important clue for the geotectonic setting and geological age of the evolution of the Bogda structural belt. Rocks like Pillow basalt, dacite, and pyroclastic are found in the Xiaopinggou region of the Dabancheng pillow lava assemblage. The weighted average age of 21 single zircon grains in dacite using LA-ICP-MS U-Pb dating data was 326.0±2.8Ma. Only the primary zircon crystallization ages in the magma chamber were represented by these ages. The Pillow lava assemblage in Xiaopingou was not diagenetically older than the early stage of the Late Carboniferous (the end-stage of the Bashkirian), according to the four single-grain zircon ages of the most recent crystallization, which yields a concordant age of 315.4±1.3Ma. This age is consistent with the geological age determined by the fossils in the Liushugou formation.

Based on the theory of water footprint and the spatial processing method of GIS data, this paper makes a comprehensive analysis on the population density, economic benefit and water resources scarcity of Zhoushan archipelago. The results show that the per capita water footprint of Zhoushan archipelago is 830.82 m³/person, which is smaller than that of Hangzhou and Jiaxing in the same latitude of Zhejiang Province. The average population density of water footprint in Zhoushan archipelago is 12.04 person/10,000 m³, showing a trend of high in north and south and low in central China. The average economic benefit of water footprint is 135.11 yuan/m³; the high value area is distributed in the south of Zhoushan archipelago, Dinghai District and Putuo District, where the economy is more developed, the average degree of water shortage in Zhoushan archipelago is 1.41, water resources are relatively scarce, and the pressure of water resources is greater. In view of this, we should improve the residents' awareness of water saving, reduce the discharge of sewage, and pay attention to the rational distribution of water resources among regions to reduce the pressure of water resources in the core urban areas.

This paper systematically describes the current situation of marine information resources management, analyzing the existing problems in the construction of marine information system. And it introduces the possibility of classification, standardization and integration of marine information resources based on knowledge service, and the construction of marine professional knowledge service framework based on big data. By studying the new knowledge service mode, it provides technical support for giving full play to the advantage of marine professional information resources.

With the acceleration of urbanization, various logistics and information flows, such as urban economy, population and buildings, are showing the phenomenon of polycentric collection. Urban fire has become an important factor in destroying the environment, people's property and life safety. To decrease the adverse effects, this research constructs a location model based on fuzzy constraints and demand points under multiple risk levels, which provides a new way of thinking for the layout optimization of fire stations.

This paper uses DeepLab v3 to extract glacier information from remote sensing images. DeepLab v3 is embedded into the stacked model to study the glacier recognition in the hinterland of Qilian Mountains in Gansu Province, learn the glacier characteristics and background characteristics respectively, and then integrate the two channels to produce the final results. With this method, glacier information can be extracted with high efficiency and accuracy.

A high-resolution wind speed estimation method (HRWSEM) is proposed for urban surfaces based on vertical stratification downscaling technique, and the usability of the method is evaluated by applicated in Shanghai as an example. The HRWSEM integrates the wind speed variation characteristics at different heights within the atmospheric boundary layer in the vertical direction and the influence of effective roughness on wind speed at different heights in the horizontal direction. It is not only able to better characterize the refined distribution characteristics of urban wind speed, but also the influence of large-scale complex terrain under different wind directions. In addition, it also can better simulate the temporal variation characteristics of wind speed and has good consistency with the observation.

Tight sandstone gas is a huge backup resource for conventional gas. The Shahezi Formation in Xujiaweizi fault depression is the main source rock in the deep basin, and the sedimentary environment is mainly lacustrine and delta facies. The source rocks have large thickness, wide distribution, high organic matter content, mature to over-mature, large amount of resources, and have broad exploration prospects. As a reservoir, it has the characteristics of close to the source, rapid sedimentary facies transition and strong reservoir heterogeneity. In this paper, a practical dessert analysis method for tight sandstone gas is established based on the detailed interpretation of seismic data and sequence division, starting from the source rock, reservoir zoning, classification and comprehensive inversion results. In general, the upper sequence of Shahezi Formation is better than the lower sequence, and the braided river delta facies is better than other sedimentary systems. The subaqueous fan delta front distributary channel of SQ4 and SQ3, the subaqueous distributary channel of braided river delta front and estuary sand bar facies are a kind of dessert development areas, which are also the most favorable facies for the next exploration. With the deepening of research and the progress of technology, the exploration of tight sandstone gas in Shahezi Formation is certain to get breakthrough.

Aiming at the problem of missing data after point cloud data filtering, this paper proposes a point cloud repair method based on the principle of Linear Maximum Entropy, which converts the Maximum Entropy Model into a linear model, and then uses linear programming to solve the model. The mathematical expectation and variance in the elevation value is used to establish linear constraints, and the weight coefficient of the sampling point is solved by the maximum entropy value to determine the elevation value of the fixed point, so as to complete the point cloud repair. By comparing with Maximum Entropy Method and Inverse Distance Weight method, the feasibility of Linear Maximum Entropy Model in point cloud data repairs is discussed. The results show that the point cloud data repaired by the Linear Maximum Entropy Model is more accurate, and a high-quality model can be established.

Nowadays, semantic segmentation is commonly used to solve various remote sensing image problems. Furthermore, it has emerged as a critical technology for remote sensing image applications. In recent years, with the widespread application of deep learning in computer vision, deep learning semantic segmentation has been appreciated by more remote sensing researchers. Its performance depends on the learning of remote sensing data by the model. However, the complexity and multiplicity of remote sensing data lead to the increased difficulty of learning remote sensing data, and also reduce the ability of the model to segment high-resolution remote sensing images. To improve the model performance for deep learning semantic segmentation, many improvement methods have been proposed, such as batch regularization (BN), data augmentation, and residual networks. However, most of the methods are optimization strategies for deep learning with natural domain images. Further exploration of the effectiveness of related methods in specific application areas is necessary. In high-resolution remote sensing images, the correlation between segmentation model hyperparameter settings and residual networks has rarely been investigated, which makes the residual methods inadequately applied to the semantic segmentation of high-resolution remote sensing images. In this paper, an optimal residual semantic segmentation network based on the U-Net architecture is proposed. The ORC-UNet not only considers the loss function configuration of the model, but also the initial filter number setting of the model. With these two hyperparameter configurations, the residual method can effectively improve the segmentation performance of the model. In order to confirm the effectiveness of ORC-UNet, comprehensive experiments are conducted on the Potsdam dataset in this paper. The experimental results show that ORC-UNet achieves the best overall segmentation performance with an F1-score value of 74.44%, and it also achieves the best performance in three categories of the dataset (Background, Building, and Roads) with F1-score values of 30.78%, 89.58%, and 86.05%, respectively. Meanwhile, ORC-UNet is also close to the best results for the remaining categories. In addition, the relevant comparison results also indicate that the cross-entropy loss function and initialization value 64 of ORC-UNet are the best configurations. Meanwhile, ORC-UNet is the best segmentation model among all residual models.

At present, in the related research of 3D line scene, the multi-scale expression method is mostly used to describe the detail degree of geographical features, and there is a lack of a method to describe objects from the perspective of granularity, so it is hard to meet the application requirements of objects in different scenarios. This paper introduces quotient space theory and proposes a representation method of highway geographic entities at different granularity levels. Firstly, the conditions that can be divided into highway geographical entities within the same granularity are determined. Then, the granularity partition model of highway geographic entities is constructed from three aspects: text similarity, literal similarity and semantic distance similarity. Finally, each quotient space is organized by the theory of quotient space particle size synthesis, and the whole geographical entity is divided into multiple granularities, and the experimental verification is carried out. The results show that the method can well meet the multi-level and multi-granularity requirements of geographic entities, and realize the hierarchical optimization of expressway scenes.

In July 2020, the 11 # tower of the 220kV Mengwen transmission line in Lixian County, Sichuan Province was significantly deformed due to a landslide, which seriously threatened the operation safety of the transmission line. Satellite optical images, satellite InSAR data, UAV images, and ground survey were integrated to investigate and evaluate the landslide. The tower is located at the shoulder of the front edge of an ancient landslide and five small soil landslides with signs of deformation are developed at the front edge of the ancient landslide. The slope presents seasonal deformation characteristics. In May 2020, there was a sudden increase in the deformation rate, and the overall deformation trend increased compared with previous years. It is recommended to monitor the deformation in the rainy season, and carry out the relocation as soon as possible after the rainy season. This case study has proved the effectiveness of spaceborne optical remote sensing technology, spaceborne InSAR, airborne photography technology, and ground survey means in the detailed investigation and risk assessment of tower landslides.

Owing to the extremely rich and sensitive information they can provide, urban high-resolution (HR) remote sensing images have a wide range of applications in smart city planning. However, conventional spectral information-based image processing methods are no longer appropriate for producing such images. Remote sensing image segmentation processes commonly apply the mean shift method, which generally uses spatial and spectral features, to maintain edge information and reduce the impact of noise. This paper proposes a novel mean shift segmentation method based on Gabor texture analysis, in which the number of dimensions in the feature space is increased by applying textural elements in addition to spatial and spectral elements to fully utilize detailed information in urban high-resolution remote images. To obtain texture information, multi-scale, multi-orientation Gabor filters are used to extract pixel-to-pixel features from an image, with local variance (LV) used to estimate the optimal texture bandwidth. QuickBird and Mass. Buildings images are used to experimentally validate the proposed method based on analyses using the global segmentation quality index (GS) and error rate (ER) as indicators. The results obtained for two different remote sensing datasets show that the proposed method is superior and more precise than the conventional mean shift method.

Drought, which frequently occurs in the major soybean producing areas in China, has led to a serious reduction in soybean yields. The objective of the present paper was to study the spectral characteristics of soybean under water stress, and propose the Soybean Water Stress Index (SWSI) to monitor the extent and area of water stress through a field simulated experiment. The experiment was carried out in the Agricultural Experimental Base of Jilin University from May to September in 2020, and canopy spectral data were collected once a week. The result showed that the spectral reflectance of soybean canopy increased in the VIS and SWIR spectral regions and decreased in the NIR with an increase of the water stress. This paper selected NDVI, RDVI, PRI, MCARI, NDWI, WI and SWSI to identify different degrees of water stress of soybean. The result suggested that the RDVI and SWSI were suitable for identifying soybean under water stress. To seek the best identifiable vegetation index, the normalized average distance of vegetation indices under different water stress degrees were calculated. The result indicated that the distance of SWSI is more than that of other indices’ in the whole growth period, illustrated that the identifiable ability of SWSI for different water stress degrees of soybean is better than other indices, then SWSI has the strong sensitivity and stability. Therefore, SWSI can be used to monitor the area and extent of drought and provide information support for disaster relief and decisions.

The continental shale oil reservoirs in the study area not only have high clay content and complex pore structure, but also have developed laminaceous foliation and strong anisotropy. Laboratory core analysis is an important means to study the rock and fluid properties of shale oil and gas reservoirs. Because shale cores swell with water and are fragile, conventional experimental methods for sandstone are no longer suitable for shale core analysis, so laboratory petrophysical analysis methods for shale oil reservoirs need to be developed. At present, the two-dimensional NMR spectrum fluid identification plate under shale oil laboratory conditions is controversial at home and abroad, so the research work of establishing two-dimensional NMR spectrum fluid identification plate is carried out for continental shale in the study area. In this paper, from the rock physical property analysis of shale reservoirs, combined with the feature of Qingshankou Formation continental shale reservoir in the study area, to carry out high temperature carbonization saturated water and saturated oil, dry samples, analysis of shale nuclear magnetic response characteristics of different pore space fluid, establish suitable for the study area of continental shale two-dimensional nuclear magnetic resonance (NMR) spectra fluid identification chart. The conclusion of this paper is that with the increase of temperature, the signals of clay interlayer, pore and macropore gradually decrease, until the carbonization temperature reaches 704℃, the nuclear magnetic signal can't be measured; The nuclear magnetic spectrum of saturated fluid in the dry distillation core shows that kerogen and hydroxyl signals disappear completely, and a large amount of self-absorbed saturated water enters clay layers, while the nuclear magnetic signals in clay layers, small holes and large holes of saturated oil cores increase obviously. The nuclear magnetic spectrum of saturated fluid after dry distillation shows that clay interlayer can be used as the storage space of crude oil, and the nuclear magnetic signals of oil and water in clay interlayer are obviously different.

With the rapid development of modern agriculture, the traditional agricultural farming model has been unable to meet the requirements of the rapid development of contemporary society for productivity. The monitoring of crop soil moisture content by satellite hyperspectral is limited by the problems of spatial resolution and data source. The rapid development of UAV has made up for the defects of satellite hyperspectral. In this manuscript, the hyperspectral data of UAV is used as the main data source, the measured spectrum of vegetation in cotton field and the measured ground moisture content data are used as auxiliary data. Different vegetation indexes are calculated by using the measured spectral curve on the ground. Meanwhile, the correlation with the measured soil moisture content and vegetation indexes is also analyzed. The quantitative relationship between vegetation canopy spectral information and ground measured soil moisture content is established and the soil moisture content is retrieved through the vegetation canopy spectral information indirectly. In order to optimize the hyperspectral data of UAV, the regression relationship between the same vegetation index of two data sources is established, and the soil moisture content model constructed by the measured spectral curve vegetation index is applied to the UAV hyperspectral image in order to complete the large-scale spatial inversion mapping of soil water content. The results showed that there was a positive correlation between soil water content and vegetation index as a whole. The correlation between soil moisture content and normalized vegetation index (NDVI), green wave vegetation index (GNDVI), soil regulated vegetation index (OSAVI) and soil ratio vegetation index (SR) reached 0.79, 0.72, 0.73 and 0.84. NDVI and SR are selected to construct the soil moisture content inversion model, and the model determination coefficients are 0.63 and 0.77, respectively. Due to the difference between the vegetation index of ground measured spectrum and hyperspectral data of UAV, the hyperspectral data are optimized through the vegetation index established by ground measured spectrum to realize the inversion mapping of soil moisture content of UAV hyperspectral data.

As one of the important indexes of the climate characteristics, the daily mean temperature in climate change research, agricultural meteorological disaster monitoring and other fields plays an important role; compared with the traditional way of monitoring and estimating the average daily temperature, the remote sensing technology has comprehensive, macroscopic, dynamic and other incomparable absolute advantages, and can accurately describe the spatial heterogeneity of the daily mean temperature. In order to improve the quality of agricultural meteorological service and increase the monitoring accuracy on agricultural disasters, we obtain the optimal inversion model of the daily mean temperature in soybean growing area. In this paper, based on the FY-3D-MERSIⅡ remote sensing data, a random forest model and multiple regression model were constructed, respectively, to inversion of spatially continuous daily mean temperature in Liaoning Province. Results are as below: (1) On the whole, the random forest model has good applicability in the daily mean temperature retrieval, the root mean square error (RMSE) and mean absolute error (MAE) of the random forest method are 0.95 °C and 1.75 °C; but the multiple regression model inversion accuracy is relatively low, RMSE is 1.24, MAE is 1.15°C. (2) Combined with the soybean growing area, data found that although the inversion results of random forest model and multiple regression model in the eastern mountains of the study area have great deviation, the proportion of soybean planting in this area is relatively low; therefore, both models have good applicability in retrieving daily average temperature in soybean growing area, and the random forest model is relatively more stable. (3) Based on the spatial interpolation, results show that the random forest model and multiple regression model in describing the spatial distribution of the daily mean temperature is more exquisite and accurate, especially in the coastal areas, and the inversion results are more consistent with the reality, which proves the feasibility of daily mean temperature inversion based on remote sensing.

Polarimetric interferometric SAR (PolInSAR) technology plays an important role in large range forest height estimation, and the estimation algorithm is a very important part in the inversion process. This paper used the simulated L-band fully polarized airborne SAR data, and used DEM difference method, complex coherence amplitude method and RVOG ground phase method to estimate tree height, and simulates five kinds of forest density to test the effect of forest density on the inversion results of RVOG ground phase method, and carries out the result statistics and accuracy verification. The results showed that the average value and RMSE of DEM difference method were 10.10m and 10.57m, the average value and RMSE of complex coherence amplitude method were 17.63m and 7.20m, and the average value and RMSE of RVoG ground phase method were 20.22m and 5.81m. The results of the complex coherence amplitude method were relatively discrete, indicating that the robustness of the algorithm was slightly weak. The inversion effect of the RVOG ground phase method was the best. With the increase of forest density in a certain range, the RMSE of RVOG ground phase inversion results changed from 6.91m to 5.72m, indicating that the inversion accuracy was higher, but the increase of accuracy was not obvious when the forest density reached a certain value, indicating that it is suitable for forest height inversion with medium and high tree density.

Buildings are one of the most important infrastructures in cities. Automatic extraction of buildings from high-resolution remote sensing imagery is of great significance for urban management and population estimation. Aiming at the problem of insufficient edge features and loss of details in the building extraction results, a network model combined with Canny edge detection and convolutional block attention module based on U²-Net was proposed in this paper. Adding the building edge feature map to the U²-Net to compensate for the problems of insufficient extracted building edge features and loss of detail. Using the convolutional block attention module to achieve effective feature extraction of buildings. Qualitative and quantitative analyses were performed on the WHU building datasets. The experimental results showed that improved U²-Net can accurately extract building regions from remote sensing images. And for the problem that deep learning network relies on training samples in the process of building extraction, this paper discusses the influence of the number of samples on the results. The experimental results showed that a reasonable setting of the number of building samples can improve the extraction accuracy of buildings, not that the greater the number of samples, the higher the accuracy.

With the development of new technologies such as 3D laser scanning, 3D reconstruction and virtual reality, digital conservation of cultural relics has been widely applied and has become an important way to record, display, monitor, digitally restore, study and protect cultural relics. Aiming at the repair of the damaged chignon of the Big Buddha in Tongnan, a virtual restoration method based on the geometric feature information of the 3D point cloud model is proposed. This method is based on the high-precision 3D model obtained by 3D laser scanning, combined with close-range photogrammetry and computer virtual reality technology, and integrates historical evidence data and statistical analysis of the 3D model data of the Buddha's chignon to carry out a variety of virtual restoration effects on the Buddha's chignon in the computer, and the experimental results show that the virtual restoration based on the high-precision 3D model is practical and feasible, and has a certain guiding significance for the actual restoration of the chignon of the Buddha.

In order to expand the application of knowledge graph in geology, this paper constructs a knowledge graph application system of "knowledge graph visualization of mineral deposits + geospatial information indexing + inference analysis." The system adopts Flask as the back-end framework and Bootstrap as the front-end framework. Using the mineral deposit knowledge graph as the database, the system realises the geospatial indexing function of mineral deposit knowledge by integrating geographic images such as remote sensing images, and realises the inference and analysis function of mineralisation knowledge by integrating graph algorithms.

Using spectrum maps to master the spectrum situation and manage spectrum resources accurately is a hot issues that researchers focus on. Firstly, we clarified the concept of spectrum mapping, sorted out its development history, and on this basis, the significance of global spectrum mapping for the application of extended spectrum map is reviewed and summarized; secondly, the research scope, role and status of global spectrum mapping are expounded, the technical system of global spectrum mapping is clarified, and its application needs are analyzed in depth; finally, the research on the design of global spectrum mapping scheme is carried out, and preliminary experimental results are given. The study shows that the global spectrum mapping can effectively realize the organization and management of global spectrum mapping data, which is of great significance for the development and enrichment of the concept and connotation of spectrum map, and can help electromagnetic spectrum use and management personnel to establish the spectrum situation in a timely and accurate manner and accurately perceive the spectrum situation.

Deformation analysis based on point cloud is one of the major applications in surveying engineering, which makes up for the shortcomings of traditional surveying methods that hardly reflect the overall variation with fewer sampling points. A deformation analysis method of roads in mining areas based on vehicle point cloud is proposed in our study, which consists of three parts: (1) fine registration of multitemporal point cloud; (2) accurate extraction of the ground and road's point cloud; (3) rapid detection of road's deformation information. To verify the feasibility of the proposed method, a road's point cloud in mining areas captured by the mobile laser scanner is selected. Results show that the method can extract the complete road deformation information and provide technical support for subsidence monitoring.

With the rapid development of UAV technology, the research topic of remote sensing image segmentation has gradually attracted more and more attention. Whether the image can be accurately segmented is a measure of the goodness of the algorithm. In recent years, machine learning methods have been applied to a large number of fields, and deep neural network technology has also been widely used in the field of UAV remote sensing image segmentation. This paper introduces the specific applications of various deep learning methods in remote sensing image segmentation, and briefly analyzes the development of neural networks in this problem according to the research status of several typical deep neural networks in UAV remote sensing image segmentation. Research shows that on the basis of proper adjustment of optimizer, learning rate and loss function, using deep learning method to segment UAV remote sensing images, the accuracy rate can reach more than 96%.

With the development of remote sensing satellite technology, the richness and data level of satellite images are rising rapidly. After the construction of provincial satellite application technology centers in China has been started one after another, the production, processing and inspection of massive image data are very time-consuming and prone to quality problems if they are produced in the traditional way. Firstly, according to the needs of massive remote sensing image data production, this paper designed the technical route and studied the key issues of data preparation, data governance and quality inspection in intelligent batch processing. It designed and completed the remote sensing image data production system. The system has innovatively developed image source filter, intelligent mosaicking seamline editing and other functions, which are very suitable for the actual needs. Through practical application, the system can improve production efficiency and ensure data quality.

In order to solve the problems of edge and plane passivation and inaccurate distribution of pixel power to vertex in 3D mesh optimization using image matching best surface estimation method, a 3D mesh method based on contour restoration and least square vertex movement is proposed to optimize the initial mesh. Firstly, large scale simplification and subdivision method are combined, the global contour of the level of detail in mesh is restored, and then the vertex gradient accumulation of the least square method is used to allocate the moving amount of the energy of the object in the triangle to the three vertices more accurately. Finally, the smoothing term based on curvature is used to replace the homogenized Laplacian smoothing. The mesh shape is retained while the matching noise is filtered to achieve more refined mesh optimization. After the experimental verification, the feasibility of this method is proved.

In order to clarify the mutation characteristics of alpine vegetation cover, a study of vegetation coverage change and its climate response relationship was carried out in Qinghai province based on AVHRR NDVI_3g, temperature, precipitation, wind speed and sunshine data from 1982-2015. The results show that the overall NDVI series in the study area had mutant in 1998 and 2005, with mutants near 1987 in wet and semi-humid areas, and then the increase rate slowed down, and a clear trend of NDVI increase in arid areas after 2007. Temperature changes had a significant effect on vegetation coverage before the change in 1998, and there was a significant time lag effect between vegetation coverage and each meteorological factor during 1982-2015, and the changes of each meteorological factor over 1-5 years had some effect on vegetation. Cumulative changes in temperature, mutants in sunshine hours and rainfall in the study area were the main causes of mutation of vegetation coverage in 1998 and 2005.

Climate change and urbanization have resulted in increased flood risk around the world. However, flood risk management varies in different countries, especially in developed and developing countries. This paper focuses on the comparison of flood control measures in India and the USA. This paper is aimed to find out a clear gap between developed and developing countries in the flood control field. This paper also utilizes a comprehensive database of global flood events. This database is used to visualize the flood trend around the world and help analyze flood severity in both India and the USA. The process mainly includes data mining, pre-processing, visualization, and data analysis (Tableau is used as the premier data visualization tool). This paper hopes to bring about awareness that flood control is of global responsibility and that the gap between developed and developing countries should be reduced as soon as possible.

Compared with observation by ground meteorological stations, satellite monitoring of nighttime sea fog can provide a wider range of fog distribution. However, there is a lack of analysis on the applicability of several classical remote sensing retrieval algorithms for nighttime sea fog in Shandong coastal areas. Based on FY-4A geostationary meteorological satellite data, the method of dual channel difference (DCD), the method of temperature difference (DT), and the method of normalized fog index (NDFI) are used to retrieve the nighttime sea fog area in Shandong coastal area between January 2019 and December 2020. And the critical success index (CSI), the hit rate (HR), the probability of detection (POD), and the false alarm ratio (FAR) are used as index parameters to verify the retrieval results and discuss the applicability of the three methods based on the ground observation data. The results are shown below. 1) The nighttime sea fog often occurs in spring, winter and summer, and the frequency of nighttime sea fog is lower in autumn. In the statistical area, the occurrence frequency of nighttime sea fog in the Yellow Sea and the East China Sea is relatively high, followed by the South China Sea. 2) The performances of the three retrieval algorithms for nighttime sea fog are different in different seasons. The northern Bohai Sea performs better in winter, while other sea areas perform better in the fog season from April to July, and worse in July to September. Among the three inversion algorithms, the DT method performs the best because of its high CSI, high HR and low FAR. Due to the high POD and high FAR, the effect of DCD method comes second, and the NDFI method performs the worst. 3) The retrieval of nighttime sea fog in the Yellow Sea by the three algorithms is slightly better than that in the Bohai Sea. Among them, the DCD method and DT method have stronger retrieval ability of sea fog at night. 4) The applicability analysis of three retrieval algorithms of nighttime sea fog show that the retrieval ability of the DT method is stronger than that of the other two algorithms. The reason may be that the DT method considers the existence of inversion conditions in the algorithm process. The NDFI method is more sensitive to threshold.

Based on the Baidu index, this article obtained the national network attention data of Guilin Lijiang Scenic Spot from 2011 to 2021, and analyzed the spatiotemporal distribution characteristics of the network attention by combining the seasonal concentration index and the Herfindahl index. The results showed that: (1) The inter-annual variation of the network attention of Guilin Lijiang Scenic Spot fluctuated greatly, with a tortuous upward trend from 2011 to 2019, a gradual downward trend from 2019 to 2021, while the seasonal and monthly changes of network attention were not significant. (2) The spatial aggregation of the network attention of Guilin Lijiang Scenic Spot is not high, it presents a spatial distribution pattern of "close and developed regions have high attention, remote and undeveloped regions have low attention."

With the development of big traffic data, bus schedules should be changed from the traditional "empirical" rough scheduling to "responsive" accurate scheduling to meet the travel needs of passengers. Based on passenger flow distribution, considering passengers' feelings of congestion and waiting time at the station, we establish a dual-cost bus scheduling optimization model (DCBSOM) with the optimization objectives of minimizing bus operation and passenger travel costs and improving the classical genetic algorithm (GA) by adaptively determining the crossover probability and variance probability of the algorithm. We use an improved double probability adaptive genetic algorithm (IDPAGA) to solve the model and consider the constraint conditions of the model in the iterative process of the IDPAGA. By solving the arithmetic example, we get: (1) the optimized departure plan is more inclined to the passenger benefits and more in line with passenger travel pattern; (2) the optimal solution can reduce the overall objective function value by 4.22%, improve the bus operation cost by 4.9%, and reduce the passenger travel cost by 13.4%. The conclusions show that although the DCBSOM built by the research increases the bus sector's operating costs to a certain extent, it can better meet the passenger travel demand, improve passenger travel satisfaction, and reduce the passenger travel cost and waiting for cost.

Since the successful opening of the first subsea tunnel in Qingdao city, the situation of "lack of communication between Qingdao and Huangdao" has been effectively resolved. The research proposes a multi-perspective analysis method based on the micro level (ML_MPA), which analyzes the characteristics of the Qingdao-Huangdao bus network system using complex networks. Based on the bus data, we construct the bus station network model and bus line network model between the urban area (Shibei District and Shinan District) and Huangdao District, and comprehensively analyze the overall characteristics of the bus system. The results show that the bus station network has a scale-free characteristic and the bus line network has a small-world characteristic. Combined with the analysis results, it is found that the station distribution of the Qingdao-Huangdao bus system is loose, but the line planning is reasonable. However, there is still the inconvenience of residents taking the bus. The research can provide a theoretical basis for the management of the Qingdao-Huangdao bus network system.