最新刊期
卷 27 ， 期 9 ， 2023
-
摘要：In the past decades, the effects of global climate change and the increase of human activities have remarkably increased the demand for remote sensing monitoring. Moreover, with the accumulation of remote sensing data from multiple platforms and multiple sensors, the quantity and quality of multitemporal images have substantially improved. Multitemporal remote sensing images Change Detection (CD) is a processing and analysis technology that aims to automatically detect, identify, and describe changes occurring in the same geographical area at different times. With the advancement of remote sensing and Artificial Intelligence (AI) technology, traditional data-driven and modal CD methods are evolving toward data-model-knowledge jointly driven direction to solve the land surface spatio-temporal CD problem in a variety of application fields in a more automatic, refined, and intelligent manner. This paper first summarizes existing problems in multitemporal remote sensing CD by analyzing the use of homogeneous and heterogenous data sources, developments from traditional to intelligent CD models, and challenges from theoretical to practical CD applications. Optical image CD is taken as an example, and the evolution of CD technology in the era of AI is examined, which can be summarized as three periods of data-driven CD, model-driven CD, and data-model-knowledge driven CD. Then, the characteristics and problems of each periods are discussed. Furthermore, for each of the three aspects (unsupervised, supervised, and weakly supervised), the characteristics and trends in the development of traditional to cutting-edge CD techniques are discussed. In the future, one can focus on breaking through key issues such as the physical interpretability, generalization, and transferability of the CD models as well as their successful implementation in cross-data, cross-scene, and cross-domain applications.关键词：remote sensing;change detection;multi-temporal analysis;artificial intelligence;machine learning;deep learning- 2531
- |
- 1831
- |
- 1
发布时间：2023-10-07 -
摘要：Remote sensing change detection can provide information on land surface change, which is important for studying man-nature interactions and facilitating sustainable development. With the advancement of remote sensing imaging technology and the rapid development of computer technology, extensive remote sensing images with various modes and spectral, spatial, and temporal resolutions have been collected, enabling the development of massive remote sensing change detection methods based on deep learning and their successful application in a wide range of fields.Unlike previous reviews, this work examines remote sensing change detection based on deep learning from the perspectives of bibliometric analysis, research scale, and critical problem exploration to provide reference materials for future remote sensing change detection research. The definition and importance of remote sensing change detection as well as the motivation for this review are briefly presented in the introduction. The literature structure and research hotspot information of existing research, such as the number of publications, distribution of journals and institutions, main researchers, common data sources, network model, and application field information, are clarified in the second section, which is combined with bibliometric analysis. In the third section, focus is on deep learning-based remote sensing change detection algorithms, which are categorized and presented on three scales: pixel, object, and scene. How to extract pixels, objects, and scenes from remote sensing images as well as how to perform network analysis are also explained. In the fourth section, the limitations of deep learning-based remote sensing change detection are covered, and the most recent research are presented to address these issues as well as future development possibilities. Next, a segment dedicated to the finale.The bibliometric analysis reveals deep learning-based change detection has progressed rapidly in the last three years, with fruitful research results and domestic institutional scholars dominating. High-resolution images and CNN are the most used data sources and network model, and extensive land use/coverage and building change detection are hot application fields. As for methods, different research scales respond to varied data features and network model structures. The object and scene technique have advantages, and they face similar issues, which are summarized below. First is the problem of detecting changes using multimodal remote sensing data. To address this, adversarial training, attention mechanisms, and feature deep fusion methods based on feature space transformation appear promising. Multimodal data fusion and other multimodal learning approaches are among the future’s emerging directions. Second, change detection under small sample and imbalanced sample settings is difficult. Semi-supervised schemes must be improved to address the problem of small sample size, and self-supervised methods are predicted to become a research hotspot. The oversampling technique and ensemble learning in deep learning models provide a new path for unbalanced samples. The third issue is obtaining diversified change information. Semantic change detection, which obtains extensive information on change types, and Transformer for time series change detection, which obtains long-term change information, are the future trends. Furthermore, deep learning-based change detection requires advances in gathering dynamic information such as time and seasonal pattern of change.This work systematically compiles and reviews the research status and progress of deep learning-based remote sensing image change detection. Multimodal heterogeneous change detection, semantic change detection, and time series change detection are future prospects as application needs and data diversity grow. In the areas of resources, the environment, and disaster relief, practical uses of existing knowledge are few. Continuously extending the in-depth study of new technologies and methods is required as is promoting wide, in-depth remote sensing change detection research and application.关键词：remote sensing;change detection;deep learning;bibliometric;methods classification;challenges and prospects;review- 2037
- |
- 2413
- |
- 1
发布时间：2023-10-07 -
摘要：Deep learning has shown great potential in the change detection of multi-temporal remote sensing images in recent years. However, the annotated datasets, which are critically required in training change detection networks, is often limited in various change detection tasks in practical applications. As land cover change usually occupies only a small portion of an image, the number of changed samples is often very small, leading to a serious imbalance between changed and unchanged samples. Therefore, it is an urgent challenge to effectively training change detection networks with small and imbalanced change detection samples. Compared to the collection of change detection samples, it is much easier to obtain land cover classification samples at a single time. Based on the adequate land cover classification samples, a well-trained land cover segmentation network can provide important prior features for change detection.Therefore, this paper proposes a method named as siamese Nested-UNet for change detection in posterior probability space (SNU-PS), which aims to reduce the dependence on change detection samples by utilizing the posterior probability information of segmentation network. The method first trains a High-Resolution Network (HRNet) based on land cover classification samples to obtain the posterior probability of the bi-temporal image. Then, the posterior probability images are input into a siamese Nested-UNet for change detection(SNU) to obtain the change detection results. In order to simplify the network complexity and reduce the training difficulty, the training of semantic segmentation network and change detection network are carried out step by step without interactions in their training stages. As the posterior probability image already contains semantic information of land cover, the requirement of the change detection samples is reduced because the change detection network does not need to extract the features in the multi-spectral images.The change detection experiments based on the SpaceNet7 and HRSCD datasets show that SNU-PS can well utilize the semantic information provided by the land cover segmentation network and maintain stable change detection accuracy when it was trained with different change detection sample sizes. Compared with Post Classification Comparison (PCC), CVAPS (Change-vector analysis in posterior probability space), and different change detection networks (FC-EF, BIT, PCFN, and SNU), SNU-PS achieved higher accuracy and better stability, especially when the sample size is small. Unfortunately, all of the compared methods failed to identify the change type due to the extreme imbalanced samples of different change types.SNU-PS method makes full use of the low-cost classification samples to train the semantic segmentation network, which helps to reduce the reliance on the change detection samples because the change detection network in SNU-PS does not undertake the feature exploration of multi-spectral images. Moreover, the semantic segmentation network and the change detection network are integrated with independent training process in SNU-PS, thus the integration of two networks does not increase the training difficulty and semantic segmentation network can be flexibly replaced with better network if available. In conclusion, the proposed SNU-PS maintains good performance under small sample size, thus has a good applicability in various change detection tasks.关键词：land cover;change detection;deep learning;small sample;sample imbalance;semantic segmentation network;Siamese network;posterior probability- 246
- |
- 967
- |
- 0
发布时间：2023-10-07
Change Detection and Deep Learning
-
摘要：Bidirectional Reflectance Distribution Function (BRDF) is a basic variable in optical quantitative remote sensing, which describes the reflection anisotropy of surface targets with different sun-target-sensor geometry. BRDF not only plays an important role in the characterization of land surface structure but also has great relevance for the research of earth energy balance. The definition, inversion, and observation technology of BRDF have made remarkable progress over the past 40 years. Moreover, with the launch of multiangular remote sensors, its BRDF products have been generated and released, which are widely used in remote sensing community.Based on the principle of BRDF inversion, the most common problems associated with BRDF inversion are first analyzed, including the ill-posed problem caused by insufficient observations, the noise of observation data, and the noise accompanied by the introduced prior knowledge that causes the uncertainties of the inversion.Then, the current BRDF inversion methods used to solve the problems above are analyzed, summarized, and classified into three: fundamental inversion methods, regularization-constrained inversion methods, and information classification and amplification inversion methods. Fundamental inversion methods are suitable when the number of observations is greater than the number of variables to be retrieved, and prior knowledge is not required. They include the least square method, the least variance method, and the robust estimation method. The least square method and the robust estimation method are only used when observations are sufficient, but the least variance method can be used even when observations are insufficient. However, prior knowledge is required for regularization-constrained inversion, information classification, and information amplification inversion methods, all of which are used to address the ill-posed problem. The regularization-constrained inversion method constrains the inversion results by regularization rules. The information classification and information amplification inversion methods include multistage target decision making, Bayesian estimation, Kalman filtering, and multisensor joint inversion. Among them, the multistage target decision-making method can allocate as much information as possible to the target parameters, and the Bayesian estimation method, the Kalman filter method, and the multisensor joint inversion method address the issue of insufficient observations by expanding data sources.The challenges of how to improve the inversion accuracy of land surface BRDF in the future were also discussed, namely, high-resolution BRDF inversion, mountainous surface BRDF inversion, and the application of artificial intelligence technology in BRDF inversion. The BRDF model suitable for low- and medium-resolution pixel scales is not suitable for high-resolution pixel scales due to the strong proximity effect and mutual occlusion effect among high-resolution pixels. With the rapid growth in high-resolution satellite data and UAV data, the development of appropriate models for high-resolution pixel-scale BRDF inversion is imminent. The second model, mountainous surface BRDF inversion, also faces challenges due to the complex terrain and a lack of remote sensing data. To solve the problem, a multisource, multiscale joint inversion method as well as the prior knowledge dataset of mountainous surface BRDF need to be created. Finally, with the accumulation of remote sensing data over the last few decades, remote sensing has entered the “Big Data Era.” Investigating how to invert surface BRDF with remote sensing based on artificial intelligence technology is worthwhile.关键词：Bidirectional Reflectance Distribution Function (BRDF);multiangle;quantitative;optical remote sensing;ill-posed;inversion principles;Inversion methods;surface energy balance- 683
- |
- 539
- |
- 0
发布时间：2023-10-07 -
摘要：Cloud is an important indicator of weather and climate change, and it is of great importance to atmospheric energy distribution and radiative transfer. Satellite remote sensing has become an indispensable technology for cloud research due to its global coverage, multiple types of information, and high repetitive frequency. In this work, the development of cloud measurement technology and the data application research of current spaceborne millimeter-wave radar and lidar are summarized. The future development requirements and trend are then proposed according to the progress of current cloud measurement technology and applications.First, the technology research and data typical application status of spaceborne millimeter-wave radar and lidar are introduced and summarized, in which the key technologies and important application of Cloud Profile Radar (CPR) on CloudSat and cloud aerosol lidar (CALIOP) on CALIPSO are highlighted. The achievements of collaborative observation by CPR and CALIOP as well as the aspects to be improved are then summarized emphatically. Based on the requirement analysis of obtaining the 3D fine structure of global cloud, the development and application potential of novel measurement technologies such as terahertz radar, high-spectral-resolution lidar, and multisensor data fusion are provided. Furthermore, the key technologies of the new cooperative observation modes, such as multisatellite system and multiple payloads on single satellite platform, are discussed. The influence of orbit altitude on the detection performance is deliberated based on the analysis of the relationship between the received intensity of echo signal and the orbit altitude.The cloud load performance of active remote sensing, cooperative observation mode, cooperation with passive remote sensing instruments, and multisource data retrieval and fusion for spaceborne cloud measurement still need to be improved according to the research status of spaceborne cloud radar and lidar. The latest advanced technology of spaceborne active remote sensing can be fully utilized for observing more accurate cloud parameters and more microphysical information, such as multiwavelength, multipolarization, Doppler technology, terahertz radar, hyperspectral lidar, and multisensor data fusion method. The collaborative observation of microwave radar, lidar, and other sensors on the same platform is necessary for better understanding the formation and evolution of cloud, to improve the ability of weather forecasting and climate monitoring further.Satellite active remote sensing for cloud measurement is employed to observe the macroscopic and microphysical characteristics of cloud accurately. Its high-quality data have been widely applied to cloud physics, weather forecasting, environment monitoring, and climate change. Moreover, it plays an important role in improving the accuracy of numerical weather forecasting and climate research. With the continuous progress of active remote sensing technology, more cloud parameters and higher performance can be realized for spaceborne cloud measurement. In addition to the high-technology applications in spaceborne cloud remote sensing, collaborative observation is an important development trend. A future perspective is then projected on the cooperative observation mode of multiple sensors with common platform and new data fusion methods.关键词：Satellite cloud measurement;active remote sensing;spaceborne millimeter wave radar;spaceborne LiDAR;cooperative observation;data fusion- 944
- |
- 508
- |
- 0
发布时间：2023-10-07
Research Progress
-
摘要：Snow depth is one of the most important physical properties of snow, and the accurate estimation of snow depth is critical to human production and life, such as water resource management, climate change research, disaster early warning, and management. At present, snow depth data observed by meteorological stations and retrieved by passive microwave remote sensing (SMMR, SSM/I, and SSMI/S) have been widely used for long time series snow depth research. To clarify the advantages and disadvantages of these data in the study of snow depth change, this paper compares the spatial distribution and interannual variation of the maximum snow depth and mean snow depth of these in China. Results show the distribution of the two types of snow depth is consistent in the stable snow cover areas, but the maximum snow depth observed by metrological stations is remarkably greater than the maximum retrieved by remote sensing in the deep snow area of more than 40 cm and the snow depth of less than 5 cm in southern China. The correlation between the two snow depth data is the best in northeast China, the second in Xinjiang, and the worst in the Qinghai-Tibet Plateau. Using the passive microwave snow depth data after 1988 is more suitable to study the snow depth changes in China because the SMMR sensor (1978.10.26—1987.8.20) has low time resolution and serious data loss in the middle and low latitudes, resulting in poor quality of the corresponding snow depth data. Furthermore, comparing the changes of the two types of snow depth in China in recent 30 years, the results show the changes of those are the same in different regions, with a substantial increase in the northeast Plain and a considerable decrease in the western and southeastern parts of the Qinghai–Tibet Plateau. Meteorological stations, influenced by their site selection, cannot reflect the high-altitude snow depth and mountain district time distribution and the change of situation. However, the snow depth retrieved by passive microwave remote sensing is affected by snow thickness, seasonal variation of snow density, liquid-water content of snowpacks, snow grain size, and other factors. Therefore, it cannot reflect the extreme snowfall events with rapid changes of snow attribute in a short time.关键词：snow depth;meteorological stations;passive microwave remote sensing;typical snow areas;comparative analysis- 740
- |
- 306
- |
- 0
发布时间：2023-10-07 -
摘要：The backscatter coefficient of Synthetic Aperture Radar (SAR) images is highly related to its incidence angle and surface characteristics. For analysis based on the backscatter coefficient, the influence of incidence angle on the backscattered signal must be corrected when analyzing wide-swath SAR images, for example, monitoring ice sheets and glaciers. The cosine square correction method is commonly used for such purpose, which assumes the snow and ice surface are Lambertian when normalizing the SAR images backscatter coefficient scattered by the ice and snow surface to a reference incidence angle. However, presuming scattering radar signal equally to all directions lying in the half space adjacent to the surface is unreasonable for the Greenland Ice Sheet (GIS) because the dry-snow zone is transparent to the C-band signal, volume scattering dominates percolation zone, and specular scattering dominates the wet-snow zone and the bare-ice zone.In this paper, a backscatter coefficient normalization algorithm is proposed based on linear regression to backscatter coefficient differences and incidence angle differences of two quasi-simultaneous observations, usually one obtained in ascending tracks and another in descending tracks. These two Sentinel-1 images share the same backscatter characteristics on the GIS, and only incidence angle differences induce backscatter coefficient differences. Considering the backscatter characteristics of the GIS surface vary with seasons and altitudes, which leads to variations of the regression coefficients, these two factors are introduced to evaluate the different regression coefficients. Then, the backscatter coefficient of Sentinel-1 dual-polarization SAR images can be normalized to a reference angle according to the regression coefficient at the given altitude and season. In the model training part in this paper, the regression coefficients are derived with Sentinel-1 images obtained in northwest Greenland, where the overlapping area between ascending and descending acquisitions is large enough to cover different glacier zones. In the testing part, our proposed backscatter coefficient correction method with the derived regression coefficients is applied to the Sentinel-1 images in IW and EW modes observing most areas of the GIS, and the backscatter coefficients at the overlapping area are compared.Results show the proposed method performs better than the cosine-square method for correcting the co-polarization images and similarly for correcting the cross-polarization images. For IW mode imagery, RMSEs are lower than 0.7, 1.0, 2.0, and 1.0 dB for Jan, Apr, Jul, and Oct, respectively. For EW mode imagery, RMSEs are lower than 1.4, 1.9, 2.9, and 2.9 dB for Jan, Apr, Jul, and Oct, respectively. Our proposed method shows lower RMSE for cross-polarization SAR images than co-polarization SAR images. Our method is performed in the same data source of NSIDC-0723, Greenland Image Mosaic from Sentinel-1A and 1B v3, and yields SAR imagery mosaics without sharp changes of backscatter coefficient among adjacent orbits.The proposed backscatter coefficient normalization method can benefit correcting the backscatter coefficient of wide-swath Sentinel-1 SAR images for the GIS and reduce the uncertainty of the subsequent applications including SAR image mosaicking and surface freeze–thaw monitoring.关键词：remote sensing;backscatter coefficient normalization;incidence angle;Sentinel-1;Greenland ice sheet;SAR;cryosphere;Glacier;image mosaicking- 2790
- |
- 207
- |
- 0
发布时间：2023-10-07 -
摘要：As a sensitive indicator of climate change, the glacial mass balance is of great relevance to regional water resource management, glacier disaster prevention and control, and global sea-level change prediction. With the intensification of global warming, the melting of glaciers in the western Qilian Mountains has accelerated since 2000. However, in recent years, not much is known about the interannual mass balance changes in this area, especially in Laohugou Number 12 glacier.In this paper, worldview optical stereo mapping, SRTM, and TanDEM-X bistatic InSAR are used to generate multisource DEM data, and the DEM difference method is used to obtain the interannual ice thickness change rate of the western Qilian Mountains from 2013 to 2014 and 2014 to 2015, and the average ice thickness change rate from 2000 to 2015. Results of the glacier mass balance for the corresponding period are obtained. On this basis, taking Laohugou Glacier Number 12 as an example, the glacier mass balance change rate during the three periods of 2013—2014, 2014—2015, and 2000—2015 is estimated, and the impact of precipitation and temperature changes on the mass balance changes are analyzed.The results show the ice thickness change rates of the western Qilian Mountains from 2013 to 2014 and 2014 to 2015 were -0.35 ± 0.034 m and -0.028±0.004 m, respectively, and the mass balance change rates were -0.27 ± 0.014 m w.e./year and -0.024 ± 0.084 m w.e./year, respectively. The average mass balance of Laohugou Number 12 Glacier from 2000 to 2015 was -0.013 ± 0.02m w.e./year, and the glacier was in a state of melting. The glacier loss rate slowed down from -0.33 ± 0.04 m w.e./year in 2013—2014 to -0.036 ± 0.09 m w.e./year in 2014—2015, which was mainly related to the increase in precipitation in 2015.This paper verifies the feasibility of high-quality optical stereo mapping satellite DEM data in solving the interannual mass balance problem of mountain glaciers.关键词：remote sensing;WorldView DEM;TanDEM-X DEM;Laohugou No. 12 Glacier;Mass balance;InSAR;Qilian Mountains- 591
- |
- 274
- |
- 0
发布时间：2023-10-07 -
摘要：Glacier identification is important for monitoring water resources and climate change in surrounding areas. Although optical images have achieved high accuracy in glacier boundary identification, optical images are affected by cloud cover, and reproducing information under the clouds is difficult. Fully polarized SAR images contain rich features, and deep learning can fully exploit image information. Therefore, using fully polarized SAR images combined with deep learning can compensate for the lack of optical images and obtain accurate glacier recognition results. In this paper, VGG16-unet (VGG16 combined with U-net) is used to identify glaciers based on ALOS2-PALSAR fully polarized images of the western part of the Himalayas. The features include the diagonal elements of the polarization coherence matrix, Freeman-Durden, H/A/α, Pauli, VanZyl, and Yamaguchi polarization decomposition parameters totaling 19 features. To make full use of the image information, these features are analyzed and combined, and the glacier recognition accuracies are compared to select the best features. Given evident differences between glacier and nonglacier topography, elevation, slope, and local incidence angle are combined with polarization features as auxiliary features.Comparing the classification accuracy of different polarization features reveals the accuracy of Pauli, Freeman-Durden, VanZyl, and Yamaguchi features based on physical characteristics is higher, among which Pauli features have the highest accuracy with an Overall Accuracy (OA) of 92.54% and an average user intersection ratio (mIoU) of 78.78%. The OA is improved to 94.34%, and the mIoU is improved to 82.35% after adding the topographic data. In order to improve the recognition accuracy of glaciers further, a feature cross-combination approach is proposed, and results show the OA of the combination reaches 94.98%, and the mIoU reaches 85.67%, which are 0.64% and 3.32% higher than the classification accuracy of Pauli features, respectively.Selecting the best feature combination method and combining with deep learning plays an important role in improving the accuracy of glacier recognition, and the use of neural networks combined with fully polarized SAR images can effectively compensate for the shortcomings of optical images in glacier boundary identification.关键词：remote sensing;glaciers;ALOS2-PALSAR;polarimetric decomposition;image segmentation;deep learning;Himalayas- 1052
- |
- 442
- |
- 0
发布时间：2023-10-07 -
摘要：With global warming, glacier surface depressions are frequently occurring in Antarctica, which is challenging for stability evaluation of glaciers. However, the shortage of high spatial and temporal resolution digital elevation model, researches on single glacier surface variations remain rather limited. 11-period REMA during 2011-2016 were used to measure surface sudden depression on Dalk Glaicer, East Antarctica. Landsat 7, 8 Wordlview-2 satellite images and so on were utilized to analyze the process and reason of the glacier surface evolution. The results show that: a serious surface depression occurred on Dalk Glacier in 2013 with maximum depression depth of 45.29 m and caused the englacially stored meltwater loss of 26.29×106 m3; then the elevation were increasing until that the elevation achieved the pre-depression level in 2016.The depression was typical of uniform settlement and intense surface melting. And the depression was closely interconnected with active ice-covered lakes drainage inside Dalk Glacier. Thus, Dalk glacier has been in a dangerous and unstable position due to melting and depressions. In addition, REMA was verified to monitor glacier surface depression, which was valuable for refine monitoring ice shelf and glacier response to climate change.关键词：the Antarctic;Dalk Glacier;surface depression;climate change- 701
- |
- 198
- |
- 0
发布时间：2023-10-07
Remote Sensing of Cryosphere
-
摘要：Multiangle remote sensing can provide richer, multidirectional features for ground object observation, improve the distinguishability between land types, and lay a solid data foundation for the accurate identification of ground cover. GF-7 is the first domestic sub meter surveying and mapping satellite after ZY-3 satellite, which brings an opportunity to solve the problem of “foreign matter homospectrum” using multiangle characteristics and to improve the identification accuracy of crops. In this paper, GF-7 forward-looking and backward-looking panchromatic and backward-looking multispectral data are used, and various features combinations are input to the support vector machine classifier to analyze the influence of multiangle features on crop recognition accuracy relative to the spectral and texture features. Results show that compared with only spectral features, with the addition of the angle difference feature, the production accuracy of garlic and winter wheat increased by 4.07% and 3.15%, respectively, and the user accuracy increased by 6.73% and 2.12%, respectively. Compared with the combination of spectral and texture features, with the addition of the angle difference feature, the production accuracy of garlic and winter wheat increased by 3.14% and 1.01%, respectively, and the user accuracy increased by 5.11% and 0.67%, respectively. Through the analysis of McNemar test, the improvement of classification accuracy is stable, angle difference feature can effectively improve the identification accuracy of crops. Tracing it to its cause, the multiangle characteristics of GF-7 satellite have unique differences in the spectral response of different crop types during multiangle observation. The difference improves the separability between crops to ensure the accuracy of crop remote sensing mapping.关键词：GF-7;SVM;angle difference;remote sensing;winter wheat;garlic;agriculture- 833
- |
- 311
- |
- 0
发布时间：2023-10-07 -
摘要：China’s temperate sparse forest grassland is a transition ecosystem between forest and grassland and is the top-level ecosystem that evolves under the unique climate and topography of northern China. Sparse forest grassland is characterized by mixed woody and herbaceous vegetation, which are difficult to directly distinguish by remote sensing even when using high-spatial-resolution satellite data. Consequently, mapping Fractional Woody and Herbaceous Vegetation Cover (FWHVC) in this ecosystem is challenging. How to precisely monitor the growth status of woody and herbaceous vegetation in sparse forest grassland on a regional scale is a popular and difficult topic in vegetation remote sensing in dryland.This study proposed a novel method of FWHVC estimation in temperate sparse forest grassland based on Unmanned Aircraft Vehicle (UAV) aerial images and high-spatial-resolution satellite images (GF-6 and Sentinel-2) via a machine learning algorithm. Training datasets of FWHVC were derived from a very-high-resolution aerial image (2.41 cm/pixel). Meanwhile, this study compared the results of FWHVC estimated from GF-6 and Sentinel-2 images.The results were as follows (1) The UAV platform could precisely capture the land cover type and provide a large number of reliable training datasets of FWHVC. (2) FWHVC could be estimated well on a regional scale based on both GF-6 and Sentinel-2 high-resolution satellite data through the machine learning algorithm. The FWHVCs derived from GF-6 images and UAV aerial images had determination coefficients R2 of 0.72 and 0.66, root-mean-square errors (RMSEs) of 6.76% and 10.69%, and estimated accuracies (EAs) of 46.31% and 77.88%, respectively. The FWHVCs derived from Sentinel-2 images and UAV aerial images had determination coefficients R2 of 0.72 and 0.81, RMSEs of 6.53% and 8.20%, and EAs of 54.30% and 83.17%, respectively. (3) The EA of the FWHVC estimated from Sentinel-2 was slightly better than that of GF-6. Meanwhile, the EA of fractional herbaceous vegetation cover estimation was higher than that of woody vegetation cover estimation for both satellite images.This paper provides a new way to estimate FWHVC in temperate sparse forest grassland on a regional scale by using multisource remote sensing data and a machine learning algorithm. The multiscale approach could provide new methodological support to accurately monitor woody and herbaceous vegetation cover in temperate sparse forest grassland. In the future, FWHVC in sparse forest grassland can be monitored dynamically by utilizing long-term and high-spatial-resolution satellite remote sensing data on a regional scale.关键词：remote sensing;Elm sparse forest grassland;Unmanned Aerial Vehicle (UAV);GF-6;Sentinel-2;Random Forest;Classification and Regression Tree(CART)- 2641
- |
- 567
- |
- 0
发布时间：2023-10-07 -
摘要：Searching for an efficient, high-precision method for mapping paddy rice planting distribution in Northeast China has important implications for accurate paddy rice yield estimation and agricultural policy making.In this paper, paddy rice planting distribution was mapped by feature optimization random forest method in Panjin City, Liaoning Province. Based on the land coverage types, 2000 samples of 1000 paddy rice samples, 250 water samples, 300 wetland samples, 150 dry land samples, and 300 construction land samples were acquired. Training samples and testing samples accounted for 70% and 30%, respectively. In addition, 36 paddy rice field validation points were obtained through field surveys. The spectrum features, vegetation indexes, water index, and red edge indexes were constructed by using the GF-6 WFV images taken in the periods of May 11, May 25, June 1, June 6, July 20, and August 22 in 2020, and these images corresponded to the trefoil stage, transplanting stage, returning green stage, booting stage, and heading stage according to the phenological phase of paddy rice in Panjin City, respectively. The returning greening stage image was covered by June 1 and June 6. The feature importances of single temporal images and time series images were calculated, and out-of-bag (OOB) estimations on different feature combination models were performed based on OOB data. The optimal input features were selected after comprehensively considering the accuracy and complexity. Then, the feature optimization random forest model was established to extract the paddy rice planting area and spatial distribution information in Panjin City in 2020.According to the testing samples and the paddy rice field validation points, the accuracy evaluation of classification results showed the following: (1) Based on the single temporal images with different phenological phases, all the classification accuracies were 94% and above. The classification result of the image in the paddy rice transplanting stage was the best that the overall accuracy, F1 score (paddy rice), Kappa coefficient, and field validation point accuracy were 97.67%, 98.84%, 0.97, and 97.22%, respectively. (2) On the basis of comparison with the classification results of single temporal images, using time-series images for land coverage classification and paddy rice information extraction effectively improved the classification accuracy and reduced misclassification and omission, and the paddy rice classification map polygons were more regular. The overall accuracy, F1 score (paddy rice), Kappa coefficient, and field validation points accuracy with time series images were 99.33%, 100%, 0.99, and 97.22%, respectively. (3) Through analyzing of the paddy rice extraction results with or without red edge bands and red edge indexes, the classification accuracy was improved by the introduction of red edge information. This paper proved that based on the feature optimization random forest model, the paddy rice information was accurately extracted by using the single temporal image of paddy rice transplanting stage. Compared with single temporal image, using time-series images improved the classification accuracy. Considering the complexity and running speed of the model, the single temporal image of paddy rice transplanting stage was used to extract paddy rice planting area to meet the accuracy requirement in practical applications. (4) Through analyzing the results of paddy rice extraction without purple band and the yellow band, this paper proved the introduction of purple and yellow bands can improve the classification accuracy, but the effect of improving the accuracy of the classification result was inferior to the red edge information.Improving the classification accuracy of paddy rice and enhancing crop recognition capabilities by red edge information, purple band, and yellow band, showed the GF-6 satellite had broad application prospects in crop precise identification and area extraction.关键词：remote sensing;Random Forest;red edge band;feature optimization;GF-6;paddy rice;purple band;yellow band- 857
- |
- 318
- |
- 0
发布时间：2023-10-07
Forestry and Agriculture
-
摘要：Remote sensing monitoring of foliar dust is one of the important methods to assess mine dust pollution. Compared with natural dust, mine dust, enriched with heavy metals, poses a more serious threat to human health and vegetation growth. Recently, most of foliar dust monitoring carried out inversion and monitoring for the amount of foliar dust, without distinguishing between mine dust and natural dust in mining areas. In this paper, methods were proposed to extract foliar dust information and to identify dust source, taking the Jiawula–Chagan Pb-Zn-Ag mining area in Inner Mongolia as an example.The Feature-oriented Principal Component Selection (FPCS) method was applied to extract the distribution and intensity of foliar dust based on the analysis of the spectral response characteristics of foliar dust fall for Sentinel-2 data. Moreover, a Dust-source Spectrum Index (DSI) was proposed based on the reflectance differences between mine foliar dust and natural foliar dust, and DSI was used to distinguish the extracted foliar dust information between mine dust and natural dust. Then, the correlation between the type of dust source, the intensity of foliar dust, and the distribution of mine objects was analyzed, and the main mine dust sources and their dust diffusion characteristics were assessed. Finally, minimum noise fraction and pixel purity index methods were applied to analyze spectral characteristics difference and mineral composition difference between natural dust and mine dust.Results showed foliar dust increased the vegetation reflectance in visible regions (B1—B4), decreased the vegetation reflectance in near infrared regions (B7—B8A), and shifted the red edge of the vegetation to shorter wavelength direction. The spectral reflectance gradually decreased with the increase of distances away from the mine dust sources, while the spectral reflectance gradually decreased in visible regions and increased in near infrared regions with the increase of distances away from dry lake dust sources. Mine dust sources and the affecting vegetation showed reflectance absorption characteristics near 864.7 nm (B8A), attributed to pyrite oxidation. The distribution and intensity of foliar dust were successfully extracted by FPCS and compared with the reported experience model. DSI could distinguish mine dust fall from natural dust fall. The extracted mine foliar dust pixels had a strong spatial correlation with the mine objects. The main mine dust sources in the study area were mine dumps and mine roads, the dust diffusion intensity of mine dumps was greater, and the dust diffusion of mine dumps spread farther.The sensitivity of Sentinel-2 reflectance to identify the intensity of foliar dust fall and to distinguish the type of dust source was verified. FPCS could be applied to extract the distribution and intensity of foliar dust, without ground sampling. The extracted mine foliar dust pixels were classified into mine dust fall and natural dust fall using the proposed DSI and achieved excellent results. This paper provided a method for the rapid assessment of dust pollution in mine areas.关键词：remote sensing;mining area;foliar dust;dust source;Sentinel-2;feature-oriented principle component selection (FPCS);dust source index (DSI)- 886
- |
- 197
- |
- 0
发布时间：2023-10-07 -
摘要：Limited by the weather during floods, the remote sensing data used for flood assessments are mostly radar images or aerial data, and the role of numerous night lights and optical image data in flood assessments needs to be further explored. This paper took Fuyang from July to August as the research area, based on the monitoring data of Sentinel-1, Sentinel-2, and Landsat 8 and extracted water body information with the help of Google Earth Engine. This paper used night light data (NPP-VIIRS DNB) to establish the total night-time light (TNL) and compounded night light (CNLI) to explore the relationship between water changes and night lights to monitor and evaluate the effect of floods. Results showed the following: (1) The distribution of water bodies in the southern part of Fuyang changed remarkably from July to August, especially the water bodies in the Mengwa Flood Diversion Project increased substantially. On July 31, the water body area reached the maximum of 323 km2, which was six times larger than the water body area before the flood, and then the coverage of water bodies was declining. This trend corresponded to the time of flood storage and discharge of Wangjiaba gate. (2) The combined analysis of Fuyang night light index TNL index and CNLI index found the change trend of the light index was opposite to that of the water body, indicating the night light index can effectively reflect the changing of flood disasters. (3) Analyzing the water body and night light index of eastern Fuyang with relatively complete data further showed night light and water body data can be used to monitor floods. This paper expanded the application range of night light data and optical images and confirmed that after rigorous data processing, multisource remote sensing data such as radar image based on Sentinel-1, optical image based on Sentinel-2, and Landsat 8 can effectively monitor the change of flood disaster and play an important role in flood monitoring in the future.关键词：Google Earth Engine (GEE);night lights;multi-source remote sensing;flood disaster;Sentinel;NPP-VIIRS DNB;Landsat- 1499
- |
- 703
- |
- 0
发布时间：2023-10-07
GeologyandDisasters
-
摘要：Hyperspectral imaging technology has the unique potential for the low-cost, large-scale, and rapid monitoring of soil heavy metals. For hyperspectral images, the number of soil image elements differs greatly from the number of soil samples, so the problem of small samples is prominent. In this paper, a soil heavy metal estimation method based on Fractional-Order Derivative (FOD) for hyperspectral images is proposed.First, the neighboring pixels of soil samples were extracted to expand the samples and increase the spectral variability. Second, FOD was used to highlight the spectral features. Then, the bands were selected by Competitive Adaptive Reweighted Sampling (CARS), and partial least squares (PLSR) was used to construct the model. Seventy-two soil samples and aerial hyperspectral images obtained from the Huangshan South mine in Hami, Xinjiang were used to estimate three heavy metals, namely, lead (Pb), zinc (Zn), and nickel (Ni).After sample expansion, the estimation accuracy of the test set was improved for three heavy metals, the test set R2 improved from 0.6128 to 0.7974 for Pb, from 0.8178 to 0.8690 for Zn, and from 0.6969 to 0.8303 for Ni, while the R2 of the training set was above 0.8. The accuracy of estimation model for three heavy metals with the best fractional-order differentiation was better than that using integer-order differentiation. CARS+PLSR obtained higher estimation accuracy than the modeling approaches of GA+PLSR and CC+PLSR. The estimation accuracies R2 were 0.7974, 0.8690, and 0.8303 for Pb, Zn, and Ni, respectively.Sample expansion alleviated the overfitting phenomenon and improved the estimation accuracy. The FOD of the optimal order could effectively enhance the spectral features and improve the estimation accuracy. CARS was more accurate than CC and GA.关键词：fractional order derivative;hyperspectral remote sensing images;Competitive Adaptive Reweighted Sampling (CARS);soil heavy metal;small sample size;visible and near-infrared band;short-ware infrared band- 685
- |
- 305
- |
- 0
发布时间：2023-10-07 -
摘要：Land surface reflectance, as a key physical parameter describing the basic properties of the land surface, is one of the key parameters for quantitative remote sensing research and applications, such as remote sensing indices computation, leaf area index retrieval, dynamic forest cover monitoring. The main representative land surface reflectance products at home and abroad are MODIS surface reflectance products, Landsat surface reflectance products and Sentinel surface reflectance products. The WFV (Wide Field of View) sensor onboard the Gaofen-1 satellite (GF-1) has high spatial resolution and wide imaging capability. GF-1 WFV data have become an important data source in the fields of resource investigation and ecological environment. At present, GF-1 WFV data have been fully opened to the public, however the GF-1 WFV data are only provided in L1-level product and lack land surface reflectance product. The aim of this paper is to produce GF-1 WFV land surface reflectance product.This paper proposes a coupled atmospheric correction algorithm that integrates the universal kriging method of MODIS AOD spatial fusion, 6S model, and dynamic look-up table and produced an open-access GF-1 WFV land surface reflectance product of 2020 in China’s land area.In terms of product accuracy validation, first, visual inspection, histogram, and spectral curve changes of typical land cover types before and after atmospheric correction were compared. Second, cross-validation was performed between GF-1 WFV and Landsat-8 OLI land surface reflectance products. Third, ground-based measurements were used for validation. The validation results show the image quality after atmospheric correction is remarkably improved and agrees with the Landsat 8 OLI land surface reflectance product. The root mean square error with the Landsat 8 OLI land surface reflectance products do not exceed 3%, and the- 1100
- |
- 370
- |
- 0
发布时间：2023-10-07 -
摘要：Airborne LiDAR bathymetry (ALB) seabed sediment classification can provide basic data for the development and utilization of marine resources, marine environmental protection, marine engineering construction, and other fields, which has great relevance to marine activities and marine scientific research. To solve the feature redundancy problem in ALB seabed sediment classification, this paper proposes a sediment classification algorithm considering optimal waveform and topographic features. Based on the extracted waveform and topographic features, the Relief-F feature optimization model is constructed, and multivariate features are optimized by calculating the contribution rate of each feature in the sediment classification. Then, random forest, support vector machine, and BP neural network classifiers are used to classify coral reefs, gravel, sand, vegetation, and coastal zones five types of sediments. The proposed method is verified using the ALB data captured around Ganquan Island in the Xisha Archipelago. The experiment results showed that after using the Relief-F algorithm for feature optimization, the classification accuracies of RF, SVM, and BP neural network improve by 1.1%, 1.1%, and 2.7%, respectively. The random forest sediment classification has higher classification accuracy, and the overall accuracy and Kappa coefficient reach 95.36% and 0.94, respectively. The research results can provide effective technical support for the seabed sediment classification in the fields of marine engineering and other fields.关键词：airborne LiDAR bathymetry;sediment classification;waveform features;topographic features;Relief-F feature optimization model;image processing;ocean- 544
- |
- 183
- |
- 0
发布时间：2023-10-07
Models and Methods
- Postal code：100190
- Tel：010-58887052 Email：nrsb@aircas.ac.cn
- Technical support is provided by Beijing Founder electronics co., LTD 京ICP备20021838号-9
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0