With the fast development and wide application of remote sensing technology, remote sensing images with higher quality are needed. However, it is difficult to directly acquire high-resolution, multispectral remote sensing images. To obtain high-quality images by integrating the information from different imaging sensors, pansharpening technology emerged. Pansharpening is an effective method used to obtain multispectral images with high spatial resolution. Many scholars have studied this method and obtained fruitful achievements. In recent years, deep learning theory has developed rapidly and has been widely applied in pansharpening. This study aims to systematically introduce the progress in pansharpening and promote its development. To this end, this study first introduced the traditional, classical pansharpening methods, followed by commonly used remote sensing satellites. Then, this study elaborated on the pansharpening methods based on deep learning from the perspective of supervised learning, unsupervised learning, and semi-supervised learning. After that, it described and analyzed loss functions. To demonstrate the superiority of the pansharpening methods based on deep learning and analyze the effects of loss functions, this study conducted remote sensing image fusion experiments. Finally, this study presented the future prospects of the pansharpening methods based on deep learning.

The rapid and accurate estimation of soil water content at different spatial and temporal scales is key research content in the fields of hydrology, environment, geology, agriculture, and climate change. However, it is still a challenge to obtain accurate soil water content presently. In the past, the traditional point-based soil sampling and analysis methods were time-consuming and laborious. By contrast, retrieving soil water content using remote sensing images has the advantages of a wide range, high timeliness, low cost, and strong dynamic contrast. In hyperspectral remote sensing, soil water content is related to the wavelength range of soil reflectance. So far, many methods have been used to describe the relationships between soil water content and hyperspectral remote sensing. This paper summarized existing methods for estimating soil water content based on hyperspectral reflectance and divided them into four categories: spectral reflectance methods, function methods, model methods, and machine learning methods. Moreover, this paper compared and analyzed the potential and limitations of different methods in terms of accuracy, complexity, auxiliary data requirements, operability under different modes, and the dependence on soil types. Finally, this study put forward corresponding suggestions for future research on the relationships between soil water content and hyperspectral reflectance.

The remote sensing information extraction of vegetation is the basis and key link for remote sensing investigation and dynamic monitoring of vegetation coverage, which is of great significance for regional ecological environment protection and sustainable development. For this purpose, the research progress on the remote sensing information extraction methods of vegetation was reviewed from prior knowledge, expert knowledge and related auxiliary information, extraction of vegetation phenological features, the fusion of multi-source remote sensing data, machine learning, and other methods. Then, the main problems and challenges existing at the present stage were pointed out, and the future development trend was put forward. The research shows that there are many methods to extract remote sensing information about vegetation, and different methods have their own advantages and disadvantages in the application. However, the research on remote sensing information extraction methods of vegetation is currently facing many challenges, such as the lack of openness of high-resolution remote sensing data, the poor stability of parameter settings in vegetation information extraction models, the prominent phenomenon of same objects with different spectra and different objects with the same spectrum, the difficulties in automatic extraction of vegetation remote sensing information based on an expert knowledge base, and the need in further research on the multiple-method fusion. Therefore, making more breakthroughs in integrating multi-source data, multiple methods and new features of multi-temporal remote sensing images will be necessary to promote the refined, automated, and intelligent development of remote sensing information extraction of vegetation.

Determining the present distribution of historically abandoned mines nationwide and carrying out orderly ecological rehabilitation of these mines are important parts in the preparation of mine ecological rehabilitation planning and serve as the main bases for the deployment of ecological rehabilitation engineering. This study proposed the technical process and method for determining the historically abandoned mines according to the definition of historically abandoned mines and the public management requirements. This technical method was proven effective through tests.

Water information extraction is an important study direction in the application of high spatial resolution remote sensing images. Conventional recognition methods only focus on the shallow features of water. Therefore, to further improve the robustness of water information extraction algorithms and increase the segmentation precision by extracting more deep information from remote sensing images, this study proposed a water classification method using the semantic segmentation model based on deep learning. First, deep neural networks were used to mine the information from high-resolution remote sensing images. Then, attention modules were used to integrate the deep information with the shallow features such as shape, structure, texture, and hue. Based on the integrated information, a new deep semantic segmentation model with higher precision and prediction efficiency than existent models was built. Finally, the ablation experiment was conducted to compare with conventional recognition methods and common semantic segmentation models. The experiment demonstrates that the proposed algorithm model yields higher overall precision and efficiency than previous methods and that the algorithm parameters are easy to set and less human intervention is required in the model. This study proved the accuracy and efficiency of deep learning and attention mechanism on water information extraction from high-resolution remote sensing images. Moreover, this study provided a possible solution for the segmentation of high-resolution remote sensing images using the deep learning method and explored the future prospect of the solution.

Atmospheric correction is an important preprocessing step for hyperspectral remote sensing images. The atmospheric correction quality determines the application degree of hyperspectral remote sensing to a certain extent. First, this study analyzed the influence of the atmosphere on radiative transfer and summarized the inversion methods of aerosol optical thickness and water vapor in the atmosphere, indicating the main atmospheric factors affecting the quality of hyperspectral remote sensing images. Then, the influence of the atmosphere was demonstrated theoretically by clarifying the derivation process of the radiative transfer equation and the action mechanism of relevant parameters, indicating the main aspects of hyperspectral atmospheric correction. Furthermore, this study summarized the hyperspectral atmospheric correction methods formed in recent years, including methods based on empirical statistics and radiative transfer, and analyzed the study advances and development trends of hyperspectral atmospheric correction. Finally, this study forecasted the development of atmospheric correction of hyperspectral remote sensing images. This study will provide a certain reference for the engineering application and study of hyperspectral remote sensing.

As a major form of soil degradation, soil salinization can greatly harm agricultural production and ecological environment. Remote sensing methods can acquire soil spectral characteristics in a rapid, macroscopic, and timely manner. Based on this, remote sensing monitoring models can be built for a wide range of soil salinization monitoring and assessment. Thus, summarizing and discussing the building methods for remote sensing monitoring models of soil salinization is of great significance to improve the precision of remote sensing monitoring of soil salinization and to monitor and control salinized soil. This study reviewed the recent literature related to remote sensing studies concerning soil salinization at home and abroad. Then, it summarized the steps such as factor selection, model building, and precision verification in the building of remote sensing monitoring models of soil salinization. Focusing on the current hot research topic, this study discussed the limitations and development trends. The main conclusions are as follows. The remote sensing monitoring models of soil salinization are important means for monitoring and forecasting salinized soil. In recent years, the hot research topic in this field is to improve the model precision using new data sources and models. Differences exist in the use of remote sensing data sources among different studies, but the modeling factors are all optimized from spectral sensitive bands, prior spectral indices, and remote sensing-derived data. The remote sensing monitoring models of soil salinization mainly include the linear regression model and the machine learning model. The remote sensing models built for different regions have different precision and applicability.

To obtain the distribution of nighttime light pollution on a city scale, this study monitors the nighttime light pollution in Nanjing City based on Luojia 1-01 nighttime light remote sensing images. The apparent radiance of the remote sensing images was converted into the surface incident luminance according to surface reflectance and building coverage ratio. Based on this and the illuminance values observed in the field, various empirical models were established to calculate the nighttime illuminance of Nanjing City. Finally, the distribution of nighttime light pollution in Nanjing City was analyzed according to the calculated nighttime illuminance. The results show that the third-order polynomial regression model had the highest accuracy, with a determination coefficient of 0.87 and a mean absolute error (MAE) of 4.71 lx. The nighttime illuminance in Nanjing City varied in the range of 0~55 lx, with obvious spatial distribution differences. In general, the areas with high illuminance were mainly concentrated in the main urban area and the illuminance showed a decreasing trend from the main urban area to the surrounding area. Light pollution was the most serious in Gulou and Qinhuai districts, where light pollution covered more than 70% in terms of area. The light pollution in the suburb was relatively weak, and the three districts with the weakest light pollution included Gaochun, Lishui, and Liuhe districts successively, where light pollution covered less than 4% in terms of area. Some areas in Nanjing City showed extremely high illuminance (> 30 lx), including large shopping malls, large factories, traffic hubs, roads, and some residential areas. It should be noted that there are many residential areas near these places except for traffic hubs and large factories. This study explored a method of monitoring urban light pollution at night based on Luojia 1-01 remote sensing data. It will provide data support for the light pollution control and management in Nanjing City and a scientific reference for the light pollution monitoring in other areas.

Snow depth and snow water equivalent are critical elements for snow cover observation and are greatly significant in fields such as cryosphere, global climate change, and water resource surveys. Microwave remote sensing is superior to both visible-light and near-infrared remote sensing in snow cover observation. This study systematically summarized the research results of the passive microwave remote sensing in the inversion of snow depth and snow water equivalent. It organized three types of snow cover observation methods, i.e., field surveys, long-term observations at ground stations, and regional observations based on satellite remote sensing, as well as major snow cover parameters to be observed. Furthermore, it summarized and evaluated three inversion algorithms, i.e., semi-empirical method, physical model, and machine learning. Finally, this study presented the results of the snow cover in the Qinghai-Tibet Plateau observed using passive microwave remote sensing, predicted the future development trend of remote sensing-based inversion of snow cover parameters, and put forward scientific suggestions for the in-depth implementation of the inversion of snow depth and snow water equivalent passive microwave remote sensing.

Ecological evaluation plays an important role in supporting urban development planning and using a remote sensing index to carry out ecological evaluation is a feasible method. Today, with the development of cloud computing, this paper explores a time-series calculation method of remote sensing ecological index suitable for Google Earth Engine, to address the problem that the calculation results of different sensors differ greatly in the process of big data calculation. Firstly, by taking Kuitun City, Xinjiang Uygur Autonomous Region, as the study area, this paper performs the de-clouded fusion process on Landsat images from 1989 to 2019. Secondly, this paper calculates the four major components of the fused images and makes preferences in the calculation of the humidity component and temperature component. Finally, this paper proposes the normalization method of the overall optimum and calculates the remotely sensed ecological index for each year on this basis. The analysis of the obtained results shows that the first principal component under the calculation by this method has a higher contribution rate, and the time series results on this basis have a higher polynomial fitting effect. It indicates that the method can specify uniform standards for different sensors, enhance the comparability of calculated results between different sensors, optimize the calculated results of remote sensing ecological indices, and ensure the interpretability of ecological evaluation grading results.

China is one of the countries with frequent landslide disasters. In recent years. In recent years, more than 70% of the catastrophic geological hazards have occurred not within the scope of known hidden danger points of geological hazards in China. Therefore, there is an urgent need for investigating large-scale landslide disasters using automatic and efficient technologies and methods for landslide identification. To quickly identify the location of landslides from massive remote sensing images, it is necessary to determine the key areas of landslides to support subsequent interpretation and research. This study investigated loess landslide identification based on GF-1 images and digital elevation model (DEM) data. First, a database of remote sensing images and DEM landslide samples was constructed. Second, the landslide samples were classified using the channel fusion convolutional neural network model. Finally, the classification results were restored to the remote sensing images according to the location information. Experimental results showed that the model yielded landslide identification accuracy of 95.7% and a recall rate of 100.0%. The model used in this study has a small number of network layers, a high convergence speed, and higher efficiency and identification accuracy. As a result, it allows for the quick identification of key landslide areas from remote sensing images in the case of a limited number of samples, thus supporting the investigation of large-scale landslide disasters.

Water conservation is one of the most important functions of an ecosystem and can maintain and provide water resources for the ecosystem and humans. According to the physical meaning of water conservation, this study used leaf area index, vegetation coverage, and evapotranspiration to represent the water conservation of the vegetation layer and used surface temperature, soil moisture content, and slope to represent the water conservation capacity of the soil layer. Then, this study developed a remote sensing monitoring and evaluation model for water conservation through principal component analysis to explore the spatial-temporal distribution characteristics of the water conservation capacity in the Three Gorges reservoir area. The results show that the water conservation index (WCI) contained the objective information of various indices, could be used to quickly and conveniently assess the water conservation function in the Three Gorges Reservoir area, and properly represented the water conservation capacity there. In 2019, the water conservation capacity was unevenly distributed in the Three Gorges reservoir area and was high downstream and low upstream. The northeastern part of Chongqing was dominated by forest ecosystems and had the strongest water conservation function. From 2013 to 2019, the WCI slightly increased in most areas, especially in some parts of Fengdu, Kaizhou, and Yunyang areas.

Intertidal zones, as important parts of coastal wetlands play a significant role in ecological and economic development. However, the dynamic interaction between seawater and land makes it difficult to accurately determine the tidal flat area using the remote sensing information extraction method based on instant remote sensing images. To solve this problem, this study developed an intertidal information extraction method based on Google Earth Engine (GEE) platform and remote sensing index. This proposed method was applied to study the coastal zone of Zhoushan Islands. First, a decision tree algorithm based on the fusion of the digital elevation model (DEM) data was built using the Landsat8 time series image data in 2021. Then, a multi-layer automatic decision tree classification model was formed using the maximum spectral index composite (MSIC) and the Otsu algorithm (OTSU). Based on this, the DEM data were fused to extract and calculate the area of the intertidal zone in Zhoushan Islands. The results show that the area of the intertidal zone in Zhoushan Islands is 35.19 km² in 2021. The evaluation based on the Google Earth high-resolution images shows that this proposed method has a general precision of 97.7% and a Kappa coefficient of 0.95, indicating good extraction precision and practical effects. This method can provide data support for sustainable management and utilization of coastal zone resources through automatic and rapid extraction of intertidal information, thus promoting regional high-quality development.

Predicting the subsequent subsidence in mining areas according to the law of mining subsidence is the key to assessing mining risks and adjusting mining planning. This study determined the available conditions of the logistic model for mining subsidence prediction through analysis and simulation experiments and proposed a method for the dynamic prediction of mining subsidence based on small baseline subset (SBAS)-interferometric synthetic aperture radar (InSAR) technology and the logistic model. Firstly, the time-series subsidence data of a mining area was obtained using the SBAS-InSAR technology. Then, taking the time series subsidence data as the data for fitting, the parameters of the logistic model were calculated pixel by pixel by using the trust region algorithm. Then, the pixel range in which the subsequent subsidence can be predicted was determined according to the available conditions of the logistic model. Finally, according to the Logistic model, the subsequent subsidence within the predictable range was predicted. This method was applied to a certain mining area in Erdos City, Inner Mongolia for tests, and the prediction results were verified using the InSAR monitoring results of corresponding dates. The predicted results after 36 days and 108 days of ming had the root mean square error (RMSE) of 0.010 1 m and 0.023 6 m, respectively, and their proportion with prediction errors of less than 0.03 m reached 98.9% and 89.3%, respectively. These results indicate that the method for dynamic prediction proposed in this study has high prediction accuracy.

With the trend towards the precise and digital planting management of orchards, apple cultivation relies more heavily on the planting management supporting technologies of orchards. In recent years, continuous breakthroughs made in spatial resolution and revisiting frequency have made remote sensing technology a major supporting technology for the precise planting management of apple orchards. However, there is an absence of reviews of the application status and prospect of this technology in the planting management of orchards. Based on the analysis of primary applications of remote sensing technology in the precise planting management of apple orchards, this study classified the applications into three major categories, namely the surveys of basic orchard information, inversions of orchard parameters, and the planting management support of orchards. Furthermore, this study reviewed the methods and performance of the applications of remote sensing technology in various fields and explored the application potential. Finally, it identified three types of problems with current research and application of remote sensing technology, namely insufficient studies on mechanisms and in some application fields, low-degree integration of multiple technologies, and the lack of large-scale application models. In addition, this study proposed four hot research and application topics in the future, namely models used to simulate the growth mechanisms of apple trees, the integrated support system for the planting management of apple trees, the single-tree monitoring based on satellite data, and the diversified services of remote sensing-based monitoring products.

Fast extraction of buildings with high accuracy from remote sensing images is an important research of remote sensing intelligent application services. To address the problems of imprecise segmentation of building edge in remote sensing images, holes in large-scale target segmentation, and a large amount of network parameters in the DeepLab model, a lightweight DeepLabv3+ model for building extraction from remote sensing images is proposed. In this method, the lightweight network MobileNetv2 is used to replace Xception, the backbone network of DeepLabv3+, so as to reduce the number of parameters and improve the training speed; The hole rate of hole convolution in ASPP is optimized to improve the effect of multi-scale semantic feature extraction. The improved model has been tested on WHU and Massachusetts data sets. The results show that the IOU and F1 score in WHU dataset are 82.37% and 92.89%, respectively, 2.71 percentage points and 2.14 percentage points higher than those of DeepLabv3+, 2.04 percentage points, and 2.32 percentage points higher than those of DeepLabv3+ in Massachusetts data set. The number of training parameters and training time is reduced, and the accuracy of the building extraction is effectively improved, which can meet the requirements of fast extraction of high-precision buildings.

The remote sensing image fusion technology can combine multi-source images containing complementary information to obtain images with richer content and higher spectral quality, thus it is the key and foundation of remote sensing applications. Aiming at the problems of spectral distortion and spatial structure distortion that are prone to occur in the process of remote sensing image fusion, the knowledge-based remote sensing image FuseNet (RSFuseNet) was constructed based on the attention mechanism and using normalized difference vegetation index (NDVI) and normalized difference water index (NDWI) as prior knowledge. Firstly, considering that the high-pass filtering can fully extract edge texture details, a high-pass filtering module was constructed to extract high-frequency details of panchromatic images. Secondly, NDVI and NDWI were extracted from multi-spectral images. Then, an adaptive squeeze-and-excitation (SE) module was constructed to recalibrate the input features. Finally, the adaptive SE module was combined with the convolution unit to perform fusion processing on the input features. The experiment was conducted using Gaofen 6 remote sensing image as the data source, and selecting Gram-Schmidt (GS) transformation, principal component analysis (PCA), a deep network architecture for pan-sharpening (PanNet), and pansharpening by convolutional neural networks (PNN) models as comparative models. The experimental results show that the peak signal to noise ratio (PSNR) index (40.5) and the structural similarity (SSIM) index (0.98) of the RSFuseNet model are better than those of comparative models, indicating that the method in this study has obvious advantages in remote sensing image fusion.

As an essential branch of surveying and mapping science, underwater topographic surveys are closely related to human operations in oceans and lakes. For underwater topography detection in shallow-water areas, conventional acoustic methods face the hull stranding risk, and passive optical methods have low survey accuracy. The airborne laser sounding is a novel means for bathymetric surveys in shallow-water areas, and its application in offshore areas can fill the gap of underwater topography data in shallow-water areas. This study presents a brief introduction to the composition and principle of the airborne laser sounding system, followed by a description of laser sounding data acquisition. Furthermore, this study highlights the critical processing technologies for airborne laser sounding data, including waveform data processing, error correction, and point cloud data processing. Finally, this study summarizes the technical difficulties and developmental trends of airborne laser sounding.

The principal component analysis is a widely used method for dimensionality reduction of hyperspectral remote sensing images. In task-oriented work, the principal component selection method based on cumulative variance contribution rate is not ideal. To address the problem of principal component selection after principal component analysis transformation, a method of principal component selection based on spatial statistics is proposed. The selection of principal components is performed by calculating the values of the semi-variogram parameter range and partial sill/sill of each principal component. The magnitude of a range is used to judge the range of spatial correlation of each principal component, and the partial sill/sill is used to judge the strength of spatial correlation of each principal component. The simulation proves that the variable range and partial sill/sill can effectively express the range and strength of spatial correlation of hyperspectral remote sensing images. Based on the experiment of real hyperspectral remote sensing images, the empirical threshold of principal component selection is determined from subjective and objective aspects, that is, the range is 2.5, and the partial sill/sill is 0.2. According to the classification results based on the support vector machine algorithm, compared with traditional methods, the principal components with better image quality can be screened by using variable range and partial sill/sill, which can not only achieve the purpose of dimensionality reduction, but also ensure high classification accuracy.

Water color represents the most intuitive visible perception of the color of water bodies that is jointly affected by substances such as suspended particulate matter, chlorophyll, and soluble organic matter. Water color is a water environmental parameter with a long history and plays a critical role in research on the ecosystem of inland and nearshore water bodies. With the progress made in colorimetric research, as well as hyperspectral imaging and satellite remote sensing techniques, the colorimetric method of water color has developed. This study systematically reviewed the colorimetric research progress of inland and nearshore water bodies and elaborated on the theories and practical applications of the colorimetric method from the angles of apparent optical properties (AOP) and inherent optical properties (IOP). Moreover, it presented the colorimetric processing method of satellite remote sensing data. The colorimetric method is a technical method for the quantitative expression of water color. It is also an important branch of water color research and an extension and supplement to the study of water color components, with a broad application prospect. To further improve the application of the colorimetric methods in inland and nearshore water bodies, it is necessary to enhance the construction of bio-optical datasets of water bodies in the future. Moreover, colorimetric studies should be conducted in two dimensions, namely AOP and IOP, and it is necessary to intensify research on domestic satellite-based colorimetric methods and increase the types of relevant water color products.

The typical crop cotton in the Ugan-Kuqa River Delta Oasis was used as the research object to study the applicability and optimization process of the deep learning method in the identification of cotton distribution in arid areas. Based on the domestic GF-2 images and the field survey data, the Unet deep learning method was adopted, in which the characteristics of the Unet network’s multiple convolution operations were fully utilized to explore the deep-level characteristics of cotton in remote sensing images, thereby improving the precision of cotton extraction. The results show that the recognition effect of the Unet model to extract cotton, corn, and peppers in the study area is better than the classification results of the object-oriented method and the traditional machine learning algorithms. The overall precision is 84.22%, and the Kappa coefficient is 0.804 7. Compared with the object-oriented method and the traditional machine learning algorithms SVM and RF, the overall precision has increased by 7.94 percentage points,11.93 percentage points, and 11.73 percentage points, respectively, and the Kappa coefficient has increased by 10.13%, 14.72%, and 14.60%, respectively. In the classification results of the Unet model, both the mapping precision and the user precision of cotton are higher than those of the other three methods, which are 94.95% and 89.07%, respectively. Therefore, it is feasible and reliable to use the Unet model to extract high-precision cotton spatial distribution information of arid areas on GF-2 high-resolution remote sensing images.

In order to realize the refined line inspection management of transmission lines, improve its operation and maintenance efficiency, realize satellite intelligent inspection, and accurately find the defects and hidden dangers of towers and transmission lines, the paper took the coordinates of transmission line towers in Kunming City, Yunnan Province as an example and proposed a method to calibrate the coordinates of transmission towers using satellite images. The method first uses the reference base-map data as the basis to match the control points and uses the digital elevation model (DEM) to perform geometric correction on the original remote sensing image. Then combined with such technologies as shadow detection and edge detection and visual interpretation, the calibrated tower coordinates are obtained. The experiment verified the geometric correction accuracy of the SuperView-1 (SV1) and Gaofen-2(GF2) satellite images in the Kunming area, and the errors in the plane after correction were 0.931 and 1.387 m, respectively. In addition, the experiment verified the calibration accuracy of the old tower coordinates on the two lines. The results show that the plane accuracy of the tower has increased from 13.811 m and 8.256 m to 5.970 m and 5.104 m, respectively, which meets the basic power grid requirements. This method can realize the calibration of the tower coordinates, reduce the workload of manual inspection, and improve the efficiency of line inspection. With the explosive growth of remote sensing image data, multi-source images from the space and ground will continue to be combined, and the technology for the positioning of transmission towers based on satellite remote sensing images will have a broader development prospect.

This study aims at the identification and potential evaluation of the mineralization elements of calcrete-hosted uranium deposits in Saudi Arabia through the exploration of calcrete-hosted uranium deposits in the uranium exploration project of China and Saudi Arabia. Based on satellite (ASTER) remote sensing data and DEM data, the uranium metallogenic conditions of three calcrete areas were compared and analyzed using methods including visual discrimination, hydrological analysis, and principal component analysis and techniques including uranium source evaluation, source-pathway-trap system division, and ore-bearing rock identification. The results show that Area 2 has the most complete uranium metallogenic conditions in terms of uranium source and source-pathway-trap conditions, Area 1 lacks a good sedimentary basin as a drainage area, and Area 3 lacks a good uranium source. Accordingly, the following conclusions were drawn. The integrity of the source-pathway-trap system is crucial and indispensable for the metallogenesis of calcrete-hosted uranium deposits. Moreover, high-quality uranium sources and sedimentary environments are conducive to the formation of large-scale calcrete-hosted uranium deposits. The duration of uranium enrichment and accumulation directly affects the scale of calcrete-hosted uranium deposits. The favorable sedimentary environment for calcrete-hosted uranium deposits is an evaporative lake (playa) with large uranium sources in the study areas of Saudi Arabia. Therefore, this study can guide the exploration of calcrete-hosted uranium deposits in similar areas.

Cross-border mining is a difficult and hot topic in the current supervision of oil and gas mines. Based on the judgement and interpretation of superficial and surface engineering, such as well sites, station sites, oil wells, gas wells, metering plants, gas gathering stations, gathering and transportation stations, patrol roads, and oil and gas pipelines within a single mining right, this study proposed for the first time that the combined information of superficial and surface engineering allow for quickly clarifying the accumulation and flow direction of oil and gas and accurately identifying and determining the production sites belonging to the same mining right and the cross-border sites. This method has been applied to a certain oil field as the test area and has been proved effective.

The remote sensing-based feature extraction of opencast mining areas is a hot topic in research on the monitoring of mining activities. However, there is a lack of systematic reviews and summaries of relevant studies. Therefore, this study first defined the features of an opencast mining area, divided the feature extraction into single- and multi-feature extractions according to feature types, and briefly described the differences between the feature extraction of opencast mining areas and general surface feature extraction and land use classification. Then, this study briefly summarized the sources and data processing platforms of remote sensing images available in relevant studies. Subsequently, this study divided the remote sensing-based methods for the feature extraction of opencast mining areas into three categories, namely visual interpretation, traditional feature-based approach, and deep learning. Then, it summarized the research status of these methods and analyzed their advantages, disadvantages, and applicability. Finally, this study proposed the future research direction of the remote sensing-based feature extraction of opencast mining areas, holding that the future developmental trend is to further promote the intelligent, fine-scale, and robust feature extraction of mining areas by effectively utilizing multi-source and multi-temporal data, networks with a stronger feature extraction capacity, and methods for the optimization of complex scenes. The results of this study can be used as a reference for the study and application of remote sensing-based feature extraction of opencast mining areas.

Vehicle detection is a hot research topic in the fields of computer vision, photogrammetry, and remote sensing. With the continuous development of deep learning technology, vehicle detection based on remote sensing images has been applied in fields such as smart city construction and intelligent transportation. This study systematically summarized existent vehicle detection algorithms based on remote sensing images and deep learning models and highlighted the classification, analysis, and comparison of one-stage and two-stage vehicle detection algorithms. Moreover, this study summarized the key technologies of vehicle detection in large-scale and complex backgrounds and analyzed the advantages and disadvantages of mainstream deep learning models of vehicle detection based on remote sensing images. Experiments were conducted to evaluate the YOLOv5, Faster-RCNN, FCOS, and SSD algorithms using DOTA and DIOR datasets. The vehicle detection precision based on the DOTA dataset was 0.695, 0.410, 0.370, and 0.251, respectively and that based on the DIOR dataset was 0.566, 0.243, 0.231, and 0.154, respectively. The experimental results show that the small target scale is still the main factor restricting the vehicle detection performance based on remote sensing images and that the application of deep learning models to the detection of small targets is to be further improved. Finally, based on public datasets and the analysis of existing algorithms, this study proposed the solution and development trend of vehicle detection based on remote sensing images in large-scale and complex backgrounds.

Using remote sensing to detect changes in urban buildings can obtain the change information of building coverage quickly and accurately. However, it is difficult to detect 3D changes quickly and accurately based on image data alone. Moreover, conventional point cloud-based methods have low automation and poor precision. To address these problems, this study used the airborne LiDAR point clouds and employed the RandLA-Net’s point cloud semantic segmentation method to improve the accuracy and automation of change detection. Meanwhile, the failure in differentiating two-period data due to point cloud disorder was overcome through point cloud projection. The standard RandLA-Net method, with the location and color information of points as features, is mainly used for semantic segmentation of street-level point clouds. In this study, urban large-scale airborne point clouds combined with the inherent reflection intensity and the spectral information of point clouds given by images were used to explore the influence of different feature information on the precision of the results. Furthermore, it was found that in addition to the point cloud intensity and spectral features, the coordinate information of points is equally important and can be converted into relative coordinates to significantly improve the result precision. The experimental findings show that the results obtained using RandLA-Net are significantly better than those using conventional methods for building extraction and change detection. This study also verified the feasibility of using deep learning methods to process LiDAR data for building extraction and change detection, which can realize reliable 3D building change detection.

Impervious surface is a key factor to measure the urban ecological environment. It is of great significance for urban development planning to grasp the dynamic changes of impervious surfaces timely and accurately. Taking the Chenggong District of Kunming City as an example, based on the Landsat images in 2007, 2011, 2015, and 2019, the comparative study of normalized difference impervious surface index (NDISI) and modified soil adjusted vegetation index(MSAVI) was carried out to analyze the spatial and temporal evolution characteristics of impervious surface. The results showed that: ①As the extraction accuracy and Kappa coefficient of NDISI were 87.01% and 0.81, respectively, which were better than MSAVI’s 81.78% and 0.75, this paper selected the NDISI method to extract impervious surfaces in the Chenggong District;② the impervious surface area extracted in this paper increased from 46.12 km² in 2007 to 72.64 km² in 2011, 146.94 km² in 2015 and 164.42 km² in 2019, especially from 2011 to 2015, the impervious surface area had the fastest growth rate and nearly doubled. The changes to the impervious surface in Chenggong District are mainly influenced by such factors as national policies, urban planning, topographic factors, and traffic development. The impervious surface area along the Dianchi Lake in the west of Chenggong District and several administrative regions in the middle of Chenggong District developed rapidly, which brings certain pressure on the prevention and control of waterlogging in urban areas and the Dianchi Lake area. In the process of future urban planning, the expansion scope and speed of impervious surfaces should be well controlled to avoid ecological and environmental problems caused by the unreasonable spatial patterns of impervious surfaces.

This study aims to achieve the dynamic and continuous monitoring of drought in Xinjiang. Based on the temperature vegetation dryness index (TVDI), as well as the Sen’s slope trend analysis, R/S, and partial correlation analysis, this study analyzed the spatial and temporal dynamics, changing trends, and future sustainable state of TVDI and the influences of seasonal precipitation and temperature on TVDI in Xinjiang during the period from 2001 to 2020. The results are as follows. ① The northern Tianshan Mountains and the Kunlun Mountains showed minimum TVDI values of less than 0.57, indicating light drought. The Tarim and Junggar basins showed TVDI values of greater than 0.86, indicating extraordinary drought. ② The TVDI values in spring decreased at a rate of 0.001 3/a. By contrast, the TVDI values in summer, autumn, and winter increased at a rate of 0.001 4/a, 0.002 0/a, and 0.000 8/a, respectively. Therefore, the increased amplitude of the TVDI values was the highest in autumn and the lowest in winter. ③ In the near future, the TVDI values in most regions of Xinjiang will increase in spring and winter, while the pixel quantity of most TVDI values will increase in summer and autumn. ④ The TVDI values were mainly negatively correlated with precipitation in spring and winter and were positively correlated with precipitation in summer and autumn. The TVDI values were mainly positively correlated with temperature in spring and were negatively correlated with temperature in autumn and winter. Moreover, the TVDI values in summer had a decreased correlation with temperature from west to east, with the correlation gradually changing from a negative to a positive correlation.

In order to explore the development trend of border cities in China and assess the city’s border defense capability, the D-LinkNet34 deep learning algorithm is used to automate the extraction of buildings and roads in Tuolin, Shiquanhe and Pulan towns in Tibet Autonomous Region, and to analyze the development trend and border defense capability of border towns based on landscape index and population size. Analysis results show that: ① The extraction method based on D-LinkNet deep learning network can effectively further classify urban construction land, with average total progress of more than 80% and IOU above 70%.② The distribution of plaques in the towns of Pulan and Shiquanhe shows a trend of aggregation, and the trend of urban expansion weakened. The distribution of plaques in Tuolin Town shows a scattered trend, and the trend of urban expansion is obvious. ③ The building area is linearly related to the resident population, and the building area of Tuolin Town increased by about 68.75%from 2002 to 2018, and the resident population increased by about 39.00%. The building area of Shiquanhe Town increased by about 70.75% from 2004 to 2020, while the resident population increased by about 68.44%. The building area of Pulan Town increased by about 68.36% from 2005 to 2018, while the resident population increased by about 25.04%. This study provides a new method for quantitative evaluation of the expansion characteristics and border defense capability of border cities, as well as a reference for building China’s border defense capability.

The exploitation of rich and unique mineral resources in Hainan Island has promoted economic growth but has also caused serious ecological environment problems. Analyzing the impacts of mining in Hainan Island and proposing suggestions on ecological restoration facilitate the protection and management of the ecological environment in Hainan Island. To this end, this study obtained the information on land destruction and ecological restoration of mines in Hainan Island using 2018 remote sensing images with high spatial resolution through image preprocessing, establishing interpretation indicators, and man-machine interactive interpretation. Specifically, with the information on land destruction and ecological restoration of mines as input, the assessment indicator system for mine geological environment was established based on 13 assessment factors of four categories, namely physical geography, basic geology, resource damage, and geological environment. Then, this study analyzed and assessed the effects of the geological environment of mines based on the analytic hierarchy process, obtaining the following results. The severely affected areas account for 0.22% of the total land area of Hainan Province and are mainly distributed in Wenchang City, Ledong Li Autonomous County, Xiuying District of Haikou City, Chengmai County, Lin’gao County, and Changjiang Li Autonomous County. The mine geological environment problems in these areas mainly include secondary geological disasters such as mining collapse of goaves and landslides caused by the mining of large-scale iron ore mines, as well as soil erosion and ecosystem degradation caused by the mining of coastal zirconium-titanium placers. The moderately severely affected areas account for 1.68% of the total land area of Hainan Province and are mainly distributed in Wenchang City, Danzhou City, Chengmai County, Qionghai City, Lin’gao County, Haikou City, and Dongfang City. The mine geological environment problems mainly include land damage caused by landslides induced by the mining of small- and medium-sized iron ore mines, as well as severe impacts on original terrain and landforms and the natural ecological environment caused by mining. The generally affected areas account for 4.93% of the total land area of Hainan Province and are mainly distributed in the coastal areas in the eastern part, the economically developed areas in the middle and northern parts, and the area with rich metallic minerals in the western part in Hainan Province. The mine geological environment problems in these areas mainly include the destruction of the surface landforms and natural vegetation caused by the mining of the scattered small nonmetal mines of building materials. This study proposed ecological restoration countermeasures targeting the different geological environment problems. For metal mines, it is suggested to primarily restore the ecosystem by natural restoration methods, supplemented by artificial restoration methods based on the elimination of geological hazards, soil improvement, and water environment management. For zirconium-titanium placers and nonmetal mines of building materials, it is recommended to restore vegetation to prevent water and soil erosion. For the coastal mine areas with severe desertification, it is recommended to gradually restore the ecosystem of the mining areas by growing crops such as watermelons and peanuts to improve soil and planting trees such as casuarina and Vatican hainanensis.

Land use can cause carbon stock changes by affecting the structural layouts and functions of terrestrial ecosystems. Therefore, research on the relationship between land use changes and carbon stock is greatly significant for optimizing regional land use patterns and making sensible ecological decisions. This study predicted the spatial-temporal changing characteristics of land use and carbon stock in Xi’an under different scenarios in the future using the PLUS and InVEST models and investigated the impact of land use changes on carbon stock. The results are as follows. From 2000 to 2015, the expansion of construction land and the transfer of high-carbon-density land reduced the carbon stock of Xi’an by 2.49×10⁶ t. From 2015 to 2030 the carbon stock continuously declined by 2.14×10⁶ t in the natural growth scenario, and the carbon stock of Xi’an will increase by 6.92×10⁵ t in the ecological protection scenario due to the measures taken for land protection and transfer control. In the cultivated land protection scenario, the cultivated land will be protected, but the high-carbon-density land such as woodland and grassland will be affected by the expansion of construction land during 2015—2030, reducing the carbon stock to 1.60×10⁸ t. As indicated by the analysis of carbon density change, ecological protection measures can increase the changing rate of carbon density. Compared with the natural growth scenario, the ecological protection scenario will increase the proportion of areas with increased carbon density (mainly high-increase areas) from 0.05% to 1.57%. By contrast, under the cultivated land protection scenario, the carbon density will decrease, and high-increase areas will be transformed into moderately-high-increase areas. Based on cultivated land protection, it is necessary to take proper ecological protection measures in the future land use planning of Xi’an to control the rapid expansion of construction land from cultivated and forest land. Optimizing land use patterns can effectively reduce the loss of carbon stock, improve the level of regional carbon stock, and achieve regional sustainable development.

The urban environment is an important issue in the whole world, and the urban heat island (UHI) effect is one of the important research topics. Owing to the expansion of the urban area and the increase in population, the urban heat island effect has also significantly changed. With the Landsat imageries as the data source and the central urban area of Tangshan City, Hebei Province as the main study area, this study analyzed the impacts of green space evolution on urban temperature change using the methods such as the radiative transfer equation algorithm, supervised classification, gravity center shift, and random sampling. The results are as follows. ① During the study period, the development direction and area of UHIs were roughly consistent with the scale and direction of rapid urban development. Moreover, the migration directions of the gravity centers of the UCI/UHIs were similar to those of the green space and urban area, with the migration distance of the gravity centers of UCIs greater than that of the UHIs. ② The urban green space (UGS) has been continuously lost during the study period, with the largest loss area of approximately 55.79 km² occurring in agricultural land. Moreover, the largest increased area occurred in urban land and was approximately 47.85 km². ③ The evolutionary trends of UCIs/UHIs were inconsistent with those of the UGS in different periods. This result may be related to the stock of green space. ④ The cooling effect on the urban surface (-0.16 ℃) induced by green space expansion was much smaller than the warming effect on the urban surface (6.37 ℃) caused by green space loss. The research results will provide a reference for urban planning in order to rationally arrange green space, retain sufficient green space, and effectively reduce the development speed of UHIs.

As a typical and important ground target, the bridge is the vital passage between transportation lines, so automatic detection of a bridge is of great social and economic significance. Deep learning has become a new way of bridge detection, but the detection accuracy for bridges obscured by cloud and mist is low. In order to solve this problem, an automatic bridge target detection method combining Random erase (RE) data enhancement and the YOLOv4 model is proposed: firstly, the scale range of the target in the data set is determined, and the candidate frame size is obtained by K-means clustering; secondly, the cloud obscuration is simulated by a combination of RE and mosaic data enhancement; thirdly, the enhanced data set is trained by YOLOv4 network; and finally, the mean Average Precision (mAP) is used to evaluate the experimental results. The experimental results show that the detection accuracy obtained by mAP is 97.06%, which is 2.99% higher than that of traditional YOLOv4, and the average detection accuracy of bridges obscured by a cloud is improved by 12%, which verifies the effectiveness and practicability of the proposed method.

Mining collapse has caused damage to soil, vegetation, and water resources. With the implementation of the national ecological restoration strategy, it is significant to effectively identify and monitor collapse areas. For this purpose, based on multi-source high-resolution remote sensing images and Sentinel-1 SAR radar images, this study identified and monitored the mining collapses of a coal mine in Baiyin City, Gansu Province using the two technologies, namely the Stacking-InSAR method for extracting ground subsidence data and the human-computer interactive interpretation of optical images of mining collapse. Moreover, this study comprehensively compared the characteristics of both techniques and explored the application prospects of both techniques in the deployment of ecological restoration engineering. The results are as follows: ① The Stacking-InSAR radar monitoring technology can better reflect the deformation during the monitoring period and can effectively identify the mining collapse areas in shallow, middle, and deep coal seams. ② The high-resolution optical image technology can better identify the mining collapse areas in shallow and middle coal seams, more accurately identify the damaged land, and can well identify the historically formed mining collapse areas and damaged land whose collapse deformation has stopped. ③ The collapse deformation and land damage of various stages can be obtained by combining the InSAR monitoring technology and the recognition method base on high-resolution remote sensing images, thus providing detailed and reliable basic data for ecological restoration engineering.

Mangrove forests are periodically inundated by tidal water. This characteristic opens up an opportunity but also poses a challenge for the information extraction of mangrove forests using remote sensing technology. To explore the contribution of the red-edge band of GF-6 satellite data in information extraction of mangrove forests under the condition of random tides, this study investigated the southeastern Dongzhaigang area-the largest mangrove forest area in Hainan Province and obtained standard samples using the GF-2 satellite data. The reflectance spectral curves of typical surface features were constructed based on the standard samples and the GF-6 satellite data. Then, a baseline was established based on the bands strongly absorbed by vegetation, and the intertidal mangrove forest index (IMFI) applicable to the GF-6 satellite data was defined using the average reflectance of bands above the baseline. Meanwhile, the red-edge normalized difference vegetation index (RENDVI) was also established. The two indices were compared with commonly used indices, such as the normalized difference vegetation index (NDVI) and the normalized difference water index (NDWI), using box-whisker plots. Then, using the decision tree model constructed based on IMFI and RENDVI, information on typical mangrove forest in the study area were extracted. The precision of the extraction results was verified through comparison with visual interpretation results of the samples extracted from the GF-2 satellite data. The results show that: ① Because mangrove forests are periodically inundated by tidal water, the reflectance spectral curves of intertidal mangrove forests were relatively scattered between the standard spectral curves of water bodies and mangrove forests; ② IMFI and RENDVI can reflect the differences in the reflectance spectra of the red-edge and near-infrared bands and thus effectively separated the intertidal mangrove forests, mangrove forests, and water bodies; ③ The decision tree model constructed based on IMFI and RENDVI can effectively extract the distribution information of the mangrove forests, with an overall accuracy of 0.95 and a Kappa coefficient of 0.90. The introduction of the red-edge band plays an important role in the information extraction of mangrove forests and has great potential for application. This study can be used as a reference for the ecological applications of red-edge data from domestic satellites.

The mariculture industry occupies an important position in the marine economy of Guangdong Province. Timely and accurate knowledge of the spatial distribution and area changing trends of mariculture areas can greatly promote the sustainable development of the mariculture industry. Conventional interpretation methods for remote sensing images have problems of poor repeatability, low applicability, and high subjective arbitrariness. By contrast, the U-Net convolutional neural network, which belongs to the deep learning network model, can better extract the features of the object with higher extraction precision. Therefore, based on the multi-temporal Landsat TM/OLI remote sensing images, this study identified the mariculture areas (enclosed-sea and open-cage aquaculture areas) in Guangdong from 1998 to 2021 using the U-Net model as the interpretation model. The area trend analysis of mariculture areas was made. The changing characteristics of mariculture areas in terms of spatial distribution patterns were studied. The results are as follows. Compared with network models such as K-Means cluster analysis and DBN, the U-Net model with higher interpretation precision is more suitable for the interpretation of mariculture areas in Guangdong. The mariculture areas in Guangdong are mainly distributed in the western portion of Guangdong, such as Zhanjiang, Jiangmen, and Yangjiang. The mariculture areas in Guangdong can be classified into three levels in terms of area. They have small changes and keep a relatively stable state. The mariculture areas in Guangdong showed a spatial trend of outward expansion from 1998 to 2014 and inward contraction from 2014 to 2021. This study will provide data and technical support for the scientific management of the mariculture areas in Guangdong.

The GXL (GeoImaging Accelerator) is a new generation of distributed processing platform for remote sensing data. It is fast, efficient, and flexible and plays an important role in the processing of domestic satellite data. This study investigated the software development kit (SDK) of GXL from the aspects of view, controller, and model based on the MVC (Model View Controller) framework of GXL. Furthermore, it developed a new algorithm processing module and employed distributed program deployment to enhance the function and algorithms of satellite data processing. An experiment was carried out to process domestic satellite (GF-1, GF-2, and ZY1-02C) data. The experiment results show that the SDK of GXL allows for flexibly expanding the processes for satellite data processing and improving the productivity of domestic satellite products. Therefore, the SDK of GXL can better satisfy the demands of various industries.

Ecological quality assessment is an important prerequisite for guaranteeing the harmony and stability of the production and life of human beings and the ecological environment and for achieving the sustainable development of regional social economy. It has become a new trend to quickly, accurately, and objectively assess the regional ecological quality using use remote sensing technology. This study used the MODIS data of the Beijing-Tianjin-Hebei region in 2001, 2010, and 2019 to extract four important indices, namely, NDVI (greenness), LSM (humidity), NDBSI (dryness), and LST (heat). Then, this study obtained the MODIS remote sensing ecological index (RSEI_M) using the principal component analysis method to conduct a comprehensive assessment and change monitoring of the ecological quality in the Beijing-Tianjin-Hebei region over the past 20 years. The results are as follows. ① The ecological quality of the Beijing-Tianjin-Hebei region shows distinct regional differences. The Yanshan Mountain in the north and the Taihang Mountain in the west have high ecological quality, while the Zhangjiakou area in the northwestern part of Hebei Province and the urban center in the southeastern part of Hebei Province suffer low ecological quality. ② In 2001, 2010, and 2019, the average RSEI_M of the Beijing-Tianjin-Hebei region was 0.556, 0.583, and 0.527, respectively, with the overall ecological quality showing a downward trend. ③ From 2001 to 2019, the area with improved and degraded ecological quality in the Beijing-Tianjin-Hebei region accounted for 20.18% and 35.69% respectively, and the ecological quality in this region showed a pattern of improvement in the northwest and degradation in the southeast. The main reasons for the ecological improvement in the northwestern part of the region are the changes in water and heat conditions, such as an increase in precipitation and temperature, and a series of man-made protection measures. The reasons for ecological degradation in the southeastern part of the Beijing-Tianjin-Hebei region mainly include the rapid advancement of urbanization and the enhancement of social and economic activities. The comprehensive assessment of regional ecological quality can be effectively achieved based on MODIS data, thus providing a reference for the green and high-quality development of regional social economy.

In order to solve the problem of poor model generalizing ability in real super-resolution reconstruction of remote sensing images, which is easily caused by the use of artificial high-low resolution image pairs, combined with the residual in residual (RIR) module of residual channel attention network (RCAN), dual regression network (DRN) is improved, and residual dual regression network (RDRN) is proposed. Ten thousand 512 × 512 pixel images from LandCover.ai and DIOR aerial image data sets were selected to form the sample data set for training and testing the network, and the reconstruction results were compared with those of other super-resolution network models. The experimental results show that RDRN has an excellent performance in both reconstruction quality and model parameters. It can achieve a better super segmentation reconstruction effect with lower model complexity and has good generalization ability for different low-resolution remote sensing images.

This study performed a multi-scale error analysis of the mean surface downward shortwave radiation flux product ERA5 (0.25° × 0.25°) of the European Centre for Medium-Range Weather Forecasts (ECMWF) using 93 pieces of solar radiation hourly data in 2013 of China. Subsequently, this study revised and analyzed the total radiation product ERA5 by training the random forest model using various relevant elements such as meteorological and geographic ones. Finally, the model was used to obtain the map of revised hourly radiation spatial distribution. As a result, the reanalyzed data can be better applied in industries such as agriculture, electric power, and urban construction. The results are as follows. ① The MAE, RMSE, and R values between the ERA5 solar radiation and the measured values of stations in 2013 were 27.60 W/m², 29.87 W/m², and 0.97 respectively. Moreover, the ERA5 values were higher than the measured values of stations. ② The accuracy was improved after the revision using the random forest model. After revision, the MAE, RMSE, and R values between the ERA5 solar radiation and the measured values of stations were 3.34 W/m², 3.85 W/m², and 1.00, respectively, indicating that correlation was significantly improved. ③ The spatial macroscopic distribution patterns of radiation before and after revision were consistent, but the ERA5 radiation value significantly decreased in local areas.

Compared with common dual-temporal satellite images, satellite time series images contain richer surface information and can alleviate the impact of foreign objects with the same spectrum and the same object with different spectra. Therefore, they play an important role in change detection. However, the change detection methods for satellite time series images are mostly based on pixels and ignore the spatial relationship between pixels and their surroundings. This causes noise in the change detection result. Accordingly, this study proposed a method of change detection based on spatial-temporal-spectral features(CDSTS) for satellite time series images. First, the temporal, spatial (textural and statistical), and spectral features of each pixel were extracted from Landsat time series images using a gray-level co-occurrence matrix and local statistical calculation methods. Then, anomalies of time series features were automatically screened according to the time series performance regularity of each pixel in different bands. These anomalies were then fused with the detection results of the continuous change detection and classification method (CCDC) to obtain high-precision changed/unchanged training sample points. Finally, the SVM classifier was trained using the training sample points and their corresponding spatial-temporal-spectral features for full graph classification. The results show that the CDSTS algorithm significantly outperforms the commonly used time series change detection algorithms CCDC and COLD (continuous monitoring of land disturbance) in terms of change detection precision, with the overall precision improved by 4.8 to 11.7 percentage points.

The terrain conditions and tea plantation structure of Wuyishan City are complex, with cloudy and rainy weather, so it is difficult to obtain satellite images here. To address the problem of difficult extraction of tea gardens from a single image source, we investigated the spectral information of Sentinel-2 images and the texture features of Google images in Xintian Town, Wuyishan City, coupled with which a tea garden classification method based on multi-source high-resolution satellite images and multidimensional convolutional neural networks (MM-CNN) was established. In this method, tea gardens and suspected tea gardens were extracted, respectively, with two models developed with images with different spatial resolutions, based on one- and two-dimensional CNN. Results obtained with the two CNN models were combined, and the high-accuracy distribution of tea gardens in the study area was generated in a relatively economical and efficient way. The results showed that the spatial distribution accuracy of the tea gardens identified by MM-CNN is better than that of the single image source method. The MM-CNN method is highly universal and robust and provides a reference method for efficiently monitoring the distribution of tea gardens in large-scale hilly areas of South China.

The current dead tree detection primarily relies on manual field surveys and, thus, is limited by forest topography, suffers a low detection efficiency, and is dangerous. Given these problems, this study proposed a YOLOv4-tiny dead tree detection algorithm based on the attention mechanism and spatial pyramid pooling (SPP) and improved the original detection model. First, the SPP structure was introduced after the Backbone part of the model to combine local and global features and enrich the feature representation capability of the model. Then, the original activation function LeakyReLU in the model was replaced with ELU, which made the activation function saturate unilaterally, thus improving the convergence and robustness of the model. Finally, the attention mechanism ECANet was introduced into the model to enhance the capacity of the network to learn important information in images, thus improving the performance of the network. The images of trees in a mountain forest of a scenic area in southern Liaoning were collected using an unmanned aerial vehicle (UAV). Then, dead trees in these images were detected using different models. The detection results show that the improved algorithm had a detection accuracy of 93.25%, which was improved by 9.58%, 12.57%, 10.54%, and 4.87% than that of the YOLOv4-tiny, YOLOv4, and SSD algorithms and an algorithm stated in literature [8], respectively, and achieved the effective detection of dead trees.

Uneven brightness and inconsistent colors are prone to occur inside and between captured images in the process of remote sensing imaging. However, the manual color conditioning combined with image processing software can no longer meet the color matching demand of geometrically increasing remote sensing images. Given this, this study proposed a kind of unsupervised channel-cycle generative adversarial network (CA-CycleGAN) integrated with the attention mechanism suitable for ground objects in complex urban areas with a high land utilization rate. Firstly, the sample data set used for color reference was manually made through histogram adjustment and Photoshop, and the appropriate urban images were selected as the sample set to be corrected. Then, the two kinds of images were cut respectively to obtain the preprocessed image sample sets. Finally, the preprocessed image set to be corrected and the image set for color reference were processed using the CA-CycleGAN. Because the attention mechanism has been added to the generator, the generated focuses can be distributed into key areas using the attention feature map in the training process of the confrontation between the generator and the discriminator, thus improving the image effects and obtaining the color correction model based on urban images and the images after color correction. Both the image correction effect and the loss function diagram show that the proposed method is optimized based on the CycleGAN and that the comprehensive performance of the CycleGAN integrated with the attention mechanism is better than that without the attention mechanism. Compared to conventional methods, the method proposed in this study greatly reduced the time for color correction and achieved more stable image color correction effects than manual color matching. Therefore, the method proposed in this study enjoys significant advantages in the color dodging of remote sensing images and has a good application prospect.

Flat landslides, typically characterized by crack grooves, are a common type of special disasters in southwestern China. However, the dense vegetation and complex terrain in disaster-developed areas limit the efficiency of conventional ground or remote sensing (RS) survey methods in the identification and extraction of disaster information. As one of the emerging remote sensing technologies, the airborne LiDAR technology and its data visualization analysis methods provide a new solution for the accurate identification of flat landslides. First, the high resolution digital elevation model (HRDEM) can be obtained using the UAV airborne LiDAR. Then, the HRDEM can be combined with visualization methods including sky view factor (SVF), hillshades, and 3D morphology simulation for the effective identification of flat landslides and their crack grooves. This study investigated the newly identified landslide hazard in the southern part of Nuoguzhai Village, Chunzai Town, Tongjiang County, northern Sichuan Province. The comprehensive RS identification method was used to realize the construction of landslide identification signs, the determination of the landslide boundary, the identification of crack groove position, and information extraction based on airborne LiDAR data. Combined with the results of field surveys, the effectiveness of the airborne LiDAR technology for the identification of flat landslides and their crack grooves in highly vegetation-covered areas was verified from both qualitative and quantitative aspects. The related study results can be used as a reference for the early identification, monitoring, and prevention of flat landslides.

Dynamic shoreline monitoring is greatly significant for the scientific management of coastal zones and the rational utilization of marine resources. Based on the Landsat remote sensing images of four periods i.e., 1990, 2000, 2010, and 2020, this study extracted the changes in the shorelines and the coastal zones within the 2 km of the buffer zone in the north of Jiaodong Peninsula from 1990 to 2020 by making comparison and using an object-oriented method. By combining the calculation method for shoreline change intensity, this study analyzed the changing rate and temporal-spatial distribution characteristics of the shorelines using the digital shoreline analysis system (DSAS). Then, this study conducted a driving analysis of changes in the shoreline by constructing a human activity intensity index (HAII) model. The results are as follows. The shorelines of the study area generally showed an upward trend and advanced slowly to the seaside. The overall length of the shorelines increased by 183.13 km. The highest increased and decreased amplitude occurred in artificial shorelines and sandy natural shorelines, respectively. The shoreline changing rates showed uneven temporal-spatial distribution. The maximum growth rate of 94.59 m/a occurred in the Jiaolai River - Jiehe River section, while the maximum erosion rate of -49.01 m/a occurred in the Jiehe River - Dagujia River section. The changes in offshore human activities were the main contributor to the temporal-spatial changes of coastlines in the study area. The lengths and types of shorelines were mainly affected by human activities through sea reclamation and port construction.

Targeting the subtropical climate characteristics of the Guangdong-Hong Kong-Macao Greater Bay Area, this study acquired the images of the experimental area from the TerraSAR-X Radar remote sensing satellite. Given the varying scale of the surface feature targets in the Radar satellite observation scenes, this study proposed an ENet convolution spatial pyramid pooling module (ENet-CSPP) model for surface feature classification. Since ordinary convolution can more effectively maintain domain information than atrous convolution, this study proposed a multi-scale feature fusion module based on convolution spatial pyramid pooling. Since there were a few training samples in the SAR remote sensing image dataset, this study combined the multi-scale feature fusion module with the lightweight convolutional neural network. The encoder of the ENet-CSPP network consisted of an improved ENet network and the convolution spatial pyramid pooling module. The decoder output surface feature classification images after the fusion of deep and shallow features. The quantitative comparison experiments were conducted on the GDUT-Nansha dataset. The ENet-CSPP model outperformed other models in three performance indices, namely pixel accuracy, average pixel accuracy, and mean intersection over union. This result indicates that the multi-scale lightweight model effectively improved the accuracy of surface feature classification.

Water is an important factor in the formation and maintenance of wetland ecosystems. Monitoring the changes in the water area of wetlands is of great significance for wetland conservation. Taking Sentinel-1A data from 2018 to 2019 as the data source, this study calculated the intra- and inter-annual synthetic aperture radar (SAR) backscattering coefficient (σ⁰) and coherence coefficient (μ⁰) images of the Zhalong Wetland. Then, this study assigned weights according to the proximity to water bodies of color optical images and extracted the weighted images of σ⁰ and μ⁰. Finally, this study extracted the wetland water bodies using the threshold segmentation method and random forest algorithm. The purpose is to monitor the dynamic variations in the wetland water area and explore the intra- and inter-annual variation rules of the wetland water body. The results are as follows. The random forest algorithm yielded the highest extraction accuracy of water bodies, with an absolute value of the mean difference of representative months was 6.69 km². The threshold segmentation method based on μ⁰ images yielded the lowest classification accuracy of water bodies, with an absolute value of the mean difference of 13.07 km². Overall, the intra-annual water area of the Zhalong Wetland showed significant seasonal variations during 2018—2019. The water area fluctuated in the ranges of 1 300~1 600 km² during late spring and early summer and 700~900 km² during late summer and early autumn. The inter-annual water area varied with conditions such as climate and temperature. In particular, the wetland water area in October and November 2019 was approximately 1 050 km² greater than that in 2018 due to large amounts of rainfall. As shown by the calculation based on effective data, the water area in 2019 was about 550 km² greater than that in 2018 in the Zhalong Wetland.

Exploration of the water body extraction model based on GF-6 satellite images can provide new data sources and technical support for water body monitoring. First, GF-6 WFV satellite data of Zonag Lake was used to analyze the reflectance difference between water and other ground objects in each band of GF-6 WFV satellite data. Based on this, a novel water index named red side water index (RSWI) was constructed. Then, the overall accuracy and Kappa coefficient generated by the confusion matrix were used to verify RSWI and the other three water extraction models, which include the single-band threshold method, normalized difference water index, and modified shade water index. Finally, six typical lakes with different types of areas larger than 100 km² in Hoh Xil were selected for analysis of general applicability. The results showed that compared with other methods, the decision tree model composed of RSWI and NIR bands effectively eliminates the influence of lake bottom sediments on water bodies and extracts shallow water bodies more completely, with an overall accuracy of 93.78% and a Kappa coefficient of 0.87. Additionally, it has been found that the stability and general applicability of RSWI are better than those of other water body models with respect to different types of lakes.