Several machine-learning algorithms have been proposed for remote sensing image classification during the past two decades. Among these machine learning algorithms, Random Forest (RF) and Support Vector Machines (SVM) have drawn attention to image classification in several remote sensing applications. This paper reviews RF and SVM concepts relevant to remote sensing image classification and applies a meta-analysis of 251 peer-reviewed journal papers. A database with more than 40 quantitative and qualitative fields was constructed from these papers. The meta-analysis mainly focuses on: (1) the analysis regarding the general characteristics of the studies, such as geographical distribution, frequency of the papers considering time, journals, application domains, and remote sensing software packages used in the case studies, and (2) a comparative analysis regarding the performances of RF and SVM classification against various parameters, such as data type, RS applications, spatial resolution, and the number of extracted features in the feature engineering step. The challenges, recommendations, and potential directions for future research are also discussed in detail. Moreover, a summary of the results is provided to aid researchers to customize their efforts in order to achieve the most accurate results based on their thematic applications.

Since the rise of deep learning in the past few years, convolutional neural networks (CNNs) have quickly found their place within the remote sensing (RS) community. As a result, they have transitioned away from other machine learning techniques, achieving unprecedented improvements in many specific RS applications. This paper presents a meta-analysis of 416 peer-reviewed journal articles, summarizes CNN advancements, and its current status under RS applications. The review process includes a statistical and descriptive analysis of a database comprised of 23 fields, including 1) general characteristics such as various applications, study objectives, sensors, and data types, and 2) algorithm specifications such as different types of CNN models, parameter settings, and reported accuracies. This review begins with a comprehensive survey of the relevant articles without considering any specific criteria to give readers an idea of general trends, then investigates CNNs within different RS applications to provide specific directions for the researchers. Finally, a conclusion summarizes potentialities, critical issues, and challenges related to the observed trends.

Hyperspectral imaging has recently emerged in the geosciences as a technology that provides rapid, accurate, and high-resolution information from lake sediment cores. Here we introduce a new methodology to infer particle size distribution, an insightful proxy that tracks past changes in aquatic ecosystems and their catchments, from laboratory hyperspectral images of lake sediment cores. The proposed methodology includes data preparation, spectral preprocessing and transformation, variable selection, and model fitting. We evaluated random forest regression and other commonly used statistical methods to find the best model for particle size determination. We tested the performance of combinations of spectral transformation techniques, including absorbance, continuum removal, and first and second derivatives of the reflectance and absorbance, along with different regression models including partial least squares, multiple linear regression, principal component regression, and support vector regression, and evaluated the resulting root mean square error (RMSE), R-squared, and mean relative error (MRE). Our results show that a random forest regression model built on spectra absorbance significantly outperforms all other models. The new workflow demonstrated herein represents a much-improved method for generating inferences from hyperspectral imagery, which opens many new opportunities for advancing the study of sediment archives.

Relative radiometric normalization is often required in time series analysis of satellite Earth observations such as land cover change detection. Normalization process reduces the radiometric differences caused by changes in the environmental conditions during the acquisition of multitemporal satellite images. In this paper, we proposed an efficient and automatic method based on Gaussian mixture model (GMM) to find a set of subjectively chosen invariant pixels. A linear model, based on Error Ellipse, was then adjusted to normalize the subject image. The proposed method involves two main steps; in the first step, invariant pixels, which are known as most probable unchanged pixels, were obtained by analyzing image differences estimated by GMMs. Then, these pixels were used to model the relationship between two multitemporal images. To evaluate the proposed method in real analysis scenarios, three multitemporal datasets acquired by different satellite sensors such as Ikonos, Quickbird, SuperView-1, and Worldview-2 were analyzed. These images were collected before and after the 2011's Japan and the 2004's Indonesia Tsunamis, and the 2017's Iran–Iraq earthquake. Experimental results demonstrated that the proposed method can considerably improve the radiometric variations between temporal images for change detection applications.

Covering 55% of Canada’s total surface area and stretching from coast to coast to coast, the Canadian boreal zone is crucial to the nation’s economic and ecological integrity. Although often viewed as relatively underdeveloped, it is vulnerable to numerous stressors such as mining, forestry, and anthropogenic climate change. Natural archives preserved in lake sediments can provide key insights by quantifying pre-disturbance conditions (pre-1850 CE) and the nature, magnitude, direction, and speed of environmental change induced by anthropogenic stressors over the past ∼150 years. Here, we paired a review of paleolimnological literature of the Canadian boreal zone with analyses of published sediment core data to highlight the effects of climate change, catchment disturbances, and atmospheric deposition on boreal lakes. Specifically, we conducted quantitative syntheses of two lake health indicators: elemental lead (Pb) and chlorophyll a. Segmented regressions and Mann–Kendall trend analysis revealed a generally increasing trend in elemental Pb across the boreal zone until ∼1970 CE, followed by a generally decreasing trend to the present. Snapshot comparisons of sedimentary chlorophyll a from recent and pre-industrial sediments (i.e., top-bottom sediment core design) revealed that a majority of sites have increased over time, suggesting a general enhancement in lake primary production across the boreal zone. Collectively, this body of work demonstrates that long-term sediment records offer a critical perspective on ecosystem change not accessible through routine monitoring programs. We advocate using modern datasets in tandem with paleolimnology to establish baseline conditions, measure ecosystem changes, and set meaningful management targets.

Lakes provide many vital ecosystem services but are highly sensitive to human impacts and climate change. While data needed to assess their long-term trajectories are generally unavailable, lake responses to such perturbations can be inferred from sediment archives, including records of chlorophyll, a proxy for primary production. However, few if any datasets exist to test hypotheses at large spatial scales about the drivers of long-term change. Here, we apply rapid hyperspectral imaging to produce high-resolution records of sediment chlorophyll (defined as the sum of chlorophyll a and b and their degradation products) in dated cores from 80 lakes across Canada and identify an overall increase since the mid-19th-century. Breakpoint analysis detected a shift whereby the average rate of chlorophyll increase was seven times greater after 1966 than before. We then compared spatiotemporal chlorophyll trends to climate variables, solar irradiance, and land-use data to assess their relationships to past changes in primary production. We identified robust correlations between chlorophyll and climate variables (temperature and ice cover) that were especially strong after 1966, suggesting continental-scale alterations to lake primary production in response to recent accelerated warming. We also detected significant associations with incident solar radiation and catchment human impacts, and show that both had the potential to influence past primary production at the continental scale.