最新刊期
卷 27 ， 期 11 ， 2023
-
摘要：Remote Sensing (RS) has become an essential information source for national resource and energy surveys, food security monitoring, ecological environmental protection, natural disaster assessment, and national defense security. Bibliometric is a helpful method to analyze the development dynamics, hotspots, and evolution in the RS discipline. It is a powerful approach to sort out and visualize the progress of RS development.This study conducts a bibliometric analysis of RS-related Science Citation Index (SCI) papers published from 1962 to 2021. The research hotspots and changes of RS in the United States, Europe, and China from 1962 to 2021 are systematically determined. The application of typical RS satellites globally and in China are compared and analyzed. The research characteristics of US, European, and Chinese scholars in the three frontier technologies (i.e., Synthetic Aperture Radar, Hyperspectral, and LiDAR) are summarized.Results show that (1) the number of SCI papers and authors in RS has shown a trend of rapid growth and accelerated growth since 1998, from 69,666 published in 2012 to 169,797 in 2021. China has surpassed the US in its annual publication to become the first since 2014, and it has been far ahead since then. It published 8,063 RS SCI papers by 2021, which accounted for 42.17% of the 19,121 global publications. (2) In terms of RS technology, RS started from multispectral imaging, and it developed rapidly to the frontier technologies of synthetic aperture radar, hyperspectral, LiDAR, unmanned aerial vehicle (UAV), high-resolution image, and deep learning. Furthermore, RS gradually played an increasingly important role in many application fields. (3) In terms of RS data application, Landsat, MODIS, Sentinel, and other foreign data have been widely used by global users. Chinese scholars highly rely on these foreign satellite data to conduct RS research. By contrast, the application of domestic satellites is relatively rare, and the international influence of domestic satellites is very weak, which is very mismatched with China’s status of RS. (4) Significant differences are observed in the hotspots of RS research between China and other countries. By relying on advanced satellite and payload technologies (e.g., Landsat and MODIS), the US developed science- and demand-driven RS research, which has been widely used in various application fields. European RS scholars attached great importance to the research and application of Sentinel satellites, which have surpassed Landsat in the number of SCI papers they published. Chinese RS has shown outstanding quantitative advantages in all research fields and applications. Meanwhile, Chinese RS scholars pay more attention to synthetic aperture radar, hyperspectral, LiDAR, deep learning, neural networks, feature extraction, and other cutting-edge technologies and algorithms.关键词：remote sensing;bibliometrics;discipline development trends;research hotspot;satellite data;application fields- 1619
- |
- 572
- |
- 0
发布时间：2023-12-08 -
摘要：Plant diversity is closely related to ecosystem productivity, stability, and resource use efficiency. The rate of plant biodiversity loss due to human activities, extreme climate, and species invasions is accelerating annually, and an urgent need is recognized for rapid and accurate collection of information of terrestrial plant diversity for biodiversity conservation.Remote sensing techniques are important methods of earth observation from space. In recent years, image data from remote sensing have been developing toward refinement and comprehensiveness, and high-quality data covering more ground information have been gradually applied. The emergence of hyperspectral remote sensing technology enables sensors to collect continuous spectral curves of ground targets in fine spectral resolution, which consequently provides massive information of ground objects and realizes the quantitative inversion of ground object parameters. Hyperspectral remote sensing technology offers a technical basis for the large-scale observation of plant diversity and functional traits. It further brings opportunities for the verification of theories of community assembly with the continuous variation in spatial scales.In this study, we review the development of hyperspectral remote sensing technology and its application in detecting plant diversity and functional traits. Two types of inversion approaches for quantifying biodiversity through hyperspectral remote sensing, namely, direct inversion and indirect inversion, are introduced. The direct inversion approach takes the spectral variation hypothesis (SVH) as its theoretical basis, which posits that biodiversity is linked to the heterogeneity of spectral image. The SVH-based approaches, also known as “spectral diversity metrics,” are to directly establish the relationship between spectral information and plant diversity. Common spectral diversity metrics include the coefficient of variation of spectral bands, the convex hull volume in spectral space, the spectral angle mapper, the divergence of spectral information, and the convex hull area. Numerous studies have proven that these spectral diversity metrics can be used to effectively track the variation in biodiversity indicators, such as species richness, Shannon index, and Rao’s Q index.The indirect inversion approach correlates spectral information with plant diversity via quantitative remote sensing. Plant functional traits can be retrieved from hyperspectral image data through empirical and physically-based models with convincing accuracy. With the quantitative retrieved traits from image data, functional diversity indices, which can be closely linked to ecosystem functioning, such as FRic (functional richness), FDiv (functional divergence), and FEve (functional evenness), can be characterized and spatially mapped. Studies also confirmed that the indirect approach can be employed to assess taxonomic and even phylogenetic diversity through the quantification of vegetation indices.Combined with existing application examples, we then discuss the technical advantages of hyperspectral remote sensing technology in the studies on species invasion, species mapping, biodiversity spatial patterns, and the large-scale biodiversity and ecosystem functioning relationship. At the end of this review, limitations of the application of hyperspectral remote sensing technology in ecological studies are analyzed. With the development of multisource remote sensing technology, hyperspectral remote sensing coordinated with other technological means (e.g., ground flux monitoring, laser radar technique, and computer visualization) will be applied more extensively in biodiversity-relevant studies.关键词：hyperspectral remote sensing;biodiversity;spectral diversity;plant functional traits;biodiversity and ecosystem functioning relationship- 4013
- |
- 999
- |
- 0
发布时间：2023-12-08
Research Progress
-
摘要：Large-scale green tides occurring in the Yellow Sea (YS) of China have become a critical eco-environmental problem, causing serious damage to marine and the coastal ecological environment, aquaculture, and tourism since 2007. Green tide biomass is a key parameter for accurate quantification of floating macroalgae, serving as an effective indicator for monitoring changes in the marine ecological environment. Satellite remote sensing technology plays a pivotal role in supporting the monitoring and assessment of green tide. Spaceborne optical sensors, in particular, offer a wealth of data that is indispensable for the fine-scale quantitative monitoring and assessment of green tide. In this study, we have established robust statistical relationships between Biomass Per Area (BPA) and various optical remote sensing indices by modeling the laboratory measurements of U. prolifera biomass (wet weight) per unit area and the corresponding spectral reflectance data. The computational methods of BPA have been carefully designed and validated for different optical data, including Moderate Resolution Imaging Spectroradiometer (MODIS), the Multispectral Instrument (MSI) onboard Sentinel-2 satellites, and the Coastal Zone Imager (CZI) onboard China’s HaiYang-1C/D (HY-1C/D) satellites. These results indicate that BPA can serve as a highly effective parameter in quantifying green tide using remote sensing data. Unlike common parameters such as pixel area or coverage area, BPA can mitigate the scale effects of spatial resolution differences from various observations, minimizing the uncertainty especially when integrating multiple remote sensing data. With the coordinated utilization of CZI and MODIS data in 2021 and the developed BPA models, the detailed intra-annual variations in green tide biomass in the YS of China were quantified. This analysis has revealed the intricate spatial distribution patterns and trends inherent in green tide biomass fluctuations. The utilization of multiple optical remote sensing data sources for the estimation of green tide biomass carries important methodological significance and serves as an accurate data reference for the precise, quantitative, and dynamic monitoring of green tide in the YS of China.关键词：Green tide biomass;optical remote sensing;HY-1C/D;CZI;MODIS- 876
- |
- 511
- |
- 0
发布时间：2023-12-08 -
摘要：We use the point-to-point method and Beitsch’s co-location comparison method to conduct a series of evaluations on the passive microwave remote sensing products (PM) for observing sea ice concentration (SIC) in the Prydz Bay, Antarctica by using two kinds of ship-based observation datasets. Considering the difference in ship-based observation data, we divide the comparison into two parts. First, according to the ship-based observation data of China’s 29th, 31st, and 37th Antarctic scientific expedition in the period of 2012—2021, eight remote sensing SIC products are classified and quantitatively compared according to the size of SIC. Results show that NSIDC/NT2 product assesses the highest correlation and the best stability in all cases. In the co-location comparison, the correlation coefficient can reach 0.926, the Root Mean Square Error (RMSE) is 12%, and the average bias is only 2%. Second, to make up for the lack of historical data of AMSR2 sensor series products, we evaluate the seasonal cycle and long-term variation signals of four remote sensing data products by using the ASPeCt ship-based observation dataset from 1992 to 2000 in the same way. The inversion accuracy of this period is lower than the case-by-case comparison result from 2012 to 2021, and a tremendous seasonal difference is observed. The bias of the four products increases from the melting period to the freezing period. During this period, the overall inversion results of CDR and bootstrap algorithms based on SSM/I sensors are better, with correlation coefficients of more than 0.8, RMSE of 16%, and bias of approximately 8%. However, a large bias remains in the low SIC region. This study shows that the accuracy of PM SIC products in a small sea area is insufficient, and it fluctuates greatly with the difference in SIC type, season, and algorithm. Therefore, the necessary considerations are to modify the resolution, use multisource data as much as possible, and classify data according to the ice conditions. Referring to Beitsch’s idea of Antarctica partitioning and comparison, we further obtain the accuracy of remote sensing products under different ice conditions in a local region. We add China’s scientific research ship-based observation data to increase the sample numbers for investigating the Prydz Bay area, which covers rich surface ice types. The regional comparison provides a reference for understanding the limitations of PM SIC products in micro-area inversion and also guarantees ice prediction and navigation safety. Considering the rapid reduction in Antarctic sea ice in recent years and the appearance of a 40-year minimum Antarctic sea ice range in February 2022, high-precision real-time PM SIC products need to be developed to determine the causes of sea ice anomalies and simulate sea ice changes in the future. Knowing the inversion accuracy of various PM SIC products under different conditions will help improve the subsequent PM SIC products and fusion algorithms. In the future, more factors that affect the accuracy of PM inversions, such as ice thickness, ice type, and other factors, should be considered to evaluate PM SIC products in other regions of the Antarctic in detail.关键词：Prydz Bay;sea ice concentration;passive microwave remote sensing;ship-based observation;data quality assessment- 290
- |
- 172
- |
- 0
发布时间：2023-12-08 -
摘要：Dongting Lake is the second largest freshwater lake in China, and the water body fluctuates greatly and frequently in dry and wet seasons. The high-frequency observation of water body changes in Dongting Lake by satellite is important for timely and accurate monitoring of its hydrological dynamic changes.This study aims to reconstruct and analyze the elaborate time series information of Dongting Lake water area from 2017 to 2020 by using Sentinel-1 and Sentinel-2 images (10 m) under the condition of high time-frequency and high spatial resolution. In addition, this study develops a method to reconstruct the whole lake water area and encrypt the time series of lake area based on the partially cloudless image of the lake area. This method uses the statistical fitting relationship between the block area and total area extracted from Sentinel-1 full image to establish an empirical model for obtaining a dense time series of Dongting Lake water area.Fitting results show that the area of each block is significantly positively correlated with the lake total area, and the average R2 value is 0.94. From 2017 to 2020, 119 Sentinel-1 images and 38 Sentinel-2 images were obtained to extract the block inundation range of Dongting Lake for reconstructing the whole lake area. After all Sentinel-1 and Sentinel-2 images were combined, the average observation images available in each month are 6. In some months, the monitoring times can reach 10 times, and the time interval is 3—6 days. It can conduct fine monitoring of the water area change of Dongting Lake. Besides, the time series of water area reconstructed in this study can accurately describe the significant seasonal fluctuations and interannual changes. The water area reaches the peak in July and the valley from November to February. The average ratio of maximum area to minimum area in a month is 1.36. The most violent fluctuation in surface waters occurred in November, with a ratio of 1.52. After Sentinel-1/2 image observation time series were integrated, the average water area of Dongting Lake from 2017 to 2020 is about 1147.13 km2.Compared with the water area time series of Dongting Lake constructed based on Sentinel-1/2 images, the water area time series combined with landsat-8 image can improve the time resolution of water observation. However, it has little effect on the monthly and annual averages of water area. This study combines sentinel series radar and optical images of high temporal and spatial resolution to develop a fine extraction method of high time-frequency water area series. In summary, this study can provide scientific and technological support for improving remote sensing monitoring and fine management of water resources in Dongting Lake and high dynamic lakes in the middle and lower reaches of the Yangtze River.关键词：remote sensing monitoring;Dongting Lake;Sentinel;Watershed Area;time series;Yangtze River Basin- 1304
- |
- 384
- |
- 0
发布时间：2023-12-08
Atmosphere and Ocean
-
摘要：The spectral library can contain the spectral information on the whole types of ground surface objects in the observation area of hyperspectral images. Thus, the optimized dictionary learning via spectral library refers to the process of constructing optimized spectral dictionary under strict theoretical derivation, in which the spectra in the spectral library are used as training samples. The abovementioned process enables the spectra in the hyperspectral image to be sparsely represented under the learned spectral dictionary. To this end, a new spectral super-resolution method using optimized dictionary learning via spectral library is proposed in this study. This method uses only one high spatial multispectral image to reconstruct high spatial hyperspectral image. The aforementioned problem is formulated in the framework of sparse representation, as an estimation of the band matching matrix, the optimized spectral dictionary, and the corresponding sparse coefficients. Specifically, a band matching method is proposed to map the common spectral library to a specific spectral library corresponding to the reconstructed high spatial hyperspectral image. Then, an optimization of spectral dictionary and its corresponding sparse coefficients is derived theoretically using the alternating direction method of multipliers (ADMM) algorithm and by utilizing the abovementioned specific spectral library and the high spatial multispectral image. Comparison results with the relative methods demonstrate that our method not only can achieve a high-quality reconstruction of the high spatial hyperspectral image but also can significantly improve the classification accuracy of multispectral images by even only using one high spatial multispectral image.We aim to reconstruct high spatial hyperspectral image only from one high spatial multispectral image with high quality.Three steps of our proposed method are discussed in detail. First, the band matching matrix is estimated using the band wavelength information. Second, the matched spectral dictionary is optimized using the matched spectral library and the high spatial multispectral image. Third, the equivalent sparse coefficient matrix with respect to the matched spectral dictionary is derived theoretically and estimated iteratively.Extensive experiments and comparative analyses of the proposed method are conducted on various datasets to demonstrate the performance and practical application value of our proposed method. The improvement in classification accuracy on the reconstructed high spatial hyperspectral images is also evaluated using some typical classification methods.A spectral super-resolution method is proposed, and it uses only one high spatial multispectral image to reconstruct high spatial hyperspectral image. A band matching matrix, which is used to map the common spectral library to a specific spectral library, is obtained by solving the minimum distance problem. A spectral dictionary and its corresponding sparse coefficient matrix are optimized from the matched spectral library and the high spatial hyperspectral image by minimizing augmented Lagrangian function using ADMM iteratively. Experiments on simulated and real datasets demonstrate that our proposed method can produce comparable results for the spectral super-resolution to the other relative state-of-the-art reconstruction or fusion-based methods using additional low spatial hyperspectral image. It can also provide higher reconstruction quality than the HIRSL method without optimization. Our proposed SODL method that uses only one multispectral image may help develop new light and small high spatial hyperspectral imaging equipment.关键词：spectral super-resolution;spectral library;sparse representation;optimized dictionary learning;landcover classification- 2657
- |
- 260
- |
- 0
发布时间：2023-12-08 -
摘要：Surface albedo plays a vital role in land surface climate and biosphere models. Many researchers and teams use different algorithms to develop high- resolution surface albedo products for better satisfying the requirements of multi-field research and broaden the application of quantitative remote sensing parameters.The proposed algorithm aims to generate albedo products of high spatial resolution based on the sensor data of GF-1 WFV and the GLASS albedo product with 500 m resolution. The idea of the algorithm is to first invert the GF-1 WFV data with a direct inversion algorithm for obtaining the primary albedo product of 16 m resolution. Then, it downscales the GLASS albedo product of 500 m resolution with the texture information of the primary product of 16 m resolution to obtain the final albedo product of 16 m resolution.The algorithm results are verified with ground observations at eight stations in the Heihe Experimental Area using the data from 2016 to 2017. The time series graphs of the measured data and the inverted fusion data show that the fused albedo product of 16 m resolution agrees well with the measured value. At the same time, through the analysis of the scatter plots of all stations from 2016 to 2017, the root mean square error of the fusion albedo is 0.02439, and the primary albedo is 0.05135. The fusion albedo is closer to the measured value than the primary albedo. The GLASS albedo product of 500 m resolution in the photovoltaic industrial park was compared with the co-located albedo product of 16 m resolution to visually illustrate the effect of the albedo product of 16 m resolution. The albedo product of 16 m resolution could better support the studies on human activities and the environment. The algorithm quantitatively fuses the texture information of 16 m resolution in the GF-1 data with the mean value information of the GLASS albedo product of 500 m resolution to obtain the albedo product of 16 m resolution. It contains two main steps: a simple direct inversion algorithm and a downscale-to-fuse algorithm. The albedo product of 16 m resolution enriches the spatial texture information on the premise that the average value is consistent with that of the GLASS albedo product of 500 m resolution.关键词：remote sensing;albedo;algorithm;Downscaling fusion;high resolution;Verification;GF-1;Hei He- 906
- |
- 244
- |
- 0
发布时间：2023-12-08 -
摘要：Processing Landsat 5 TM through COSI-Corr software can better quantify the migration mode and speed of sand dunes. This way can also provide specific guidance and suggestions for the implementation of sand prevention and control projects. At present, few studies on the error source analysis and timing inversion of Landsat 5 TM migration field are available. This work takes the northwestern part of the Mu Us Sand Land as the study area to obtain the dune migration time series and annual average velocity field from 1991 to 2000. Results show that the main errors of the Landsat 5 TM migration field are the orbit error, the attitude angle error, and the miscorrelated noise. After the error is eliminated, the accuracy has been increased by 23%—34%, 4%—20%, and 2%—5% successively, and it can be increased by 13%—14% after inversion by the least square method. The monitoring results show that the sand dunes in the northwest of the Mu Us Sandy Land continued to advance to the southeast from 1991 to 2000, and the maximum speed could exceed 6 m/yr. However, under the action of the northwest-southeast wind, periodic back-and-forth movement occurred along the northwest-southeast direction. The study area has a velocity of 0—1 m/year accounting for 63.9%, and the velocity of more than 3 m/a is only 4.2%. The error elimination and time sequence inversion of Landsat 5 TM can effectively improve the accuracy of the migration field, which results in a more accurate and reliable time sequence. Different from previous studies, the error elimination and time sequence inversion of the dune migration field can reflect the actual wind energy environment on the dune surface in more detail.关键词：Landsat 5 TM;COSI-Corr;error analysis;dune migration;Mu Us sandy- 760
- |
- 565
- |
- 0
发布时间：2023-12-08 -
摘要：HypeSspectral Image Classification (HSIC) is a pixel-level classification problem, and it involves classifying each pixel in the hyperspectral image and confirming the pixel category. However, discriminative features in HSIC task are difficult to acquire and learn, and the extraction of sufficient and effective features directly affects the classification results. In the past few years, Convolutional Neural Networks (CNNs) have achieved better results in HSIC, but the high dimensionality of hyperspectral images and the equal processing of all bands by CNNs have limited the performance of CNN. This study proposes an end-to-end Spectral-Spatial Attention Bilateral Network (SSABN) for HSIC. The network directly uses 3D blocks of the original image as input data without the complicated preprocessing. First, the original data are processed through the spectral-spatial attention module to enhance the useful bands or pixels for classification and suppress invalid information. Then, the spatial and context paths of the bilateral network are designed. The spatial path has three layers, and each layer is composed of convolution, batch normalization, and Relu activation function to extract spatial information. The context path is composed of three downsampling and attention refining modules. The downsampling is used to provide receptive field, and the attention refinement module is used to refine downsampling features. Finally, a feature fusion module is designed to fuse different levels of features through maximum pooling and average pooling for generating discriminative features. Compared with common CNN, SSABN can adaptively enhance effective information, extract more abstract discriminative features, and consume less training time. Experimental results show that SSABN has good fitting ability in different training sample ratios. In the results of ablation experiments, the accuracy of the spectral-spatial attention mechanism is 1%—2% higher than those of other mainstream attention mechanisms, and the feature fusion module can improve the discrimination of extracted features. In the experiments of three public datasets, the classification accuracy of SSABN is higher than 99%, and the training time is less than those of other methods. The classification performance of SSABN is better than those of other hyperspectral image classification algorithms, while reducing its training time can more effectively improve accuracy and efficiency.关键词：remote sensing;convolutional neural network;deep learning;Feature fusion;Indian Pine dataset;Pavia University dataset;Salinas dataset- 1131
- |
- 447
- |
- 0
发布时间：2023-12-08 -
Remote sensing image semantic segmentation method combining cosine annealing with atrous convolution
摘要：This study aims to capture the rich context information and multiscale feature information in remote sensing images, improve the integrated model strategy, and enhance the accuracy of semantic segmentation. Thus, this study proposes a high-resolution remote sensing image semantic segmentation method using cosine annealing with increasing period and multiscale atrous convolution.The multiscale parallel atrous convolution helps the network capture context information in a larger range and improves the ability of the network to recognize multiscale objects without increasing parameters. The method in this study uses the atrous convolution while discarding the pooling operation to maintain the spatial resolution. Meanwhile, the method adopts the fully connected conditional random field to add spatial and edge context information for making up for part of the position information missed by the atrous convolution. As a result, the outline of extraction objects by semantic segmentation fits the ground truth better. Moreover, the cosine annealing strategy with increasing period is introduced to adjust the learning rate and obtain a suitable number of local optimal solutions. We integrate the local optimal solutions in the method to further improve the pixel classification ability of the network.The overall accuracy and kappa coefficient of the proposed model, which are 86.6% and 81.8%, respectively, are better than those of the current advanced semantic segmentation models.The experimental results performed on the Gaofen image dataset show that the fusion of image context information and multiscale feature information can effectively identify objects with complex structures. Moreover, the model coupled with the period-increasing cosine annealing strategy could obtain better semantic segmentation accuracy than and less inference time than that coupled with the equal-period cosine annealing strategy.关键词：high-resolution remote sensing image;semantic segmentation;cosine annealing with increasing period;multi-scale parallel atrous convolution;target extraction;in-context learning;conditional random field;multi-scale learning- 2883
- |
- 327
- |
- 0
发布时间：2023-12-08 -
摘要：Semantic segmentation of high-resolution remote sensing image has important theoretical and practical value in the field of aerial image analysis. However, the traditional segmentation methods are prone to edge blur, loss of detail information, and low resolution due to the richness of building semantics and the complexity of image background in high-resolution remote sensing images.An end-to-end convolutional neural network called Dilated-UNet (D-UNet) is proposed to solve the problem of fuzzy boundary and information loss in high-resolution satellite image semantic segmentation. First, the U-Net network structure is improved and the multiscale dilated convolution module of four channels is expanded using the division technology. Each channel uses different convolution expansion rates to identify the multiscale semantic information for extracting richer detailed information. Second, a joint loss function of cross entropy and Dice coefficient is designed to achieve the desired segmentation effect.The model is comprehensively evaluated and tested on the Inria aerial image dataset. Experimental results show that the proposed remote sensing image segmentation method can effectively segment urban buildings at pixel level from high-resolution remote sensing images, and the segmentation accuracy is higher and is therefore better than those of other methods.Our proposed D-UNet can deliver automatic building segmentation from high-resolution remote sensing images with high accuracy. Thus, it is a useful tool for practical application scenarios.关键词：remote sensing images;semantic segmentation;Multiscale;dilated convolution;image processing- 3270
- |
- 399
- |
- 0
发布时间：2023-12-08 -
摘要：Hyperspectral remote sensing technology, as a new type of earth observation technology, provides rich spectral information of features and can identify and finely classify feature targets. A single pixel in hyperspectral images contains multiple features as limited by the spatial resolution. As a result, the mixed pixels become widespread. Ultimately, the accuracy of pixel-level applications is difficult to improve. Nonnegative Matrix Factorization (NMF), with its clear physical meaning, lays the foundation for the development of unsupervised linear spectral unmixing. Thus, traditional NMF often uses Euclidean distance as a similarity measure method. On the one hand, hyperspectral data have manifold distribution. Thus, simple linear measurement between two points cannot accurately represent the distance between data. This problem makes the sample internal features weakly correlated, which results in the NMF algorithm having an inaccurate prediction of the high-dimensional spatial inaccurate prediction of the translational noise in high-dimensional space. On the other hand, the objective function constructed based on this method ignores the correlation characteristics in the image space, which inhibits the performance of the algorithm.Method Considering the correlation between data manifolds and features, this study proposes a nonnegative matrix factorization unmixing algorithm based on Sinkhorn distance and graph regularization constraint (SDGNMF). On the basis of fully exploiting the advantages of EMD, the algorithm imposes entropy regularization constraint on EMD, improves EMD to Sinkhorn distance, and takes it as the standard of measuring error, which effectively reduces the computational complexity. In addition, EMD with entropy regularization constraint, that is, the representation of the model by Sinkhorn distance, can better model the relationship between different dimensional features and fully utilize the correlation of features. In particular, this study introduces the graph regularity constraint based on the Sinkhorn distance to further characterize the manifold structure of data. Compared with the unmixing model constructed by Euclidean distance, SDGNMF is relatively insensitive to the noise in hyperspectral data and can better extract the internal structural information of the data, which improves the unmixing accuracy.Result An experiment was conducted on simulated and real datasets. Experimental results prove that the proposed algorithm proposed has achieved excellent subspace learning results and has good robustness. Compared with several other algorithms, SDGNMF can retain the similar structure after iteration. The correlation between the endmember features is also fully considered in SDGNMF. Thus, the similar substances distributed in adjacent regions can be separated. Therefore, SDGNMF can better display the details of local abundance and obtain a more realistic and perfect abundance map.Conclusion In general, the proposed unmixing model can overcome noise and consider the correlation of features and data manifold structure simultaneously. Experimental results show that the proposed algorithm can effectively improve the unmixing accuracy of most hyperspectral remote sensing data, especially those with high feature correlation. However, the proposed algorithm has high computational complexity. In addition, the algorithm only considers the prior knowledge of abundance. Therefore, future work will focus on solving these problems.关键词：hyperspectral unmixing;nonnegative matrix factorization (NMF);Sinkhorn distance;entropy regularization;graph regularization- 636
- |
- 326
- |
- 0
发布时间：2023-12-08
Models and Methods
-
摘要：Short-Rotation Plantations (SRPs) as the main economic forests have an important impact on ecological environmental protection and social economic development, but detailed information on the temporal and spatial distribution of SRPs is lacking. SRPs have nearly half a century of plantation history and extensive distribution in South China. This study aims to extract the long-term temporal and spatial distribution information of SRPs and analyze its changing trends and driving factors.The Guangxi Zhuang Autonomous Region, where SRPs are most widely grown in China, was used as the research area in this study. Based on the Google Earth Engine cloud platform and Landsat image data from 1986 to 2019, the 34-year Normalized Burn Ratio (NBR) long-term series data were first reconstructed by per pixel composite method. Then, the LandTrendr time series trajectory segmentation algorithm was used to segment and fit the NBR time series data for extracting the spatiotemporal distribution information of SRPs. Finally, Google Earth high-resolution images were used to select samples for verifying the accuracy of classification and extraction and analyzing the spatiotemporal characteristics and related factors of planting area changes in SRPs.(1) The accuracy of the SRP information extraction results was evaluated by the confusion matrix: the overall accuracy of the binary classifications reached 80.52%, the mapping accuracy of SRPs was 79.6%, the user accuracy of SRPs was 81.2%, and the kappa coefficient was over 0.6, which indicate that the classification model has a good classification effect. (2) The planting area of SRPs in Guangxi has been increasing steadily and rapidly year by year in the past 30 years. The planting area was only 1.93×105 ha in 1990, and it reached 4.04×106 ha by 2019, with an average annual growth rate of 1.33×105 ha. (3) In terms of spatial distribution, SRPs are concentrated in eastern and southern Guangxi. They are mainly distributed in low-altitude areas below 500 m and slopes with a surface slope of about 20°. Among them, Hechi City is the prefecture-level city with the largest planting and distribution area of SRPs in Guangxi. (4) A strong correlation exists between the change trend of SRP planting area and forestry output value (r=0.830894, p<0.001), which suggests that SRPs are an important affecting factor of forestry economy.The method proposed in this study based on the LandTrendr time series trajectory segmentation algorithm of SRP spatiotemporal information extraction is proven to be very effective. The mapping and analysis of the spatiotemporal distribution of SRPs can provide decision support for forestry management and provide basic data for the research on forest carbon cycle.关键词：remote sening;Short-Rotation Plantations;LandTrendr;Landsat;Spatiotemporal distribution mapping;long time series;Time series trajectory segmentation- 815
- |
- 201
- |
- 0
发布时间：2023-12-08 -
摘要：Crop stubble cover is an important method of conservation tillage. Obtaining the distribution of different corn stubble cover modes quickly and accurately is vital to the implementation status monitoring and effect evaluation of conservation tillage. Microwave remote sensing has characteristics of all-weather and strong penetration. Thus, it not only can ensure the acquisition of data in a short period for stubble monitoring but also can be sensitive to the information of surface roughness and crop residue structure, which provides rich information for the identification of stubble modes. Some studies consider the stubble monitoring with microwave data, but they mainly focus on the estimation of stubble coverage, and the identification of different stubble modes is rarely explored. In addition, the microwave backscattering coefficient is affected by many factors, such as soil moisture and roughness. Thus, the accuracy using microwave data simply to monitor stubble is limited.In this study, an identification method for corn stubble modes by removing soil backscatter is proposed using Sentinel-1 SAR data as the main data source. Based on the autumn field sample data in 2019 in Lishu County, Jilin Province, the backscattering model of the corn stubble is designed to separate the corn stubble scattering contribution from the soil scattering contribution and reduce the interference of soil scattering contribution on the identification of the corn stubble modes. A new Fusion Radar Index (FRI), which is produced with Sentinel-1 SAR data and Sentinel-2 optical image, is combined with traditional commonly used SAR features such as radar index and SAR textures. It is used to analyze the backscattering coefficient characteristic of field surface with different stubble modes. The best feature combination for stubble recognition is selected through the analysis of identification ability. A convolution neural network model based on 1D CNN is constructed using the optimal feature set selected to identify the corn stubble modes. The corn stubble modes are also mapped for the study area. Results show that (1) the overall accuracy of stubble identification is above 83% based on VH polarized data, FRI, and GLCM1–GLCM6 with backscattering values, which proves that the feature set obtained from Sentinel-1 radar scattering characteristics is feasible and effective for identification of the corn stubble modes. (2) The identification performance of the corn stubble modes based on data without the soil backscatter contribution improves significantly. The OA and kappa coefficients are 89.28% and 0.84, respectively. Compared with those before removing the influence of soil scattering, the recognition accuracy and kappa coefficient are improved by 5.44% and 0.09. Therefore, separating the soil scattering contribution from the total scattering contribution based on the stubble radar backscattering model can effectively reduce the influence of soil factors on the monitoring of corn stubble and improve the accuracy of the corn stubble mode recognition.This study demonstrates the great potential of Sentinel-1 SAR data and backscattering models to access the distribution map of corn stubble modes. It also provides a new idea for the wide application of Sentinel-1 SAR image in the research of corn stubble.关键词：remote sensing;Sentinel-1 SAR data;corn stubble;recognition of stubble modes;backscatter model;optimal feature set- 962
- |
- 209
- |
- 0
发布时间：2023-12-08 -
摘要：The subalpine coniferous forest in west Sichuan is located in southwest China, which is affected by cloudy, rainy, and foggy conditions. Thus, conducting vegetation classification in the area by using satellite images is difficult.(Objective)Therefore, this work selects Wanglang Nature Reserve, which is a typical area of subalpine coniferous forest in western Sichuan, as the study area. A multi-rotor UAV is used to acquire high-resolution RGB images of the northern part of the study area, and it is combined with a convolutional neural network model for vegetation classification.(Method)This study selects the semantic segmentation method (U-Net) for classification, constructs vegetation classification models based on UAV images of different spatial resolutions and sample sets under different tile sizes, and establishes a forest fingerprint library to further explore the potential of convolutional neural networks on UAV remote sensing images.(Result)(1) The combination of UAV visible images and convolutional neural network model for classification can obtain classification results of high accuracy, which reached the optimum at a spatial resolution of 5 cm and a size of 256×256. The overall accuracy was 93.21%, and the kappa coefficient was 0.90. (2) The increase in ultrahigh spatial resolution had limited improvement on the model accuracy. When the spatial resolution was increased from 10 cm to 5 cm, the overall accuracy of the model improved by 0.02 and the Kappa coefficient improved by 0.03, and the classification accuracy of the model did not improve significantly. (3) Choosing the appropriate size can improve the classification accuracy of the model. Under the spatial resolution of 5 cm, the overall accuracy of the model with the size of 128×128 was 82.30% and the kappa coefficient was 0.76, and the overall accuracy of the model with the size of 256×256 was 93.21% and the kappa coefficient was 0.90. (4) For the vegetation types that were underrepresented in the region, the influence of spatial resolution and tile size was much higher than that of the dominant tree species, especially the influence of spatial resolution was the highest. The producer and user accuracies for deciduous shrubs at a spatial resolution of 20 cm were below 70%.(Conclusion)This study shows that vegetation classification of subalpine coniferous forests in western Sichuan using UAV high-resolution RGB images combined with convolutional neural networks can achieve high-precision classification results. The effects of UAV spatial resolution and tile sizes on the accuracy of convolutional neural network models are explored, which further details the potential of convolutional neural networks on UAV high-resolution RGB images to provide an automatic and accurate research method for vegetation classification in this region.关键词：UAV RGB imagery;Convolutional Neural Network(CNN);subalpine coniferous forest in western Sichuan;vegetation classification;tile size;spatial resolution- 755
- |
- 267
- |
- 0
发布时间：2023-12-08 -
摘要：Many remote sensing indices have been developed for land surface phenology monitoring, but the ability of different remote sensing indices to represent the seasonal changes of land surface vegetation differs. At present, the evaluations of remote sensing indices for land surface phenology monitoring are mostly conducted under different standards, which results in poor comparability among research results. Thus, the best remote sensing indices cannot be selected according to different regions based on the aforementioned research results, which would affect the large-scale (e.g., hemispheric and even global) land surface phenology monitoring. Taking 406 records from 75 carbon flux tower stations and 482 records from 129 phenological camera stations as the reference standard, this study systematically evaluated the application of 10 remote sensing indexes in monitoring land surface phenology in the middle and high latitudes of the northern hemisphere. In addition, the best remote sensing indices and their accuracy under different conditions were compared and analyzed from two evaluation perspectives (including phenological extraction accuracy and phenological change trend consistency) and four dimensions (including vegetation type, geographical environment, phenological type, and phenological event).Although some remote sensing indices are the best in most conditions, the best remote sensing indices for different vegetation types, geographical environment, phenological types (functional phenology, structural phenology), and phenological events (spring and autumn) do not focus on a few of remote sensing indices but are scattered among all kinds of them. Even with the best remote sensing index, the error of land surface phenology monitoring is still large in some conditions. From different evaluation perspectives, the remote sensing indices with a high accuracy of phenology extraction are not exactly the same as those with a high consistency of phenological change trend, which suggests that the best remote sensing index should be selected according to the objects. The results of this study can provide the best remote sensing index selection basis for land surface phenology monitoring under different conditions, which will be helpful to improve the accuracy of large-scale land surface phenology monitoring and evaluate its uncertainty.关键词：remote sensing index;land surface phenology;vegetation type;geographical environment;structural phenology;functional phenology- 924
- |
- 543
- |
- 0
发布时间：2023-12-08
Forestry and Agriculture
- 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