In recent years, the data science and remote sensing communities have started to align due to user-friendly programming tools, access to high-end consumer computing power, and the availability of free satellite data. In particular, publicly available data from the European Space Agency’s Sentinel missions have been used in various remote sensing applications. However, there is a lack of studies that utilize these data to assess the performance of machine learning algorithms in complex boreal landscapes. In this article, I compare the classification performance of four non-parametric algorithms: support vector machines (SVM), random forests (RF), extreme gradient boosting (Xgboost), and deep learning (DL). The study area chosen is a complex mixed-use landscape in south-central Sweden with eight land-cover and land-use (LCLU) classes. The satellite imagery used for the classification were multi-temporal scenes from Sentinel-2 covering spring, summer, autumn and winter conditions. Using stratified random sampling, each LCLU class was allocated 1477 samples, which were divided into training (70%) and evaluation (30%) subsets. Accuracy was assessed through metrics derived from an error matrix, but primarily overall accuracy was used in allocating algorithm hierarchy. A two-proportion Z-test was used to compare the proportions of correctly classified pixels of the algorithms and a McNemar’s chi-square test was used to compare class-wise predictions. The results show that the highest overall accuracy was produced by support vector machines (0.758 ± 0.017), closely followed by extreme gradient boosting (0.751 ± 0.017), random forests (0.739 ± 0.018), and finally deep learning (0.733 ± 0.0023). The Z-test comparison of classifiers showed that a third of algorithm pairings were statistically different. On a class-wise basis, McNemar’s test results showed that 62% of class-wise predictions were significant from one another at the 5% level or less. Variable importance metrics show that nearly half of the top twenty Sentinel-2 bands belonged to the red edge (25%) and shortwave infrared (23%) portions of the electromagnetic spectrum, and were dominated by scenes from spring (38%) and summer (40%). The results are discussed within the scope of recent studies involving machine learning and Sentinel-2 data and key knowledge gaps identified. The article concludes with recommendations for future research.

Increasing aridity is one major consequence of ongoing global climate change and is expected to cause widespread changes in key ecosystem attributes, functions, and dynamics. This is especially the case in naturally vulnerable ecosystems, such as drylands. While we have an overall understanding of past aridity trends, the linkage between temporal dynamics in aridity and dryland ecosystem responses remain largely unknown. Here, we examined recent trends in aridity over the past two decades within global drylands as a basis for exploring the response of ecosystem state variables associated with land and atmosphere processes (e.g., vegetation cover, vegetation functioning, soil water availability, land cover, burned area, and vapor‐pressure deficit) to these trends. We identified five clusters, characterizing spatiotemporal patterns in aridity between 2000 and 2020. Overall, we observe that 44.5% of all areas are getting dryer, 31.6% getting wetter, and 23.8% have no trends in aridity. Our results show strongest correlations between trends in ecosystem state variables and aridity in clusters with increasing aridity, which matches expectations of systemic acclimatization of the ecosystem to a reduction in water availability/water stress. Trends in vegetation (expressed by leaf area index [LAI]) are affected differently by potential driving factors (e.g., environmental, and climatic factors, soil properties, and population density) in areas experiencing water‐related stress as compared to areas not exposed to water‐related stress. Canopy height for example, has a positive impact on trends in LAI when the system is stressed but does not impact the trends in non‐stressed systems. Conversely, opposite relationships were found for soil parameters such as root‐zone water storage capacity and organic carbon density. How potential driving factors impact dryland vegetation differently depending on water‐related stress (or no stress) is important, for example within management strategies to maintain and restore dryland vegetation.

The crossing of aridity thresholds triggers abrupt changes in multiple functional and structural ecosystem attributes across global drylands. While we understand the consequences associated with aridity thresholds, the key factors influencing dryland vegetation resistance when crossing them remain unclear. Here, we used field observations from 58 dryland sites across five continents and satellite remote sensing data (2000-2022) to show that plant richness, soil moisture dynamics and texture, and bare soil fraction are important variables contributing to vegetation resistance. Additionally, drought history (frequency and magnitude of past droughts) is important in interaction with plant richness and soil texture. Interestingly, plant species richness was negatively related to vegetation resistance, except in areas with higher drought history and in grasslands. Our results highlight that vegetation resistance depends on local environmental conditions. Enhancing our understanding of the factors important for vegetation resistance is an important step towards dryland conservation efforts and sustainable management strategies.

Savannas cover a wide climatic gradient across large portions of the Earth’s land surface and are an important component of the terrestrial biosphere. Savannas have been undergoing changes that alter the composition and structure of their vegetation such as the encroachment of woody vegetation and increasing land-use intensity. Monitoring the spatial and temporal dynamics of savanna ecosystem structure (e.g., partitioning woody and herbaceous vegetation) and function (e.g., aboveground biomass) is of high importance. Major challenges include misclassification of savannas as forests at the mesic end of their range, disentangling the contribution of woody and herbaceous vegetation to aboveground biomass, and quantifying and mapping fuel loads. Here, we review current (2010–present) research in the application of satellite remote sensing in savannas at regional and global scales. We identify emerging opportunities in satellite remote sensing that can help overcome existing challenges. We provide recommendations on how these opportunities can be leveraged, specifically (1) the development of a conceptual framework that leads to a consistent definition of savannas in remote sensing; (2) improving mapping of savannas to include ecologically relevant information such as soil properties and fire activity; (3) exploiting high-resolution imagery provided by nanosatellites to better understand the role of landscape structure in ecosystem functioning; and (4) using novel approaches from artificial intelligence and machine learning in combination with multisource satellite observations, e.g., multi-/hyperspectral, synthetic aperture radar (SAR), and light detection and ranging (lidar), and data on plant traits to infer potentially new relationships between biotic and abiotic components of savannas that can be either proven or disproven with targeted field experiments.