In recent years, several powerful machine learning (ML) algorithms have been developed for image classification, especially those based on ensemble learning (EL). In particular, Extreme Gradient Boosting (XGBoost) and Light Gradient Boosting Machine (LightGBM) methods have attracted researchers’ attention in data science due to their superior results compared to other commonly used ML algorithms. Despite their popularity within the computer science community, they have not yet been well examined in detail in the field of Earth Observation (EO) for satellite image classification. As such, this study investigates the capability of different EL algorithms, generally known as bagging and boosting algorithms, including Adaptive Boosting (AdaBoost), Gradient Boosting Machine (GBM), XGBoost, LightGBM, and Random Forest (RF), for the classification of Remote Sensing (RS) data. In particular, different classification scenarios were designed to compare the performance of these algorithms on three different types of RS data, namely high-resolution multispectral, hyperspectral, and Polarimetric Synthetic Aperture Radar (PolSAR) data. Moreover, the Decision Tree (DT) single classifier, as a base classifier, is considered to evaluate the classification’s accuracy. The experimental results demonstrated that the RF and XGBoost methods for the multispectral image, the LightGBM and XGBoost methods for hyperspectral data, and the XGBoost and RF algorithms for PolSAR data produced higher classification accuracies compared to other ML techniques. This demonstrates the great capability of the XGBoost method for the classification of different types of RS data.

Wetlands across Canada have been, and continue to be, lost or altered under the influence of both anthropogenic and natural activities. The ability to assess the rate of change to wetland habitats and related spatial pattern dynamics is of importance for effective and meaningful management and protection, particularly under the current context of climate change. The availability of cloud-based geospatial platforms has allowed for the production of wetland maps at scales previously unfeasible due to technical limitations, yet the assessment of changes to wetlands at the level of the wetland class (bog, fen, swamp, and marsh) has yet to be implemented across Canada. Class-level change information is important when considering changes and impacts to wetland functions and services. To demonstrate this possibility, this study assessed 30 years of change to wetlands across the province of Newfoundland using Landsat imagery, spectral indices, and Random Forest classification within the Google Earth Engine (GEE) cloud-computing platform. Overall accuracies were high, ranging from 84.37% to 88.96%. In a comparison of different classifiers, Random Forest produced the highest over accuracy results and allowed for the estimation of variable importance, when compared Classification and Regression Tree (CART) and Minimum Distance (MD). The most important variables include the thermal infrared band (TIR), elevation, the difference vegetation index (DVI), the shortwave infrared bands (SWIR), and the normalized difference vegetation index (NDVI). Change detection analysis shows that bog, followed by swamp and fen, are the most common wetland classes across all time periods generally, and marsh wetlands are the least common wetland classes across all time periods respectively. The analysis also shows a general instability of wetland classes, though this is largely due to conversion from one wetland class to another. Future work may integrate RADAR data and consider weather patterns. The results of this study elucidate for the first time patterns of wetland class change across Newfoundland from 1985 to 2015 and demonstrate the potential of the GEE and Landsat historical imagery to assess change at provincial and national scales.

Oil spills are one of the most hazardous disasters with significant short- and long-term effects on fragile marine ecosystems. Synthetic Aperture Radar (SAR) has been considered an effective technology for mapping and monitoring oil spills in the marine environment, primarily thanks to its weather-, illumination-, and time-independent capabilities. To cope with serious oil spill threats, researchers have developed various analytical methodologies utilizing key advantages of SAR imagery to identify the occurrence of oil spills and discriminate lookalikes. Choosing the appropriate SAR specifications and investigating the effects of field conditions are challenging for oil spill monitoring and should be investigated further. This paper presents a comprehensive review study on maritime surveying and oil slick detection using SAR imagery through indexed research studies’ compilation and analysis. To this end, a total of 230 peer-reviewed papers, published in various remote sensing (RS) journals and 78 conference papers in the International Society for Photogrammetry and Remote Sensing (ISPRS) archive and the IEEE International Geoscience and Remote Sensing Symposium (IGARSS) proceedings were reviewed. Our review study represents a meta-analysis investigation of these papers focusing on several features, including data, sensor type, imaging mode, microwave carrier frequency (e.g., L-, C-, and X-bands), polarization option (i.e., single-pol, dual-pol, full-pol, and compact-pol), incidence angle, and wind speed. Furthermore, it provides an overview of the RS techniques developed to deal with the oil spill detection task. This paper can be a guideline for two groups of audiences; those interested in oil spill monitoring who want to get an overview of the problem and how to address it, and those already working in the field who want to understand the scope of the work being accomplished. Consequently, the current paper contributes both to academic RS research and to practical applications.

Despite their importance to ecosystem services, wetlands are threatened by pollution and development. Over the last few decades, a growing number of wetland studies employed remote sensing (RS) to scientifically monitor the status of wetlands and support their sustainability. Considering the rapid evolution of wetland studies and significant progress that has been made in the field, this paper constitutes an overview of studies utilizing RS methods in wetland monitoring. It investigates publications from 1990 up to the middle of 2022, providing a systematic survey on RS data type, machine learning (ML) tools, publication details (e.g., authors, affiliations, citations, and publications date), case studies, accuracy metrics, and other parameters of interest for RS-based wetland studies by covering 344 papers. The RS data and ML combination is deemed helpful for wetland monitoring and multi-proxy studies, and it may open up new perspectives for research studies. In a rapidly changing wetlands landscape, integrating multiple RS data types and ML algorithms is an opportunity to advance science support for management decisions. This paper provides insight into the selection of suitable ML and RS data types for the detailed monitoring of wetland-associated systems. The synthesized findings of this paper are essential to determining best practices for environmental management, restoration, and conservation of wetlands. This meta-analysis establishes avenues for future research and outlines a baseline framework to facilitate further scientific research using the latest state-of-art ML tools for processing RS data. Overall, the present work recommends that wetland sustainability requires a special land-use policy and relevant protocols, regulation, and/or legislation.

Over the last decade, developments in hyperspectral sensors have caused an increase in the use of hyperspectral images (HSIs) due to fine spectral resolution, which has lead to the precise extraction and identification of the materials of the observed scene. Change detection (CD) is one of the most fundamental applications of remote sensing, which provides timely change information about Earth's surface. This article presents unsupervised binary and multiple CD approaches using bitemporal HSIs based on spectral unmixing (SU) and proposes a new formulation for CD without using threshold selection methods. SU is a powerful technique for analyzing hyperspectral data. CD, by incorporating unmixing, has the potential to obtain subpixel-level information from data. The proposed framework is organized in the context of four steps: Step 1) estimating the number of endmembers, extracting the endmember's signature, calculating abundances, and applying similarity assessment to identify corresponding endmembers; Step 2) discriminating pure changed pixels; Step 3) discriminating mixed changed pixels, and combining results by pure changed pixels to produce a binary change map (BCM); and Step 4) labeling changed pixels based on the proposed formulation and producing a multiple change map. This article presents new strategies for binary and multiple CD, which provide a better figure of merit as they do not need the threshold selection methods and are less time-consuming. The experiments on real hyperspectral datasets revealed that both CD accuracy and computational performance benefited from these strategies. Experimental results proved the high performance of the proposed approaches for detecting changes in both binary and multiple cases.

Wetland is one of the most productive resources on earth, and it provides vital habitats for several unique species of flora and fauna. Over the last decade, mapping and monitoring wetlands by utilizing deep learning (DL) models and remote sensing data gained popularity due to the importance of wetland preservation. In general, DL-based methods have shown astonishing achievement in wetland classification, but some practical issues, such as limited training samples, still need to be addressed. Moreover, the performance of most of the DL approaches is decreased when moderate-resolution images with few features are used as input data. One solution to breaking the performance bottleneck of a single model is to fuse two or more of them. To this end, we strive to investigate and develop a multimodel DL algorithm for wetland classification based on the combination of a graph convolutional network (GCN) and a shallow convolutional neural network (CNN), which is called the Wet-GC algorithm hereinafter. In doing this, moderate-resolution Sentinel-1 (S1) synthetic aperture radar (SAR) and Sentinel-2 (S2) multispectral optical imagery are fed into the GCN and CNN models, respectively. As we know from the literature, the synergistic use of S1 SAR and S2 optical imagery can be used to extract different types of wetland features and increase the class discrimination possibility. Hence, wetland mapping by jointly using GCN and CNN has the ability to boost the wetland classification task. Findings indicate that the efficiency of Wet-GC with an overall accuracy (OA) of 88.68% outperforms the results obtained from random forest (OA = 84.88%), support vector machine (OA = 82.86%), extreme gradient boosting (OA = 86.55%), and ResNet50 (OA = 86.93%). The outcomes reveal that the Wet-GC architecture proposed in this article has an excellent capability to be applied over large areas with minimal need for training samples and can perform acceptably in supporting regional wetland mapping.