Search

Satellite normalized difference vegetation index (NDVI) time series, essential for ecological and environmental applications, is still limited by a trade-off between the spatiotemporal resolution and time coverage despite various global products. The Advanced Very High-Resolution Radiometer (AVHRR) instrument can provide the longest continuous time series since 1982, but with the drawback of coarse spatial resolution and poor data quality. To address this issue, a spatiotemporal fusion-based long-term NDVI product (STFLNDVI) since 1982 was generated in this study at a 1-km spatial resolution with monthly intervals, by fusing with the Moderate Resolution Imaging Spectroradiometer (MODIS) data. A multi-step processing fusion framework, containing temporal filtering, normalization, spatiotemporal fusion, and residual error correction, was employed to combine the superior characteristics of the two products, respectively. Simulated comparison with MODIS data and real-data assessments with true 1 km AVHRR data both confirm the ideal accuracy of the fusion product in spatial distribution and temporal variation, providing stable long-term results similar to MODIS data. We believe that the STFLNDVI product will be of great significance in characterizing the spatial patterns and long-term variations of global vegetation and the historical radiometric calibrations in AVHRR data gaps around the Arctic and instrument differences between MODIS and AVHRR should be further considered in the future.

Abstract Existing assessments might have underappreciated ozone-related health impacts worldwide. Here our study assesses current global ozone pollution using the high-resolution (0.05°) estimation from a geo-ensemble learning model, with key focuses on population exposure and all-cause mortality burden. Our model demonstrates strong performance, achieving a mean bias of less than -1.5 parts per billion against in-situ measurements. We estimate that 66.2% of the global population is exposed to excess ozone for short term (> 30 days per year), and 94.2% suffers from long-term exposure. Furthermore, severe ozone exposure levels are observed in Cropland areas, particularly over Asia. Importantly, the all-cause ozone-attributable deaths significantly surpass previous recognition from specific diseases worldwide. Notably, mid-latitude Asia (30°N) and the western United States show high mortality burden, contributing substantially to global ozone-attributable deaths. Our study highlights current significant global ozone-related health risks and may benefit the ozone-exposed population in the future.

Global and local modeling is essential for image super-resolution tasks. However, current efforts often lack explicit consideration of the cross-scale knowledge in large-scale earth observation scenarios, resulting in suboptimal single-scale representations in global and local modeling. The key motivation of this work is inspired by two observations: 1) There exists hierarchical features at the local and global regions in remote sensing images, and 2) they exhibit scale variation of similar ground objects (e.g. cross-scale similarity). In light of these, this paper presents an effective method to grasp the global and local image hierarchies by systematically exploring the cross-scale correlation. Specifically, we developed a Cross-scale Self-Attention (CSA) to model the global features, which introduces an auxiliary token space to calculate cross-scale self-attention matrices, thus exploring global dependency from diverse token scales. To extract the cross-scale localities, a Cross-scale Channel Attention (CCA) is devised, where multi-scale features are explored and progressively incorporated into an enriched feature. Moreover, by hierarchically deploying CSA and CCA into transformer groups, the proposed Cross-scale Hierarchical Transformer (CHT) can effectively explore cross-scale representations in remote sensing images, leading to a favorable reconstruction performance. Comprehensive experiments and analysis on four remote sensing datasets have demonstrated the superiority of CHT in both simulated and real-world remote sensing scenes. In particular, our CHT outperforms the state-of-the-art approach (TransENet) in terms of PSNR by 0.11 dB on average, but only accounts for 54.8% of its parameters.

Accurate crop mapping plays a critical role in optimizing agricultural monitoring and ensuring food security. Although data-driven deep learning methods have demonstrated success in crop mapping with satellite image time series (SITS) data, their promising performances heavily depend on labeled training samples. Nevertheless, the difficulty of annotating crop types often results in labeled data scarcity, leading to a decline in the model’s performance. Self-supervised learning (SSL) is a novel technique for crop mapping with limited labels. However, the existing SSL methods applied to SITS data typically explore masking solely on temporal dimension, which cannot guarantee strong spatial representation and therefore hinders the accurate prediction of complex crop fields. Furthermore, these methods sequentially extract spatial and temporal information without fully integrating information across different dimensions. In this study, we propose a spatiotemporal masking strategy for pre-training a SpatioTemporal Collaborative Learning Network (STCLN) to extract informative spatial and temporal representations from SITS data. Additionally, we design a SpatioTemporal Attention (STA) module in STCLN that integrates representations from spatial and temporal dimensions. The experimental results on two crop type mapping benchmarks encompassing various crop types demonstrate the outperformance of our proposed method. STCLN_wp outperforms the previous state-of-the-art (SOTA) methods with 6.49% higher mIoU on PASTIS dataset and 4.04% higher mIoU on MTLCC dataset. The ablation experiments on pre-training, masking strategies, and the STA module validate the effectiveness of our methodological design. Additionally, experiments conducted under varying sizes of the training set highlight the superior generalization ability of our method for crop type mapping in label-scarce situations. The code of our method is available at https://github.com/XiaoleiQinn/STCLN.

Deep convolutional neural networks (DCNNs) have been successfully used in semantic segmentation of high-resolution remote sensing images (HRSIs). However, this task still suffers from intra-class inconsistency and boundary blur due to high intra-class heterogeneity and inter-class homogeneity, considerable scale variance, and spatial information loss in conventional DCNN-based methods. Therefore, a novel boundary-enhanced dual-stream network (BEDSN) is proposed, in which an edge detection branch stream (EDBS) with a composite loss function is introduced to compensate for boundary loss in semantic segmentation branch stream (SSBS). EDBS and SSBS are integrated by highly coupled encoder and feature extractor. A lightweight multilevel information fusion module guided by channel attention mechanism is designed to reuse intermediate boundary information effectively. For aggregating multiscale contextual information, SSBS is enhanced by multiscale feature extraction module and hybrid atrous convolution module. Extensive experiments have been tested on ISPRS Vaihingen and Potsdam datasets. Results show that BEDSN can achieve significant improvements in intra-class consistency and boundary refinement. Compared with 11 state-of-the-art methods, BEDSN exhibits higher-level performance in both quantitative and visual assessments with low model complexity. The code will be available at https://github.com/lixinghua5540/BEDSN.

Optical and synthetic aperture radar (SAR) images provide complementary information to each other. However, the heterogeneity of same-ground objects brings a large difficulty to change detection (CD). Correspondingly, transformation-based methods are developed with two independent tasks of image translation and CD. Most methods only utilize deep learning for image translation, and the simple cluster and threshold segmentation leads to poor CD results. Recently, a deep translation-based CD network (DTCDN) was proposed to apply deep learning for image translation and CD to improve the results. However, DTCDN requires the sequential training of the two independent subnetwork structures with a high computational cost. Toward this end, a concatenated deep-learning framework, multitask change detection network (MTCDN), of optical and SAR images is proposed by integrating the CD network into a complete generative adversarial network. This framework contains two generators and discriminators for optical and SAR image domains. Multitask refers to the combination of image identification by discriminators and CD based on an improved UNet++. The generators are responsible for image translation to unify the two images into the same feature domain. In the training and prediction stages, an end-to-end framework is realized to reduce cost. The experimental results on four optical and SAR datasets prove the effectiveness and robustness of the proposed framework over eight baselines.

This article addresses the problem of building segmentation for rural areas with high-resolution remote sensing images. Due to the irregular spatial distribution of rural buildings, it is often challenging to perform pixel-wise dense prediction to entire areas like the usual segmentation task to extract buildings. Specifically, we present a multilevel decoding network model that classifies the input image on the patch and image levels according to the distribution of buildings. A scene head module is used to identify scenes defined as patches that contain buildings. Depending on the scene classification results, a decode gate is taken to determine the level of prediction. This hierarchical extraction strategy reduces the amount of inference time. Experiments on our constructed rural building dataset with large-scale images validate the high efficiency of the proposed method.

Foundation models have reshaped the landscape of remote sensing (RS) by enhancing various image interpretation tasks. Pretraining is an active research topic, encompassing supervised and self-supervised learning methods to initialize model weights effectively. However, transferring the pretrained models to downstream tasks may encounter task discrepancy due to their formulation of pretraining as image classification or object discrimination tasks. In this study, we explore the multitask pretraining (MTP) paradigm for RS foundation models to address this issue. Using a shared encoder and task-specific decoder architecture, we conduct multitask supervised pretraining on the segment anything model annotated remote sensing segmentation dataset, encompassing semantic segmentation, instance segmentation, and rotated object detection. MTP supports both convolutional neural networks and vision transformer foundation models with over 300 million parameters. The pretrained models are finetuned on various RS downstream tasks, such as scene classification, horizontal, and rotated object detection, semantic segmentation, and change detection. Extensive experiments across 14 datasets demonstrate the superiority of our models over existing ones of similar size and their competitive performance compared to larger state-of-the-art models, thus validating the effectiveness of MTP.

High resolution satellite images are becoming increasingly available for urban multi-temporal semantic understanding. However, few datasets can be used for land-use/land-cover (LULC) classification, binary change detection (BCD) and semantic change detection (SCD) simultaneously because classification datasets always have one time phase and BCD datasets focus only on the changed location, ignoring the changed classes. Public SCD datasets are rare but much needed. To solve the above problems, a tri-temporal SCD dataset made up of Gaofen-2 (GF-2) remote sensing imagery (with 11 LULC classes and 60 change directions) was built in this study, namely, the Wuhan Urban Semantic Understanding (WUSU) dataset. Popular deep learning based methods for LULC classification, BCD and SCD are tested to verify the reliability of WUSU. A Siamese-based multi-task joint framework with a multi-task joint loss (MJ loss) named ChangeMJ is proposed to restore the object boundaries and obtains the best results in LULC classification, BCD and SCD, compared to the state-of-the-art (SOTA) methods. Finally, a large spatial-scale mapping for Wuhan central urban area is carried out to verify that the WUSU dataset and the ChangeMJ framework have good application values.

Satellite video is an emerging surface observation data that has drawn increasing interest due to its potential in spatiotemporal dynamic analysis. Single object tracking of satellite videos allows the continuous acquisition of the positions and ranges of objects and establishes the correspondences in the video sequence. However, small-sized objects are vulnerable to rotation and non-rigid deformation. Moreover, the horizontal bounding box of most trackers has difficulty in providing accurate semantic representations such as object position, orientation, and spatial distribution. In this article, we propose a unified framework, named rotation equivariant Siamese network enhanced by probability segmentation (REPS), to enhance the tracking accuracy and semantic representations simultaneously. First, to deal with the inconsistency of representations, we design a rotation equivariant (RE) Siamese network architecture to detect the rotation variations of objects right from the start frame, achieving the RE tracking. Second, a pixel-level (PL) refinement is proposed to refine the spatial distribution of objects. In addition, we proposed an adaptive Gaussian fusion that synergizes tracking and segmentation results to obtain compact outputs for satellite object representations. Extensive experiments on satellite videos demonstrate the superiority of the proposed approach. The code will be available at https://github.com/YZCU/REPS

The global population is on the rise, leading to an increased demand for food resources. Accurate mapping and monitoring of paddy rice fields have become crucial for effective food management and yield estimation. Several large-scale paddy rice mapping products with medium-to-high resolution have been made by using threshold-based phenological methods. The main shortcoming is that paddy rice varies in different regions due to differences in climate and planting habits, which results in the limitations of fixed phenological indicator thresholds for large-scale paddy rice mapping. Therefore, paddy rice samples from different regions are needed to account for regional differences between paddy rice and improve paddy rice mapping accuracy. To address two key shortcomings in existing sample generation methods (the non-robustness of DTW distance and the limitation of fixed phenological indicator thresholds for generating samples) and the three difficulties of cross-regional sample generation (different planting times, growth cycles, and planting intensities), this study proposed a novel approach for cross-regional paddy rice sample generation based on spatial domain transfer, achieving cross-regional paddy rice mapping. The accuracy results (OA of 95.16–97.54%) in three study regions with different climatic and topographic conditions in China and improved accuracy (OA improvement of 5.89–19.06%) compared with the method of NESEA-Rice10, proved the effectiveness of the proposed method and the potential of the proposed method in large-scale paddy rice mapping. The proposed method can obtain high-accuracy paddy rice mapping results in regions without paddy rice samples and be applied to the mapping of other crops and land cover types.

The objective of image super-resolution is to reconstruct a high-resolution (HR) image with the prior knowledge from one or several low-resolution (LR) images. However, in the real world, due to the limited complementary information, the performance of both single-frame and multi-frame super-resolution reconstruction degrades rapidly as the magnification increases. In this paper, we propose a novel two-step image super resolution method concatenating multi-frame super-resolution (MFSR) with single-frame super-resolution (SFSR), to progressively upsample images to the desired resolution. The proposed method consisting of an L0-norm constrained reconstruction scheme and an enhanced residual back-projection network, integrating the flexibility of the variational model-based method and the feature learning capacity of the deep learning-based method. To verify the effectiveness of the proposed algorithm, extensive experiments with both simulated and real world sequences were implemented. The experimental results show that the proposed method yields superior performance in both objective and perceptual quality measurements. The average PSNRs of the cascade model in set5 and set14 are 33.413 dB and 29.658 dB respectively, which are 0.76 dB and 0.621 dB more than the baseline method. In addition, the experiment indicates that this cascade model can be robustly applied to different SFSR and MFSR methods.

Paddy rice mapping is crucial for cultivation management, yield estimation, and food security. Guangdong, straddling tropics and subtropics, is a major rice-producing region in China. Mapping paddy rice in Guangdong is essential. However, there are 2 main difficulties in tropical and subtropical paddy rice mapping, including the lack of high-quality optical images and differences in paddy rice planting times. This study proposed a paddy rice mapping framework using phenology matching, integrating Sentinel-1 and Sentinel-2 data to incorporate prior knowledge into the classifiers. The transplanting periods of paddy rice were identified with Sentinel-1 data, and the subsequent 3 months were defined as the growth periods. Features during growth periods obtained by Sentinel-1 and Sentinel-2 were inputted into machine learning classifiers. The classifiers using matched features substantially improved mapping accuracy compared with those using unmatched features, both for early and late rice mapping. The proposed method also improved the accuracy by 6.44% to 16.10% compared with 3 other comparison methods. The model, utilizing matched features, was applied to early and late rice mapping in Guangdong in 2020. Regression results between mapping area and statistical data validate paddy rice mapping credibility. Our analysis revealed that thermal conditions, especially cold severity during growing stages, are the primary determinant of paddy rice phenology. Spatial patterns of paddy rice in Guangdong result from a blend of human and physical factors, with slope and minimum temperature emerging as the most important limitations. These findings enhance our understanding of rice ecosystems’ dynamics, offering insights for formulating relevant agricultural policies.

Wildfires frequently occur around the world, which seriously threaten the ecology, environment, economic development, even human safety. In this work, we propose a novel framework for near-real-time and early-stage wildfire detection using Himawari-8 satellite 10-min data. Different from most of the existing methods, the proposed framework jointly combines spatial, temporal and spectral information for wildfire detection. The integrated time-series spatial variance, temporal difference and spectral difference can comprehensively judge the wildfire points and exclude disturbed points. Dispense with cloud detection and setting too many manual thresholds, a spatial–temporal–spectral recurrent neural network (STS-RNN) is developed to adaptively learn the time-series spatial–temporal–spectral curves. Compared with JAXA’s wildfire products, the proposed framework can more accurately detect the small, early-stage, day-time, night-time and forest wildfire points in three experimental scenarios. Especially for the early-stage wildfire detection, the proposed framework may provide the rapid alarm for the local fire department and emergency management agency. This greatly breaks through the limitations of existing wildfire detection methods.

High resolution of global land cover dynamic is indicative for understanding the influence of anthropogenic activity on environmental change. However, most of the land cover products are based on Landsat image that only has 30 m resolution, which is insufficient to distinguish the heterogenous urban structure; while very high spatial resolution image usually has low temporal resolution, which is difficult to monitor the urban dynamic. Deep-learning-driven super-resolution mapping is a prevailing way of achieving very-high-resolution land cover dynamic products in aspect of alleviating the mixed pixel problem of Landsat image. However, two limitations are obvious: 1) the fixed grid of kernel during the upsampling process favors spatial homogeneity and suppresses the learning of spatial heterogeneity of urban composition and 2) geometric or radiation variation over large spatial and long temporal extent in remote sensing images makes the super-resolution mapping approach difficult to transfer for application. Here, we attempt to solve these two limitations: 1) a progressive edge-guided super-resolution architecture is designed to allow nonuniformed kernel specific at the low-confidence edge region and intensify the learning of heterogenous compositions’ patterns and 2) an alternating optimization strategy is designed to minimize the resultant entropy and modulate the classification hyperplane to accommodate to the manifold of the discrepant region. Validation experiments are investigated based on a fine-grained and large-extent super-resolution (FLAS) dataset constructed in this study, and it is found that our approach remarkably enhances rich detailed patterns of heterogenous region and outperforms other state-of-the-art algorithms. Besides, we applied DETNet to the large spatial extent of 28 metropolises in China (>40,000 km2) and the large temporal extent of continuous 21-year (2000–2020) in Wuhan city to examine transferability. From the land cover areas variation, we find that the expansion rate of cropland is faster than the urban expansion over the past 10 years, which are gradually becoming the principal source for the encroachment of forest and lakes. From detailed urban dynamic reflected by the 21-year products, we find that urban-villages between the old city zone and the outer high-tech development zone are gradually disappeared. The captured dynamic is consistence with the urban-village renovation policy during this period, which is meant to redistribute the spatial configuration of the city for a more sustainable urban structure. We believe that the proposed method can facilitate a seamless and fine-grained observation system that can fill the weakness of the existing land cover activities and provide a brand-new insight into the urban dynamic and its underlying mechanism.

The spatial resolution of land cover mapping has been increasing with the evolution of Earth observation technology. However, the higher spatial resolution makes it more laborious to collect training samples for efficient land-cover product updating. Fortunately, the existing historical products with a lower spatial resolution can be used as labels to achieve cross-resolution mapping with the latest images of a higher resolution. Although cross-resolution mapping can generate a large number of low-cost training labels, the labels can be noisy due to resolution mismatch or semantic errors. Furthermore, the existing deep learning based classification models have difficulty in maintaining the output resolution. In this article, a cross-resolution land-cover mapping framework with noisy label learning is proposed to complete high-resolution land-cover mapping based on noisy lower-resolution labels. The proposed method contains three parts: 1) To solve the resolution mismatch problem, the training labels are refined by modeling the spectral similarity and the spatial adjacency between the labels and the images using a conditional random field model. 2) To avoid the problem of resolution loss with the deep fully convolutional neural networks, a high-resolution deep semantic segmentation network is proposed to achieve deep feature extraction while maintaining the output resolution. 3) To eliminate the influence of semantic errors in the generated labels, a class-conditional label correction method is proposed, which detects and corrects the abnormal incorrect labels to facilitate network training. A national-scale land-cover mapping experiment for China was carried out. The 10-m spatial resolution land-cover map of China for 2020 was produced with improved accuracy based on the 30-m spatial resolution historical product, which shows the practicability of the proposed CRLC for rapid land-cover mapping.

Local climate zone (LCZ) has become a new standard classification scheme in urban landscapes and showed great potential in urban climate research. Traditional classifiers and ordinary neural networks only consider the spectral or local spatial features of the pixel, ignoring the effect of nonlocal information on the LCZ classification. The graph convolutional network (GCN) has been used to exploit the relationship between adjacent and global land covers owing to the ability to conduct flexible convolution over graphs. In this work, we integrated a convolutional neural network and two GCNs into an end-to-end hybrid framework and generated LCZs directly from the original images. Local-, regional-, and global-level features were extracted and grouped complementarily to foster better performance. Experiments were conducted in six cities around the world to verify the effectiveness of our method. Results showed that the average classification accuracy of the six cities reached 0.956 and performed better than any other comparable model. Ablation experiments also demonstrated the mutual promotion of the different modules. Finally, the small sample experiment provided a practical reference for the LCZ classification in the absence of samples in future.

Hyperspectral anomaly detection (HAD) is an important hyperspectral image application. HAD can find pixels with anomalous spectral signatures compared with their neighbor background without any prior information. While most of the existed researches are related to statistic-based and distance-based techniques, by summarizing the background samples with certain models, and then, finding the very few outliers by various distance metrics, this review focuses on the HAD based on machine learning methods, which have witnessed remarkable progress in the recent years. In particular, these studies can generally be grouped into the traditional machine learning and deep-learning-based methods. Several representative HAD methods, including both traditional machine and deep-learning-based methods, are then conducted on four real HSIs in the experiments. Finally, conclusions regarding HAD are summarized, and prospects and future development direction are discussed.

With the hyperspectral imaging technology, hyperspectral data provides abundant spectral information and plays a more important role in the geological survey, vegetation analysis, and military reconnaissance. Different from normal change detection, hyperspectral anomaly change detection (HACD) helps to find those small but important anomaly changes between multitemporal hyperspectral images (HSI). In previous works, most classical methods use linear regression to establish the mapping relationship between two HSIs and then detect the anomalies from the residual image. However, the real spectral differences between multi-temporal HSIs are likely to be quite complex and of nonlinearity, leading to the limited performance of these linear predictors. In this article, we propose an original HACD algorithm based on autoencoder (ACDA) to give a nonlinear solution. The proposed ACDA can construct an effective predictor model when facing complex imaging conditions. In the ACDA model, two siamese autoencoder networks are deployed to construct two predictors from two directions. The predictor is used to model the spectral variation of the background to obtain the predicted image under another imaging condition. Then the mean square error between the predictive image and corresponding expected image is computed to obtain the loss map, where the spectral differences of the unchanged pixels are highly suppressed and anomaly changes are highlighted. Ultimately, we take the minimum of the two loss maps of two directions as the final anomaly change intensity map. The experiments results on public “Viareggio 2013” datasets demonstrate the efficiency and superiority over traditional methods.

With the maturation of satellite technology, Hyperspectral Remote Sensing (HRS) platforms have developed from the initial ground-based and airborne platforms into spaceborne platforms, which greatly promotes the civil application of HRS imagery in the fields of agriculture, forestry, and environmental monitoring. China is playing an important role in this evolution, especially in recent years, with the successful launch and operation of a series of civil hyperspectral spacecraft and satellites, including the Shenzhou-3 spacecraft, the Gaofen-5 satellite, the SPARK satellite, the Zhuhai-1 satellite network for environmental and resources monitoring, the FengYun series of satellites for meteorological observation, and the Chang’E series of spacecraft for planetary exploration. The Chinese spaceborne HRS platforms have various new characteristics, such as the wide swath width, high spatial resolution, wide spectral range, hyperspectral satellite networks, and microsatellites. This paper focuses on the recent progress in Chinese spaceborne HRS, from the aspects of the typical satellite systems, the data processing, and the applications. In addition, the future development trends of HRS in China are also discussed and analyzed.

Satellite images have been widely used for urban heat island (UHI) monitoring in recent studies, among which the summer UHI has attracted more attention. However, the studies based on high spatial resolution images have to use single-day land surface temperature (LST) to analyze the summer UHI, due to the low temporal resolution, which is not representative of the summer and leads to incomparability in the time series. The studies based on low spatial resolution images can generate a time series of representative LSTs for summer (e.g., summer mean LSTs), due to the high temporal resolution, but these LSTs lack sufficient spatial details for a refined analysis. To fill these gaps, we propose to integrate the respective advantages of the above approaches to generate a comparable and fine-scale LST time series with a high spatiotemporal resolution. By normalizing the LSTs between the different satellite images via robust fitting with Huber's M-estimator and moment matching, the comparability is ensured. Furthermore, the high-spatial resolution and high-temporal resolution are combined via the spatiotemporal fusion. Overall, we propose a procedure to produce a comparable time series of annual and fine-scale summer mean LSTs, which can serve as a basis for elaborate analysis of the thermal environment.

Hyperspectral image (HSI) fusion can effectively improve the spatial resolution of HSIs by integrating high-resolution multispectral images (MSIs). Considering the spatial and spectral degradation relationship between a fused image and input images, a physics-based GAN is proposed to fuse HSI and MSI. A physical model estimating degradation of image is introduced in the generator and in the discriminators. For the generator, a set of recursive modules including a physical degradation model and a multiscale residual channel attention fusion module integrate the spectral-spatial difference information between input images and estimated degradation images to restore the details of the fused image. Subsequently, the residual spatial attention fusion module is used to combine the results of all recursions to obtain the final reconstructed result. As for the discriminators, three networks with the final fused image, estimated LR HSI and estimated MSI as inputs share the same architecture. Finally, the loss function that contains adversarial losses and L1 losses of the fused image and estimated degradation images is used to optimize network parameters. The experimental results demonstrate that the proposed method outperforms state-of-the-art methods.

The past years have witnessed great progress on remote sensing (RS) image interpretation and its wide applications. With RS images becoming more accessible than ever before, there is an increasing demand for the automatic interpretation of these images. In this context, the benchmark datasets serve as an essential prerequisites for developing and testing intelligent interpretation algorithms. After reviewing existing benchmark datasets in the research community of RS image interpretation, this article discusses the problem of how to efficiently prepare a suitable benchmark dataset for RS image interpretation. Specifically, we first analyze the current challenges of developing intelligent algorithms for RS image interpretation with bibliometric investigations. We then present the general guidances on creating benchmark datasets in efficient manners. Following the presented guidances, we also provide an example on building RS image dataset, i.e., Million Aerial Image Dataset (Online. Available: https://captain-whu.github.io/DiRS/), a new large-scale benchmark dataset containing a million instances for RS image scene classification. Several challenges and perspectives in RS image annotation are finally discussed to facilitate the research in benchmark dataset construction. We do hope this article will provide the RS community an overall perspective on constructing large-scale and practical image datasets for further research, especially data-driven ones.

Building footprint information is one of the key factors for sustainable urban planning and environmental monitoring. Mapping building footprints from remote sensing images is an important and challenging task in the earth observation field. Over the years, convolutional neural networks have shown outstanding improvements in the building extraction field due to their ability to automatically extract hierarchical features and make building predictions. However, as buildings are various in different sizes, scenes, and roofing materials, it is hard to precisely depict buildings of varied sizes, especially in large areas (e.g., nationwide). To tackle these limitations, we propose a novel deep-supervision convolutional neural network (denoted as DS-Net) for extracting building footprints from high-resolution remote sensing images. In the proposed network, we applied deep supervision with an extra lightweight encoder, which enables the network to learn representative building features of different scales. Furthermore, a scale attention module is designed to aggregate multiscale features and generate the final building prediction. Experiments on two publicly available building datasets, including the WHU Building Dataset and the Massachusetts Building Dataset, show the effectiveness of the proposed method. With only a 0.22-M increment of parameters compared with U-Net, the proposed DS-Net achieved an IoU of 90.4% on the WHU Building Dataset and 73.8% on the Massachusetts Dataset. DS-Net also outperforms the state-of-the-art building extraction methods on the two datasets, indicating the effectiveness of the proposed deep supervision and scale attention.

As the COVID-19 epidemic began to worsen in the first months of 2020, stringent lockdown policies were implemented in numerous cities throughout the world to control human transmission and mitigate its spread. Although traffic density reduction inside the city was felt subjectively, there has thus far been no objective and quantitative study of its variation to reflect the intracity population flows and their corresponding relationship with lockdown policy stringency from the view of remote sensing (RS) images with the high resolution under 1 m. Accordingly, we here provide a quantitative investigation of the traffic density reduction before and after lockdown was implemented in six cities (Wuhan, Milan, Madrid, Paris, New York, and London) around the world heavily affected by the COVID-19 epidemic, which is accomplished by extracting vehicles from the multitemporal high-resolution RS images. A novel vehicle detection model combining unsupervised vehicle candidate extraction and deep learning identification was specifically proposed for the images with the resolution of 0.5 m. Our results indicate that traffic densities were reduced by an average of approximately 50% (and as much as 75.96%) in these six cities following lockdown. The influences on traffic density reduction rates are also highly correlated with policy stringency, with an ${{\boldsymbol{R}}^2}$ value exceeding 0.83. Even within a specific city, the traffic density changes differed and tended to be distributed in accordance with the city's land-use patterns. Considering that public transport was mostly reduced or even forbidden, our results indicate that city lockdown policies are effective at limiting human transmission within cities.

Global land cover mapping activities are of great important for retrieving the environment we are living in. However, the trade-off between spatial and temporal resolution makes it difficult to obtain the continuous fine-scale land cover product for detail and frequent land surface analysis. To overcome the difficulty, this study proposed a modified maximum a posteriori (MAP) based spatiotemporal sub-pixel mapping (SSM), for a long-term and fine-scale land cover mapping. The proposed approach consists of three parts, i.e., data fidelity term, spatial regularization term, and enhanced temporal regularization term, which can make use of historical fine-scale image for a better super-resolution of the current coarse image. One regional experiment was implemented for the algorithm validation, and one real experiment was investigated, by generating continuous land cover mapping products (2010–2015) at Landsat-like spatial resolution from time series of MODIS images. Compared with traditional methods, the proposed approach reconstructed more pleasant spatial details and had greater quantitative accuracy, outperforming the state-of-the-art spatiotemporal sub-pixel mapping method by averagely 2.58% for all the inspected dates in OA metric. This study not only provides a generalized research framework for generating continuous fine-scale land cover product from daily available MODIS images, but also validates the usability of the time series products for detail land cover monitoring, and reveals a finding that the study area of Wuhan is a well practitioner of the sustainable development policy during its urbanization process.

In order to mitigate the spread of COVID-19, Wuhan was the first city to implement strict lockdown policy in 2020. Even though numerous researches have discussed the travel restriction between cities and provinces, few studies focus on the effect of transportation control inside the city due to the lack of the measurement and available data in Wuhan. Since the public transports have been shut down in the beginning of city lockdown, the change of traffic density is a good indicator to reflect the intracity population flow. Therefore, in this paper, we collected time-series high-resolution remote sensing images with the resolution of 1 m acquired before, during and after Wuhan lockdown by GF-2 satellite. Vehicles on the road were extracted and counted for the statistics of traffic density to reflect the changes of human transmissions in the whole period of Wuhan lockdown. Open Street Map was used to obtain observation road surfaces, and a vehicle detection method combing morphology filter and deep learning was utilized to extract vehicles with the accuracy of 62.56%. According to the experimental results, the traffic density of Wuhan dropped with the percentage higher than 80%, and even higher than 90% on main roads during city lockdown; after lockdown lift, the traffic density recovered to the normal rate. Traffic density distributions also show the obvious reduction and increase throughout the whole study area. The significant reduction and recovery of traffic density indicates that the lockdown policy in Wuhan show effectiveness in controlling human transmission inside the city, and the city returned to normal after lockdown lift.

Hyperspectral imaging is a significant remote sensing technology for deep space exploration to understand the planetary geological evolution. However, hyperspectral images (HSIs) usually suffer from various noises because of the complicated environment and equipment limitations, which leads to inconvenience to subsequent applications. In this work, a comprehensive and systematic investigation on planetary noise categories is summarized initially, and an intercomparison among state-of-the-art terrestrial joint spatial-spectral restoration models is performed to test their capability on planetary datasets. The Compact Reconnaissance Imaging Spectrometer for Mars and Observatoire pour la Minéralogie, l'Eau, les Glaces et l'Activité are adopted as examples. An improved nonreference quantitative evaluation method based on the high-resolution imaging science experiment imagery is proposed. The processed high-resolution classification result of Russell Dune can be used as the reference of unmixing after denoising. Then, spectral and spatial fidelity can be assessed indirectly. Experimental results emphasize that denoising approaches with modeling for non-independently and identically (non i.i.d.) noise characteristics are more suitable for planetary HSIs because of the diversity and complexity of their noises. This kind of method is flexible under practical circumstances and maintains the intrinsic information of HSIs better.

Graph learning is an effective manner to analyze the intrinsic properties of data. It has been widely used in the fields of dimensionality reduction and classification for data. In this paper, we focus on the graph learning-based dimensionality reduction for a hyperspectral image. Firstly, we review the development of graph learning and its application in a hyperspectral image. Then, we mainly discuss several representative graph methods including two manifold learning methods, two sparse graph learning methods, and two hypergraph learning methods. For manifold learning, we analyze neighborhood preserving embedding and locality preserving projections which are two classic manifold learning methods and can be transformed into the form of a graph. For sparse graph, we introduce sparsity preserving graph embedding and sparse graph-based discriminant analysis which can adaptively reveal data structure to construct a graph. For hypergraph learning, we review binary hypergraph and discriminant hyper-Laplacian projection which can represent the high-order relationship of data.

This paper deals with the massive point cloud segmentation processing technology on the basis of machine vision, which is the second essential factor for the intelligent data processing of three dimensional conformation in digital photogrammetry. In this paper, multi-model fitting method is used to segment the point cloud according to the spatial distribution and spatial geometric structure of point clouds by fitting the point cloud into different geometric primitives models. Because point cloud usually possesses large amount of 3D points, which are uneven distributed over various complex structures, this paper proposes a point cloud segmentation method based on multi-model fitting. Firstly, the pre-segmentation of point cloud is conducted by using the clustering method based on density distribution. And then the follow fitting and segmentation are carried out by using the multi-model fitting method based on split and merging. For the plane and the arc surface, this paper uses different fitting methods, and finally realizing the indoor dense point cloud segmentation. The experimental results show that this method can achieve the automatic segmentation of the point cloud without setting the number of models in advance. Compared with the existing point cloud segmentation methods, this method has obvious advantages in segmentation effect and time cost, and can achieve higher segmentation accuracy. After processed by method proposed in this paper, the point cloud even with large-scale and complex structures can often be segmented into 3D geometric elements with finer and accurate model parameters, which can give rise to an accurate 3D conformation.

This paper presents the scientific outcomes of the 2018 Data Fusion Contest organized by the Image Analysis and Data Fusion Technical Committee of the IEEE Geoscience and Remote Sensing Society. The 2018 Contest addressed the problem of urban observation and monitoring with advanced multi-source optical remote sensing (multispectral LiDAR, hyperspectral imaging, and very high-resolution imagery). The competition was based on urban land use and land cover classification, aiming to distinguish between very diverse and detailed classes of urban objects, materials, and vegetation. Besides data fusion, it also quantified the respective assets of the novel sensors used to collect the data. Participants proposed elaborate approaches rooted in remote-sensing, and also in machine learning and computer vision, to make the most of the available data. Winning approaches combine convolutional neural networks with subtle earth-observation data scientist expertise.

Missing data reconstruction for remote sensing images, such as dead-pixel recovery and cloud removal, is important for remote sensing data applications. Missing information reconstruction is well known as being an ill-posed inverse problem. In this paper, a weighted low-rank tensor regularization model is proposed to handle the problem. The proposed model fully utilizes the correlations in the spatial, spectral, and temporal components of the remote sensing images to adaptively deal with the varied missing data problems, including the Aqua Moderate Resolution Imaging Spectroradiometer (MODIS) band 6 dead line problem, the Landsat scan-line corrector failure (SLC-off) problem, and cloud contamination. A double-weighted treatment is developed to balance the contributions from the different dimensions and preserve the different structures and textures in remote sensing images. The experiments undertaken confirmed the good performance of the proposed method, and the reconstruction results of the proposed method, in both visual effect and quantitative evaluation, were superior to those of the other methods.

Fine particulate matter (PM _2.5 ) is widely concerned for its harmful impacts on global environment and human health, making air pollution monitoring so crucial and indispensable. As the world’s first open, real-time, and historical air quality platform, OpenAQ collects and provides government measurement and research-level data from various channels. However, despite OpenAQ’s innovation in providing us with ground-measured PM _2.5 worldwide, we find significant data gaps in time series for most of the sites. The incompleteness of the data directly affects the public perception of PM _2.5 concentration levels and hinders the progress of research related to air pollution. To address these issues, a two-step hybrid model named ST-SILM, i.e. spatio-temporal model with single exponential smoothing-inverse distance weighted (SES-IDW) and long short-term memory (LSTM), is proposed to repair the missing data from PM _2.5 sites worldwide collected from OpenAQ from 2017 to 2019. Both spatio-temporal correlation and neighborhood fields are considered and established in the model. To be specific, SES-IDW were firstly used to repair missing values, and secondly, the LSTM network was employed to reconstruct the time series of continuous missing data. After the global ground-measured PM _2.5 was reconstructed, the light gradient boosting machine model was applied to remote sensing estimation of the original ground-measured PM _2.5 and of the reconstructed ground-measured PM _2.5 to further verify the performance of ST-SILM. Experiment results show that the estimation accuracy of the reconstructed dataset is better ( R ^2 from 2017 to 2019 increased by 0.02, 0.02, and 0.01 compared with the original dataset). Therefore, it is concluded that the proposed model can effectively reconstruct data from PM _2.5 sites worldwide.

Research on unmanned aerial vehicles (UAV) has been increasingly popular in the past decades, and UAVs have been widely used in industrial inspection, remote sensing for mapping & surveying, rescuing, and so on. Nevertheless, the limited autonomous navigation capability severely hampers the application of UAVs in complex environments, such as GPS-denied areas. Previously, researchers mainly focused on the use of laser or radar sensors for UAV navigation. With the rapid development of computer vision, vision-based methods, which utilize cheaper and more flexible visual sensors, have shown great advantages in the field of UAV navigation. The purpose of this article is to present a comprehensive literature review of the vision-based methods for UAV navigation. Specifically on visual localization and mapping, obstacle avoidance and path planning, which compose the essential parts of visual navigation. Furthermore, throughout this article, we will have an insight into the prospect of the UAV navigation and the challenges to be faced.

Coronavirus disease 2019 (COVID-19) pandemic has led to a rare reduction in human activities. In such a background, data from ground-based environmental stations, satellites, and reanalysis materials are utilized to conduct a comprehensive analysis of the global air quality changes during the COVID-19 outbreak. The results showed that under the impact of the COVID-19 outbreak, a significant decrease in particulate matter (PM _x ) and nitrogen dioxide (NO _2 ) occurred in more than 40% of the world’s land area, with NO _2 (PM _x ) decreasing by ∼30% (∼20%). The mobility, meteorological factors, and the response speed to COVID-19 outbreaks were examined. It was further found that in quick-response cities, lockdowns produced a sharp decline in mobility and had a dominant impact on air quality. In contrast, in slow-response cities, mobility dropped gradually since the confirmation of the first COVID-19 case (FCC) and he impact of the FCC, lockdowns, and meteorological factors were comparable.

Supervised classification is the commonly used method for extracting ground information from images. However, for supervised classification, the selection and labelling of training samples is an expensive and time-consuming task. Recently, automatic information indexes have achieved satisfactory results for indicating different land-cover classes, which makes it possible to develop an automatic method for labelling the training samples instead of manual interpretation. In this paper, we propose a method for the automatic selection and labelling of training samples for high-resolution image classification. In this way, the initial candidate training samples can be provided by the information indexes and open-source geographical information system (GIS) data, referring to the representative land-cover classes: buildings, roads, soil, water, shadow, and vegetation. Several operations are then applied to refine the initial samples, including removing overlaps, removing borders, and semantic constraints. The proposed sampling method is evaluated on a series of high-resolution remote sensing images over urban areas, and is compared to classification with manually labeled training samples. It is found that the proposed method is able to provide and label a large number of reliable samples, and can achieve satisfactory results for different classifiers. In addition, our experiments show that active learning can further enhance the classification performance, as active learning is used to choose the most informative samples from the automatically labeled samples.

Learning efficient image representations is at the core of the scene classification task of remote sensing imagery. The existing methods for solving the scene classification task, based on either feature coding approaches with low-level hand-engineered features or unsupervised feature learning, can only generate mid-level image features with limited representative ability, which essentially prevents them from achieving better performance. Recently, the deep convolutional neural networks (CNNs), which are hierarchical architectures trained on large-scale datasets, have shown astounding performance in object recognition and detection. However, it is still not clear how to use these deep convolutional neural networks for high-resolution remote sensing (HRRS) scene classification. In this paper, we investigate how to transfer features from these successfully pre-trained CNNs for HRRS scene classification. We propose two scenarios for generating image features via extracting CNN features from different layers. In the first scenario, the activation vectors extracted from fully-connected layers are regarded as the final image features; in the second scenario, we extract dense features from the last convolutional layer at multiple scales and then encode the dense features into global image features through commonly used feature coding approaches. Extensive experiments on two public scene classification datasets demonstrate that the image features obtained by the two proposed scenarios, even with a simple linear classifier, can result in remarkable performance and improve the state-of-the-art by a significant margin. The results reveal that the features from pre-trained CNNs generalize well to HRRS datasets and are more expressive than the low- and mid-level features. Moreover, we tentatively combine features extracted from different CNN models for better performance.

Scene classification, which consists of assigning images with semantic labels by exploiting the local spatial arrangements and structural patterns inside tiled regions, is a key problem in the automatic interpretation of optical high-spatial resolution remote sensing imagery. Many state-of-the-art methods, e.g., the bag-of-visual-words model and its variants, the topic models and unsupervised feature learning-based approaches, share similar procedures: patch sampling, feature learning and classification. Patch sampling is the first and a key procedure, and it has a considerable influence on the results. In the literature, many different sampling strategies have been used, e.g., random sampling and saliency-based sampling. However, the sampling strategy that is most suitable for the scene classification of optical high-spatial resolution remote sensing images remains unclear. In this paper, we comparatively study the effects of different sampling strategies under the scenario of scene classification of optical high-spatial resolution remote sensing images. We divide the existing sampling methods into two types: random sampling and saliency-based sampling. Here, we consider the commonly-used grid sampling to be a specific type of random sampling method, and the saliency-based sampling consists of keypoint-based sampling and salient region-based sampling. To compare their performances, we rely on a standard bag-of-visual-words model to learn the global features for testing because of its simplicity, robustness and efficiency. In addition, we conduct experiments using a Fisher kernel framework to validate our conclusions. The experimental results obtained on two commonly-used datasets using different feature learning methods show that random sampling can provide comparable and even better performance than all of the saliency-based strategies.

It is a challenging problem to efficiently interpret the large volumes of remotely sensed image data being collected in the current age of remote sensing “big data”. Although human visual interpretation can yield accurate annotation of remote sensing images, it demands considerable expert knowledge and is always time-consuming, which strongly hinders its efficiency. Alternatively, intelligent approaches (e.g., supervised classification and unsupervised clustering) can speed up the annotation process through the application of advanced image analysis and data mining technologies. However, high-quality expert-annotated samples are still a prerequisite for intelligent approaches to achieve accurate results. Thus, how to efficiently annotate remote sensing images with little expert knowledge is an important and inevitable problem. To address this issue, this paper introduces a novel active clustering method for the annotation of high-resolution remote sensing images. More precisely, given a set of remote sensing images, we first build a graph based on these images and then gradually optimize the structure of the graph using a cut-collect process, which relies on a graph-based spectral clustering algorithm and pairwise constraints that are incrementally added via active learning. The pairwise constraints are simply similarity/dissimilarity relationships between the most uncertain pairwise nodes on the graph, which can be easily determined by non-expert human oracles. Furthermore, we also propose a strategy to adaptively update the number of classes in the clustering algorithm. In contrast with existing methods, our approach can achieve high accuracy in the task of remote sensing image annotation with relatively little expert knowledge, thereby greatly lightening the workload burden and reducing the requirements regarding expert knowledge. Experiments on several datasets of remote sensing images show that our algorithm achieves state-of-the-art performance in the annotation of remote sensing images and demonstrates high potential in many practical remote sensing applications.

As the largest developing country, China has experienced dramatic urban expansion since the “reform and opening-up” policy started at the end of the 1970s. In this paper, we monitor three decades of urban expansion in China’s 36 major cities, based on the spectral mixture analysis of remotely sensed satellite images. The results demonstrated that these major cities have expanded by 5.85 times from 1986 to 2015, with 15.51 km2 average expansion area per city per year. We found the urban expansion trajectories showed three different modes, i.e., exponential, linear and s-shaped, which were closely related to the city development level. In the old city zones, however, there was an interesting common tendency of the impervious surface area (ISA) first increasing and then decreasing, which could be largely attributed to the phenomenon of urban village reconstruction in China’s cities. Based on the Granger Causality Test (GCT), the interaction between urban ISA and Gross Domestic Product (GDP) per capita (GDPPC) suggested that the former was the driver of the latter. Meanwhile, taking the Yangtze River as the division between north and south China, there exists a north−south territorial differentiation for the interaction between ISA and total population at the year-end (TP).

This study proposes a novel method for multichannel image gray level co-occurrence matrix (GLCM) texture representation. It is well known that the standard procedure for the automatic extraction of GLCM textures is based on a mono-spectral image. In real applications, however, the GLCM texture feature extraction always refers to multi/hyperspectral images. The widely used strategy to deal with this issue is to calculate the GLCM from the first principal component or the panchromatic band, which do not include all the useful information. Accordingly, in this study, we propose to represent the multichannel textures for multi/hyperspectral imagery by the use of: (1) clustering algorithms; and (2) sparse representation, respectively. In this way, the multi/hyperspectral images can be described using a series of quantized codes or dictionaries, which are more suitable for multichannel texture representation than the traditional methods. Specifically, K-means and fuzzy c-means methods are adopted to generate the codes of an image from the clustering point of view, while a sparse dictionary learning method based on two coding rules is proposed to produce the texture primitives. The proposed multichannel GLCM textural extraction methods were evaluated with four multi/hyperspectral datasets: GeoEye-1 and QuickBird multispectral images of the city of Wuhan, the well-known AVIRIS hyperspectral dataset from the Indian Pines test site, and the HYDICE airborne hyperspectral dataset from the Washington DC Mall. The results show that both the clustering-based and sparsity-based GLCM textures outperform the traditional method (extraction based on the first principal component) in terms of classification accuracies in all the experiments.

In this paper, a blind restoration method is presented to remove the blur in remote sensing images. An alternating minimization (AM) framework is employed to simultaneously recover the image and the point spread function (PSF), and an adaptive-norm prior is used to apply different constraints to smooth regions and edges. Moreover, with the use of the knife-edge features in remote sensing images, an automatic knife-edge detection method is used to obtain a good initial PSF for the AM framework. In addition, a no-reference (NR) sharpness index is used to stop the iterations of the AM framework automatically at the best visual quality. Results in both simulated and real data experiments indicate that the proposed AM-KEdge method, which combines the automatic knife-edge detection and the AM framework, is robust, converges quickly, and can stop automatically to obtain satisfactory results.

In this study, a new clustering-based feature extraction algorithm is proposed for the spectral-spatial classification of hyperspectral imagery. The clustering approach is able to group the high-dimensional data into a subspace by mining the salient information and suppressing the redundant information. In this way, the relationship between neighboring pixels, which was hidden in the original data, can be extracted more effectively. Specifically, in the proposed algorithm, a two-step process is adopted to make use of the clustering-based information. A clustering approach is first used to produce the initial clustering map, and, subsequently, a multiscale cluster histogram (MCH) is proposed to represent the spatial information around each pixel. In order to evaluate the robustness of the proposed MCH, four clustering techniques are employed to analyze the influence of the clustering methods. Meanwhile, the performance of the MCH is compared to three other widely used spatial features: the gray-level co-occurrence matrix (GLCM), the 3D wavelet texture, and differential morphological profiles (DMPs). The experiments conducted on four well-known hyperspectral datasets verify that the proposed MCH can significantly improve the classification accuracy, and it outperforms other commonly used spatial features.

Cloud cover is a common problem in optical satellite imagery, which leads to missing information in images as well as a reduction in the data usability. In this paper, a thick cloud removal method based on stepwise radiometric adjustment and residual correction (SRARC) is proposed, which is aimed at effectively removing the clouds in high-resolution images for the generation of high-quality and spatially contiguous urban geographical maps. The basic idea of SRARC is that the complementary information in adjacent temporal satellite images can be utilized for the seamless recovery of cloud-contaminated areas in the target image after precise radiometric adjustment. To this end, the SRARC method first optimizes the given cloud mask of the target image based on superpixel segmentation, which is conducted to ensure that the labeled cloud boundaries go through homogeneous areas of the target image, to ensure a seamless reconstruction. Stepwise radiometric adjustment is then used to adjust the radiometric information of the complementary areas in the auxiliary image, step by step, and clouds in the target image can be removed by the replacement with the adjusted complementary areas. Finally, residual correction based on global optimization is used to further reduce the radiometric differences between the recovered areas and the cloud-free areas. The final cloud removal results are then generated. High-resolution images with different spatial resolutions and land-cover change patterns were used in both simulated and real-data cloud removal experiments. The results suggest that SRARC can achieve a better performance than the other compared methods, due to the superiority of the radiometric adjustment and spatial detail preservation. SRARC is thus a promising approach that has the potential for routine use, to support applications based on high-resolution satellite images.

Flow measurements in open channels have often utilized velocity-area methods. Thus, estimations of the average velocity in a cross-section of rural canals play an important role in the flow measurement of an irrigation district. This paper derives a model for calculating depth average velocity. This model considers the classical logarithmic formula describing the velocity distribution and flow partitioning theory, which is aimed at finding out a location that represents the depth average velocity (LDAV) along the vertical line from boundary to water surface. Subsequently, the average flow velocity of the whole channel can be further determined by using the velocity-area method in different regions. Moreover, the LDAV has different expressions in different sub-regions according to flow partitioning theory under various aspect ratios. The results are verified by experiments under different experimental conditions, and the formula is highly applicable and has a high theoretical significance and practical value.

Vegetation water content (VWC) is recognized as an important parameter in vegetation growth studies, natural disasters such as forest fires, and drought prediction. Recently, the Global Navigation Satellite System Interferometric Reflectometry (GNSS-IR) has emerged as an important technique for monitoring vegetation information. The normalized microwave reflection index (NMRI) was developed to reflect the change of VWC based on this fact. However, NMRI uses local site-based data, and the sparse distribution hinders the application of NMRI. In this study, we obtained a 500 m spatially continuous NMRI product by integrating GNSS-IR site data with other VWC-related products using the point–surface fusion technique. The auxiliary data in the fusion process include the normalized difference vegetation index (NDVI), gross primary productivity (GPP), and precipitation. Meanwhile, the fusion performance of three machine learning methods, i.e., the back-propagation neural network (BPNN), generalized regression neural network (GRNN), and random forest (RF) are compared and analyzed. The machine learning methods achieve satisfactory results, with cross-validation R values of 0.71–0.83 and RMSEs of 0.025–0.037. The results show a clear improvement over the traditional multiple linear regression method, which achieves R (RMSE) values of only about 0.4 (0.045). It indicates that the machine learning methods can better learn the complex nonlinear relationship between NMRI and the input VWC-related index. Among the machine learning methods, the RF model obtained the best results. Long time-series NMRI images with a 500 m spatial resolution in the western part of the continental U.S. were then obtained. The results show that the spatial distribution of the NMRI product is consistent with a drought situation from 2012 to 2014 in the U.S., which verifies the feasibility of analyzing and predicting drought times and distribution ranges by using the 500 m fusion product.

Multi-angle remote sensing images are acquired over the same imaging scene from different angles, and share similar but not identical information. It is therefore possible to enhance the spatial resolution of the multi-angle remote sensing images by the super-resolution reconstruction technique. However, different sensor shooting angles lead to different resolutions for each angle image, which affects the effectiveness of the super-resolution reconstruction of the multi-angle images. In view of this, we propose utilizing adaptive weighted super-resolution reconstruction to alleviate the limitations of the different resolutions. This paper employs two adaptive weighting themes. The first approach uses the angle between the imaging angle of the current image and that of the nadir image. The second is closely related to the residual error of each low-resolution angle image. The experimental results confirm the feasibility of the proposed method and demonstrate the effectiveness of the proposed adaptive weighted super-resolution approach.

By using the high spectral resolution, hyperspectral images (HSIs) provide significant information for target detection, which is of great interest in HSI processing. However, most classical target detection methods may only perform well based on certain assumptions. Simultaneously, using limited numbers of target samples and preserving the discriminative information is also a challenging problem in hyperspectral target detection. To overcome these shortcomings, this paper proposes a novel adaptive information-theoretic metric learning with local constraints (ITML-ALC) for hyperspectral target detection. The proposed method firstly uses the information-theoretic metric learning (ITML) method as the objective function for learning a Mahalanobis distance to separate similar and dissimilar point-pairs without certain assumptions, needing fewer adjusted parameters. Then, adaptively local constraints are applied to shrink the distances between samples of similar pairs and expand the distances between samples of dissimilar pairs. Finally, target detection decision can be made by considering both the threshold and the changes between the distances before and after metric learning. Experimental results demonstrate that the proposed method can obviously separate target samples from background ones and outperform both the state-of-the-art target detection algorithms and the other classical metric learning methods.