This PDF file contains the front matter associated with SPIE Proceedings Volume 13223, including the Title Page, Copyright information, Table of Contents, and Conference Committee information.

Mudanjiang has become a disaster-prone area due to the influence of various factors such as landform, lithology of formations, geological structure, meteorology, and human activities. According to the field survey, Mudanjiang has been identified with 15 collapses, 10 landslides, 7 debris flows, and 37 potential unstable points. The survey shows that under conditions of high-intensity and sustained rainfall, the interface between rock and soil experiences the infiltration of water, which leads to the softening of the area. This softening creates potential sliding surfaces or zones that have the potential to trigger landslides. Mudanjiang has abundant rainfall, and rainstorm is concentrated and frequent. This unique meteorological condition significantly contributes to the stimulation and formation of debris flows in the region. This paper aims to analyze the frequency and characteristics of regional disasters using multi-year rainfall data in Mudanjiang. Onsite deployment of landslide monitoring equipment was conducted to monitor various elements of landslides, with a particular focus on analyzing landslide displacement characteristics. In order to gain a deeper understanding of the internal mechanism and rheological characteristics of debris flows, rheological shear tests were conducted on debris flows using a rotational viscometer. Using ANSYS FLUENT to simulate the rheological behavior of debris flows.

Ji'nan rock mass is a mesobasic magmatic rock invaded in the Mesozoic Cretaceous, which not only forms rich geothermal, mineral water, iron ore, gypsum ore and other minerals around it, but also plays an important role in controlling Ji'nan springs. Studying its spatial distribution characteristics and upwelling mode is the basis for rational development and utilization of resources and spring water protection. Through comprehensively analyzing the 225 medium-deep drilling holes, 3000 shallow holes around the rock mass, and geophysical data, this work carried out detailed characterization of the boundary information and vertical characteristics of Ji'nan rock mass, multiple geological profiles were drawn to establish a three-dimensional geological model. The spatial morphology of rock mass and magma upwelling channel in Ji'nan are deeply studied. The results show that: (1) The latest estimate of the Ji 'nan pluton is 58.38km long from east to west, 36.26km wide from north to south, with an area of 352.02km² and a vertical depth of 2.3km; (2) The magma upwelling channel is located in the area of Xindingzhuang in the west of Ji'nan.

To address the issues of inaccurate coefficient estimates for geological disaster risk assessment factors and unreasonable risk zoning, this paper proposes a grid-based machine learning model for geological disaster risk assessment, taking a multi-factor and multi-characteristic approach to geological disasters. Firstly, geological disaster data from Wenzhou city is collected and organized. Using GIS spatial analysis and attribute calculation techniques, seven factors including landform type, terrain slope, road construction, geological structure, soil type, river distribution, and vegetation coverage are used as evaluation factors for geological disasters. A neural network model is introduced to optimize the calculation of factor weights. Finally, a geological disaster risk factor evaluation map is derived using weighted analysis. The research results show that the use of a machine learning model to automatically calculate evaluation factors and weight coefficients, broken down by grid units, enhances the rationality of geological disaster risk assessment and can provide a scientific reference for regional geological disaster prevention and mitigation management.

Desertification prevention and control is the main task of desertification control, and Mu Us Sandy Land is one of the four major sandy lands in China, with frequent droughts and very fragile ecological environment. Climate change is the main factor leading to drought, and it may face more severe drought in the future. Normalized Difference Vegetation Index and Land Surface Temperature data were used to calculate the Temperature Vegetation Drought Index By using Sen+Mann- Kendall trend analysis, Hurst Index, partial correlation analysis and lag analysis methods, the spatial and temporal change characteristics, future change trends and time-delay effects of TVDI on climate factors in Mu Us Sandy Land from 2001 to 2020 have been explored. The results showed that: (1) the average TVDI was 0.6, and the drought gradually eased from west to east. In autumn (Z=1.99) 38.5% of the pixels showed a significant drying trend, and in the other three seasons (Z mean =2.95) and the whole year (Z=3.47), the trend was very significant drying trend. (2) The relationship between annual TVDI and precipitation is mainly not significant negative correlation, accounting for 63.4%, which is concentrated in the southern part of Shaanxi Province with higher altitude and the southeastern part of Yanchi County. (3) In summer and autumn, TVDI is mainly not significantly positively correlated with temperature and evapotranspiration, and mainly not significantly negatively correlated with precipitation.

Based on the continuous observation data of 5 gravity instruments in Shandong gravity network, we analyze the Characterization of the co-seismic response to earthquakes with different epicentral distances and magnitudes and the maximum epicenter distance of seismic response capability. The results show that: 1) The instrumental stability of the Shandong gravity network is good, the co-seismic waveform duration and wave amplitude are highly positively correlated, and the earthquake magnitude is positively proportional to the duration and wave amplitude. 2) In terms of coseismic response ability, the Jiaxiang Terrace gravity meter is superior to the Tai'an Terrace, and the Jinan Terrace gravity meter is superior to the Yantai Terrace and the Tancheng Terrace. 3) The equations corresponding to the magnitude (M_S) and the maximum epicenter distance (Δ) of the mapped earthquakes from the Shandong Gravity Network Instrument were obtained.

The rock types of the tight sandstone reservoir in the Lianggaoshan Formation of the Sichuan Basin are complex. Clarifying the micro pore structure characteristics and main controlling factors of physical properties of the reservoir is beneficial for evaluating the sweet spots of tight oil and selecting favorable areas. This article selects samples from the tight sandstone reservoir of the Lianggaoshan Formation, and uses methods such as XRD, field emission scanning electron microscopy, and Cast Thin Section to reveal the microscopic pore structure of the tight sandstone reservoir and explore the influence of diagenesis on the physical properties of the tight sandstone reservoir. The results show that the pores of the tight sandstone in the Lianggaoshan Formation are mainly secondary pores, and the reservoir space is mainly composed of residual intergranular pores, intergranular dissolution pores, and intragranular dissolved pores. When the clay content increases, the intergranular pores of clay minerals are more developed, visible micro-fracture. The heterogeneity of the physical properties of tight sandstone is mainly controlled by diagenesis, with compaction and cementation playing a destructive role. Compaction is the most important reason for the decline of the physical properties of the Lianggaoshan Formation reservoir. In addition, the primary pores are cemented by calcium, which can also lead to a deterioration of the physical properties; The organic acids produced by hydrocarbon generation interact with mineral particles to create dissolved pores, which can greatly improve the physical properties of the reservoir. When chlorite develops, a chlorite rim is formed, and quartz growth is limited. Autogenic chlorite inhibits the growth of autogenic quartz, thereby maintaining the original pores and improving the physical properties.

Regarding the issue of variability in aboveground biomass (AGB) of plants caused by topographical factors in mountainous areas affecting water and thermal distribution, this paper explores an aspect-oriented modeling approach for AGB inversion using satellite remote sensing NDVI indices with support from ground-measured AGB data. Firstly, based on DEM slope analysis, three aspects—sunny slope, shady slope, and half-sunny half-shady slope—are classified. Secondly, the located sample data is overlaid with Sentinel-2 NDVI data and aspect data to obtain {aspect-NDVI-AGB} sample data. Subsequently, fitting models are established separately for the three aspects. Finally, with the support of aspect data, these models are used to calculate the overall AGB distribution data based on the input Sentinel-2 NDVI. Experimental results indicate that in Yunnan's Yuanmou County, the aspect-oriented inversion modeling method based on sunny slope, shady slope, and half-sunny half-shady slope achieves higher accuracy than a single modeling method. When modeling for the three aspects, exponential models exhibit higher fitting accuracy compared to linear models. The AGB remote sensing estimation model, developed based on remote sensing data, ground-measured biomass data, and different aspect terrain features, is feasible and provides technical reference for estimating regional grassland resources and even other plant AGB. Additionally, the consideration of incorporating more factors such as elevation and geographical location in classification modeling methods needs further exploration.

Comprehensive vegetation cover in grasslands is a crucial indicator of grassland health and ecological balance, holding significant importance for scientifically sound grassland management and ecological environment monitoring. Satellite remote sensing inversion methods can provide full coverage for the scientific assessment of comprehensive vegetation cover in a given area. However, the sampling for inversion modeling often relies on conventional field survey methods, which are not only labor-intensive but also subject to high subjectivity, making it difficult to achieve satisfactory modeling results. This paper explores the use of multispectral low-altitude unmanned aerial vehicle (UAV) aerial photographs for sample point positioning, and employs three methods—RGB image clustering extraction, NRG image clustering extraction, and NDVI threshold extraction—to quantitatively estimate grassland comprehensive vegetation cover. The experimental results for Yuanmou County in Yunnan indicate that for areas with very high comprehensive vegetation cover, all three methods cannot achieve high assessment accuracy due to the interference of yellowed vegetation. In regions with lower comprehensive vegetation cover, NRG image clustering extraction and NDVI threshold extraction can achieve higher accuracy, with NRG images being more conducive to visual interpretation, and the NDVI threshold extraction method being simpler and more efficient.

In order to understand the degree of influence of mines and their surrounding geological environment and conduct a reasonable evaluation, this article establishes a mining geological environment evaluation model based on a combination of subjective and objective evaluation methods. The article takes Binzhou City, Shaanxi Province as an example, combined with the natural geographical situation and development data of the mining area, and uses AHP-EWM-TOPSIS to establish a comprehensive evaluation model for the geological environment of mines. The evaluation of the geological environment impact of 10 mines in Binzhou City is relatively similar to the actual investigation situation, which objectively reflects the impact of mining activities on the geological environment of mines. The results indicate that the AHP-EWM-TOPSIS model can be effectively applied to the geological environment evaluation of multiple mining areas, providing strong technical support for the evaluation of regional mining geological environment.

At present, the scope of landslide geological disaster monitoring is limited, which leads to the decline of monitoring accuracy and monitoring stability coefficient. Therefore, the application of UAV tilt photography three-dimensional model in landslide geological disaster monitoring is proposed. Firstly, the paper analyzes the parameters and characteristics of unmanned aerial vehicle (UAV) oblique photogrammetry, normalizes the UAV oblique photogrammetry data and case data, saves the standardized data in network form, and obtains the preprocessing results of UAV oblique photogrammetry data; based on this, further setting the risk assessment level of landslide geological hazards and achieving landslide geological hazard monitoring. Set the risk assessment grade of landslide geological disaster to realize the monitoring of landslide geological disaster. The experimental results show that the monitoring accuracy of this method is above 99% at all monitoring points, and the stability coefficient is above 3, which shows that this method has higher reliability and stability in the monitoring process.

As an important part of China's grassland ecosystem, it is crucial to use remote sensing for high-resolution, long-term ecological monitoring and management of the Xilingol Grassland. This study uses Landsat data from May to September for the years 2013 to 2022 to estimate GPP, with a spatial resolution of 30m, and utilizes Slope and F-test methods to discern the spatial-temporal variations of GPP in Xilingol Grassland. The results indicate that over the past decade: 1) The multi-year average GPP value for Xilingol Grassland stands at 458.55 g C∙m^-2∙year^-1, with high-value areas mainly located in the northeastern and southern parts of the grassland, and low-value areas mainly located in the southwestern region. 2) The inter-annual average changes in GPP in Xilingol Grassland showed a pattern of initial decline followed by an increase. Among them, the highest proportion without significant change is 54.92%. 3) In particular, the meadow grassland mainly shows an increase or no significant change, accounting for 37.28% and 52.82% respectively. The typical grassland mostly shows no significant change, accounting with 61.8%. And desert steppes mainly showed a decreasing trend with 62.83%.

With the continuous behavior of human beings, the use and protection of soil have undergone tremendous changes ( Land Use and Land Cover Change, LUCC ), which has led to the diversity of the spatial structure of the earth and its surrounding environment. In order to understand the impact mechanism of land use change on carbon balance and explore how to reduce carbon emission through effective land use strategies, research on carbon emission estimation caused by LUCC change has been conducted. In order to estimate the carbon emissions caused by LUCC changes in Shenyang city, this paper carries out land use classification based on support vector machine classification in PIE remote sensing cloud platform, analyzes land use changes using GIS platform, and calculates the carbon emissions caused by land cover changes in Shenyang city through carbon emission modeling. During the period of 2000 to 2020, the terrestrial ecosystems of Shenyang city lost carbon stocks of about 2.8×106t of carbon. Therefore, for cities like Shenyang, effective land use strategies are needed to reduce carbon emissions and protect the health of surrounding ecosystems.

Hyperspectral water quality monitoring presents unique technical advantages such as real-time acquisition of water quality monitoring parameters, dynamic notification of water environmental quality status, and all-weather tracking of water quality abnormal information. It is an effective way to achieve surface water environmental pollution monitoring. The accuracy of conventional water quality inversion models is often based on large samples, which thereby restricts the advantages of hyperspectral water quality inversion in rapidness and efficiency. This paper discusses the inversion effects of PLS and SVM on TP and TN contents in a scene with only 12 samples. The results show that the optimal inversion models R² for TP and TN in the SVM method are 0.9 and 0.8, respectively. The PLS method has better inversion capabilities as the optimal inversion models R² of TP and TN can reach 0.95 and 0.89, respectively. At the same time, the regression equation in the PLS method can display specific characteristic wavelengths, which have a clear physical meaning. Through this research, the inversion model types are further enriched for the water quality spectrum online monitoring system.

In view of the irrational mining and low safety associated with joint mining in close distance coal seams, the rational staggered distance between working faces of the upper and lower coal seams during joint mining in close distance coal seams is explored in the geological context of coal 12 and 13^upper in Jinbao Coal Mine, Northwest China. Stable pressure zone (SPZ) mining and reduced pressure zone (RPZ) mining are studied using theoretical analysis and numerical simulation. Through calculations for SPZ mining and RPZ mining, it is determined that for joint mining in close distance coal seams, the safe staggered distance between working faces is 25.93–31.93 m. Numerical simulation of SPZ and RPZ mining methods based on Universal Distant Element Code (UDEC) shows that 25–45 m is the rational staggered distance between joint mining faces in close distance coal seams and it guarantees high mining safety. The numerical simulation result is basically consistent with theoretical analysis. This finding can provide useful reference for mining operation in coal seams with similar geological conditions.

Understanding the temporal and spatial evolution characteristics of meteorological droughts holds significant value for assessing disaster risk. This study analyzes the temporal and spatial features of drought in Shanxi Province from 1960 to 2022, based on the Standardized Precipitation Evapotranspiration Index. Mutation and trend analyses, were employed for this purpose. We also utilized correlation analyses to explore the relationship between meteorological drought and atmospheric circulation. The results show that: (1) Drought in Shanxi Province shows a clear trend of enhancement on both the annual scale and the seasonal scale. The time of drought mutation is predominantly concentrated in the 1970s and 1980s. (2) Droughts in the southern region depict an amplifying trend during the autumn, while those in the northern region exhibit an intensifying trend during all other seasons and on an annual scale. (3) Droughts in the summer and winter have a higher degree of correlation with the circulation index than those in the spring and autumn.

In order to accurately identify the characteristics of regional soil erosion, this paper proposes a method for identifying the characteristics of regional soil erosion based on unmanned aerial vehicle low altitude remote sensing technology. By installing high-resolution sensors on drones, multispectral remote sensing images of the target area are obtained, and advanced image processing and analysis methods are used to extract features related to soil erosion. The experimental results show that the regional soil erosion feature recognition method based on drone low altitude remote sensing technology can accurately identify the types of soil erosion, providing strong support for soil and water conservation and ecological environment protection. The conclusion indicates that this technology has the advantages of efficiency, accuracy, and flexibility, and can be widely applied in regional soil erosion monitoring and assessment, providing scientific basis for sustainable development.

Taking 8 major provincial capital cities and municipalities such as Changchun and Shanghai as the research objects, based on the Landsat8 L1 product in 2022-2023, we use the atmospheric correction method to carry out the inversion of summer and winter surface temperatures in the target urban administrative areas, and analyse the characteristics of the urban surface heat field differentiation; using SPSS, the data are analysed through stepwise regression, and the coupling between the urban heat field and the influencing factors is explored. The data were analysed by stepwise regression using SPSS to explore the coupling between urban thermal field and influencing factors. The results show that: ① the summer temperature between the east and west cities is lower in the east and higher in the west, and the temperature between the cities of similar latitude tends to be the same in winter; the temperature difference between the north and south cities is significant, which is more prominent in winter. ② The temperature in the city centre is higher than that in the suburbs in summer, the near-water area is lower than that in the far-water area, the cooling effect of the water system decreases with the decrease of latitude, and the temperature in the city centre is significantly lower than that in the outside in winter, except for Lhasa and Changchun. (iii) The eight independent variables in the summer model explain 69.8% of the summer urban heat field divergence characteristics, and the seven independent variables in the winter model explain 70.8% of the winter urban heat field divergence characteristics. In summer, the temperature decreases by 0.02℃ for every 1m increase in height, 0.003℃ for every 1km² increase in impermeable layer area, and 0.001℃ for every 1km² increase in water body area. In winter, the temperature increases by 0.01℃ for every 1 person/km² increase in population density, 0.05℃ for every 1km² increase in water body area, and 0.05℃ for every 1km² increase in vegetation cover area. 1km², the temperature decreases by 0.0125℃. in order to provide theoretical support for the study of urban heat field.

The article researches on the spatial morphology of traditional Bai villages in Dali based on GIS and Fractal theory. Firstly, six traditional villages on the west bank of Erhai Lake in Dali were selected for data collection. Secondly, these data were used to calculate, analyze, and discuss the fractal dimension of traditional villages by ArcGIS software. Finally, the results include: 1)the fractal dimension values at different scale levels depend on the self similarity between village boundaries and the environment. 2)the similarity of village boundaries, buildings, and roads determines the fractal dimension values of different scale levels in each village. And conclusions are summed up that GIS technology has expanded its application scope in traditional Bai villages in Dali, and has quantitatively analyzed the spatial form data of traditional villages.

Urban water network landscape and hydrological stability are directly related to the healthy development of urban landscape patterns and ecosystems. In this paper, major rivers within the fourth Ring road of Shenyang City are selected as the research object. Based on the high-resolution remote sensing image data of Landsat8 in 2013 and Sentinel-2 in 2021 and the elevation data of DEM in 2021, the landscape and hydrological changes of the water network in Shenyang City are analyzed with the technical support of geographic information system (GIS). The results show : the landscape structure of the Shenyang urban water network is not reasonable, and the water network has low linear connection characteristics and ring connection characteristics. The actual hydrological data of Shenyang changed greatly compared with the hydrological extraction results of DEM: a large number of natural river channels in the lower reaches of Hunhe River were artificially reconstructed, Puhe River received natural reconstruction many times, and the number of artificial ditches in Tiexi District was quite large. Based on the above analysis, combined with the current situation of the Shenyang water network, this paper puts forward reasonable optimization suggestions and corresponding protection and management remedies, to provide guarantee for the sustainable development of Shenyang city, in order provide reference for the planning and construction of water network in other cities.

Geologic hazards are frequent in the southwestern region of China, posing a serious threat to the safe operation of the transmission and substation system. Therefore, it is crucial to evaluate the susceptibility of geologic hazards along power transmission lines for effective hazard prevention and mitigation within the transmission system. This study focused on five districts and counties along the high-voltage transmission lines in Chongqing City, China. Thirteen evaluation factors were considered, encompassing elevation, slope, slope aspect, plane curvature, profile curvature, lithology, distance to fault, land cover, geomorphology, rainfall, NDVI, distance to waterway, and distance to road. These factors served as the basis for weight of evidence (WOE), random forest (RF), and support vector machine (SVM) evaluation indexes for geologic hazards susceptibility evaluation and visualization analysis. The results showed that (1) the AUC values of the area under the Receiver Operating Characteristic curve (ROC) for the WOE, RF, and SVM models were 0.870, 0.883, and 0.880, respectively, with RF having the highest accuracy. (2) The proportions of high and very high susceptibility zones obtained by the three models of WOE, RF, and SVM were 58.43%, 41.92%, and 50.50%, respectively. These proportions closely aligned with the distribution of geologic hazards. The evaluation outcomes provide a more precise reflection of the potential pattern of future geologic hazards incidence in the study area.

The Sky View Factor (SVF) and Green Visual Index (GVI) have significant impacts on the urban thermal environment and pedestrian comfort, serving as crucial parameters for studying the urban thermal environment. Baidu Street View can be used to acquire high-reliability panoramic street view images, enabling low-cost environmental assessments of urban environments. This article proposes a method to measure SVF and GVI at the street level based on semantic segmentation processing. By analyzing the SVF and GVI of each street, we investigate their spatial distribution characteristics. Additionally, by introducing Local Climate Zones (LCZ), we take Xi'an as an example to explore the characteristics of urban thermal environments in different LCZ types. Furthermore, the large-scale and precise SVF and GVI data are combined with specific land use and architectural conditions of each block for in-depth studies of block-level thermal environments. The results indicate the following: 1) There is an inverse relationship between SVF and GVI in the main urban area of Xi'an, indicating unbalanced development in road greening. There is a significant gap in GVI among different roads. 2) The buildings in the study area are primarily low to medium-rise, with a predominance of open types. Dense, low to medium-rise buildings (LCZ1-3) exhibit higher heat island intensities compared to open, low to medium-rise buildings (LCZ4-6)) There are significant differences in the relationships between SVF, GVI, and urban heat island intensity in various regions. Additionally, SVF and GVI in different types of urban land use and cover areas have varying impacts on heat island intensity.

The fractures produced by hydraulic fracturing will close to a certain extent under the action of formation stress, which will significantly affect the testing and productivity prediction of tight reservoirs. It is necessary to build an accurate model considering the fracture closure to predict the production characteristic of tight reservoirs. In this study, we propose a method to predict the productivity of staged fractured horizontal wells considering fracture closure. To address the research objectives, In most cases, there are many methods to solve the model, such as Stehfest numerical calculation, Pederosa algorithm, etc., which can realize the analysis and application of the model from different angles. These methods facilitate the analysis of the closing pressure's impact on fracture conductivity and the evaluation of fracture parameters' influence on well productivity. The results show that in the actual production process, the value of hydraulic fracturing parameters is not very high, but has a reasonable value range. Among the measures to increase the production of tight reservoirs, the most important ones are the preparation of fracturing method and fracturing design. One of the most important indicators to affect the fracturing effect is the fracture pressure in the fracturing system and the fitting curve of closed fracturing. If the fracture closure pressure increases, the fracture conductivity decreases more obviously. The key parameters of artificial fractures, especially the conductivity, will be a critical point. Of course, it is necessary to fully improve reservoir productivity and optimize various parameters of fractures.

The Rizhao section of Yishu fault zone has the storage conditions of medium-to-high temperature geothermal ,but the geological structure related to geothermal is not clear. In order to find out the development of geological structures in this region, and consider the difference characteristics of stratum density and magnetism in the study area, gravity and magnetic exploration were carried out on the Rizhao section of Yishu fault zone, with 1,618 high-precision gravity and magnetic survey points were completed respectively, covering an exploration area of about 80 km²,and the development of fault, spatial distribution and distribution of rock mass in the area were ascertained, and six fault zones (F1-F6) were found in the northeast and middle-southeast regions, and F4 was inferred to be closely related to geothermal, and the results of the gravity and magnetic exploration have a good consistency, which is of certain guiding significance for the investigation and evaluation of geothermal resources in this region.

Typhoon disasters have resulted in severe loss of life and property in Zhejiang Province. In order to further assess the risk level of typhoon disaster losses, this study takes Typhoon Lekima as the research object, considering risk indicators such as the risk factors of catastrophic factors, conceived disaster environment, and acceptor, and constructs a high-resolution typhoon disaster dataset based on kilometer grid units. Secondly, based on the disaster damage data of Typhoon Lekima, risk levels are delineated as output variables. Finally, a typhoon disaster risk assessment model is established using a convolutional neural network combined with the SENet attention mechanism. The results show that the model's validation accuracy exceeds 80%, and the introduction of the attention mechanism can also increase the model's evaluation accuracy by 1.46%. This achieves kilometer-scale disaster risk assessment of typhoon disasters, providing model data support for disaster prevention and mitigation efforts.

By collecting and sorting out the historical documents recorded in the 1949 period of Gansu Province to 2020, And according to the annual, district and county-level administrative division classification, With the use of mathematical statistics and interval method, Reestablished the history of earthquake events in Gansu Province, The spatial and temporal distribution characteristics of disaster-causing historical earthquakes in Gansu Province after the founding of the People's Republic of China are comprehensively investigated, Find out the high-incidence areas of earthquake disasters in Gansu Province and the key areas of comprehensive prevention and control, It can provide reliable historical disaster information support and guarantee for the national comprehensive disaster risk survey. The results of the analysis showed that, A total of 36 historical earthquakes in Gansu province from 1949 to 2020 (among which, Thirty three historical earthquakes were centered in Gansu Province, The epicenter of the earthquake was located in neighboring province), And the magnitude of 4.3 magnitude (including 4.3 level) above; From the time scale, The disasters were the worst in 2008 and 2013; From the spatial scale, Zhangye City has the most earthquake distribution in the disaster history, But the worst hit is in southeastern Gansu Province, Wenxian County of Longnan City and Minxian County of Dingxi City are the most prominent disasters, And distributed in the Lintan-Dangchang fault zone near the disaster history of the most earthquake; The results of the interval method showed that, The disaster historical earthquakes in Gansu province include 2 extremely heavy earthquakes, 4 severe earthquakes, 8 relatively heavy earthquakes and 22 general earthquake disasters.

Due to the unsatisfactory results of existing method for steel plants segmentation, we proposed a new method that integrates high-resolution remote sensing images, point of interest data (POI data) and OpenStreetMap data (OSM data) with the DeeplabV3+ model. This method was applied to Tangshan city, yielding the segmentation results for the steel plants in 2017 and 2022. Further analysis was conducted on the changes in Tangshan’s steel plants and the variation in temperature distribution within these plants. The results indicate that the DeeplabV3+ model, integrating remote sensing imagery, road network data and POI, can effectively increase the extraction of steel plants and maintaining a controlled decline in the recall rate. Low-temperature steel plants in Tangshan are all located in coastal areas, while mid-to-high temperature steel plants are primarily distributed in the inland regions of Tangshan.

Crop classification is a crucial step in crop monitoring and yield estimation. Integrating multi-source and multi-temporal remote sensing images can effectively enhance the accuracy of crop classification, but it leads to an increase in the dimensionality of features. It is necessary to employ appropriate feature selection methods and classification algorithms to improve the precision and efficiency of the classification. Therefore, this study takes the Sanjiang Plain as an example, combines multi-temporal features of Sentinel-1/2 images, and conducts research on crop classification methods using the XGBoost ensemble learning algorithm through feature selection methods. The research results show that the feature selection combined with the XGBoost algorithm yields the optimal classification results, reducing the feature dimensions from 195 to 33, with spectral features being the most important. The Kappa coefficient reached 0.941, which is 0.024 and 0.059 higher than RF and SVM, respectively. The error in crop area extraction results compared to the official statistical data is less than 14%. This method can accurately and efficiently perform crop classification, providing a reference for crop classification in the Sanjiang Plain and similar regions.

Since 2007, the continuous large-scale outbreaks of green tides in the Yellow Sea have caused severe impacts on the marine ecological environment and marine economy. Satellite remote sensing monitoring information provides crucial support for emergency decision-making in green tide disasters. The existing monitoring envelope for green tide distribution is drawn using buffer analysis, but the excessive number of vertices poses a significant computational burden for drift prediction. Therefore, this paper developed a method for green tide distribution envelope mapping based on remote sensing images. Firstly, the green tide coverage information was extracted using the NDVI threshold method. Then, the green tide distribution envelope was mapped based on coverage using buffer analysis. Finally, the envelope simplification algorithm based on edge length and vertex angle characteristics (SM-SAC) proposed in this paper was applied to simplify the distribution envelope. This model (SM-SAC) can effectively simplify the vertices, thereby improving the efficiency of green tide distribution drift prediction and providing technical support for developing emergency decision-making plans for green tide disasters.

In recent years, with the push of the "South-to-North Water Transfer Middle Route project a continuous northward flow of clean water" project, more and more attention has been paid to the protection of water source area, and further studies have been conducted on landslide in the geological disasters in the blocked river basin. In order to identify the importance of landslide spatial form for scientific landslide control, this paper takes Taotaoya shear layer landslide as the research object. Based on the fine tracing investigation of landslide landform and surface landmark features, combined with the analysis methods of geomechanics, rock structure and microscopic composition, the spatial morphological characteristics such as landslide perimeter and slope Angle are studied. By using ABAQUS finite element numerical simulation, the landslide evolution process was restored and its development trend was analyzed and predicted. The results show that the landslide has the formation conditions of easy soft rock formation, soft structural plane (zone) along slope and occurrence of groundwater. The mechanical evolution mechanism can be summarized as strength softening, gravity dislocation and shear slip. ABAQUS finite element studies the distribution of stress, displacement and plastic region of the whole structure to further verify the accuracy of development trend and fill the blank of finite element numerical simulation. Compared with conventional research methods, the above research results are unique, providing new reference ideas for the study of the same type of rocky slope and theoretical basis for engineering management, guiding the disaster prevention and avoidance of similar construction projects, and reducing the investment of similar projects.

Geological survey methods have been importantly applied and attempted in the field of mineral exploration. This paper compared and analyzed several typical geological survey methods and proposes a mineral exploration method based on UAV remote sensing technology and deep learning image recognition. By combining the super-pixel segmentation algorithm and CNN model, the efficiency and accuracy of remote sensing image feature recognition were improved, and the finding efficiency of mineral exploration was enhanced. The practical application results showed that the proposed method had a good performance in terms of mineral finding accuracy and applicability, and could make up for the shortcomings of traditional geological survey methods in mineral exploration.

The disaster of landslide seriously threatens the safety of people's life and property. Landslide deformation monitoring is the key to disaster prevention and reduction. We used Unmanned aerial vehicle (UAV) photogrammetry and interferometry synthetic aperture radar (InSAR), digital elevation of difference before and after the landslide, to analyze topographic changes, trace the evolution of deformation before the landslide, and monitor the deformation after the landslide. The result shows that The percentages of elevation lowering and elevation raising by volume are 55% and 45%. In addition, the evolution before landslide is a deformation accumulation process. The surface deformation after landslide show that landslide still deform. The cures of monitoring points distinct slow deformation stage and accelerated deformation stage.

Defects often emerged on diaphragm wall of pit foundation during construction period, which is a key problem that causes engineering disasters. A finite element model considering the leakage on diaphragm wall is established based on numerical method, to study soil settlement and lateral displacement of diaphragm wall caused by excavation and leakage under complicated geological conditions. The research shows that when diaphragm wall has is leakage hole, the soil increases significantly due to the consolidation of soil layer above the leakage hole, and the deeper the leakage hole is, the greater the maximum settlement resulted. When the depth of the leakage hole is 2.5m and 18m, the maximum settlement increases by 69% and 235% compared with that without leakage. Under the influence of increasing effective consolidation stress of soil layer above the leakage hole, an increase of horizontal displacement on diaphragm wall resulted, as the depth of the leakage hole increased from 2.5m to 18m, maximum horizontal displacement increases approximately 18% and 44% compared with no leakage conditions.

Low earth orbit (LEO) satellite communication system has been recognized as a key enabler for 6G. In LEO satellite communication systems, uplink performance is a key performance indicator (KPI) for evaluating the system performance. This paper studies uplink waveform performance for 6G. Since new applications give rise to new design challenges from different points of view. In particular, the uplink waveform shall be carefully designed so that different requirements from different applications can be met. In this paper, we first present three uplink waveforms, which are the discrete Fourier transform spread orthogonal frequency division multiplexing (DFTsOFDM), the zero tail DFTsOFDM (ZT-DFTsOFDM), as well as the unique word DFTsOFDM (UW-DFTsOFDM). Then, we detail the evaluation metrics, which are the system spectral efficiency, the waveforms’ peak-to-average power ratio (PAPR), as well as the waveforms’ out-of-band emission (OOBE). Finally, system performance is evaluated in two application scenarios, which are the low earth orbit (LEO) satellite communication scenario and maritime wireless communication scenario, showing that a flexible uplink waveform is desired for 6G.

Analyzing the GPR spectrum characteristics of underground loose bodies can help identify diseases in such bodies and provide technical basis for underground hazard control. Due to the concealment, diversity, and complexity of loose bodies in practical scenarios, previous research has mainly focused on experiments and software simulations. This article applies a combination of model simulation and measured data analysis to study different types of loose bodies and analyze their spectral response patterns. The study results show that the GPR reflection waves of loose bodies have strong top reflection characteristics, and the top reflection waves also have multiple wave characteristics, with chaotic internal waveforms; The better the looseness of the loose body, the stronger the reflected wave, the larger the scale of the loose body, and the stronger the GPR reflected wave. Conclusion: It is feasible and effective to use GPR to identify underground loose body.

Shiyan Reservoir is located in a typical high-density built-up area in Shenzhen, and its water environment quality and water ecological health are key to the safety and security of Shenzhen's drinking water. The 13-way water quality and water ecological indicators of Shiyan Reservoir were monitored for 14 months. The chlorophyll a concentration in Shiyan Reservoir ranged from 21.25 to 88.77 μg/L. The temporal heterogeneity of the chlorophyll a concentration was obvious, and the variation was drastic; the highest chlorophyll a concentration was from April to July, which was maintained at about 85 μg/L, and the risk of algal bloom was high. The annual average water temperature of Shiyan Reservoir is 25.31℃, and the water temperature from May to September is about 30℃ for a long time, so the risk of algal bloom is high. The COD of the reservoir is 1.71~3.08 mg/L, and the TOC is 2.22~5.13 mg/L. The organic pollution of the reservoir is very low, and it reaches the environmental quality of surface water Ⅰ standard. Transparency is 0.73~1.16m, relative to 2010~2015, the transparency of Shiyan Reservoir increased significantly. Shiyan Reservoir TN pollution is serious, during the investigation period, TN is 1.66~2.49 mg/L, for Class V water quality. In recent years, Shenzhen has carried out a systematic water environment comprehensive treatment project, TN decreased significantly, but it is still the primary pollutant in Shiyan Reservoir. Shiyan Reservoir TP 0.01 ~ 0.043 mg / L, total phosphorus concentration is low, to achieve the environmental quality of surface water Ⅰ ~ Ⅱ class standards. The monthly average integrated trophic state of the reservoir is located between 43.13 and 53.81, with an annual average value of 50.49. The monthly average integrated trophic state of April and November 2021 and April and May 2022 is less than 50, which is in the state of light eutrophication, and the other months are in the state of moderate eutrophication. The correlation analysis showed that water temperature was the primary influence factor of phytoplankton in flood season, and the correlation coefficient was as high as 0.69. The main correlation factors of phytoplankton in flood season were WT, TOC, nitrate nitrogen, and SD, and the main correlation factors of phytoplankton in dry season were TOC, DO, and WT. Throughout the whole year, phytoplankton was highly positively correlated with WT, TOC, moderately positively correlated with COD, pH, and turbidity, and moderately negatively correlated with transparency (SD). transparency (SD) was moderately negatively correlated. Two-factor ANOVA showed that the variance of each environmental factor was better explained by the period and sampling site and the interaction effect between the two (R² ranging from 23.7% to 78.1%). The main influence of environmental factors in Shiyan Reservoir was period, and the temporal heterogeneity of environmental factors was significantly higher than the spatial heterogeneity. A total of 21 genera of cyanobacteria, 24 genera of diatoms, 39 genera of chlorophyta, 3 genera of euglenophyta, 2 genera of pyrrophyta, 1 genus of chrysophyta and 2 genera of cryptobranchs were detected in Shiyan Reservoir. Cyanobacteria dominated from May to October, and diatoms dominated from November to February. There were 9 species of cyanobacteria were dominant in Shiyan Reservoir, including Raphidiopsis, Oscillatoria, Phormidium, Planktothrix, Anabaena, Merismopedia, Chroococcus, Microcystis and Snowella.

The basic volcanic rock gas reservoir in Xushen gas field is controlled by volcanic bodies, and the reservoir has the characteristics of poor horizontal continuity, thin vertical thickness, and small planar development range. Part of the volcanic bodies No. 7, No. 10, and No. 15 are developed within the confirmed boundary of Block X23. There is only one well, namely X23, on the No. 10 volcanic body within the boundary, and the lack of well data makes it difficult to integrate geological engineering modeling. To this end, the actual drilling on the three volcanic bodies outside the block boundary was all included in the research area, and the basic volcanic rock gas reservoir structure modeling, attribute modeling, fracture modeling, and stress modeling were carried out through the combination of well and seismic control, forming a geological engineering integrated modeling technology that combines single and composite modeling, and integrates multiple information. When the well is deployed inside a single volcanic body, the single geological model is mainly used, When the well is deployed at the contact area of the volcanic body, the composite model is mainly used. The integrated model can guide well deployment and drilling guidance, making well deployment more scientific and reasonable, drilling guidance more accurate, and improving reservoir drilling rate from multiple angles. The pilot test well P1 has a reservoir drilling rate of 73.6%, which is 27.1 percentage points higher than the average. The integrated model can also guide the design of fracturing schemes and improve the fracturing effect. A total of 40 clusters were fractured in the P1 well, with 35-38 clusters opened and an opening rate of 87.5% -95%. The fracturing effect is significant. The integrated modeling technology of geological engineering has achieved good application results in the development of basic volcanic gas reservoirs in Block X23. The initial daily production of P1 well was 120000 cubic meters/day, which is 3-4 times that of similar gas wells. This research approach can be borrowed from other medium basic volcanic gas reservoirs to improve the development efficiency of gas reservoirs.

The potential of solar energy that buildings accept is one basis for building and evaluating a solar city. Improving the accuracy of assessment results at the medium to long-term scale holds greater significance. This article develops a framework with the viewshed model as its core for estimating the solar energy potential of urban building roofs. Experimental validation conducted at Wuhan University of Technology Nanhu New Campus demonstrates that the estimation framework achieves a balance between accuracy and efficiency.

Aiming at the lack of detailed comparison of existing airborne LIDAR point cloud filtering algorithms. In this paper, taking the surface subsidence area of a working face in the western mining area as the experimental area, three classical filtering algorithms, progressive morphological filtering, triangulation progressive encryption filtering and filtering based on slope threshold, are selected to denoise, and the applicability of the algorithms is analyzed. The experimental results show that in the mining area of northern Shaanxi, where the surface fluctuation is obvious and the vegetation is sparse, the point cloud modeling effect after denoising is the best by using triangulation progressive encryption filtering. The test results can provide important technical means for fine modeling of mining subsidence in mining areas.

The way urban climates are classified affects both sustainable urban development and environmental planning. Local Climate Zone (LCZ) classification offers a comprehensive framework to classify different urban areas based on their climate-related characteristics. This paper investigates the application of deep learning techniques for LCZ categorization using multispectral Sentinel-2 satellite images. Sentinel-2's capacity to record optical data over a wide range of spectrum bands makes it an invaluable tool for understanding variations in urban climate. This study uses a deep learning model called convolutional neural networks (CNNs) to effectively extract and learn spatial attributes from the multispectral Sentinel images. The work uses a labeled dataset with Sentinel images for training the model and classifications of LCZ. During the training phase, the model parameters are tuned to enhance the interpretability of climate-related patterns in urban environments. Using a validation dataset, classification metrics such as accuracy, precision, recall, and F1 score are used to evaluate the model performance. These conclusions offer useful information to environmental scientists, urban planners, legislators, and those involved in climate-resilient urban design. This demonstrates the efficacy of using multispectral and SAR images for precise LCZ categorization, advancing our understanding of the variability of urban climate and assisting planners in making well-informed decisions regarding urban development strategies.

Currently, most of the domestic numerical weather prediction systems in China only assimilate satellite microwave sounding observations over sea. There is a need to further explore the potential applications of satellite observations in the sensitive sounding channels over land. This is in line with the practical operational development needs of numerical weather prediction in China, and several approaches were proposed in this paper. These approaches aimed to solve the basic problems of accurate inversion of surface emissivity and cloud detection, including a dynamic surface emissivity calculation method constrained by wavelet data set, and an intelligent cloud detection algorithm based on GPR (Gaussian Process Regression). These methods were developed to enable the practical application of assimilating land-over satellite data in the surface-sensitive sounding channels. The research results show that: The number of land-over data entering the assimilation system was found to be equivalent to the number of sea-surface observations, significantly expanding the assimilation data source. The 24-hour and 48-hour forecast effect of heavy precipitation was found to be better than that of the operational model.

Remote sensing image segmentation of land and sea is of great significance for the development and monitoring of coastlines. Traditional methods of coastline recognition and extraction cannot accurately extract image features when faced with irregularly textured remote sensing images, and the segmentation results are not ideal. This paper uses convolutional neural networks to solve the problem of land-sea segmentation in remote sensing images, and innovates on the basis of the classical U-net structure network model. In the encoder, the global response normalization is added to improve the edge weight in the image, which significantly improves the segmentation accuracy. The experimental results of muddy shoreline using landsat-8 remote sensing image data in the study area show that the segmentation effect of the improved network model is better than that of U-net, Deeplabv3 and U2net.

This paper adopts GEO5 geotechnical series software and focuses on parametric BIM modelling with typical push-type soil landslides. By combining the numerical simulation technology of BIM software with the on-site situation of the landslide control project, the key contents of the corresponding engineering treatment scheme are obtained according to the numerical simulation results. It reflects the important value of BIM technology in landslide control. It is helpful to explore how BIM technology transforms from simple model image displays to solving practical engineering problems based on parametric modelling.

The types of wetland resources in lake areas are diverse and complex, and the differences in spectral characteristics of different features can easily lead to a decrease in recognition and monitoring capabilities when extracting wetland remote sensing image information. This article is based on Landsat remote sensing image data. Through image preprocessing and enhancement, spectral and texture features are extracted. Artificial neural networks, maximum likelihood method, random forest and other algorithms are used to extract and classify wetland information, and the classification accuracy and effectiveness of the three methods are compared. The experimental results show that the random forest method reaches 85.18% overall recognition accuracy and 0.8149 Kappa coefficient, which is the most effective in extracting and classifying the features in the lake area. Further extracting the multivariate features of remote sensing data such as spectral features and texture features, the experimental results show that the overall classification accuracy reaches 91.45% and the Kappa coefficient reaches 0.8927, and the classification accuracy is effectively improved.

This article conducts numerical simulation research on the instability problem of dangerous rock masses in Zhayeba Temple based on Rockfall software. Firstly, a three-dimensional geological model is constructed through unmanned aerial vehicle oblique photogrammetry. Secondly, Rockfall is used for kinematic simulation of the unstable rock mass to analyze its motion characteristics. Based on the simulation results, two typical Areas are divided for secondary simulation. The results indicate that the movement of falling rocks in region 1 is a rolling collapse, with a bouncing height of 0 meters and a maximum kinetic energy of 52.55 Kj. Most of the falling rocks in Area 2 are jumping collapses, with the highest probability of bouncing at a height of 1.19 meters and the highest probability of kinetic energy around 101.72-183.10 Kj. Due to the impact of the collapse that has occurred in the upper part, some falling rocks have an energy of 712.05 Kj and a bouncing height of 25 meters. Although the impact force of falling rocks in Area 1 is relatively small, as the rolling distance increases, the threat to Area 2 is much greater than the impact on BT2. This study provides a reference for the development of targeted governance measures and is of great significance for disaster prevention and reduction engineering.

Inclined photogrammetry, as one of the main data collection methods for new mapping, can not only obtain texture images and geometric information of feature elevations, which cannot be obtained by conventional aerial photography, but also quickly establish three-dimensional models. In this paper, the UAV tilt photography method was adopted, and software was used to realize the real-life 3D modeling of the study area, and the whole process of applying the technology was investigated and the quality of the 3D model was assessed. The results show that the model meets the relevant specification requirements, and it can be used for the topographic map surveying and mapping. The overall effect of the 3D model modelling is good.

The Standardized Precipitation Index (SPI) accurately reflects the drought phenomenon, closely matches the real situation, and proves to be a reliable drought monitoring tool. This paper employs geodetector methodology to explore the drivers of spatial heterogeneity in the SPI, calculated using 1-km resolution Global Precipitation Measurement (GPM) data, within the Yellow River Basin from 2001 to 2019. Besides precipitation, which emerges as the predominant driver influencing the SPI, the individual explanatory power of all other factors is notably weak, each contributing less than 20%. However, it is observed that the combined influence of multiple factors on SPI₁₂ generally surpasses that of any single factor. Notably, the interaction between altitude and precipitation exhibits the most substantial effect. These findings hold significant implications for the efficient management of water resources and the preservation of the ecological environment within the Yellow River basin.

Image-based wildlife recognition is a crucial means for wildlife monitoring. With the advancement of computer vision technology, image-based wildlife recognition using deep learning has become an important method. However, due to the scarcity of wildlife image data, especially when collecting images of rare animals, applications of deep learning-based image wildlife recognition face issues such as limited samples and difficulty in model convergence. In response to this problem, in this paper we propose an image-based balanced hierarchical relationship learning (BHRL) method for fewshot wildlife object detection. This method overcomes the challenge of scarce image samples for detecting wildlife targets in outdoor environments, enhancing the accuracy of image-based wildlife target detection under small sample conditions. The experimental results demonstrate that when using only 20 reference images, the proposed method achieves a MAP50 accuracy of 73.51% for detecting Eurasian spoonbill, male Elliot's pheasant, red fox, tufted deer, and macaque. Compared to the few-shot learning meta-DETR method, the proposed method achieves an improvement of 12.23%, while Faster R-CNN and YOLOv5, which rely on abundant samples, achieve improvements of 35.1% and 28.58%, respectively. The experimental results further validate the challenge of applying deep learning-based target detection methods under small sample conditions. The proposed few-shot learning method based on balanced hierarchical relationship learning enhances the model's reasoning ability, making it suitable for wildlife target detection under small sample conditions. This method holds potential practical value in the fields of wildlife monitoring and ecological research.

Revealing the spatial and temporal evolution characteristics of drought in the southwest region is vitally important for drought early warning and ecological environmental protection. Based on the standardized precipitation evapotranspiration index (SPEI) for the 1982-2020 period of the southwest region, we analyzed the interannual and seasonal temporal characteristics of drought by means of the Theil-Sen median trend analysis, Mann–Kendall test and wavelet analysis methods. The results showed that (1) the interannual SPEI in the southwest region has decreased over the past 39 years, showing a drying trend; the seasonal SPEI showed a decreasing trend, and the rate of decrease in winter was 0.196/10a, which is significantly higher than that in the other seasons. (2) The main cycles of drought at the annual scale were 39a, 30a, and 14a.The primary periods of each season exhibit variability and differ across various temporal scales.

This paper proposes a deep neural network named SAR Latent Matching Network (SARLMN) for SAR 3D reconstruction from a single SAR image. SARLMN consists of three components: point clouds encoder, point clouds decoder, and SAR images encoder. The encoder is responsible for mapping the images and the point clouds to the latent space, and the decoder is responsible for reconstructing the 3D point clouds from the latent space. The network realizes the 3D reconstruction of SAR images by narrowing the distance between the image and point cloud in the latent space. Meanwhile, in order to solve the problem of insufficient data, we make a simulation dataset for the network training and verify the capability of SARLMN on 3D reconstruction.

Albedo is a key factor characterizing and driving climate change and surface mass loss in Antarctica. To obtain a spatial-temporally continuous Antarctic albedo product and verify its quality, this paper carries out a filling procedure for the MCD43A3V061 albedo product for the summer half of the 2001-2020 period (October to March each year). Considering the correlation between albedo and snow grain size, elevation, and meteorological factors, this paper uses the random forest method to establish a regression model between them, so as to use the spatial-temporally continuous snow grain size, elevation, and meteorological data to calculate the vacant value of the albedo product. Therefore, a spatial-temporally complete and continuous Antarctica twenty-year albedo product can be reconstructed. The validation results show that the random forest-based albedo (RF-albedo) exhibits good agreement and correlation (RMSE=0.027, R=0.864) with the original MCD43A3V061 product. Compared with in situ data, the RF-albedo quality (RMSE = 0.048, R = 0.569, Bias = - 0.024) is nearly the same as MCD43A3V061 (RMSE = 0.047, R = 0.517, Bias= - 0.026).

The heatwave (HW) magnitude varies with the combination of HW indicators, and quantifying the HW magnitude is crucial for HW disaster risk mitigation and sustainable economic and social development. In this study, a Comprehensive Heatwave Index (CHI) was constructed to quantify the HW magnitude on the Tibetan Plateau (TP) from 1979 to 2022 by considering the collective utility and equilibrium utility of HW indicators. The results demonstrate that the CHI effectively illustrates the spatiotemporal characteristics of HWs on the TP, with a significant increase in HW magnitude, particularly in the western TP glacial region, which is the hotspot and significant intensification region of HW magnitude. The research findings provide scientific insights for implementing more effective measures to respond to global climate change and prevent extreme disaster risks on the TP.

As a common oceanic phenomenon, Mesoscale eddies play a significant role in oceanic heat exchange and energy transfer in the South China Sea. However, satellite remote sensing is primarily utilized for measuring surface ocean data, posing challenges in detecting subsurface eddies. This study proposes a data-driven deep learning approach to address the deficiency of subsurface oceanic data. This method leverages satellite surface data to establish a mapping model between surface satellite data and subsurface temperature. The proposed Unet-Eddy model can successfully invert the three-dimensional temperature of mesoscale eddies at depths of 0-1000 m. Experimental results demonstrate the potential of this approach in inverting the three-dimensional subsurface temperature of mesoscale eddies and partially compensating for the shortage of internal oceanic data.

With the frequent occurrence of extreme natural disasters, power grid systems are facing increasingly severe challenges. The aim of this study is to analyze the impacts of geohazards on the safe operation of power systems, especially the threats posed to transmission lines. To this end, we constructed a hazard vulnerability map using a logistic regression model based on seven key internal factors, such as DEM, slope direction, slope gradient, road, water system, soil type, and NDVI. In addition, considering precipitation as an external trigger, this paper further combines internal and external factors to construct a geohazard risk mapping. By combining the geographic locations of transmission lines in New England with these maps, we evaluated the hazard factor of each transmission line. The results show that the geohazard risk faced by transmission lines can be effectively predicted and quantified by this method, which provides an important reference for grid planning and disaster prevention and control.

This paper focuses on the global environmental issue of light pollution. Using remote sensing data and related side data, the light pollution risk assessment models are established for five latitudes: economy (ECO), health (HEA), environment and sustainability (SAE), climate and topography (CAT), and regional light index (RLI). First, Topsis method is used to construct a light pollution risk assessment model. In order to develop the light pollution risk coefficient system, five first-level indicators and 11 second-level indicators are identified based on the five latitudes identified above. Considering the limitation of single weight calculation method, entropy weight method and coefficient of variation method are combined to obtain comprehensive index weight. Secondly, based on the grey prediction model optimized by BP neural network, the light pollution degree of Shanghai, China and Mount Nimba, Guinea in the next 7 years is predicted, and specific optimization strategies are proposed. Finally, the robustness of the model is proved by sensitivity analysis.

The simulation of various land use change scenarios holds significant importance for achieving dual carbon objectives. This study zeroed in on Qinghai Province, exploring the spatial and temporal evolution as well as future trends of carbon storage through a blend of the InVEST model's carbon storage module and the PLUS model, utilizing land use data from 2000, 2010, and 2020. Our results highlighted the dominance of grassland and unused land in Qinghai, with minimal changes in land use patterns over the past two decades. Notably, construction land witnessed the steepest increase, while cropland and grassland acreages underwent the most significant decrease. Under a natural development scenario, cropland, grassland, and construction land decreased, whereas forest land, water bodies, and unused land expanded. Conversely, in the cropland protection scenario, cropland acreage increased, thereby safeguarding food security. The ecological protection scenario led to an increase in forest land and grassland, effectively preserving the ecosystem. Remarkably, the total carbon storage in Qinghai Province remained stable across the years 2000, 2010, and 2020, with grassland emerging as the top contributor. Higher carbon storage concentrations were observed in the eastern and southern regions. Projecting forward to 2030, the total carbon storage in Qinghai Province exhibited varying trends of decline or increase depending on the scenario, with the ecological protection scenario emerging as the most favorable for future development. These insights provide a crucial understanding of the spatial and temporal patterns of carbon storage in Qinghai Province, serving as a scientific reference for implementing sustainable development strategies, accelerating ecological construction, and charting the dual carbon strategy in Qinghai Province.

Soil salinization seriously affects the agricultural construction and development in arid and semi-arid areas. The feasibility of inverting a soil salt content model for accurate monitoring of regional soil salinization was investigated through spectral transformation and feature optimization of hyperspectral data in Ganzhou district, Zhangye City, Gansu Province. The original spectral reflectance (R) of soil samples with varying sodium sulfate content was measured using a spectral radiometer, and 11 different types of spectral transformation and optimization techniques were applied to R. The subsequent steps involved the utilization of partial least squares regression (PLSR) and support vector machine (SVM) techniques to develop hyperspectral data inversion models for varying soil salt content. The model prediction results demonstrate that the utilization of hyperspectral data after spectral transformation and feature optimization as input variables significantly enhances the predictive capability of both PLSR and SVM models. The R²_v of the PLSR model under R" processing is improved by 0.416, and the accuracy of the SVM model under √ R" processing is the best and R²_v =0.897. The hyperspectral data inversion model, combined with spectral transformation and feature optimization, effectively predicts soil salt content and establishes a solid research foundation for quantitative estimation and accurate monitoring of soil salinization in this region.

Floating/emergent aquatic vegetation (FEAV) and algal blooms (AB) play different roles in regulating lake ecosystems. High spatiotemporal information is crucial for assessing lake status. However, distinguishing between AB and FEAV using satellite remote sensing poses challenges due to their similar spectral characteristics. To address this, we propose an automatic classification algorithm (BF algorithm), incorporating multiple spectral features, based on Sentinel-2 imagery. Validation in Lake Taihu yielded an average overall classification accuracy of 93.00% for the BF algorithm. Subsequently, the algorithm was successfully applied for long-term monitoring (2019-2022) of AB and FEAV in Lake Taihu. This algorithm holds prospects for long-term monitoring in small to medium-sized eutrophic lakes.

This paper employs the SBAS-InSAR analysis technique and Sentinel-1 satellite data to monitor deformation in the vicinity of Highway, spanning roughly 2 kilometers. It discovers significant settlement deformation in localized areas due to coal mining voids, with the maximum deformation velocity reaching 182 mm/year and the largest deformation at 171 mm. Identified near ZYT City and EGH City, close to highways, this deformation poses a risk to highway safety. Utilizing InSAR technology, widely applied for measuring deformation fields, this research aims to assess the stability and deformation around the highway near the collapse zones. It seeks to support decision-making and enhance highway construction and maintenance safety by providing critical deformation data.

Channel is the foundation of water transportation. With the progress of society, the development of industrial enterprises, the scale of urbanization construction, and the continuous expansion of waterway route construction, the extension of the service cycle of waterways, the water environment of waterways is becoming more and more severely damaged, and it is inevitable that some silt blockages will occur, affecting the normal use of waterways. Channel dredging is very important, and underwater terrain measurement plays an important role in channel dredging. This article mainly introduces the application of underwater terrain measurement in channel dredging. On the basis of analyzing underwater measurement technology and the development history of unmanned measurement ships, this paper introduces the data collection and processing methods of unmanned measurement ships. Taking the use of unmanned survey ships for underwater terrain measurement in a certain waterway dredging project as an example, a comparison was made between the terrain before and after the dredging process. The results indicate that unmanned measurement ships have certain feasibility in channel dredging measurement. Regularly using unmanned survey ships to conduct underwater topographic surveys of dredging conditions, mastering the changes in sediment accumulation in the waterway, is beneficial for the maintenance and management of the waterway.

The stability of foundation pits is crucial for the large-scale development and construction of underground spaces and high-density buildings. Analyzing the settlement amount and rate of structures can offer scientific guidance for construction. Through the utilization of small baseline set temporal interferometric radar (SBAS-InSAR) technique, we were able to determine the average settlement rate and vertical cumulative settlement of University Chengbei subway Station in Panyu District, Guangzhou over a 3-year period (April 7, 2019 to April 3, 2022). The findings indicate that the deformation rate in this area is at 6mm per year. Verification results demonstrate that the measured level data support the monitoring outcomes, confirming the feasibility of this technology in foundation pit monitoring.

Pine Wilt Disease (PWD) poses a serious threat to forest resources. This study proposes an improved YOLOv5 algorithm aimed at the remote sensing image detection of this disease. By introducing a multi-scale detection mechanism and attention mechanism, the improved model significantly enhances the recognition capability for small targets. Experimental results show that the accuracy of the improved model reaches 89.84%, which is a significant improvement over traditional methods and the original YOLOv5. This method provides effective technical support for the rapid and accurate detection of PWD, which is of great significance for forest disease prevention and control.

In view of the large amount of production data and high degree of automation, there is an urgent need for a 3D geographic entity quality inspection tool to solve the problems of large workload of data input, more manual operation and low efficiency of quality inspection. This paper from the content of 3D geographical entity quality inspection, quality inspection method, quality inspection information output way, studied a based on visual rendering and AI intelligence system, most quality elements, and automatic output quality inspection report, improve the efficiency of quality inspection, ensure efficient and accurate quality inspection task.

The safety of geological storage of carbon dioxide, as one of the key technologies for reducing greenhouse gas emissions, has received widespread attention. The injection of CO₂ may cause fault instability and lead to CO₂ leakage, so it is necessary to evaluate the fault stability of CO₂ storage. In order to investigate the fault stability of CO₂ geological storage, various injection rates are set up, and the pore pressure, von Mises stress and shear stress are used as evaluation indexes to investigate the fault stability based on the Mohr-Coulomb criterion. It is found that the fault remains stable when the static friction coefficient of Mohr-Coulomb criterion is 1, and the fault becomes unstable when the static friction coefficient is 0.47, and the instability of the fault becomes more obvious with the increase of the injection rate, which provides theoretical references for the assessment of the stability of the fault of the geological storage of CO₂.

Quickly and accurately obtaining water body information and its distribution is helpful for rational planning and management of water resources, and has important value for national resource management, planning and development, and rapid disaster assessment. With the continuous development of remote sensing technology, its application in water extraction is becoming increasingly widespread and in-depth. However, further comparative research is needed to improve the accuracy and efficiency of applying remote sensing technology to water extraction. In order to seek accurate and fast water extraction methods, the Sentinel-2 satellite remote sensing data and UAV aerial photography data were respectively used to extract water information, and the accuracy evaluation was carried out in combination with Google Earth images. The results show that Sentinel-2 satellite remote sensing data combined with normalized differential water index (NDWI)^[1] and improved normalized differential water index (MNDWI) can extract large water bodies more accurately, but the performance of small water bodies is not good. In contrast, UAV aerial images combined with maximum likelihood classification algorithm show high precision water information extraction ability, especially in the identification of small water bodies.

Studying the activity of salt diapirs is a highly complex task. Here, we addressed this challenge by utilizing eight years of Sentinel-1 satellite ascending and descending orbit data to decompose two-dimensional annual subsidence deformations in both the east-west and vertical directions. Employing statistical methods, we extracted and analyzed the pixel values of two-dimensional deformation in ten salt diapirs. Leveraging prior conclusions of activity in seven of these diapirs, we predicted and validated the activity of the remaining three diapirs with unknown activity. Ultimately, we confirmed that all three of these diapirs exhibit features indicative of activity.

GaoFen satellite data, with its excellent data quality and resolution, has become the preferred data for remote sensing applications and has been widely applied [1]. Remote sensing monitoring of mines in China has achieved quarterly or monthly monitoring of key mines nationwide. This article uses GaoFen data to carry out mining land information extraction work in Laos. By comparing the extraction results of different years, the application purpose of satellite dynamic observation of changes in mining activities was achieved. The results indicate that the remote sensing dynamic observation effect of GaoFen satellite data is also good overseas.

Land remediation planning is a prerequisite and essential foundation for implementing land remediation work. Traditional two-dimensional planning maps are difficult to accurately represent changes in height, depth, and perspective, which may lead to misunderstandings or oversights in spatial relationships during planning and design. Conventional manual 3D modeling is time-consuming, labor-intensive, and may lack accuracy in depicting real scenes. In recent years, low-altitude unmanned aerial vehicle (UAV) aerial photography for real scene 3D modeling has been widely applied. In the context of land remediation planning, a method based on UAV real scene 3D for rapid construction of Building Information Modeling (BIM) components is proposed. Firstly, tilted aerial photography data is collected using a drone, and real scene 3D modeling is carried out in ContextCapture software. Next, the simplified real scene 3D model is imported into Autodesk InfraWorks for spatial registration. Finally, taking the rapid modeling of irrigation pipes as an example, rapid construction of BIM components is achieved. Practical results show that this method effectively enhances the representational capabilities of land remediation scenes. More importantly, it significantly improves the efficiency of 3D modeling for planning BIM components, providing valuable guidance for similar project work.

Currently, there is a rising tide of digitization and intelligence in the field of transportation. Concurrently, novel surveying technologies have brought about vast amounts of high-precision terrain data. However, domestic three-dimensional software development started later and progressed more slowly. The construction of three-dimensional scenes for large-scale projects such as roads, bridges, and tunnels is not yet well-developed, and it fails to fully utilize the data information brought about by surveying technologies. The difficulty of real-time roaming and rendering of large-scale terrain models is exacerbated by the abundance of data. To address this challenge, a solution is proposed based on the Delaunay triangulation. This solution centers around high-precision data and involves the implementation of surrounding elevation point supplementation, aiming to establish a high-quality, high-precision, and high-level three-dimensional terrain. On the other hand, a terrain Level of Detail (LOD) structure is established based on quadtree segmentation, enhancing model interaction calculation speed and improving rendering and roaming effects. This solution successfully constructs an efficient and reliable three-dimensional terrain model, driving the digitization and intelligence development of large-scale engineering projects. The practical application of this approach has yielded excellent results

With the development of science and technology and the popularization of informationization, geological informatization has become an important development direction of the current geological industry. In this paper, by studying the specific practical application and technical basis of geological informatization in the field of geology, we put forward the problems and challenges faced by geological informatization at present, and based on the application of big data and cloud computing in geological informatization, especially in the integrated management of geological informatization data, we constructed a simple integrated data management software for geological information data by using computer language program. Geological information was also entered into the route of the Sichuan-Tibet Railway - Linzhi section based on the developed software, and the integration of geological data on the probability of occurrence of mudslide disasters in various areas of the railroad in the Linzhi section was completed. Finally, according to the development trend of international and domestic geological field, the research on the development and construction of geological informatization and geological informatization software is prospected to support the sustainable development and economic construction of geological industry.

Desertification is a major global ecological problem affecting human survival and development. China is one of the most desertified countries in the world, and the desertified land is mainly distributed in the three northern regions. In this paper, remote sensing technology is used to study the annual images of Bayannur from 2018 to 2022 from different angles, and annual data of desertification area, vegetation change area and net ecosystem productivity of Bayannur in recent years are extracted. Finally, three different methods were used to extract the data of Bayannur shelterbelt and forecast the future trend of Bayannur shelterbelt. This reflects the relevant situation of Bayannur Shelterbelt in 2024-2026 to a certain extent, and provides help for the country to cope with the desertification crisis.

The North China Plain is one of the major grain producing areas in China, and most of the farmland in the grain growing areas has been in the state of "fragmentation and heterogeneity" for a long time, which makes it more difficult to recognize and extract typical crop features in the remotely sensed images. This study takes Gaofen-6 remote sensing images as the research object, takes Beijing-Tianjin-Hebei as the research area, combines the characteristics of farmland landscape zoning, and utilizes artificial intelligence algorithms to realize the remote sensing intelligent interpretation of winter wheat. The remote sensing identification of crops is affected by the distribution of crops, and this paper screens four indices that can characterize the spatial distribution of crops from the spatial structure of the landscape, spatial characteristics, and landscape diversity indices. Combined with the k-mean mean for cluster analysis, the study area was divided into four different farmland type zones. Random Forest (RF), U-Net and Deformable Full Convolutional Neural Network (DFCN) were utilized to extract winter wheat from the four farmland type areas, determine the optimal decoding method for each farmland type area, and finally combine the optimal models of the four farmland type areas to achieve the optimal decoding method. The optimal models of the four types of farmland areas are combined to realize the global optimal decoding of the study area. The research results show that: (1) the quantitative description index of crop distribution can be realized by four landscape indices, namely: shape factor LSI, area-weighted average patch dimension FRAC_AM, agglomeration index AI and uniformity index SHEI; through the k-mean clustering classification method, all landscape units can be clustered and analyzed, which can be classified into four typical farmland zoning types. types of farmland zoning. (2) DFCN, U-Net and RF were used to decipher winter wheat in each farmland type area. According to the comparison of pixel accuracy, the optimal model for winter wheat recognition in the first, third and fourth types of areas was DFCN, with an accuracy of 95.8%, 94.1% and 95.3%, respectively; and the optimal model for winter wheat recognition in the second type of areas was RF, with an accuracy of 93.6%. The idea of partitioning before decoding in this paper improves the decoding accuracy of winter wheat compared with the remote sensing decoding of winter wheat with the same method for the whole study area, and provides reference value for crop classification under the complex terrain conditions.

Mapping accuracy is crucial in determining the quality of unmanned aerial vehicle (UAV) photogrammetric results, and empirical research is necessary before large-scale applications for the newly released UAVs. Based on DJI Terra and Pix4Dmapper, the absolute and relative positioning accuracy of the Mavic 3 industrial UAV was verified and analyzed in detail using real-time kinematic (RTK) equipment. The results show that: (1) In RTK mode and no ground control points (GCPs), the output images of the two platforms can meet the requirements of 1:500 large-scale mapping accuracy. The absolute/relative positioning error did not surpass 0.05 m/0.15 m in horizontal and vertical directions. Additional GCPs could not have distinctly improved the absolute positioning accuracy. (2) Pix4Dmapper outperforms DJI Terra regarding mapping accuracy but is inferior to the latter in processing efficiency. The results can serve as a reference for large-scale mapping tasks based on the DJI Mavic 3 Enterprise Series.

Forest fires are one of the major disasters in China and can cause serious losses to forest resources and the social economy. In this paper, Xichang City, Sichuan Province was used as the study area, and Sentinel 2 data and Landsat 8 data were used to monitor the forest fires and ecological environment of the area dynamically, respectively. Based on the Sentinel 2 data, the overfire area extracted by the normalized fire burn index method is about 825 ha, which is close to the investigated damaged forest area of 791.6 ha. Based on the Landsat8 data, the ecological environment is monitored by the change of RSEI ecological index. The study shows that the use of remote sensing technology can effectively monitor forest fires and ecological environment, which can provide some theoretical basis for the prevention and control of forest fires and ecological restoration.

Multi-source remote sensing data fusion could compensate for the lack of spatial or spectral resolution of single-source remote sensing images. In this paper, we took the rocky area as the research object and carried out a machine learning lithology classification study based on combining spatial features and spectral features, and proposed an automatic identification method of surface lithology based on limit learning and watershed segmentation. By comparing and analyzing typical machine learning algorithms, the combination of spectral features and spatial features was incorporated into the algorithm to form a set of high-precision lithology automatic identification methods. The experimental results showed that the method in this paper performed better than the traditional SVM classification method based on spectral features in terms of classification accuracy and Kappa coefficient, especially after the extreme learning fusion watershed segmentation algorithm.

Using rapid and effective methods to detect pine diseases in unmanned aerial vehicle remote sensing helps maintain forest ecological security. To enhance the precision in detecting pine diseases, we employed an improved ShuffleNetV2 module, building a lightweight detection network that swiftly identifies these pine trees through UAV remote sensing. Integrate the SE attention mechanism into the spatial down sampling module of ShuffleNetV2. The GSConv-PAN structure is adopted for feature fusion, which enhances the utilization of multi-dimensional information in the core space and reduces the computational load. IM-ShuffleNetV2 network has 3.2%, 2.4%, 6.2% mAP50 improvement compared to YOLOv5s、 YOLOv6n and PP-PicoDet, respectively. Compared to ShuffleNetV2, MobileNetV3, and GhostNet, the improved ShuffleNetV2 as the backbone of the IM-ShuffleNetV2 improves the mAP₅₀ by 3.5%, 4.9%, and 7.6%, respectively.

Because the sea-land segmentation of remote sensing image is of great significance and application value in many aspects such as marine management target detection, safety monitoring and remote sensing image processing. However, sea-land segmentation has some defects, such as poor processing of complex scenes and ignoring the relationship between pixels. In recent years, convolution neural networks have been widely used in remote sensing image processing, however, these existing network models are difficult to realize real-time processing under limited conditions. Therefore, this paper proposes a lightweight convolution network RDF-Net for land-sea segmentation of remote sensing images. The overall architecture of the network is encoder-decoder structure, and deep convolution and asymmetric convolution are adopted in the network to achieve real-time processing with limited resources due to parameter quantity and computational complexity. Experimental results show that a better balance between accuracy and efficiency is achieved.

Remote sensing technology, with its efficient and lightweight modeling, provides a new technical means for the digital conservation of traditional villages. This study investigates the digitization of traditional villages in Guangxi. Data information is collected through UAV photography and real-time modeling of the digitization of real scenes is carried out. The method realizes high efficiency in constructing the real scene model, which is suitable for the rapid construction of the 3D model of traditional villages.

The high cost and complexity of obtaining pixel-level labels have hindered the development of the deep learning semantic segmentation field. To tackle the label refinement problem in remote sensing, we propose an Incremental Supervision- Driven Network (ISD-Net). In the ISD-Net, we adopt an incremental supervision strategy to continuously guide the model in learning more accurate semantic information to enhance model performance and robustness. A non-parametric background attention-guided module (N-BAG) is designed, which is combined with a conditional random field (CRF) module to better distinguishes buildings from the background. The level set loss function is used to enhance the perception of building features in coarse label regions, and effectively separating noise in pixel-level box-label. We conduct experiments on the WHU dataset to simulate the coarse label scenario. The experimental results demonstrate that the proposed algorithm achieves a maximum improvement of 17.15% and 25.68% in mIoU and F1-score, respectively, compared to the original coarse labels, respectively, proving the effectiveness of this approach in separating noise and identifying building edges. This research has a broad application prospect in the direction of building contour refinement in remote sensing images.

This article takes Qingxiu District, Nanning City as a typical urban research area. Based on the Landsat 8 images of Qingxiu District in the third phase, surface temperature inversion was carried out on the three remote sensing images of the same month in 2015, 2019, and 2021. The spatial characteristics of the obtained surface temperature results were analysed. The results showed that the trend of the distribution of low temperature ranges in the research area was that the area gradually decreased. The trend of changes in the distribution of medium and low temperature ranges is that the area continues to decrease and the proportion gradually decreases, with more significant changes. The trend of changes in the distribution of intermediate temperature ranges is to first increase and then decrease. The trend of variation in the distribution of medium and high temperature ranges is constantly increasing, and the increase is extremely significant. The trend of the distribution of high temperature ranges is constantly increasing, and the increase is gradually becoming more apparent. The research area should scientifically coordinate and arrange the relationship between urban development and green area based on local policies, combined with relevant factors such as economic development, urban expansion, natural environment changes, and urbanization level, to promote the sustainable development of the city's economy and environment.

This paper introduces a deep learning-based method and system for processing marine environmental data, aiming to address data collection and fusion issues in marine environment monitoring. We collect marine environmental parameters, such as seawater temperature, conductivity, flow velocity, and wind speed, through multiple sensors, generating observational data. To ensure data accuracy, we employ a deep learning-based data cleaning method and the MTConnect data transformation method for data preprocessing to achieve cleaning and transformation. To further improve the effectiveness of data fusion, we utilize a swarm intelligence-based optimization algorithm, Particle Swarm Optimization (PSO), to optimize data from both temporal and spatial dimensions. This will provide more accurate and comprehensive support for marine science research, environmental protection, and resource development, offering a robust foundation for decision-making and marine management.

Ground vehicle gravity survey is a mature method for census and local encryption survey of the Earth's gravity field. Nevertheless, Previous dynamic gravity measurement methods heavily depend on GNSS information, greatly limiting the implementation areas and measurement efficiency of gravity surveys. To address the aforementioned issues, we introduce a novel SINS/DR/BALT vehicle gravity measurement approach and demonstrate its effectiveness through a single-line sports car test. The results show that the accuracy of SINS/DR/BALT method is equivalent to that of SINS/GNSS method.

Remote sensing water depth inversion has the advantages of large-scale comprehensive coverage, based on pixel scale, and uniform spatial distribution. It overcomes the shortcomings of traditional depth measurement methods and compensates for the disadvantages of traditional measurements in terms of uncontinuous, multi-point, and real-time measurement of vast water areas. This paper proposes a technique for geographically adaptive multispectral remote sensing water depth inversion based on BP neural networks, and tests it using data from Resource 1 satellite (02E) to Gaoya Reservoir in Shandong Province. The results show that the error in elevation is ±1.04m, and the error distribution shows randomness.

Seismic velocity spectrum picking is an important task in seismic exploration. With the development of geological exploration technology, the data of seismic velocity spectrum has experienced explosive growth. Manual picking of seismic velocity spectrum alone gradually cannot meet the needs of practical production. In order to improve efficiency and reduce time costs, there is an urgent need to find a fast and effective method to achieve an automated velocity spectrum picking process. This paper proposes an intelligent picking method based on neural networks. The method abstracts the task of seismic velocity spectrum picking into a visual task, processes it, and outputs candidate boxes through model training. The center points of the candidate boxes are used as the "time-velocity" sequences of seismic velocity spectrum images. To improve the accuracy of model training, a YOLO-based improved YOLO-DCANet network is proposed. In the backbone stage, deformable convolutional modules are introduced, and a Coordinate Attention (CA) mechanism is integrated at different stages of the process. Meanwhile, to learn the features of energy clusters at different levels in the image, a multi-scale model is introduced to match energy cluster information at different levels. The experiments show that the mAP accuracy of the proposed YOLO-DCANET is improved by about 8% compared with the traditional two-stage model, and it’s also improved by about 5% compared with the original YOLO series model. It shows that the seismic velocity spectrum picking algorithm proposed in this paper is significantly improved compared with the traditional deep learning method. In addition, the generalization test on other working areas proves that the proposed method has strong generalization ability and robustness.

Rapidly obtaining soil organic carbon content data of the land returned to forest is of great significance in clarifying the implementation effect of the returning farmland to forest project. In this study, soil organic carbon (SOC) and ground hyperspectral data were obtained from the soils of fallow forests in Daning County. The ground hyperspectral data are transformed into five kinds respectively: original spectrum (R), reciprocal logarithm (RL), reciprocal logarithm’s first-order differential (FRL), first-order differential (FD), and continuum removal (CR), and correlation analysis was performed with the SOC content to select the characteristic bands, and then three estimation models of SOC were established, namely, the principal component regression (PCR), the partial least square regression (PLSR), and the support vector regression (SVR), respectively. The results showed that SOC content and spectral reflectance were negatively correlated, and the spectral curve generally showed an increasing trend; Both RL, FRL, FD and CR transformation can improve the correlation between SOC and reflectance, among which the FRL, and the CR had the best effect; the comparison of the three modelling methods revealed that the accuracy of the SVR method was better, and the SVR model was based on the FRL as the independent variable. The SVR model with the FRL as the independent variable has the highest accuracy, and the R² of the modelling set and validation set are 0.780 and 0.707, respectively. The results show that hyperspectral technique is a useful method for estimating SOC rapidly and accurately.

Through the evaluation of the IMERG_Late run from the GPM satellite and the TMPA_3B42RT from the TRMM satellite at 22 stations in Gansu Province, it has been determined that the IMERG_Late run of the GPM satellite exhibits a higher level of consistency with measured precipitation data compared to the TMPA_3B42RT of the TRMM satellite, along with superior deviation performance. The IMERG_Late run demonstrates noticeable spatial variances and increasing trends in correlation coefficients, whereas the correlation coefficient of TMPA_3B42RT is relatively less pronounced. The agreement between the two satellite precipitation products remains steady across seasons and precipitation levels, with root mean square errors generally lower in the northwest region compared to the southeast. Furthermore, the Probability of Detection (POD) and Critical Success Index (CSI) of the IMERG_Late run generally surpass those of TMPA_3B42RT, with slightly lower POD values. Particularly under varying precipitation intensities, the IMERG_Late run outperforms TMPA_3B42RT significantly, especially within the 1~10mm/d range. Despite the occurrence of notable false alarms at specific stations for both products, these instances are predominantly concentrated in the northwest of the study area. In conclusion, the daily precipitation product of the IMERG_Late run demonstrates advantages in terms of accuracy and applicability when compared to TMPA_3B42RT.

Air pollution affects public health, destroys ecological environment and even aggravates climate change, and PM_2.5 is an important factor causing pollution. Therefore, accurate and continuous spatiotemporal PM_2.5 concentration inversion is of great significance for air pollution control. Nevertheless, current high temporal resolution inversions are predominantly conducted using Top-of-atmosphere reflectance, nighttime concentration inversion cannot be realized in the visible band. Moreover, existing hybrid models often rely on linear integrated, increasing the risk of overfitting. Therefore, we propose an improved stacking model that contains XGBoost and CatBoost model, stacked by elastic net regression. By fusing features from multi-source data and accounting for spatiotemporal heterogeneity, the model enables the simultaneous inversion of daytime and nighttime PM_2.5 concentrations. Compared to using Top-of-atmosphere reflectance alone, the daytime inversion results demonstrate a 1.08% increase in R² and a reduction of 1.34μg/m³ in RMSE. Meanwhile, the nighttime inversion results show a 2.04% improvement in R² and reduce the MAE to 9.54μg/m³ when spatiotemporal features are used. Applying the model to the Beijing-Tianjin-Hebei region enables spatiotemporal continuous inversion of PM_2.5 concentrations in the region.

The settlement deformation of subway tunnels during the construction and operation stages is a relatively complex nonlinear process. The original settlement deformation monitoring data will be affected by some noise during the collection process. In order to reduce the impact of noise and errors in the subsequent prediction process, a joint denoising model based on the improved empirical mode decomposition (ICEEMDAN) algorithm, wavelet threshold denoising and NLMS adaptive filtering was constructed. The model first uses ICEEMDAN to decompose the original data. After decomposing the intrinsic modal component IMF, it divides it into high-frequency and low-frequency components. Then the wavelet threshold is used to remove the components with high correlation coefficients in the high-frequency and low-frequency components. Noise and NLMS adaptive filtering are processed, and finally the additive reconstruction is performed to obtain the denoised data. Experimental analysis results show that compared with ICEEMDAN, wavelet threshold denoising and NLMS adaptive filtering algorithms, the joint denoising model has better denoising effect, higher correlation coefficient, signal-to-noise ratio and smaller mean square root error.

With the continuous advancement of technology, engineering surveyors can utilize drone remote sensing technology for rapid and precise measurements of engineering projects. This technology enables the acquisition of surface information and real-time monitoring, offering a novel solution for engineering surveying. This paper discusses the basic principles of drone remote sensing technology, its limitations, and operational procedures. Through in-depth case studies and analysis, the paper aims to provide a better understanding of the practical application effects of drone remote sensing technology in engineering surveying, offering valuable references and materials for related research and practice in the field of engineering surveying.

Dissolved organic carbon (DOC) transport is an important part of the terrestrial and marine organic carbon cycle. However, limited by the quantity and quality of the global river database, the magnitude of DOC and its transportation source and sink pattern are still unclear. Therefore, based on an updated GEMS-GLORI database, we used empirical and random forest models to re-analyze the DOC flux of 197 rivers worldwide. The results showed that DOC flux from terrestrial transportation to the ocean is about 193-204 Tg C yr^-1, by combining gross primary productivity and reconstructed nitrogen deposition data, we found that vegetation growth and nitrogen deposition played a positive contribution to DOC transport in most rivers. The findings help refine our understanding of the impact of carbon transport in rivers around the world.

In polar marine environments, the distribution and morphology of floating ice are significant for navigation, the safety of marine engineering, and the study of polar climate change. Due to the difficulties and dangers associated with field observations, numerical simulation based on existing observation data has become an effective research method. In this paper, based on key characteristics such as ice field density and equivalent diameter obtained from aerial photography and remote sensing in specific sea areas, a parameterized ice field with irregular geometry is generated using the Tyson polygon method and the parametric design tool Grasshopper. Combined with genetic algorithm, user-defined equivalent diameter quantization module and spatial optimization method based on virtual force, the generated floating ice field model is quantified and optimized, and it is also applicable to uniform / non-uniform floating ice field. Compared to the single shape composition of traditional generation methods, the floating ice field shapes obtained by this method are diverse and correlated. Meanwhile, the equivalent diameter of floating ice follows a power-law distribution, which is consistent with actual observation results. The research findings of this paper can provide a more accurate reference for studying the distribution of floating ice in polar waters and predicting the sailing load of polar ships.

The vectorized extraction of building footprints from aerial imagery plays an important role in enhancing urban planning accuracy, a task that requires precise delineation of building boundaries. Previous methods often struggle with the complex post-processing procedure, inefficiency, and occlusion issues. Recognizing these limitations, we propose a novel approach built upon BoundaryFormer, a state-of-the-art method for instance segmentation with polygon boundary, to extract precise vectorized building footprints efficiently. We design a corner classification scheme composed of a Corner Classification Module (CCM) and the Corner Matching Strategy (CMS) for simplifying building polygons. The CCM outputs classification scores to suppress vertices unlikely to be corners. Considering the gap between predicted vertices and ground truth corners, we present the CMS coupled with CCM to complete the training pipeline. Additionally, we leverage Focal Loss to address the class imbalance between corners and non-corners during training. Our experiments on the CrowdAI dataset prove the effectiveness of the proposed corner classification scheme and demonstrate the superiority of our method over existing methods across several Average Precision (AP) metrics.

Remote sensing has long been an effective method for water quality monitoring. However, traditional empirical and analytical methods may suffer from poor accuracy for non-optically active parameters. Taking Nansi Lake as the study area, the research conducts chlorophyll-a and total nitrogen concentrations assessment using the K-Nearest Neighbor (KNN) and Backpropagation (BP) neural network methods based on the measured water quality data in 2020 and Landsat remote sensing images. The results indicate that (1) Spectral features exhibiting stronger correlations with chlorophyll-a concentration include Blue/Green, (Blue-Green)/Green, and (Blue-Green)/Red, while those exhibiting stronger associated with total nitrogen concentration encompass Brightness, Green-Red, and (Green-Red)/Blue. (2) Through rigorous model evaluation, the K-nearest neighbor method emerged as the most effective approach. The determination coefficients for chlorophyll-a and total nitrogen model test sets were 0.60 and 0.56, respectively, with corresponding root mean square errors of 1.34 μg/L and 0.86 mg/L. (3) On February 18, 2020, relatively higher chlorophyll-a concentrations were observed in Nanyang Lake and Zhaoyang Lake, while higher total nitrogen levels were recorded in Dushan Lake.

Rice sheath blight is one of the main diseases of rice, seriously affecting the safety of rice production. This study integrates multisource data such as meteorological, spectral, and microwave data, and combines various species distribution models (SDMs) to establish a habitat suitability analysis model for rice sheath blight, thus evaluating the suitability of habitats for sheath blight. The results indicate that integrating multiple Species Distribution Models (SDMs) into an aggregative model yielded AUC values of 0.959 and 0.963 for the years 2019 and 2020 respectively. The True Skill Statistics (TSS) were 0.817 and 0.831 respectively. The overall habitat suitability for rice sheath blight in 2019 was slightly higher than that in 2020. This research provides a method for evaluating the habitat suitability of rice sheath blight, thereby offering support for the scientific prevention and control of crop diseases.

Although traditional air pollution monitoring methods can provide detailed local pollution information, they have limitations such as limited coverage and slow data updates. As an emerging monitoring method, night light remote sensing technology has the advantages of wide coverage, fast speed, and large information content, providing a new way for air pollution monitoring. Based on night-light remote sensing technology, this study conducted an in-depth analysis of the spatiotemporal changes in atmospheric pollution in Jiangxi by acquiring remote sensing satellite data. Nightlight data were used to reflect the development level of urbanization and industrialization, and combined with air quality monitoring data, the correlation between air pollution and human activities was explored. Research has found that there is a certain positive correlation between night light intensity and air pollution levels in Jiangxi, especially in industrial areas. Through time series analysis, the changing trends of air pollution in different seasons and years are revealed, providing a scientific basis for future environmental governance. This study provides empirical research on the application of nighttime light remote sensing in air pollution monitoring and provides an important reference for local governments to formulate environmental protection policies.

The microwave brightness temperature observed in areas covered by clouds and rain is strongly influenced by the atmosphere. In order to obtain accurate surface information, atmospheric correction is required. Most studies cannot achieve atmospheric correction for areas covered by rainfall. This article develops an atmospheric correction method based on the radiative transfer model. Through analyzing the impact of clouds and rain on microwave radiation transmission, it was found that clouds and rain are not sensitive to the low-frequency bands of 6.9 GHz, 7.3 GHz, and 10.7 GHz, while they have varying degrees of attenuation at 18.7 GHz, 23.8 GHz, 36.5 GHz, and 89.0 GHz. Based on this, both cloud and rain information are added to the model for atmospheric correction. By comparing the results before and after atmospheric correction, it was found that this algorithm can achieve atmospheric correction in the 18.7GHz, 23.8GHz, 36.5GHz, and 89.0GHz bands. However, it is not suitable for areas with strong water vapor absorption in the 23.8GHz band and rainfall areas in the 89.0GHz band.

In 6G, maritime wireless communications will play an important role for providing high data-rate services ubiquitously. However, due to the complex and rapidly changing marine environment, an accurate channel estimation is usually hard to achieve for maritime wireless communication systems. In this paper, a deep neural network (DNN) aided receiver design is proposed for the orthogonal frequency division multiplexing (OFDM) based maritime wireless communications. Due to the complicated and rapidly changing marine environment, the channel conditions are hard to be accurately estimated. Hence, the attainable performance of the OFDM based maritime wireless systems may be severely degraded. In order to tackle this issue, in this paper, a DNN aided marine channel estimation scheme is conceived for the OFDM based maritime wireless system, where SGD, Adam, SignSGD optimizers are considered. The attainable system performance is validated by simulations, showing that the proposed DNN-aided channel estimation scheme is capable of attaining a competitive system performance compared to the existing schemes. Furthermore, the proposed DNN-aided channel estimation scheme is able to achieve a better system performance at the low signal to noise ratio (SNR) regime in comparison to the existing channel estimation schemes.