Search

Rapid impacts from both natural and anthropogenic sources on wetland ecosystems underscore the need for updating wetland inventories. Extensive up-to-date field samples are required for calibrating methods (e.g., machine learning) and validating results (e.g., maps). The purpose of this study is to design a dataset generation approach that extracts training data from already existing wetland maps in an unsupervised manner. The proposed method utilizes the LandTrendr algorithm to identify areas least likely to have changed over a seven-year period from 2016 to 2022 in Minnesota, USA. Sentinel-2 and Sentinel-1 data were used through Google Earth Engine (GEE), and sub-pixel water fraction (SWF) and normalized difference vegetation index (NDVI) were considered as wetland indicators. A simple thresholding approach was applied to the magnitude of change maps to identify pixels with the most negligible change. These samples were then employed to train a random forest (RF) classifier in an object-based image analysis framework. The proposed method achieved an overall accuracy of 89% with F1 scores of 91%, 81%, 88%, and 72% for water, emergent, forested, and scrub-shrub wetland classes, respectively. The proposed method offers an accurate and cost-efficient method for updating wetland inventories as well as studying areas impacted by floods on state or even national scales. This will assist practitioners and stakeholders in maintaining an updated wetland map with fewer requirements for extensive field campaigns. [ABSTRACT FROM AUTHOR]

Immediately after an earthquake, rapid disaster management is the main challenge for relevant organizations. While satellite images have been used in the past two decades for building-damage mapping, they have rarely been utilized for the timely damage monitoring required for rescue operations. Unmanned aerial vehicles (UAVs) have recently become very popular due to their agile deployment to sites, super-high spatial resolution, and relatively low operating cost. This paper proposes a novel deep-learning-based method for rapid post-earthquake building damage detection. The method detects damages in four levels and consists of three steps. First, three different feature types—non-deep, deep, and their fusion—are investigated to determine the optimal feature extraction method. A "one-epoch convolutional autoencoder (OECAE)" is used to extract deep features from non-deep features. Then, a rule-based procedure is designed for the automatic selection of the proper training samples required by the classification algorithms in the next step. Finally, seven famous machine learning (ML) algorithms—including support vector machine (SVM), random forest (RF), gradient boosting (GB), extreme gradient boosting (XGB), decision trees (DT), k-nearest neighbors (KNN), and adaBoost (AB)—and a basic deep learning algorithm (i.e., multi-layer perceptron (MLP)) are implemented to obtain building damage maps. The results indicated that auto-training samples are feasible and superior to manual ones, with improved overall accuracy (OA) and kappa coefficient (KC) over 22% and 33%, respectively; SVM (OA = 82% and KC = 74.01%) was the most accurate AI model with a slight advantage over MLP (OA = 82% and KC = 73.98%). Additionally, it was found that the fusion of deep and non-deep features using OECAE could significantly enhance damage-mapping efficiency compared to those using either non-deep features (by an average improvement of 6.75% and 9.78% in OA and KC, respectively) or deep features (improving OA by 7.19% and KC by 10.18% on average) alone. [ABSTRACT FROM AUTHOR]

Polarimetric synthetic aperture radar (PolSAR) images contain useful information, which can lead to extensive land cover interpretation and a variety of output products. In contrast to optical imagery, there are several challenges in extracting beneficial features from PolSAR data. Deep learning (DL) methods can provide solutions to address PolSAR feature extraction challenges. The convolutional neural networks (CNNs) and graph convolutional networks (GCNs) can drive PolSAR image characteristics by deploying kernel abilities in considering neighborhood (local) information and graphs in considering long-range similarities. A novel dual-branch fusion of CNN and mini-GCN is proposed in this study for PolSAR image classification. To fully utilize the PolSAR image capacity, different spatial-based and polarimetric-based features are incorporated into CNN and mini-GCN branches of the proposed model. The performance of the proposed method is verified by comparing the classification results to multiple state-of-the-art approaches on the airborne synthetic aperture radar (AIRSAR) dataset of Flevoland and San Francisco. The proposed approach showed 1.3% and 2.7% improvements in overall accuracy compared to conventional methods with these AIRSAR datasets. Meanwhile, it enhanced its one-branch version by 0.73% and 1.82%. Analyses over Flevoland data further indicated the effectiveness of the dual-branch model using varied training sampling ratios, leading to a promising overall accuracy of 99.9% with a 10% sampling ratio. [ABSTRACT FROM AUTHOR]

Forest above-ground biomass (AGB) estimation provides valuable information about the carbon cycle. Thus, the overall goal of this paper is to present an approach to enhance the accuracy of the AGB estimation. The main objectives are to: 1) investigate the performance of remote sensing data sources, including airborne light detection and ranging (LiDAR), optical, SAR, and their combination to improve the AGB predictions, 2) examine the capability of tree-based machine learning models, and 3) compare the performance of pixel-based and object-based image analysis (OBIA). To investigate the performance of machine learning models, multiple tree-based algorithms were fitted to predictors derived from airborne LiDAR data, Landsat, Sentinel-2, Sentinel-1, and PALSAR2/PALSAR SAR data collected within New York’s Adirondack Park. Combining remote sensing data from multiple sources improved the model accuracy (RMSE: 52.14 Mg ha-1 and R² : 0.49). There was no significant difference among gradient boosting machine (GBM), random forest (RF), and extreme gradient boosting (XGBoost) models. In addition, pixel-based and object-based models were compared using the airborne LiDAR-derived AGB raster as a training/testing sample. The OBIA provided the best results with the RMSE of 33.77 Mg ha-1 and R² of 0.81 for the combination of optical and SAR data in the GBM model. [ABSTRACT FROM AUTHOR]

A common problem with matching algorithms, in photogrammetry and computer vision, is the imperfection of finding all correct corresponding points, so-called inliers, and, thus, resulting in incorrect or mismatched points, so-called outliers. Many algorithms, including the well-known randomized random sample consensus (RANSAC)-based matching, have been developed focusing on the reduction of outliers. RANSAC-based methods, however, have limitations such as increased false positive rates of outliers, and, consequently resulting in fewer inliers, an unnecessary high number of iterations, and high computational time. Such deficiencies possibly result from the random sampling process, the presence of noise, and incorrect assumptions of the initial values. This paper proposes a modified version of RANSAC-based methods, called Empowered Locally Iterative SAmple Consensus (ELISAC). ELISAC improves RANSAC by utilizing three basic modifications individually or in combination. These three modifications are (a) to increase the stability and number of inliers using two Locally Iterative Least Squares (LILS) loops (Basic LILS and Aggregated-LILS), based on the new inliers in each loop, (b) to improve the convergence rate and consequently reduce the number of iterations using a similarity termination criterion, and (c) to remove any possible outliers at the end of the processing loop and increase the reliability of results using a post-processing procedure. In order to validate our proposed method, a comprehensive experimental analysis has been done on two datasets. The first dataset contains the commonly-used computer vision image pairs on which the state-of-the-art RANSAC-based methods have been evaluated. The second dataset image pairs were captured by a drone over a forested area with various rotations, scales, and baselines (from short to wide). The results show that ELISAC finds more inliers with a faster speed (lower computational time) and lower error (outlier) rates compared to M-estimator SAmple Consensus (MSAC). This makes ELISAC an effective approach for image matching and, consequently, for 3D information extraction of very high and super high-resolution imagery acquired by space-borne, airborne, or UAV sensors. In particular, for applications such as forest 3D modeling and tree height estimations where standard matching algorithms are problematic due to spectral and textural similarity of objects (e.g., trees) on image pairs, ELISAC can significantly outperform the standard matching algorithms. [ABSTRACT FROM AUTHOR]

Forest aboveground biomass (AGB) provides valuable information about the carbon cycle, carbon sink monitoring, and understanding of climate change factors. Remote sensing data coupled with machine learning models have been increasingly used for forest AGB estimation over local and regional extents. Landsat series provide a 50-year data archive, which is a valuable source for historical mapping over large areas. As such, this paper proposed a machine learning-based workflow for historical AGB estimation and its change analysis from 2001 to 2019 for the New York State's forests using Landsat historical imagery, airborne LiDAR, and forest plot data. As the object-based image analysis (OBIA) is able to incorporate spectral, contextual, and textural features into the regression model, the proposed method utilizes an OBIA approach and a random forest (RF) regression model implemented on the Google Earth Engine (GEE) cloud computing platform. Results demonstrated that there is a considerable decrease of 983.79 × 106 Mg/ha in the AGB of deciduous forests from 2001 to 2006, followed by an increase of 618.28 × 106 Mg/ha from 2006 to 2011, continued with an increase of 229.12 × 106 Mg/ha of deciduous forests from 2011–2016. Finally, the results demonstrated a slight change in AGB from 2016 to 2019. The transferability of the proposed framework provides a practical solution for monitoring forests in other states or even on a national scale. [ABSTRACT FROM AUTHOR]

As a leading natural wealth, forests play an essential role in the development and prosperity of countries. Hence, monitoring their changes can lead to proper management and planning in conserving these resources. This study presents a novel unsupervised forest change detection method comprising two main steps: (1) generating a reliable difference image, i.e. sensitive to forest changes, and (2) producing a change map in which forest changes and their details (e.g. edges) are well characterized. In step (1), the vegetation indices- and spectral-based difference images were first calculated using a novel weighted angular operator. Afterwards, the difference images were combined using the 2D-multiresolution singular value decomposition (2D-MSVD) fusion approach to generate a noise-resistant difference image, in which forest changes are accurately highlighted. In step (2), the expectation-maximization gaussian mixture model (EMGMM) was first applied to the fused difference image to reach an initial binary change map. Next, an edge-aware MRF (EAMRF) model was initialized by the EMGMM-derived change map and then was adopted to achieve the final change map. Experimental results were achieved by utilizing five bi-temporal images acquired by the Landsat 5 and 8, and Sentinel 2 satellite sensors. The results indicated the efficacy of the proposed fused difference image in reflecting the forest changes. Compared with the traditional MRF method, the boundaries and geometrical shapes of changed regions were well preserved in the change maps obtained by EAMRF. The edge penalty function embedded in EAMRF also made this model converge in less running time compared to the traditional MRF algorithm. Furthermore, EAMRF outperformed the other change detection methods in terms of quantitative and qualitative results, demonstrating its high potential for forest change detection applications. The source code of the proposed change detection method with some samples of the datasets has been made available on to support related future works in remote sensing. [ABSTRACT FROM AUTHOR]

Given the key role wetlands play in climate regulation and shoreline stabilization, identifying their spatial distribution is essential for the management, restoration, and protection of these invaluable ecosystems. The increasing availability of high spatial and temporal resolution optical and synthetic aperture radar (SAR) remote sensing data coupled with advanced machine learning techniques have provided an unprecedented opportunity for mapping complex wetlands' ecosystems. A recent partnership between the National Aeronautics and Space Administration (NASA) and the Indian Space Research Organization (ISRO) resulted in the design of the NASA‐ISRO SAR (NISAR) mission. In this study, the capability of L‐band simulated NISAR data for wetland mapping in Yucatan Lake, Louisiana, is investigated using two object‐based machine learning approaches: Support vector machine (SVM) and random forest (RF). L‐band Unmanned Aerial Vehicle SAR (UAVSAR) data are exploited as a proxy for NISAR data. Specifically, we evaluated the synergistic use of different polarimetric features for efficient delineation of wetland types, extracting 84 polarimetric features from more than 10 polarimetric decompositions. High spatial resolution National Agriculture Imagery Program imagery is applied for image segmentation using the mean‐shift algorithm. Overall accuracies of 74.33% and 81.93% obtained by SVM and RF, respectively, demonstrate the great possibility of L‐band prototype NISAR data for wetland mapping and monitoring. In addition, variable importance analysis using the Gini index for RF classifier suggests that H/A/ALPHA, Freeman‐Durden, and Aghababaee features have the highest contribution to the overall accuracy. Plain Language Summary: By illuminating the surface, SAR signals can provide meaningful information on the shape, geometry, and roughness of the surface. In particular, polarimetric decompositions bring a measure of the relative contribution of backscatter from different scattering mechanisms that can be used for wetland delineations, classification, and monitoring. Given the availability of various polarimetric decompositions, the selection of appropriate decomposition based on the application and SAR sensor configuration is crucial. In this study, we investigated the performance of various polarimetric decompositions for delineating wetlands classes over Yucatan Lake in Louisiana. The adopted machine learning classification workflow was applied to the L‐band simulated NISAR data that are acquired by the UAVSAR platform to evaluate the performance of planned L‐band NISAR data. Our investigation showed that H/A/ALPHA, Freeman‐Durden, and Aghababaee features have the highest contribution to the overall accuracy. Key Points: High spatial resolution Earth Observation (EO) data and machine learning techniques have provided opportunities for preservation of wetlandsL‐band simulated NISAR was captured with UAVSAR as a proxy for evaluating the planned NISAR for application of wetland monitoringUsing 84 polarimetric features and support vector machine and random forest classifiers, overall accuracies of 74.33% and 83.93% were obtained [ABSTRACT FROM AUTHOR]

Shrub willow is considered an important dedicated energy crop in temperate climates for the production of bioenergy, biofuels, and bio-based products. A methodology to rapidly and accurately estimate above-ground biomass (AGB) is essential for understanding potential biomass supply, identifying potential growth limitations, and making management decisions. The main objective of this study was to investigate different statistical, machine learning, and deep learning models to estimate shrub willow AGB at a site in Camillus, NY using multi-spectral unmanned aerial system (UAS) imagery. The efficiency of the convolutional neural network (CNN) deep learning algorithm was compared to the well-known methods including linear regression, decision tree (DT), random forest (RF), and support vector regression (SVR). The RF model estimated the AGB with the root mean square error (RMSE) of 1.73 Mg/ha and R2 of 0.95, and outperformed other methods. The next most effective method was CNN with the RMSE of 2.69 Mg/ha and R2 of 0.89. Feature importance analysis indicated that normalized difference vegetation index (NDVI), ratio vegetation index (RVI), and difference vegetation index (DVI) had the greatest contribution to AGB estimation. This study compared shrub willow AGB estimation models using UAS imagery which will streamline bioenergy/biofuel development compared to the existing methods. [ABSTRACT FROM AUTHOR]

Copyright of Canadian Journal of Remote Sensing is the property of Taylor & Francis Ltd 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.)

Copyright of Canadian Journal of Remote Sensing is the property of Taylor & Francis Ltd 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.)

Detailed information on the spatial distribution of wetlands is crucial for sustainable management and resource assessment. Furthermore, regularly updated wetland inventories are of particular importance given that wetlands comprise a dynamic, rather than permanent, land condition. Accordingly, satellite-derived wetland maps are greatly beneficial, as they capture a synoptic and multi-temporal view of landscapes. Leveraging state-of-the-art remote sensing data and tools, this study produces a high-resolution 10-m wetland inventory map of Canada, covering an approximate area of one billion hectares, using multi-year, multi-source (Sentinel-1 and Sentinel-2) Earth Observation (EO) data on the Google Earth Engine™ cloud computing platform. The whole country is mapped using a large volume of reference samples using an object-based random forest classification scheme with an overall accuracy approaching 80% and individual accuracies varying from 74% to 84% in different provinces. This nationwide wetland inventory map illustrates that 19% of Canada's land area is covered by wetlands, most of which are peatlands dominate in the northern ecozones. Importantly, the resulting ever-demanding wetland inventory map of Canada provides unprecedented details on the extent, status, and spatial distribution of wetlands and thus, is useful for many stakeholders, including federal and provincial governments, municipalities, NGOs, and environmental consultants. [ABSTRACT FROM AUTHOR]

There are currently various classification algorithms, each with its own advantages and limitations. It is expected that fusing different classifiers in a way that the advantages of each are selected can boost the accuracy in the classification of complex land covers, such as wetlands, compared to using a single classifier. Classification of wetlands using remote-sensing methods is a challenging task because of considerable similarities between wetland classes. This fact is more important when utilizing synthetic aperture radar (SAR) data, which contain speckle noise. Consequently, discriminating wetland classes using only SAR data is generally not as accurate as using some other satellite data, such as optical imagery. In this study, a new Multiple Classifier System (MCS), which combines five different algorithms, was proposed to improve the classification accuracy of similar land covers. This system was then applied to classify wetlands in a study area in Newfoundland, Canada, using multi-source and multi-temporal SAR data. The results demonstrated that the proposed MCS was more accurate for the classification of wetlands in terms of both overall and class accuracies compared to applying one specific algorithm. Therefore, it is expected that the proposed system improves the classification accuracy of other complex landscapes. [ABSTRACT FROM AUTHOR]

The synergistic use of spatial features with spectral properties of satellite images enhances thematic land cover information, which is of great significance for complex land cover mapping. Incorporating spatial features within the classification scheme have been mainly carried out by applying just low-level features, which have shown improvement in the classification result. By contrast, the application of high-level spatial features for classification of satellite imagery has been underrepresented. This study aims to address the lack of high-level features by proposing a classification framework based on convolutional neural network (CNN) to learn deep spatial features for wetland mapping using optical remote sensing data. Designing a fully trained new convolutional network is infeasible due to the limited amount of training data in most remote sensing studies. Thus, we applied fine tuning of a pre-existing CNN. Specifically, AlexNet was used for this purpose. The classification results obtained by the deep CNN were compared with those based on well-known ensemble classifiers, namely random forest (RF), to evaluate the efficiency of CNN. Experimental results demonstrated that CNN was superior to RF for complex wetland mapping even by incorporating the small number of input features (i.e., three features) for CNN compared to RF (i.e., eight features). The proposed classification scheme is the first attempt, investigating the potential of fine-tuning pre-existing CNN, for land cover mapping. It also serves as a baseline framework to facilitate further scientific research using the latest state-of-art machine learning tools for processing remote sensing data. [ABSTRACT FROM AUTHOR]

Speckle degrades the radiometric quality of a Synthetic Aperture Radar (SAR) image. Previous methods for speckle reduction have used a fixed-size window for filtering the entire image. This, however, may not be effective for the entire image, as land covers of different sizes require different filtering windows. In this paper, a novel method is proposed by which each pixel in the image is filtered with a window appropriate for the size of object within it. The real in-phase and the imaginary quadrature components of the SAR images determine the best window size and the pixels in the intensity image are filtered using their own optimal windows. The proposed method is presented for both single- and multi-polarized SAR images, and the results of several common filters that were modified are presented. This approach is applied to two RADARSAT-2 images: one over San Francisco, California, USA and the other over St. John’s, Newfoundland and Labrador, Canada, producing results that were similar to, or outperformed, comparable filters while retaining details and suppressing speckle effectively. While the method was successful for single-look intensity data, it offers great potential for multi-look and amplitude data as well. [ABSTRACT FROM AUTHOR]

The production of spatially detailed quantitative maps of water level variations in flooded vegetation, and the detection of flow patterns and discontinuities in both managed and natural wetland ecosystems provide valuable information for monitoring these unique environments. Hydrological monitoring of wetlands is also critical for maintaining and preserving the habitat of various plant and animal species. Over the last two decades, advances in remote sensing technologies have supported wetland monitoring and management in several aspects, including classification, change detection, and water level monitoring. In particular, Interferometric Synthetic Aperture Radar (InSAR) has emerged as a promising tool for hydrological monitoring of wetland water bodies. However, a comprehensive review of the status, trends, techniques, advances, potentials, and limitations of this technique is lacking. In this study, we evaluate the use of InSAR for hydrological monitoring of wetlands, discuss the main challenges associated with this technique, recommend possible solutions to mitigate the main problems identified in the literature, and present opportunities for future research. RÉSUMÉ La production de cartes quantitatives détaillées de la variation du niveau d'eau de la végétation submergée et la détection des régimes d'écoulement et des discontinuités fournissent des renseignements importants sur les milieux humides, tant à l'état naturel que dans les zones aménagées. La surveillance hydrologique des milieux humides est essentielle au maintien et à la préservation de l'habitat de la faune et de la flore. Au cours des deux dernières décennies, les progrès techniques en télédétection ont contribué à la surveillance et à la gestion des milieux humides à plusieurs égards, dont la classification, la détection des changements et le suivi du niveau d'eau. En particulier, l'interférométrie par radar à synthèse d'ouverture (InSAR) s'est avérée un instrument prometteur pour la surveillance hydrologique des étendues d'eau en milieu humide. Il manque cependant une revue exhaustive de l'état, des tendances, des techniques, des avancés, du potentiel et des limitations de cette approche. Dans la présente étude, nous évaluons l'utilisation de l'InSAR pour la surveillance hydrologique des milieux humides, nous discutons des défis liés à cette technique, nous recommandons des solutions potentielles pour mitiger les problèmes principaux identifiés dans la littérature et nous présentons des possibilités de recherche future. [ABSTRACT FROM AUTHOR]

Speckle noise significantly degrades the radiometric quality of PolSAR image and, consequently, decreases the classification accuracy. This article proposes a new speckle reduction method for PolSAR imagery based on an adaptive Gaussian Markov Random Field model. We also introduce a new span image, called pseudo-span, obtained by the diagonal elements of the coherency matrix based on the least square analysis. The proposed de-speckling method was applied to full polarimetric C-band RADARSAT-2 data from the Avalon area, Newfoundland, Canada. The efficiency of the proposed method was evaluated in 2 different levels: de-speckled images and classified maps obtained by the Random Forest classifier. In terms of de-speckling, the proposed method illustrated approximately 19%, 43%, 46%, and 50% improvements in equivalent number of looks values, in comparison with SARBM3D, Enhanced Lee, Frost, and Kuan filter, respectively. Also, improvements of approximately 19%, 9%, 55%, and 32% were obtained in the overall classification accuracy using de-speckled PolSAR image by the proposed method compared with SARBM3D, Enhanced Lee, Frost, and Kuan filter, respectively. This new adaptive de-speckling method illustrates to be an efficient approach in terms of both speckle noise suppression and details/edges preservation, while having a great influence on the overall wetland classification accuracy. [ABSTRACT FROM AUTHOR]

Despite the fact that vast portions of Newfoundland and Labrador (NL) are covered by wetlands, currently there is no provincial inventory of wetlands in the province. In this study, we analyzed multi-temporal synthetic aperture radar (SAR) data for wetland classification at 4 pilot sites across NL. Object-based image analysis (OBIA) using a segmentation method based on optical data (RapidEye image in this study), and well-adjusted to SAR images, was first compared to pixel-based classification. Next, multi-date object-based wetland maps using the Random Forest classifier were compared to single-date classification. Finally, ratio and textural features were evaluated for wetland classification. The OBIA method demonstrated superior results, and the multi-date classification performed better than single-date classification with accuracies ranging from 75% to 95%. The multi-date results showed that the images acquired in August are the most appropriate for classifying wetlands, while the October images are of less value. Also, covariance matrix is a valuable feature set for wetland mapping. Moreover, ratio and textural features slightly increase the overall accuracy when the initial overall accuracy is relatively low. It can be concluded that multi-date SAR classification, with the proposed segmentation method, shows great potential for mapping wetlands and can be applied throughout the province. [ABSTRACT FROM AUTHOR]

Synthetic aperture radar (SAR) compact polarimetry (CP) systems are of great interest for large area monitoring because of their ability to acquire data in a wider swath compared to full polarimetry (FP) systems and a significant improvement in information content compared to single or dual polarimetry (DP) sensors. In this study, we compared the potential of DP, FP, and CP SAR data for wetland classification in a case study located in Newfoundland, Canada. The DP and CP data were simulated using full polarimetric RADARSAT-2 data. We compared the classification results for different input features using an object-based random forest classification. The results demonstrated the superiority of FP imagery relative to both DP and CP data. However, CP indicated significant improvements in classification accuracy compared to DP data. An overall classification accuracy of approximately 76% and 84% was achieved with the inclusion of all polarimetric features extracted from CP and FP data, respectively. In summary, although full polarimetric SAR data provide the best classification accuracy, the results demonstrate the potential of RADARSAT Constellation Mission for mapping wetlands in a large landscape. [ABSTRACT FROM AUTHOR]

Newfoundland and Labrador (NL) is the only province in Atlantic Canada that does not have a wetland inventory system. As a consequence, both classifying and monitoring wetland areas are necessary for wetland conservation and human services in the province. In this study, wetlands in 5 pilot sites, distributed across NL, were classified using multi-source and multi-temporal optical remote sensing images. The procedures involved the application of an object-based method to segment and classify the images. To classify the areas, 5 different machine learning algorithms were examined. The results showed that the Random Forest (RF) algorithm in combination with an object-based approach was the most accurate method to classify wetlands. The average producer and user accuracies of wetland classes considering all pilot sites were 68% and 73%, respectively. The overall classification accuracies, which considered the accuracy of all wetland and non-wetland classes varied from 86% to 96% across all pilot sites confirming the robustness of the methodology despite the biological, ecological, and geographical differences among the study areas. Additionally, we assessed the effects of the tuning parameters on the accuracy of results, as well as the difference between pixel-based and object-based methods for wetland classification in this study. [ABSTRACT FROM PUBLISHER]

Many ecosystems, particularly wetlands, are significantly degraded or lost as a result of climate change and anthropogenic activities. Simultaneously, developments in machine learning, particularly deep learning methods, have greatly improved wetland mapping, which is a critical step in ecosystem monitoring. Yet, present deep and very deep models necessitate a greater number of training data, which are costly, logistically challenging, and time-consuming to acquire. Thus, we explore and address the potential and possible limitations caused by the availability of limited ground-truth data for large-scale wetland mapping. To overcome this persistent problem for remote sensing data classification using deep learning models, we propose 3D UNet Generative Adversarial Network Swin Transformer (3DUNetGSFormer) to adaptively synthesize wetland training data based on each class's data availability. Both real and synthesized training data are then imported to a novel deep learning architecture consisting of cutting-edge Convolutional Neural Networks and vision transformers for wetland mapping. Results demonstrated that the developed wetland classifier obtained a high level of kappa coefficient, average accuracy, and overall accuracy of 96.99%, 97.13%, and 97.39%, respectively, for the data in three pilot sites in and around Grand Falls-Windsor, Avalon, and Gros Morne National Park located in Canada. The results show that the proposed methodology opens a new window for future high-quality wetland data generation and classification. The developed codes are available at https://github.com/aj1365/3DUNetGSFormer. • Large-scale complex wetland mapping methodology. • High performances and transferability for complex wetland mapping. • The use of feature extractor considerably improves wetland classification accuracy. • Validated in three study areas in Newfoundland in Canada through image interpretation and field data. [ABSTRACT FROM AUTHOR]

Well sites, including both well pads and exploratory core holes, are small polygonal landscape disturbance features approximately one half to one hectare (0.5–1 ha) in area, resulting from oil and gas exploration activities. Automatic extraction and monitoring of such small features using remote-sensing technology at regional scales has always been desirable for wildlife habitat monitoring and environmental planning and modelling. Due to the vast disturbances of well sites in a province like Alberta, Canada, high-resolution imagery is not practical for well site extraction. For operational purposes, mid-resolution and cost-effective satellite imagery such as Landsat is the choice. However, automatic well site extraction using mid-resolution satellite imagery is a challenging task. Wells are typically less than three pixels in width and length in a Landsat multispectral image. Furthermore, the spectral contrast between the well site pixels and the surrounding areas is low due to vegetation regrowth and the spectral complexity of the surrounding environment. This article presents a novel methodology for automatic extraction of well sites from Landsat-5 TM imagery. The method combines both pixel- and object-based image analyses and contains three major steps: geometric enhancement, segmentation, and well site extraction. The method was applied to Landsat-5 TM images acquired over Fort McMurray, Alberta, Canada. For accuracy assessment, four regions of interest were selected and the results of the proposed automatic method were evaluated against visual inspection of the Landsat-8 pan-sharpened image. The method results in a total average correctness, completeness, and quality measures of about 80, 96, and 77%, respectively over the four sites. In addition, the method is very fast as an entire Landsat scene is processed in less than 10 minutes. The method is an operational approach for automatic detection of well sites over the entire province and can dramatically reduce the labour cost of manual digitization for monitoring and updating well site maps. [ABSTRACT FROM PUBLISHER]

Copyright of Canadian Journal of Remote Sensing is the property of Taylor & Francis Ltd 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.)

This paper aims at exploiting the advantages of pixel-based and object-based image analysis approaches for urban land cover classification of very high resolution (VHR) satellite imagery through a combined object- and pixel-based image analysis framework. The framework starts with segmenting the image resulting in several spectral and spatial features of segments. To overcome the curse of dimensionality, a wavelet-based feature extraction method is proposed to reduce the number of features. The wavelet-based method is automatic, fast, and can preserve local variations in objects' spectral/spatial signatures. Finally, the extracted features together with the original bands of the image are classified using the conventional pixel-based Maximum Likelihood classification. The proposed method was tested on the World View-2, QuickBird, and Ikonos images of the same urban area for comparison purposes. Results show up to 17 percent, 10 percent, and 11 percent improvement in kappa coefficients compared to the case in which only the original bands of the image are used for WV-2, QB, and IK, respectively. Furthermore the objects' spectral features contribute more to increasing classification accuracy than spatial features. [ABSTRACT FROM AUTHOR]

Due to anthropogenic activities and climate change, many natural ecosystems, especially wetlands, are lost or changing at a rapid pace. For the last decade, there has been increasing attention towards developing new tools and methods for the mapping and classification of wetlands using remote sensing. At the same time, advances in artificial intelligence and machine learning, particularly deep learning models, have provided opportunities to advance wetland classification methods. However, the developed deep and very deep algorithms require a higher number of training samples, which is costly, logistically demanding, and time-consuming. As such, in this study, we propose a Deep Convolutional Neural Network (DCNN) that uses a modified architecture of the well-known DCNN of the AlexNet and a Generative Adversarial Network (GAN) for the generation and classification of Sentinel-1 and Sentinel-2 data. Applying to an area of approximately 370 sq. km in the Avalon Peninsula, Newfoundland, the proposed model with an average accuracy of 92.30% resulted in F-1 scores of 0.82, 0.85, 0.87, 0.89, and 0.95 for the recognition of swamp, fen, marsh, bog, and shallow water, respectively. Moreover, the proposed DCNN model improved the F-1 score of bog, marsh, fen, and swamp wetland classes by 4%, 8%, 11%, and 26%, respectively, compared to the original CNN network of AlexNet. These results reveal that the proposed model is highly capable of the generation and classification of Sentinel-1 and Sentinel-2 wetland samples and can be used for large-extent classification problems. [ABSTRACT FROM AUTHOR]

Algae serves as a food source for a wide range of aquatic species; however, a high concentration of inorganic nutrients under favorable conditions can result in the development of harmful algal blooms (HABs). Many studies have addressed HAB detection and monitoring; however, no global scale meta-analysis has specifically explored remote sensing-based HAB monitoring. Therefore, this manuscript elucidates and visualizes spatiotemporal trends in HAB detection and monitoring using remote sensing methods and discusses future insights through a meta-analysis of 420 journal articles. The results indicate an increase in the quantity of published articles which have facilitated the analysis of sensors, software, and HAB proxy estimation methods. The comparison across multiple studies highlighted the need for a standardized reporting method for HAB proxy estimation. Research gaps include: (1) atmospheric correction methods, particularly for turbid waters, (2) the use of analytical-based models, (3) the application of machine learning algorithms, (4) the generation of harmonized virtual constellation and data fusion for increased spatial and temporal resolutions, and (5) the use of cloud-computing platforms for large scale HAB detection and monitoring. The planned hyperspectral satellites will aid in filling these gaps to some extent. Overall, this review provides a snapshot of spatiotemporal trends in HAB monitoring to assist in decision making for future studies. [ABSTRACT FROM AUTHOR]

Land cover classification of very high resolution (VHR) imagery over urban areas is an extremely challenging task. Impervious land covers such as buildings, roads, and parking lots are spectrally too similar to be separated using only the spectral information of VHR imagery. Additional information, therefore, is required for separating such land covers by the classifier. One source of additional information is the vector data, which are available in archives for many urban areas. Further, the object-based approach provides a more effective way to incorporate vector data into the classification process as the misregistration between different layers is less problematic in object-based compared to pixel-based image analysis. In this research, a hierarchical rule-based object-based classification framework was developed based on a small subset of QuickBird (QB) imagery coupled with a layer of height points called Spot Height (SH) to classify a complex urban environment. In the rule-set, different spectral, morphological, contextual, class-related, and thematic layer features were employed. To assess the general applicability of the rule-set, the same classification framework and a similar one using slightly different thresholds applied to larger subsets of QB and IKONOS (IK), respectively. Results show an overall accuracy of 92% and 86% and a Kappa coefficient of 0.88 and 0.80 for the QB and IK Test image, respectively. The average producers' accuracies for impervious land cover types were also 82% and 74.5% for QB and IK. [ABSTRACT FROM AUTHOR]

Because of the sub-meter spatial resolution of very high resolution (VHR) optical satellite imagery, vehicles can be identified in this type of imagery. Further, because there is a time lag in image collection between the Panchromatic (Pan) and multispectral (MS) sensors onboard VHR satellites, a moving vehicle is observed by the satellite at slightly different times. Consequently, its velocity information including speed and direction can be determined. The higher spatial resolution and more spectral bands of WorldView-2 (WV2) imagery, compared to those of previous VHR satellites such as QuickBird and GeoEye-1, together with the new sensors' configuration of WV2, i.e., 4 bands on each side of the Pan sensor (MS1 and MS2), adds an opportunity to improve both moving vehicles extraction and the velocity estimation. In this paper, a novel processing framework is proposed for the automatic extraction of moving vehicles and determination of their velocities using single-pass WV2 imagery. The approach contains three major components: a) object-based road extraction, b) moving vehicle extraction from MS1 and MS2, and c) velocity estimation. The method was tested on two different areas of a WV2 image, a high speed and a low speed traffic zone. The method resulted in a correctness of 92% and a completeness of 77% for the extraction of moving vehicles. Furthermore, the estimated speeds and directions are very realistic and are consistent with the speed limits posted on the roads. The results demonstrate a promising potential for automatic and accurate traffic monitoring using a single image of WV2. [ABSTRACT FROM PUBLISHER]