Search

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.)

Google Earth Engine (GEE) is a cloud-based geospatial processing platform for large-scale environmental monitoring and analysis. The free-to-use GEE platform provides access to (1) petabytes of publicly available remote sensing imagery and other ready-to-use products with an explorer web app; (2) high-speed parallel processing and machine learning algorithms using Google's computational infrastructure; and (3) a library of Application Programming Interfaces (APIs) with development environments that support popular coding languages, such as JavaScript and Python. Together these core features enable users to discover, analyze and visualize geospatial big data in powerful ways without needing access to supercomputers or specialized coding expertise. The development of GEE has created much enthusiasm and engagement in the remote sensing and geospatial data science fields. Yet after a decade since GEE was launched, its impact on remote sensing and geospatial science has not been carefully explored. Thus, a systematic review of GEE that can provide readers with the "big picture" of the current status and general trends in GEE is needed. To this end, the decision was taken to perform a meta-analysis investigation of recent peer-reviewed GEE articles focusing on several features, including data, sensor type, study area, spatial resolution, application, strategy, and analytical methods. A total of 349 peer-reviewed articles published in 146 different journals between 2010 and October 2019 were reviewed. Publications and geographical distribution trends showed a broad spectrum of applications in environmental analyses at both regional and global scales. Remote sensing datasets were used in 90% of studies while 10% of the articles utilized ready-to-use products for analyses. Optical satellite imagery with medium spatial resolution, particularly Landsat data with an archive exceeding 40 years, has been used extensively. Linear regression and random forest were the most frequently used algorithms for satellite imagery processing. Among ready-to-use products, the normalized difference vegetation index (NDVI) was used in 27% of studies for vegetation, crop, land cover mapping and drought monitoring. The results of this study confirm that GEE has and continues to make substantive progress on global challenges involving process of geo-big data. [ABSTRACT FROM AUTHOR]

Abstract The separability of wetland types using different spectral bands is an important subject, which has not yet been well studied in most countries. This is particularly of interest in Canada because it contains approximately one-fourth of the total global wetlands. In this study, the spectral separability of five wetland classes, namely Bog, Fen, Marsh, Swamp, and Shallow Water, was investigated in Newfoundland and Labrador (NL), Canada, using field data and multi-source optical Remote Sensing (RS) images. The objective was to select the most useful spectral bands for wetland studies from four commonly used optical satellites: RapidEye, Sentinel 2A, ASTER, and Landsat 8. However, because the ultimate objective was the classification of wetlands in the province, the separability of wetland classes was also evaluated using several other features, including various spectral indices, as well as textural and ratio features to obtain a high level of classification accuracy. For this purpose, two separability measures were used: The T-statistics, calculated from the parametric t -test method, and the U-statistics, derived from the non-parametric Mann-Whitney U-test. The results indicated that the Near Infrared (NIR) band was the best followed by the Red Edge (RE) band for the discrimination of wetland class pairs. The red band was also the third most useful band for separation of wetland classes, especially for the delineation of the Bog class from the other types. Although the Shortwave Infrared (SWIR) and green bands demonstrated poor separability, they were comparatively more informative than the Thermal Infrared (TIR) and blue bands. This study also demonstrated that ratio features and some spectral indices had high potential to differentiate the wetland species. Finally, wetlands in five study areas in NL were classified by inserting the best spectral bands and features into an object-based Random Forest (RF) classifier. By doing so, the mean Overall Accuracy (OA) and Kappa coefficient in the study areas were 86% and 0.82, respectively. [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]

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]

Wetlands are important ecosystems that are linked to climate change mitigation. As 25% of global wetlands are located in Canada, accurate and up-to-date wetland classification is of high importance, nationally and internationally. The advent of deep learning techniques has revolutionized the current use of machine learning algorithms to classify complex environments, specifically in remote sensing. In this paper, we explore the potential and possible limitations to be overcome regarding the use of ensemble deep learning techniques for complex wetland classification and discusses the potential and limitation of various solo convolutional neural networks (CNNs), including DenseNet, GoogLeNet, ShuffleNet, MobileNet, Xception, Inception-ResNet, ResNet18, and ResNet101 in three different study areas located in Newfoundland and Labrador, Canada (i.e., Avalon, Gros Morne, and Grand Falls). Moreover, to improve the classification accuracies of wetland classes of bog, fen, marsh, swamp, and shallow water, the results of the three best CNNs in each study area is fused using three supervised classifiers of random forest (RF), bagged tree (BTree), Bayesian optimized tree (BOT), and one unsupervised majority voting classifier. The results suggest that the ensemble models, in particular BTree, have a valuable role to play in the classification of wetland classes of bog, fen, marsh, swamp, and shallow water. The ensemble CNNs show an improvement of 9.63–19.04% in terms of mean producer's accuracy compared to the solo CNNs, to recognize wetland classes in three different study areas. This research indicates a promising potential for integrating ensemble-based learning and deep learning for operational large area land cover, particularly complex wetland type classification. [ABSTRACT FROM AUTHOR]

Thanks to increasing urban development, it has become important for municipalities to understand how ecological processes function. In particular, urban wetlands are vital habitats for the people and the animals living amongst them. This is because wetlands provide great services, including water filtration, flood and drought mitigation, and recreational spaces. As such, several recent urban development plans are currently needed to monitor these invaluable ecosystems using time- and cost-efficient approaches. Accordingly, this study is designed to provide an initial response to the need of wetland mapping in the City of St. John's, Newfoundland and Labrador (NL), Canada. Specifically, we produce the first high-resolution wetland map of the City of St. John's using advanced machine learning algorithms, very high-resolution satellite imagery, and airborne LiDAR. An object-based random forest algorithm is applied to features extracted from WorldView-4, GeoEye-1, and LiDAR data to characterize five wetland classes, namely bog, fen, marsh, swamp, and open water, within an urban area. An overall accuracy of 91.12% is obtained for discriminating different wetland types and wetland surface water flow connectivity is also produced using LiDAR data. The resulting wetland classification map and the water surface flow map can help elucidate a greater understanding of the way in which wetlands are connected to the city's landscape and ultimately aid to improve wetland-related conservation and management decisions within the City of St. John's. • Producing the first high-resolution wetland map of the City of St. John's using remote sensing data and tools. • Producing the first wetland connectivity map of the City of St. John's. • Applying an advanced machine learning algorithm in an object based framework. • Using very high resolution multi-spectral and airborne LiDAR data in order to produce wetland map at 1 m resolution. [ABSTRACT FROM AUTHOR]