Search

Despite the extensive body of literature focused on remote sensing applications for land cover mapping and the availability of high-resolution satellite imagery, methods for continuously updating classification maps in real-time remain limited, especially when training data is scarce. This paper introduces the recursive Bayesian classifier (RBC), which converts any instantaneous classifier into a robust online method through a probabilistic framework that is resilient to non-informative image variations. Three experiments are conducted using Sentinel-2 data: water mapping of the Oroville Dam in California and the Charles River basin in Massachusetts, and deforestation detection in the Amazon. RBC is applied to a Gaussian mixture model (GMM), logistic regression (LR), and our proposed spectral index classifier (SIC). Results show that RBC significantly enhances classifier robustness in multitemporal settings under challenging conditions, such as cloud cover and cyanobacterial blooms. Specifically, balanced classification accuracy improves by up to 26.95% for SIC, 12.4% for GMM, and 13.81% for LR in water mapping, and by 15.25%, 14.17%, and 14.7% in deforestation detection. Moreover, without additional training data, RBC improves the performance of the state-of-the-art DeepWaterMap and WatNet algorithms by up to 9.62% and 11.03%. These benefits are provided by RBC while requiring minimal supervision and maintaining a low computational cost that remains constant for each time step regardless of the time-series length. [Display omitted] [ABSTRACT FROM AUTHOR]

Forest cover in Indonesia reached 95.97 million hectares or 51.2% of the total land area. Fluctuations in forest cover from 2014 to 2022 highlight the need for accurate land mapping to understand factors influencing landscape patterns. Decreases in forest cover increase vulnerability to natural disasters, particularly floods, causing significant economic and environmental losses. This study focuses on strategies for land cover mapping, especially in under-monitored regions, leveraging recent advancements in satellite and drone data for rapid and precise mapping. Deep learning methods like convolutional neural networks (CNN) have shown high accuracy in vegetation cover detection, which can be further improved with new activation functions. This study proposes the new modification activation function, with adding a linear component to enhance CNN accuracy and effectiveness. The proposed demonstrated a significant increase in accuracy, maintaining high values during testing (90.07%) on the Vegetation Density of Peatland Drone dataset. Statistical analysis showed high data distribution with significant performance variation. Despite slower computation times, the proposed excelled in convergence speed and effectiveness compared to other activation. ANOVA and Friedman analyses confirmed significant performance and computation time differences among activation methods. The proposed shows excellent potential in enhancing CNN model accuracy and stability for land cover classification. [ABSTRACT FROM AUTHOR]

This study was designed to develop a 30 m resolution land cover dataset to improve the performance of regional weather forecasting models in East China. A 10-class land cover mapping scheme was established, reflecting East China's diverse landscape characteristics and incorporating a new category for plastic greenhouses. Plastic greenhouses are key to understanding surface heterogeneity in agricultural regions, as they can significantly impact local climate conditions, such as heat flux and evapotranspiration, yet they are often not represented in conventional land cover classifications. This is mainly due to the lack of high-resolution datasets capable of detecting these small yet impactful features. For the six-province study area, we selected and processed Landsat 8 imagery from 2015–2018, filtering for cloud cover. Complementary datasets, such as digital elevation models (DEM) and nighttime lighting data, were integrated to enrich the inputs for the Random Forest classification. A comprehensive training dataset was compiled to support Random Forest training and classification accuracy. We developed an automated workflow to manage the data processing, including satellite image selection, preprocessing, classification, and image mosaicking, thereby ensuring the system's practicality and facilitating future updates. We included three Weather Research and Forecasting (WRF) model experiments in this study to highlight the impact of our land cover maps on daytime and nighttime temperature predictions. The resulting regional land cover dataset achieved an overall accuracy of 83.2% and a Kappa coefficient of 0.81. These accuracy statistics are higher than existing national and global datasets. The model results suggest that the newly developed land cover, combined with a mosaic option in the Unified Noah scheme in WRF, provided the best overall performance for both daytime and nighttime temperature predictions. In addition to supporting the WRF model, our land cover map products, with a planned 3–5-year update schedule, could serve as a valuable data source for ecological assessments in the East China region, informing environmental policy and promoting sustainability. [ABSTRACT FROM AUTHOR]

Cloud coverage poses a significant challenge to optical image interpretation, degrading ground information on Earth's surface. Synthetic aperture radar (SAR), with its ability to penetrate clouds, provides supplementary information to optical data. However, existing optical-SAR fusion methods predominantly focus on cloud-free scenarios, neglecting the practical challenge of semantic segmentation under cloudy conditions. To tackle this issue, we propose CloudSeg, a novel framework tailored for land cover mapping in the presence of clouds. It addresses the challenges posed by cloud cover from two aspects: reducing semantic ambiguity in areas of the cloudy image that are obscured by clouds and enhancing effective information in the unobstructed portions. Specifically, CloudSeg employs a multi-task learning strategy to simultaneously handle low-level visual task and high-level semantic understanding task, mitigating the semantic ambiguity caused by cloud cover by acquiring discriminative features through an auxiliary cloud removal task. Additionally, CloudSeg incorporates a knowledge distillation strategy, which utilizes the knowledge learned by the teacher network under cloud-free conditions to guide the student network to overcome the interference of cloud-covered areas, enhancing the valuable information from the unobstructed parts of cloud-covered images. Extensive experiments conducted on two datasets, M3M-CR and WHU-OPT-SAR , demonstrate the effectiveness and superiority of the proposed CloudSeg method for land cover mapping under cloudy conditions. Specifically, CloudSeg outperforms the state-of-the-art competitors by 3.16% in terms of mIoU on M3M-CR and by 5.56% on WHU-OPT-SAR , highlighting its substantial advantages for analyzing regions frequently obscured by clouds. Codes are available at https://github.com/xufangchn/CloudSeg. [ABSTRACT FROM AUTHOR]

Accurate land cover mapping is crucial for various applications, from environmental monitoring to urban planning. Traditional methods often struggle with high-dimensional data and complex landscape features. This study integrates RCNN (Region-based Convolutional Neural Network) and ANT Colony Optimization (ACO) to enhance land cover mapping accuracy. RCNN is utilized for precise segmentation of high-resolution satellite imagery, while ACO is employed for effective feature extraction, leveraging the algorithm's ability to identify and optimize features in the presence of complex patterns. Our method was evaluated using a dataset of 500 km², achieving a segmentation accuracy of 92.5% and a feature extraction precision improvement of 18.3% compared to conventional techniques. The integration of RCNN and ACO demonstrates significant advancements in capturing detailed land cover information and improving overall mapping accuracy. [ABSTRACT FROM AUTHOR]

Recent advancements in deep learning have spurred the development of numerous novel semantic segmentation models for land cover mapping, showcasing exceptional performance in delineating precise boundaries and producing highly accurate land cover maps. However, to date, no systematic literature review has comprehensively examined semantic segmentation models in the context of land cover mapping. This paper addresses this gap by synthesizing recent advancements in semantic segmentation models for land cover mapping from 2017 to 2023, drawing insights on trends, data sources, model structures, and performance metrics based on a review of 106 articles. Our analysis identifies top journals in the field, including MDPI Remote Sensing, IEEE Journal of Selected Topics in Earth Science, and IEEE Transactions on Geoscience and Remote Sensing, IEEE Geoscience and Remote Sensing Letters, and ISPRS Journal Of Photogrammetry And Remote Sensing. We find that research predominantly focuses on land cover, urban areas, precision agriculture, environment, coastal areas, and forests. Geographically, 35.29% of the study areas are located in China, followed by the USA (11.76%), France (5.88%), Spain (4%), and others. Sentinel-2, Sentinel-1, and Landsat satellites emerge as the most used data sources. Benchmark datasets such as ISPRS Vaihingen and Potsdam, LandCover.ai, DeepGlobe, and GID datasets are frequently employed. Model architectures predominantly utilize encoder–decoder and hybrid convolutional neural network-based structures because of their impressive performances, with limited adoption of transformer-based architectures due to its computational complexity issue and slow convergence speed. Lastly, this paper highlights existing key research gaps in the field to guide future research directions. [ABSTRACT FROM AUTHOR]

Airborne hyperspectral images can be used to map the land cover in large urban areas, thanks to their very high spatial and spectral resolutions on a wide spectral domain. While the spectral dimension of hyperspectral images is highly informative of the chemical composition of the land surface, the use of state-of-the-art machine learning algorithms to map the land cover has been dramatically limited by the availability of training data. To cope with the scarcity of annotations, semi-supervised and self-supervised techniques have lately raised a lot of interest in the community. Yet, the publicly available hyperspectral data sets commonly used to benchmark machine learning models are not totally suited to evaluate their generalization performances due to one or several of the following properties: a limited geographical coverage (which does not reflect the spectral diversity in metropolitan areas), a small number of land cover classes and a lack of appropriate standard train/test splits for semi-supervised and self-supervised learning. Therefore, we release in this paper the Toulouse Hyperspectral Data Set that stands out from other data sets in the above-mentioned respects in order to meet key issues in spectral representation learning and classification over large-scale hyperspectral images with very few labeled pixels. Besides, we discuss and experiment self-supervised techniques for spectral representation learning, including the Masked Autoencoder (He et al., 2022), and establish a baseline for pixel-wise classification achieving 85% overall accuracy and 77% F1 score. The Toulouse Hyperspectral Data Set and our code are publicly available at www.toulouse-hyperspectral-data-set.com and www.github.com/Romain3Ch216/tlse-experiments/ , respectively. [ABSTRACT FROM AUTHOR]

High-resolution land cover mapping is crucial in various disciplines but is often hindered by the lack of accurately matched labels. Our study introduces an innovative deep learning methodology for effective land cover mapping, independent of matched labels. The approach comprises three main components: (1) An advanced fully convolutional neural network, augmented with super-resolution features, to refine labels; (2) The application of an instance-batch normalization network (IBN), leveraging these enhanced labels from the source domain, to generate 2-m resolution land cover maps for test sites in the target domain; (3) Noise assessment tests to evaluate the impact of varying noise levels on the model's mapping accuracy using external labels. The model achieved an overall accuracy of 83.40% in the target domain using endogenous super-resolution labels. In contrast, employing exogenous, high-precision labels from the National Land Cover Database in the source domain led to a notable accuracy increase of 2.55%, reaching 85.48%. This improvement highlights the model's enhanced generalizability and performance during domain shifts, attributed significantly to the IBN layer. Our findings reveal that, despite the absence of native high-precision labels, the utilization of high-quality external labels can substantially benefit the development of precise land cover mapping, underscoring their potential in scenarios with unmatched labels. [ABSTRACT FROM AUTHOR]

Over the coming century, climate change and sea-level rise are predicted to cause widespread change to coastal wetlands. Estuarine vegetated wetlands can adapt to sea-level rise through both vertical development (i.e., biophysical feedbacks and sedimentation) and upslope/horizontal migration. Quantifying changes to estuarine vegetated wetlands over time can help to inform current and future decisions regarding land management and resource stewardship. In this study, we show how coastal land cover maps readily available in the US can be used to assess and understand estuarine vegetated wetland changes. This assessment involves two steps: (1) identifying the net gain/loss of estuarine vegetated wetlands and (2) determining which land cover types contribute to the net gain/loss. From this information, we developed estuarine vegetated wetland change scenarios that evaluate whether estuarine vegetated wetland gain kept up with loss and whether the contribution was from: (1) estuarine vegetated wetland migration or tidal restoration; (2) land building (i.e., development); or (3) both. We assessed changes from 1996 to 2016 for: (1) the conterminous US; (2) each major US coastline; and (3) focal estuaries with the most change per coast. We found that the change scenario (1, 2, or 3) varied across coastlines. Moving forward, national coastal land cover programs can be informed by utilizing methodologies that leverage contemporary information for delineating the estuarine zone from upslope/adjacent wetlands. We highlight approaches that could be used to address this challenge and provide complementary information related to wetland condition changes. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Architectural Sciences & Applications (JASA) is the property of Journal of Architectural Sciences & Applications and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The images of the Sentinel-2 constellation can help the verification process of farmers' declarations, providing, among other things, accurate spatial explicit maps of the agricultural land cover. The aim of the study is to design, develop, and evaluate two deep learning (DL) architectures tailored for agricultural land cover and crop type mapping. The focus is on a detailed class scheme encompassing fifteen distinct classes, utilizing Sentinel-2 imagery acquired on a monthly basis throughout the year. The study's geographical scope covers a diverse rural area in North Greece, situated within southeast Europe. These architectures are a Temporal Convolutional Neural Network (CNN) and a combination of a Recurrent and a 2D Convolutional Neural Network (R-CNN), and their accuracy is compared to the well-established Random Forest (RF) machine learning algorithm. The comparative approach is not restricted to simply presenting the results given by classification metrics, but it also assesses the uncertainty of the classification results using an entropy measure and the spatial distribution of the classification errors. Furthermore, the issue of sampling strategy for the extraction of the training set is highlighted, targeting the efficient handling of both the imbalance of the dataset and the spectral variability of instances among classes. The two developed deep learning architectures performed equally well, presenting an overall accuracy of 90.13% (Temporal CNN) and 90.18% (R-CNN), higher than the 86.31% overall accuracy of the RF approach. Finally, the Temporal CNN method presented a lower entropy value (6.63%), compared both to R-CNN (7.76%) and RF (28.94%) methods, indicating that both DL approaches should be considered for developing operational EO processing workflows. [ABSTRACT FROM AUTHOR]

Artificial intelligence (AI) and machine learning (ML) have been applied to solve various remote sensing problems. To fully leverage the power of AI and ML to tackle impactful remote sensing problems, it is essential to enable researchers and practitioners to understand how AI and ML models actually work and thus to improve the model performance strategically. Accurate and timely land cover maps are essential components for informed land management decision making. To address the ever-increasing need for high spatial and temporal resolution maps, this paper developed an interactive and open-source online tool, in Python, to help interpret and improve the ML models used for land cover mapping with Google Earth Engine (GEE). The tool integrates the workflow of both land cover classification and land cover change dynamics, which requires the generation of a time series of land cover maps. Three feature importance metrics are reported, including impurity-based, permutation-based, and SHAP (Shapley additive explanations) value-based feature importance. Two case studies are presented to showcase the tool's capability and ease of use, enabling a globally accessible and free convergent application of remote sensing technologies. This tool may inspire researchers to facilitate explainable AI (XAI)-empowered remote sensing applications with GEE. [ABSTRACT FROM AUTHOR]

Understanding the spatial distribution of irrigated croplands is crucial for food security and water use. To map land cover classes with high-spatial-resolution images, it is necessary to analyze the semantic information of target objects in addition to the spectral or spatial–spectral information of local pixels. Deep convolutional neural networks (DCNNs) can characterize the semantic features of objects adaptively. This study uses DCNNs to extract irrigated croplands from Sentinel-2 images in the states of Washington and California in the United States. We integrated the DCNNs of 101 layers, discarded pooling layers, and employed dilation convolution to preserve location information; these are models which were used based on fully convolutional network (FCN) architectures. The findings indicated that irrigated croplands may be effectively detected at various phases of crop growth in the fields. A quantitative analysis of the trained models revealed that the three models in the two states had the lowest values of Intersection over Union (IoU) and Kappa, i.e., 0.88 and 0.91, respectively. The deep models' temporal portability across different years was acceptable. The lowest values of recall and OA (overall accuracy) from 2018 to 2021 were 0.91 and 0.87, respectively. In Washington, the lowest OA value from 10 to 300 m resolution was 0.76. This study demonstrates the potential of FCNs + DCNNs approaches for mapping irrigated croplands across large regions, providing a solution for irrigation mapping. The spatial resolution portability of deep models could be improved further by designing model architectures. [ABSTRACT FROM AUTHOR]

Remote sensing images are important data sources for land cover mapping. As one of the most important artificial features in remote sensing images, buildings play a critical role in many applications, such as population estimation and urban planning. Classifying buildings quickly and accurately ensures the reliability of the above applications. It is known that the classification accuracy of buildings (usually indicated by a comprehensive index called F1) is greatly affected by image quality. However, how image quality affects building classification accuracy is still unclear. In this study, Boltzmann entropy (an index considering both compositional and configurational information, simply called BE) is employed to describe image quality, and the potential relationships between BE and F1 are explored based on images from two open-source building datasets (i.e., the WHU and Inria datasets) in three cities (i.e., Christchurch, Chicago and Austin). Experimental results show that (1) F1 fluctuates greatly in images where building proportions are small (especially in images with building proportions smaller than 1%) and (2) BE has a negative relationship with F1 (i.e., when BE becomes larger, F1 tends to become smaller). The negative relationships are confirmed using Spearman correlation coefficients (SCCs) and various confidence intervals via bootstrapping (i.e., a nonparametric statistical method). Such discoveries are helpful in deepening our understanding of how image quality affects building classification accuracy. [ABSTRACT FROM AUTHOR]

Satellite hyperspectral imagery is an important data source for large-scale refined land cover classification and mapping, but the high spatial heterogeneity and spectral variability at low spatial resolution and the high computation cost for massive data remain challenges in the research community. In recent years, convolutional neural network (CNN) models with the capability for feature extraction have been widely used in hyperspectral image classification. However, incomplete feature extraction, inappropriate feature fusion, and high time consumption are still the major problems for CNN applications in large-scale fine land cover mapping. In this study, a Spatial-Convolution Spectral-Transformer Interactive Network (SCSTIN) was proposed to integrate 2D-CNN and Transformer into a dual-branch network to enhance feature extraction capabilities by exploring spatial context information and spectral sequence signatures in a targeted manner. In addition, spatial-spectral interactive fusion (SSIF) units and category-adaptive weighting (CAW) as two feature fusion modules were also adopted between and after the two feature extraction branches to improve efficiency in feature fusion. The ZY1-02D hyperspectral imagery was collected to conduct the experiments in the study area of the eastern foothills of the Helan Mountains (EFHLM), covering an area of about 8800 km2, which is the largest hyperspectral dataset as far as we know. To explore the potential of the proposed network in terms of accuracy and efficiency, SCSTIN models with different depths (SCSTIN-4 and SCSTIN-2) were performed. The results suggest that compared with the previous eight advanced hyperspectral image classifiers, both SCSTIN models achieved satisfactory performance in accuracy and efficiency aspects with low complexity, where SCSTIN-4 achieved the highest accuracy and SCSTIN-2 obtained higher efficiency. Accordingly, the SCSTIN models are reliable for large-scale fast refined land cover classification and mapping. In addition, the spatial distribution pattern of diverse ground objects in EFHLM is also analyzed. [ABSTRACT FROM AUTHOR]

Land cover maps are a vital input variable to many types of environmental research and management. While they can be produced automatically by machine learning techniques, these techniques require substantial training data to achieve high levels of accuracy, which are not always available. One technique researchers use when labelled training data are scarce is domain adaptation (DA)—where data from an alternate region, known as the source domain, are used to train a classifier and this model is adapted to map the study region, or target domain. The scenario we address in this paper is known as semi-supervised DA, where some labelled samples are available in the target domain. In this paper we present Sourcerer, a Bayesian-inspired, deep learning-based, semi-supervised DA technique for producing land cover maps from satellite image time series (SITS) data. The technique takes a convolutional neural network trained on a source domain and then trains further on the available target domain with a novel regularizer applied to the model weights. The regularizer adjusts the degree to which the model is modified to fit the target data, limiting the degree of change when the target data are few in number and increasing it as target data quantity increases. Our experiments on Sentinel-2 time series images compare Sourcerer with two state-of-the-art semi-supervised domain adaptation techniques and four baseline models. We show that on two different source-target domain pairings Sourcerer outperforms all other methods for any quantity of labelled target data available. In fact, the results on the more difficult target domain show that the starting accuracy of Sourcerer (when no labelled target data are available), 74.2%, is greater than the next-best state-of-the-art method trained on 20,000 labelled target instances. [ABSTRACT FROM AUTHOR]

Satellite imagery is the only feasible approach to annual monitoring and reporting on land cover change. Unfortunately, conventional pixel-based classification methods based on spectral response only (e.g., using random forests algorithms) have shown a lack of spatial and temporal stability due, for instance, to variability between individual pixels and changes in vegetation condition, respectively. Machine learning methods that consider spatial patterns in addition to reflectance can address some of these issues. In this study, a convolutional neural network (CNN) model, U-Net, was trained for a 500 km × 500 km region in southeast Australia using annual Landsat geomedian data for the relatively dry and wet years of 2018 and 2020, respectively. The label data for model training was an eight-class classification inferred from a static land-use map, enhanced using forest-extent mapping. Here, we wished to analyse the benefits of CNN-based land cover mapping and reporting over 34 years (1987–2020). We used the trained model to generate annual land cover maps for a 100 km × 100 km tile near the Australian Capital Territory. We developed innovative diagnostic methods to assess spatial and temporal stability, analysed how the CNN method differs from pixel-based mapping and compared it with two reference land cover products available for some years. Our U-Net CNN results showed better spatial and temporal stability with, respectively, overall accuracy of 89% verses 82% for reference pixel-based mapping, and 76% of pixels unchanged over 33 years. This gave a clearer insight into where and when land cover change occurred compared to reference mapping, where only 30% of pixels were conserved. Remaining issues include edge effects associated with the CNN method and a limited ability to distinguish some land cover types (e.g., broadacre crops vs. pasture). We conclude that the CNN model was better for understanding broad-scale land cover change, use in environmental accounting and natural resource management, whereas pixel-based approaches sometimes more accurately represented small-scale changes in land cover. [ABSTRACT FROM AUTHOR]

High-resolution satellite images can provide abundant, detailed spatial information for land cover classification, which is particularly important for studying the complicated built environment. However, due to the complex land cover patterns, the costly training sample collections, and the severe distribution shifts of satellite imageries caused by, e.g., geographical differences or acquisition conditions, few studies have applied high-resolution images to land cover mapping in detailed categories at large scale. To fill this gap, we present a large-scale land cover dataset, Five-Billion-Pixels. It contains more than 5 billion labeled pixels of 150 high-resolution Gaofen-2 (4 m) satellite images, annotated in a 24-category system covering artificial-constructed , agricultural , and natural classes. In addition, we propose a deep-learning-based unsupervised domain adaptation approach that can transfer classification models trained on labeled dataset (referred to as the source domain) to unlabeled data (referred to as the target domain) for large-scale land cover mapping. Specifically, we introduce an end-to-end Siamese network employing dynamic pseudo-label assignment and class balancing strategy to perform adaptive domain joint learning. To validate the generalizability of our dataset and the proposed approach across different sensors and different geographical regions, we carry out land cover mapping on five megacities in China and six cities in other five Asian countries severally using: PlanetScope (3 m), Gaofen-1 (8 m), and Sentinel-2 (10 m) satellite images. Over a total study area of 60,000 km2, the experiments show promising results even though the input images are entirely unlabeled. The proposed approach, trained with the Five-Billion-Pixels dataset, enables high-quality and detailed land cover mapping across the whole country of China and some other Asian countries at meter-resolution. [ABSTRACT FROM AUTHOR]

Pan Sharpening is normally applied to sharpen a multispectral image with low resolution by using a panchromatic image with a higher resolution, to generate a high resolution multispectral image. The present study aims at assessing the power of Pan Sharpening on improvement of the accuracy of image classification and land cover mapping in Landsat 8 OLI imagery. In this respect, different Pan Sharpening algorithms including Brovey, Gram-Schmidt, NNDiffuse, and Principal Components were applied to merge the Landsat OLI panchromatic band (15 m) with the Landsat OLI multispectral: visible and infrared bands (30 m), to generate a new multispectral image with a higher spatial resolution (15 m). Subsequently, the support vector machine approach was utilized to classify the original Landsat and resulting Pan Sharpened images to generate land cover maps of the study area. The outcomes were then compared through the generation of confusion matrix and calculation of kappa coefficient and overall accuracy. The results indicated superiority of NNDiffuse algorithm in Pan Sharpening and improvement of classification accuracy in Landsat OLI imagery, with an overall accuracy and kappa coefficient of about 98.66% and 0.98, respectively. Furthermore, the result showed that the Gram-Schmidt and Principal Components algorithms also slightly improved the accuracy of image classification compared to original Landsat image. The study concluded that image Pan Sharpening is useful to improve the accuracy of image classification in Landsat OLI imagery, depending on the Pan Sharpening algorithm used for this purpose. [ABSTRACT FROM AUTHOR]

Accurate land cover mapping is challenging in Southeast Asia where cloud coverage is prevalent and landscape is heterogenous. Object-based mapping, multi-temporal images and combined use of optical and microwave data, provide abundant features in spectral, spatial, temporal, geometric and polarimetric dimensions. And random forest is usually employed due to robustness and efficiency in handling high-dimensional and noisy data. This study assesses whether feature selection and ensemble analysis, which are rarely adopted, yield improved result. Recursive feature elimination decreases original 568 features into a subset of 53 features, achieving the optimal combination of features. Ensemble analysis of random forest, support vector machine, and K-nearest neighbors leads to an overall accuracy of 0.816. Comparison experiments demonstrated the merits of the multi-temporal, multi-source approach, feature elimination and ensemble analysis. [ABSTRACT FROM AUTHOR]

Land cover mapping provides spatial information on the physical properties of the Earth's surface for various classes of wetlands, artificial surface and constructions, vineyards, water bodies, etc. Having reliable information on land cover is crucial to developing solutions to a variety of environmental problems, such as the destruction of important wetlands/forests, and loss of fish and wildlife habitats. This has made land cover mapping become one of the most widespread applications in remote sensing computational imaging. However, due to the differences between modalities in terms of resolutions, content, and sensors, integrating complementary information that multi-modal remote sensing imagery exhibits into a robust and accurate system still remains challenging, and classical segmentation approaches generally do not give satisfactory results for land cover mapping. In this paper, we propose a novel dynamic deep network architecture, AMM-FuseNet that promotes the use of multi-modal remote sensing images for the purpose of land cover mapping. The proposed network exploits the hybrid approach of the channel attention mechanism and densely connected atrous spatial pyramid pooling (DenseASPP). In the experimental analysis, in order to verify the validity of the proposed method, we test AMM-FuseNet with three datasets whilst comparing it to the six state-of-the-art models of DeepLabV3+, PSPNet, UNet, SegNet, DenseASPP, and DANet. In addition, we demonstrate the capability of AMM-FuseNet under minimal training supervision (reduced number of training samples) compared to the state of the art, achieving less accuracy loss, even for the case with 1/20 of the training samples. [ABSTRACT FROM AUTHOR]

Land cover can reflect global environmental changes if their associated transitions are quantitatively and correctly analysed, thus helping to assess the drivers and impacts of climate change and other applied research studies. It is highly important to acquire accurate spatial land cover information to perform multidisciplinary analyses. This work aims at estimating the accuracy of three widely used land cover products, the Moderate Resolution Imaging Spectroradiometer (MODIS) land cover product (MCD12Q1), the European Space Agency Climate Change Initiative land cover (ESA-CCI-LC), and the EU CORINE land cover (CLC), all for the reference year of 2018, by comparing them against a fine resolution land cover dataset created for this study with combined ground surveys and high-resolution Large Scale Orthophotography (LSO 25/2015). Initially, the four datasets had their land cover classes harmonized and all were resampled to the same spatial resolution. The accuracy metrics used to conduct the comparisons were Overall Accuracy, Producer's Accuracy, User's Accuracy, and the Kappa Coefficient. Comparisons with the reference dataset revealed an underestimation of the forested areas class in all three compared products. Further analysis showed that the accuracy metrics were reasonably high for the broad classes (forest vs. non-forest), with an overall accuracy exceeding 70% in all examined products. On the contrary, in the detailed classification (total land cover mapping), the comparison of the reference dataset with the three land cover products highlighted specific weaknesses in the classification results of the three products, showing that CLC depicted more precisely the landscape characteristics than the two other products, since it demonstrated the highest overall accuracy (37.47%), while MODIS and ESA-CCI-LC revealed a percentage that did not exceed 22%. [ABSTRACT FROM AUTHOR]

Data fusion has shown potential to improve the accuracy of land cover mapping, and selection of the optimal fusion technique remains a challenge. This study investigated the performance of fusing Sentinel-1 (S-1) and Sentinel-2 (S-2) data, using layer-stacking method at the pixel level and Dempster-Shafer (D-S) theory-based approach at the decision level, for mapping six land cover classes in Thu Dau Mot City, Vietnam. At the pixel level, S-1 and S-2 bands and their extracted textures and indices were stacked into the different single-sensor and multi-sensor datasets (i.e. fused datasets). The datasets were categorized into two groups. One group included the datasets containing only spectral and backscattering bands, and the other group included the datasets consisting of these bands and their extracted features. The random forest (RF) classifier was then applied to the datasets within each group. At the decision level, the RF classification outputs of the single-sensor datasets within each group were fused together based on D-S theory. Finally, the accuracy of the mapping results at both levels within each group was compared. The results showed that fusion at the decision level provided the best mapping accuracy compared to the results from other products within each group. The highest overall accuracy (OA) and Kappa coefficient of the map using D-S theory were 92.67% and 0.91, respectively. The decision-level fusion helped increase the OA of the map by 0.75% to 2.07% compared to that of corresponding S-2 products in the groups. Meanwhile, the data fusion at the pixel level delivered the mapping results, which yielded an OA of 4.88% to 6.58% lower than that of corresponding S-2 products in the groups. [ABSTRACT FROM AUTHOR]

This paper proposed a new land cover mapping algorithm for remote sensing images under the presence of clouds. Here, we modelled the observed image as the weighted sum between the reflectances from the cloud-free image and clouds where weights depend on the degree of cloud contamination called the 'cloud thickness' map. Next, we represented the cloud thickness map using a level set function. Similarly, we also represented the land cover map as functions of different level set functions whose values indicate the class label. Next, the maximum a posteriori (MAP) criterion is employed where the most likely land cover and cloud thickness maps are chosen. Under the MAP criterion, the corresponding energy function whose values depend on the level set functions can be derived where the minimum energy point corresponds to the optimum land cover and cloud thickness maps. Since the level set functions are continuous, the optimum solution can be obtained by applying the calculus of variation where the gradient descent algorithm is employed. From the observation model, the cloud-free data can only be reconstructed only if thin to medium clouds are present since thick clouds can completely block the reflectance from the land cover materials. Thus, our algorithm is more suitable for thin to medium than thick cloud covers. Our synthesis and real cloud-contamination examples support our statement since our algorithm achieves significantly higher overall accuracies over other classification techniques, especially, in the thin to medium cloud contamination, and fails to recover the true land cover classes over the thick clouds. [ABSTRACT FROM AUTHOR]

Marshlands in arid and semi-arid areas are considered constantly changing environments due to unsecured water supplies as a result of high evapotranspiration and limited and highly variable rainfall. Classification of marshlands in these regions and mapping of their land cover is not an easy task and maps need to be upgraded frequently. Satellites provide enormous amounts of information and data for the continuous monitoring of changes. The aim of this paper is to introduce an approach using multispectral satellite imagery that was adopted to classify and monitor the Al Hammar Marsh (Iraq) over several years and to suggest a relationship between the Normalized Difference Vegetation Index (NDVI), the Normalized Difference Moisture Index (NDMI), and the Normalized Difference Water Index (NDWI), using Landsat 8 data with a resolution of 30 m × 30 m, validated with Sentinel-2 datasets at 10 m × 10 m. Six land cover classes were used: (1) open water, (2) dry area, (3) dense vegetation, (4) medium-density vegetation, (5) sparse vegetation, and (6) wet soil. Three indices, NDWI, NDMI, and NDVI, were chosen for the automatic classification of each pixel and the creation of a time series of land cover maps. The proposed method can efficiently classify and monitor marshlands and can be used to study different marshlands by adjusting the thresholds for NDVI, NDMI, and NDWI. Overall, the correlation for all classes (R) between Landsat 8 and Sentinel-2 is about 0.78. Thus, this approach will help to preserve marshes through improved water management. [ABSTRACT FROM AUTHOR]

One of the main applications of machine learning (ML) in remote sensing (RS) is the pixel-level classification of satellite images into land cover types. Although classes with different spectral signatures can be easily separated, e.g. aquatic and terrestrial land cover types, others have similar spectral signatures and are hard to separate using only the information within a single pixel. This work focused on the separation of two cover types with similar spectral signatures, cocoa agroforest and forest, over an area in Pará, Brazil. For this, we study the training and application of several ML algorithms on datasets obtained from a single composite image, a time-series (TS) composite obtained from the same location and by preprocessing the TS composite using simple TS preprocessing techniques. As expected, when ML algorithms are applied to a dataset obtained from a composite image, the median producer's accuracy (PA) and user's accuracy (UA) in those two classes are significantly lower than the median overall accuracy (OA) for all classes. The second dataset allows the ML models to learn the evolution of the spectral signatures over 5 months. Compared to the first dataset, the results indicate that ML models generalize better using TS data, even if the series are short and without any preprocessing. This generalization is further improved in the last dataset. The ML models are subsequently applied to an area with different geographical bounds. These last results indicate that, out of seven classifiers, the popular random forest (RF) algorithm ranked fourth, while XGBoost (×GB) obtained the best results. The best OA, as well as the best PA/UA balance, were obtained by performing feature construction using the M3GP algorithm and then applying XGB to the new extended dataset. [ABSTRACT FROM AUTHOR]

Synthetic aperture radar (SAR) data has been an alternative for monitoring ground targets, especially in areas with cloud cover. This study evaluates the potential of Sentinel-1A attributes for mapping land use and land cover (LULC) in a region of the Brazilian Amazon, using two different machine learning classifiers: Random Forest (RF) and Support Vector Machine (SVM). Different scenarios were used that combined backscattering, polarimetry, and interferometry to the classification process, which was divided into two phases to improve the results. The RF shows superiority over the SVM for almost all scenarios for the two phases of the mapping. The scenario with all data, presented the best results with both classifiers. The final maps with RF and SVM, obtained a global accuracy of 82.7% and 74.5%, respectively. This study demonstrated the potential of Sentinel-1 to map LULC classes in the Amazon region using a classification in two phases. [ABSTRACT FROM AUTHOR]

• The annual land cover maps of Beijing from 2001 to 2020 were generated. • The land cover dynamics of Beijing during the past two decades were investigated. • The classification method combining CCDC algorithm and MRF model was proposed to achieve the temporally and spatially consistent land cover mapping. Compared to change detection using two-dates satellite images, monitoring the changes at high temporal frequencies using dense observations can provide more comprehensive understanding of the land cover dynamics for a given place. Beijing, the capital of China, has undergone fast urban growth in the past decades. The existing studies on Landsat-derived land cover dynamics in Beijing mainly focus on 5- or 10-year intervals, or annual mapping of single land cover type; however, the dynamics of all-types land cover in Beijing at one-year scale were rarely investigated. To fill this research gap, we presented a time-series land-cover mapping approach by combining the Continuous Change Detection and Classification (CCDC) algorithm with Markov random field (MRF) model to explore the annual dynamics of land cover in Beijing from 2001 to 2020 using Landsat time series. First, the annual land cover maps for Beijing were generated using CCDC algorithm. Then, the MRF model was applied to annual land cover maps to alleviate the salt and pepper noise arising from the classification of CCDC at the pixel level. Results showed that CCDC-MRF proposed in this study could produce temporally and spatially consistent results which have higher annual average overall accuracy (81.93%) than the results derived from CCDC (79.18%). In addition, the accuracy of annual land cover changes for CCDC-MRF was 92.50% in spatial domain and 80.49% in temporal domain, which were higher than the results for CCDC with 89.36% in spatial domain and 78.38% in temporal domain. The major land cover change in Beijing over the last two decades was characterized by urban expansion with the replacement of cultivated land, leading to 13.53% of cultivated land being replaced by artificial impervious surface, mainly occurring between the fifth and sixth ring roads. The method proposed in this study could generate accurate land cover maps at high temporal frequencies and the findings of this research could provide a better understanding for sustainable urban development and management. [ABSTRACT FROM AUTHOR]

The importance of Land Cover (LC) classification is recognized by an increasing number of scholars who employ LC information in various applications (i.e., address global climate change and achieve sustainable development). However, studying the roles of balancing data, image integration, and performance of different machine learning algorithms in various landscapes has not received as much attention from scientists. Therefore, the present study investigates the performance of three frequently used Machine Learning (ML) algorithms, including Extreme Learning Machines (ELM), Support Vector Machines (SVM), and Random Forest (RF) in LC mapping at six different landscapes. Moreover, the Geometric Synthetic Minority Over-sampling Technique (G-SMOTE) was adopted to deal with the class imbalance problem. In this work, the time-series of Sentinel-1 and Sentinel-2 data were integrated to improve LC mapping accuracy, taking advantage of both data. Moreover, Support Vector Machine-Recursive Feature Elimination (SVM-RFE) was implemented to distinguish the most informative features. Based on the results, the RF integrated with G-SMOTE showed the best result for four landscapes (coastal, cropland, desert, and semi-arid). SVM integrated with G-SMOTE had the highest accuracy in the remaining two landscapes (plain and mountain). Applied ML algorithms showed good performances in various landscapes, ranging Overall Accuracy (OA) from 85% to 93% for RF, 83% to 94% for SVM, and 84% to 92% for ELM. The outcomes exhibit that although applying G-SMOTE may slightly decrease OA values, it generally boosts the results of LC classification accuracies in various landscapes, particularly for minority classes. [ABSTRACT FROM AUTHOR]

As remote sensing (RS) data obtained from different sensors become available largely and openly, multimodal data processing and analysis techniques have been garnering increasing interest in the RS and geoscience community. However, due to the gap between different modalities in terms of imaging sensors, resolutions, and contents, embedding their complementary information into a consistent, compact, accurate, and discriminative representation, to a great extent, remains challenging. To this end, we propose a shared and specific feature learning (S2FL) model. S2FL is capable of decomposing multimodal RS data into modality-shared and modality-specific components, enabling the information blending of multi-modalities more effectively, particularly for heterogeneous data sources. Moreover, to better assess multimodal baselines and the newly-proposed S2FL model, three multimodal RS benchmark datasets, i.e., Houston2013 – hyperspectral and multispectral data, Berlin – hyperspectral and synthetic aperture radar (SAR) data, Augsburg – hyperspectral, SAR, and digital surface model (DSM) data, are released and used for land cover classification. Extensive experiments conducted on the three datasets demonstrate the superiority and advancement of our S2FL model in the task of land cover classification in comparison with previously-proposed state-of-the-art baselines. Furthermore, the baseline codes and datasets used in this paper will be made available freely at https://github.com/danfenghong/ISPRS_S2FL. [ABSTRACT FROM AUTHOR]

Given the importance of accuracy in land cover (LC) maps, several methods have been adopted to predict per-pixel land cover accuracy (PLCA) of classified remote sensing images. Such a PLCA map provides spatially-explicit accuracy information and is of paramount importance for both producers and end-users of LC maps to thoroughly understand the spatial distribution of accuracy. In this study, we proposed a simple yet powerful random forest (RF) based approach for PLCA mapping with limited reference sample data. The main assumption of the proposed approach is that the LC's misclassifications do not occur randomly, but rather exhibit some detectable characteristics which can be retrieved via the built model. With this approach, RF attempts to establish a nonlinear relationship between the accuracy and the same spectral bands used in LC classification. To confirm the proposed method as a consistent and practical approach for a variety of different settings, we evaluated it on five different classified remote sensing images derived from Landsat-8, Ikonos, and three Sentinel-2 images across different parts of Iran. In this manner, to validate the predictive capability of the RF-based method, we calculated the area under the receiver operating characteristic curve (AUROC) and several other statistical metrics, including sensitivity (SN), specificity (SP), positive predictive value (PPV), negative predictive value (NPV), and accuracy (ACC). Analysis of the average values of these metrics (AUROC = 0.88, SN = 95%, SP = 68%, PPV = 96%, NPV = 72%, and ACC = 95%) derived from the limited sample size datasets showed that the proposed model performs well in all case studies. The performance of the proposed model was further assessed through comparison against two benchmark methods, namely Gaussian kernel interpolation (GKI) and linear kernel interpolation (LKI). In conclusion, although our comprehensive evaluations revealed that RF, GKI, and LKI methods are promising approaches for PLCA mapping, RF outperformed both GKI and LKI in all of the experimental sites. [ABSTRACT FROM AUTHOR]

Reference data for large-scale land cover map are commonly acquired by visual interpretation of remotely sensed data. To assure consistency, multiple images are used, interpreters are trained, sites are interpreted by several individuals, or the procedure includes a review. But little is known about important factors influencing the quality of visually interpreted data. We assessed the effect of multiple variables on land cover class agreement between interpreters and reviewers. Our analyses concerned data collected for validation of a global land cover map within the Copernicus Global Land Service project. Four cycles of visual interpretation were conducted, each was followed by review and feedback. Each interpreted site element was labelled according to dominant land cover type. We assessed relationships between the number of interpretation updates following feedback and the variables grouped in personal, training, and environmental categories. Variable importance was assessed using random forest regression. Personal variable interpreter identifier and training variable timestamp were found the strongest predictors of update counts, while the environmental variables complexity and image availability had least impact. Feedback loops reduced updating and hence improved consistency of the interpretations. Implementing feedback loops into the visually interpreted data collection increases the consistency of acquired land cover reference data. [ABSTRACT FROM AUTHOR]

Land cover mapping of marshland areas from satellite images data is not a simple process, due to the similarity of the spectral characteristics of the land cover. This leads to challenges being encountered with some land covers classes, especially in wetlands classes. In this study, satellite images from the Sentinel 2B by ESA (European Space Agency) were used to classify the land cover of Al-Hawizeh marsh/Iraq-Iran border. Three classification methods were used aimed at comparing their accuracy, using multispectral satellite images with a spatial resolution of 10 m. The classification process was performed using three different algorithms, namely: Maximum Likelihood Classification (MLC), Artificial Neural Networks (ANN), and Support Vector Machine (SVM). The classification algorithms were carried out using ENVI 5.1 software to detect six land cover classes: deep water marsh, shallow water marsh, marsh vegetation (aquatic vegetation), urban area (built-up area), agriculture area, and barren soil. The results showed that the MLC method applied to Sentinel 2B images provides a higher overall accuracy and the kappa coefficient compared to the ANN and SVM methods. Overall accuracy values for MLC, ANN, and SVM methods were 85.32%, 70.64%, and 77.01% respectively. [ABSTRACT FROM AUTHOR]

To curb land degradation, a series of ecological restoration projects have been carried out since 1999, leading to dramatic land cover change (LCC) in the agricultural pastoral ecotone of northern China (APENC). To date, there is still lack of timely and accurate land cover (LC) information for management and assessment actions. This paper presents a LC mapping scheme to map annual LC information based on dynamic time warping (DTW) approach and time‐series MODIS‐NDVI product for the APENC. The DTW approach was optimized firstly based on the change of vegetation phenology phase. Next, typical reference time‐series curves set of different LC types were established and pixel‐wise similarity was examined to identify the LC type. Before that, the reconstruction of time‐series MODIS‐NDVI was performed using harmonic analysis of time‐series and reconstruction accuracy was evaluated quantitatively at a regional scale. Results showed that the gap error was relatively large for the high vegetation cover areas while it might be ignored for the moderate and low vegetation cover areas. The overall accuracy of LC mapping in year 2010 (LC‐2010) was 77.89% and kappa coefficient reached 0.71, which suggested that LC mapping scheme could extract LC information with high accuracy. The producer's accuracy ranged between 69.83% (grasslands) and 100% (water body), while the user's accuracy changed from 58% (built‐up) to 86% (forest). The overall spatial agreement between LC‐2010 and GlobaLand30‐2010 was about 64.44%. The total area of forests, croplands, water body and built‐up exhibited an increasing trend while the contrary trend were found for the grasslands and bare land during the 2001–2017. These LC maps are valuable for research the land degradation investigation and monitoring, benefit evaluation of ecological restoration projects and effect of LCC on surface hydrothermal process in APENC. [ABSTRACT FROM AUTHOR]

Archaeological site mapping is important for both understanding the history and protecting the sites from excavation during developmental activities. As archaeological sites are generally spread over a large area, use of high spatial resolution remote sensing imagery is becoming increasingly applicable in the world. The main objective of this study is to map the land cover of the Itanos area of Crete and of its changes, with specific focus on the detection of the landscape's archaeological features. Six satellite images were acquired from the Pleiades and WorldView-2 satellites over a period of 3 years. In addition, digital imagery of two known archaeological sites was used for validation. An object-based image analysis classification was subsequently developed using the five acquired satellite images. Two rule sets were created, one using the standard four bands which both satellites have and another for the two WorldView-2 images with their four extra bands included. Validation of the thematic maps produced from the classification scenarios confirmed a difference in accuracy amongst the five images. Comparing the results of a 4-band rule set versus the 8-band rule set showed a slight increase in classification accuracy using extra bands. The resultant classifications showed a good level of accuracy exceeding 70%. Yet, separating the archaeological sites from the open spaces with little or no vegetation proved to be challenging. This was mainly due to the high spectral similarity between rocks and the archaeological ruins. The high resolution of the satellite data allowed for the accuracy in defining larger archaeological sites, but still there was difficulty in distinguishing smaller areas of interest. The digital image data provided a very good 3D representation for the archaeological sites, assisting as well as in validating the satellite-derived classification maps. To conclude, our study provides further evidence that use of high resolution imagery may allow for archaeological sites to be located, but only where the archaelogical features are of an adequate size. [ABSTRACT FROM AUTHOR]

In this work, we present a new semi-supervised learning framework to cope with satellite image time series (SITS) classification in a data paucity scenario, considering extreme low levels of supervision. The proposed methodology, referred as S 3 ITS (Semi-Supervised Satellite Image Time Series classification method), is based on temporal convolutional neural networks and it takes advantage of both labelled and unlabelled information. S 3 ITS enforces the data to be projected in a discriminative manifold via contrastive learning, in order to produce a data representation where samples belonging to the same category are closer than the ones belonging to different ones. Pseudo-labelling is employed on unlabelled samples to take the most out of the available unlabelled information. Experiments on two study sites described by SITS of Sentinel-2 images highlight the quality of the proposed method with respect to common SITS-based classification methods and recent machine learning approaches especially tailored for the semi-supervised classification of multi-variate time series data. [ABSTRACT FROM AUTHOR]

Crop production estimation is of crucial concern in Taiwan, and the government has invested much effort including employing manpower and technologies. Artificial intelligence or data mining technology have been successfully applied on land-cover recognition of remote-sensing imagery. Most studies are employing a pixel-based classification approach to generate the thematic map of land covers. Few studies consider the spatial correlation of adjacent pixels of the same category. Mixed pixel issues usually degrade the prediction accuracy according to previous studies. It is thus the main goal of the present study to explore the spatial effect of adjacent pixels on land-cover mapping. The study region is a WorldView-2 satellite image in Jianan Plain, Taiwan, taken in 2014. Support Vector Machine is used as the underlying classifier. In addition to the eight spectral band intensities, normalized difference vegetation index and grey-level co-occurrence matrix (GLCM) textures are included as ancillary attributes. Furthermore, grey relational analysis (GRA) is employed to assist in classifying croplands. The findings of this study can be summarized as follows: (1) GLCM texture information improves classification accuracy marginally and renders slightly better thematic map, (2) GRA is used to acquire the most important factors concerning discriminating land covers, (3) grey relational grade threshold, a metric designed through GRA, can be used to locate uncertain region of a specified crop, which is possibly caused by mixed pixels. [ABSTRACT FROM AUTHOR]

Web maps represent an effective source for land cover mapping in capturing human activities. However, due to spatial heterogeneity, previous research has mainly focused on generating land cover maps in partial areas. Inferring spatial distribution patterns in Web maps may provide an alternative perspective on improving map production on a larger scale. This paper represents a novel approach to investigating the spatial distribution in Web maps for land cover mapping. First, linear features from Web maps are utilised to delineate parcels with insufficient Web map data for classification. Then, spatial factors are constructed from point and polygon features to identify the spatial variety of Web maps, with an artificial neural network classifier being adopted to classify land cover automatically. Land cover mapping is finally proposed by combining classified parcels and existing polygon features. The proposed method is applied in Guangzhou, Guangdong Province, using a Web map from AutoNavi. The results show an approximately 88% classification accuracy and an overall mapping accuracy of 85.06%. The results indicate that the proposed approach has the potential to be utilised in land cover mapping, and the constructed spatial factors are effective at characterising land cover information. [ABSTRACT FROM AUTHOR]

• Analysis of land cover changes of Darling Range. • Fourteen time series images between 1988 and 2014 were used. • Compares the performances of three machine learning algorithms. • The analysis outputs of random forest algorithm were validated. The Darling Range in Western Australia is a major bauxite producing region. Clearing, excavation and rehabilitation activities related to bauxite mining have influenced land cover within this region since mining commenced in the 1960s. This paper presents a study that used machine learning and time series visualisation to analyse the land cover changes of the Darling Range using time-lapse multispectral images, with the aim of understanding the impact of the mining activities and land rehabilitation patterns of the region. Land cover changes were analysed using 14 Landsat Thematic Mapper (TM) images between 1988 and 2014. The spatial distribution of land cover was classified automatically using machine learning algorithms, and their temporal changes were visualized for analysis. Supervised classification was carried out based on the six spectral features of the images using three machine learning algorithms, namely support vector machine (SVM), random forest (RF) and Naïve Bayes (NB). The results showed that the RF algorithm achieved overall accuracy, (ratio of correctly classified samples divided by the total number of samples), greater than 95% for all years. The temporal changes of land cover distribution over the study period were visualized using a change map. These changes were compared with land clearing and rehabilitation records of a mining company operating in that region. A close correlation was observed between the automated analysis outputs and the company's records. This work demonstrates the potential use of machine analysis in improving the accuracy of spatial data related to land clearing; and in monitoring vegetation recovery of closed and rehabilitated mines. [ABSTRACT FROM AUTHOR]

[Display omitted] • AGTML identifies optimal focal radii for different land cover types, enabling the generation of reliable training samples. • A new Spectral Distance Index (SED) is introduced to automatically update training samples, enhancing the classification process. • The AGTML approach proves to be adaptable to various regions and sensors, making it highly effective for automated land cover classification. The timely, accurate, and automatic acquisition of land cover (LC) information is a prerequisite for detecting LC dynamics and performing ecological analyses. Cloud computing platforms, such as the Google Earth Engine, have substantially improved the efficiency and scale of LC classification. However, the lack of sufficient and representative training samples hinders automatic and accurate LC classification. In this study, we propose a new approach that integrates the automatic generation of training samples and machine learning algorithms (AGTML) for LC classification in Heilongjiang Province, China. After optimal focal radii were determined for different LC types using Landsat 8 based on focal statistics and unique phenology. Then target training samples were automatically generated based on the improved distance measure SED (a composite of Spectral angle distance (SAD) and Euclidean distance (ED)). Furthermore, LC classification was performed using four feature combinations and three machine learning algorithms. According to independent validation data, the automatically generated training samples demonstrated good representativeness and stability among all three classifiers, with an overall accuracy (OA) of classification higher than 86%, and showed high consistency in the landscape pattern of classification. RF yielded the highest classification accuracy (92.99% OA). AGTML outperformed GLC-FCS30 in identifying large fragmentation and small patch regions in the landscape types. The AGTML approach was subsequently applied to the Guanzhong Plain using different satellite imagery. Results were consistent and accurate (>96.50% OA), demonstrating that the AGTML approach can be applied to various regions and sensors, and has immense potential for automated LC classification across regional and global scales. [ABSTRACT FROM AUTHOR]

Abstract The success of a large number of real-world applications depends on the efficiency with which land use land cover classes are extracted from Remotely Sensed (RS) imagery. This study aims at analysing the performance of Maximum Likelihood Classifier (MLC) over multispectral remote sensor data characterized by the overlapping of spectral classes and presence of mixed pixels, investigating the impact of prior class probabilities on the overall classification performance, the impact of the number of land cover classes on classification accuracy, and identifying the decision parameters for assigning a pixel to a land cover class. A quantitative analysis of decision parameters of MLC for pixel assignment is presented for five randomly selected pixels in the study area. Landsat 8 multispectral data of North Canara District was collected from USGS website and is considered for the research. Seven land use land cover classes were identified over the study area. Results obtained indicate that use of prior probability estimates improves the classification accuracy. The study also presents the variation in classification accuracy for four different cases of total number of land cover classes. The conducted research is useful in understanding the performance of MLC under multiple scenarios. [ABSTRACT FROM AUTHOR]

Fine-resolution land cover mapping at the regional or global scale is usually time-consuming and involves a lot of manual participation. This article proposes a novel and automatic land cover mapping approach called the SP atial-t E mporal spe C tral Lib rary (SPECLib), which can produce a land cover map with a 30-m spatial resolution. First, the SPECLib is developed by extracting reference spectra for uniform, typical samples of each land cover type by combining the GlobCover2009 land cover product and the 8-day composite MODIS Surface Reflectance Product (MOD09 A1). SPECLib has a temporal resolution of 8 days and contains most of the land cover types in each 1.5 ° × 1.5 ° geographic cell. For an arbitrary temporal stage, for each geographical cell of the SPECLib, there are a total of about 5000 MODIS spectra. Secondly, an operational approach for automatic land cover mapping using Landsat data is developed based on SPECLib. For each Landsat scene to be classified, the reference spectra of the land cover types in the corresponding geographic cell were automatically extracted and normalized. The Landsat scene was then classified by using the normalized reference spectra from SPECLib in an integrated classification approach that combined MLC (Maximum Likelihood Classification) and SAM (Spectral Angle Mapper). Finally, the SPECLib-based approach is tested using 510 single-date Landsat 8 OLI scenes to produce a land cover map of China. The map was validated using 11,366 interpretation samples and was shown to have an overall accuracy of 77.34% and a kappa coefficient of 0.729. Therefore, the preliminary results indicate that the SPECLib-based operational classification approach is a promisingly automatic method for global land cover mapping at a resolution of 30 m. [ABSTRACT FROM AUTHOR]

Copyright of Methods in Ecology & Evolution is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Pixel-based and object-based classifications are two commonly used approaches in extracting land cover information from remote sensing images. However, they each have their own inherent merits and limitations. This study, therefore, proposes a new classification method through the integration of pixel-based and object-based classifications (IPOC). Firstly, it employs pixel-based soft classification to obtain the class proportions of pixels to characterize the land cover details from pixel-scale properties. Secondly, it adopts area-to-point kriging to explore the class spatial dependence between objects for each pixel from object-based soft classification results. Thirdly, the class proportions of pixels and the class spatial dependence of pixels are fused as the class occurrence of pixels. Last, a linear optimization model on objects is built to determine the optimal class label of pixels within each object. Two remote sensing images are used to evaluate the effectiveness of IPOC. The experimental results demonstrate that IPOC performs better than the traditional pixel-based hard classification and object-based hard classification methods. Specifically, the overall accuracy of IPOC is 7.64% higher than that of pixel-based hard classification and 4.64% greater than that of object-based hard classification in the first experiment, while the overall accuracy improvements in the second experiment are 3.59% and 3.42%, respectively. Meanwhile, IPOC produces less salt and pepper effect than the pixel-based hard classification method and generates more accurate land cover details and small patches than the object-based hard classification method. [ABSTRACT FROM AUTHOR]

High resolution raster data for land cover mapping or change analysis are normally acquired through satellite or aerial imagery. Apart from the incurred costs, the available files might not have the required temporal resolution. Moreover, cloud cover and atmospheric absorptions may limit the applicability of existing algorithms or reduce their accuracy. This paper presents a novel technique that is capable of mapping garrigue and/or phrygana vegetation as well as karst or ground-armour terrain in photos captured by a digital camera. By including a reference pattern in every frame, the automated method estimates the total area covered by each land type. Pixel based classification is performed by supervised decision tree methods. Although the intention is not to replace traditional surface cover analysis, the proposed technique allows accurate land cover mapping with quantitative estimates to be obtained. Since no expensive hardware or specialised personnel are required, vegetation monitoring of local sites can be carried out cheaply and frequently. The developed proof of concept is tested on photos taken in thirteen different sites across the Maltese Islands. [ABSTRACT FROM AUTHOR]