Satellite Images Research Articles

UNet-GCViT: a UNet-based framework with global context vision transformer blocks for building damage detection

ABSTRACT Natural and man-made disasters take place around the world and cause significant financial and human losses. An accurate and fast post-disaster building damage mapping could play a crucial role in rapid rescue planning and operations. Remote sensing satellite images are the main source of building damage map generation. Usually, both pre-disaster and post-disaster satellite images are used for the generation of building damage maps, which encounter some challenges such as registration errors, noise, and atmospheric conditions. This study proposed a new method for building damage detection based only on post images with the UNet architecture network and Global Context Vision Transformer blocks. The deep learning network proposed in this research is automatic without any further processing. The proposed method comprises four main steps: (1) pre-processing, (2) network training, (3) building damage map generation, and (4) accuracy assessment for the final damage map. This network is applied to three different natural and man-made disaster datasets. The first dataset is the post-satellite image of the 2023 Turkey earthquake, the second one is the post-satellite image of the 2021 Bata explosion and the last one is the post-satellite image of the 2011 haiti earthquake. Results of the final building damage map indicate that the proposed method is highly effective with OA above 96%, which is superior to the other deep learning methods.

Interpretable LAI Fine Inversion of Maize by Fusing Satellite, UAV Multispectral, and Thermal Infrared Images

Interpretable LAI Fine Inversion of Maize by Fusing Satellite, UAV Multispectral, and Thermal Infrared Images

War-induced soil disturbances in north-eastern Ukraine (Kharkiv region): Physical disturbances, soil contamination and land use change.

The war in Ukraine is having a dramatic impact on the physical, chemical and biological soil properties. A comprehensive study of the war-affected soils during the ongoing war is a challenging task owing to the many constrains that arise during fieldworks. Remote sensing data is the best solution for overall analysis of physical soil disturbances. With the advancement in the very-high spatial resolution satellite imagery, it has become feasible to identify precise indicators of the war-induced physical soil disturbances. Here, we present a comprehensive methodology for assessing the war-affected soils tested in the Kharkiv region (north-eastern Ukraine). The methodology combines analysis of remote sensing data, soil mapping and subsequent soil modelling, field research and laboratory measurements. High-resolution satellite images, which acquired via WorldView and GeoEye, have enabled the identification of 420,615 craters and 3411ha of soils, which impacted by the movement of heavy military vehicles. Soil bombturbation covers 4212ha and is represented by entire or partial soil mixing, as well as the soil burial under the ejected material. The research showed that craters become the sites of secondary geomorphic processes that lead to crater filling and degradation of the surrounded areas. Furthermore, all sampled craters showed enrichment in heavy metals, especially cadmium, copper, lead and zinc. However, their concentration is predominantly only 1.1-1.5 times more than background value and almost does not exceed threshold limit values in Ukraine. Conversely, the sites of military helicopter crash and 'fresh' missile falling site are displayed critical contamination with heavy metals. The paper also discusses the feasibility of combining data from soil physical disturbances with land use change in order to assess the risk of landmines and the presence of unexploded ordnance, environmental effect of which is remain unclear in Ukraine.

Assessing the potential effects of climate change on the morphodynamics of the tropical coral reef islands in the Gulf of Mannar, Indian Ocean.

Low-lying and small tropical coral reef islands around the world are extremely vulnerable to the effects of global environmental change caused by the combination of anthropogenic climate change and escalating extreme hydrodynamic events. Erosion and inundation are anticipated to physically destabilize the tropical coral reef islands, rendering them uninhabitable within the next century. Therefore, it is crucial to assess the repercussions of these hazardous events on the delicate reef island ecosystem in order to conserve and ensure sustainable management. Multitemporal remotely sensed Landsat satellite imageries were utilized to investigate the net and decadal morphological transformation of tropical coral reef islands in the Gulf of Mannar, Indian Ocean. Over the past half-century, these islands have consistently adapted to global environmental changes, even while local sea levels rise at a rate of 3.38mm per year. Advanced statistical techniques, such as net shoreline movement (NSM), end point rate (EPR), and linear regression rate (LRR), were employed for estimating the shoreline change rate using a Digital Shoreline Analysis System (DSAS). In addition, the GIS-based overlay analysis methods were applied to examine the net and decadal areal (planform) changes and also utilized for estimating the inundation trajectories of reef islands under the sea level rise scenarios of 1m and 2m. Furthermore, time series analysis was performed to analyze the variability of critical climate-induced factors using archived reanalysis oceanographic data. In addition, linear and polynomial statistical techniques were applied to investigate the driving factors behind the coral reef island morphological transition. The findings show that two islands have already disappeared, while others have experienced a dramatic reduction in their footprint. Approximately 62.64% of the shoreline experienced significant erosion, while 36.91% witnessed gradual accretion. The Tuticorin group confronted the severe reduction in island footprint, with a significant decrease of 83.04%, followed by Keelakarai groups (33.35%), Mandapam groups (29.60%), Vembar groups (28.14%), and Rameswaram islands (3.43%). The study also predicts that the island footprint could submerge in an area of 627.30ha and 1284.21ha within the next century, with an expected sea level rise (SLR) of 1m and 2m, respectively. The study emphasizes that the combination of human-induced factors and regional coastal processes such as sea level rise and swells are the key drivers engendering the stress on the physical resilience of the coral reef islands. Urgent and continual monitoring of the reef islands is crucial for a better understanding of their dynamic trajectories and for developing nature-based solutions to catastrophic erosion. These nature-based solutions (NbS) for minimizing island erosion are initiatives that use natural ecosystems to safeguard islands while enhancing biodiversity, climate resilience, and community livelihoods. The interactions between nature-based solutions (NbS) for combating erosion, reef island resilience, and Sustainable Development Goals are evaluated based on the positive correlation, our expert knowledge, and Griggs et al.'s 2017 seven-point scale framework. The outcomes of this study may provide comprehensive insights to decision-makers and administrators for formulating and implementing policies for long-term resilience building and sustainable island management.

Monitoring the rehabilitation process of the windthrow area using UAS images and performance comparison of Sentinel-2A based different vegetation indexes

Windthrows significantly disrupt forest ecosystems, impacting biotic community life cycles. To ensure the reformation of the ecosystem chain, it is essential to rehabilitate the windthrow area as soon as possible. Therefore, it is mandotory to determine the success of the rehabilitation processes. In this study, the rehabilitation process of windthrow that occurred in the Düzce Tatlıdere Forest District (DTFD) was identified using vegetation indices calculated from Unmanned Aircraft System (UAS) images and Sentinel-2A satellite images between 2017 and 2022. The Normalized Difference Red Edge Index (NDRE), Plant Senescence Reflectance Index (PSRI), and Normalized Difference Vegetation Index (NDVI) were calculated from Sentinel-2A satellite images, and the most successful index for detecting reforested areas was identified. UAS images were used to create training data, and this data was used to classify Sentinel-2A images with the Random Forest (RF) algorithm. The classification’s accuracy was assessed using the Kappa Coefficient and Overall Accuracy (%). Results showed that NDVI had the lowest accuracy in both years, whereas NDRE succesfully detected windthrow area borders. PSRI was most successful in monitoring rehabilitation progress and detecting reforested areas between 2017 and 2022. This study, he effectiveness and limitations of the NDRE, PSRI and NDVI indices in the rehabilitation process of the windthrow area have been detected, and the most important Sentinel-2A bands were determined based on the results of the RF classification. This study is pioneering in the use of NDRE and PSRI to detect reforested areas post-windthrow.

‘DeltaCAN - A new data set of Canadian Arctic and subarctic coastal deltas’

Arctic coasts constitute the critical interface between land and sea, and are subject to rapid changes caused by a warming climate. Current trends throughout the Arctic show increasing erosion trends, while other parts of the coast are experiencing prograding trends. Until now, a vast majority of our knowledge of Arctic coastal evolution is confined to site-specific studies with limited geospatial representation. Here, we present DeltaCAN, a novel data set on the locations of Canadian deltas larger than 500 m in width derived by visual interpretation of freely available satellite imagery. DeltaCAN is Canada’s first nationwide coastal detection covering 250.000 km of coastline in the Arctic, identifying 2712 deltas. The inventory is based on inspection of remotely-sensed satellite imageries, developed through an expert-based mapping approach where we implemented a quality control mechanism to assess the completeness of the data set. The DeltaCAN data set allows for assessing changes at an unprecedented spatial extent, improving our understanding of delta morphodynamics.

Assessing water quality restoration measures in Lake Pampulha (Brazil) through remote sensing imagery.

Urban reservoirs are frequently exposed to impacts from high population density, polluting activities, and the absence of environmental control measures and monitoring. In this study, we investigated the use of satellite imagery to assess restoration measures and support decision-making in a hypereutrophic urban reservoir. Since 2016, Lake Pampulha (Brazil) has undergone restoration measures, including the application of Phoslock®, to mitigate its poor water quality conditions. Satellite images from Landsat-8 (L8) and Sentinel-2 (S2) and historical monitoring data were used to estimate total suspended matter (TSM), chlorophyll-a (Chla), and Secchi disk depth (SDD). We explored both established models from existing literature and novel alternatives, the latter employing machine learning approaches. Model performance was assessed through the coefficient of determination (R ) during calibration and using the root mean square error (RMSE) and its normalization by the observed mean (nRMSE) during validation. Random forest regressor presented superior performance for the three water quality parameters and both satellites. The performance metrics for Landsat-8 were TSM (n=44) R of 0.62; Chla (n=48) R of 0.74; and SDD (n=23) R of 0.67. For Sentinel-2, TSM (n=53) R of 0.72; Chla (n=54) R of 0.64; SDD (n=33) R of 0.89. The selected models were then applied to cloud-free images from 2013 to 2022 to provide modeled TSM, Chla, and SDD values. These were used for spatiotemporal analysis, encompassing cluster analysis, sample comparisons, and trend analysis. Three distinct regions within Lake Pampulha were identified, with the upstream region displaying poorer conditions for TSM, Chla, and SDD. Restoration measures have contributed positively to improving water quality in Lake Pampulha; however, rainy periods pose a challenge due to water quality deterioration associated with urban runoff. Remote sensing has proven to be a robust tool for assisting managers in assessing recovery measures in urban lakes.

Optimizing Deep Learning Models for Fire Detection, Classification, and Segmentation Using Satellite Images

Optimizing Deep Learning Models for Fire Detection, Classification, and Segmentation Using Satellite Images

Spatiotemporal variation in biomass abundance of different algal species in Lake Hulun using machine learning and Sentinel-3 images

Climate change and human activities affect the biomass of different algal and the succession of dominant species. In the past, phytoplankton phyla inversion has been focused on oceanic and continental shelf waters, while phytoplankton phyla inversion in inland lakes and reservoirs is still in the initial and exploratory stage, and the research results are relatively few. Especially for mid-to-high latitude lakes, the research is even more blank. Therefore, this study proposes a machine learning method based on OLCI/Sentinel-3 satellite imagery to retrieve algal biomass abundance. Remote sensing models were developed to estimate the biomass abundance of three major algal groups: Cyanophyta, Chlorophyta, and Bacillariophyta. This study compared and evaluated 6 commonly used machine learning models, including extreme gradient boosting (XGBoost), support vector regression (SVR), backpropagation neural network (BP), gradient boosting decision tree (GBDT), random forest (RF), and categorical boosting (CatBoost). The results indicated that XGBoost exhibited the highest accuracy (R2 = 0.92, RMSE = 1.78%, MAPE = 9.96%) in estimating Cyanophyta’s biomass abundance. The RF model demonstrated the highest accuracy for estimating Chlorophyta’s biomass abundance (R2 = 0.72, RMSE = 6.57%, MAPE = 50.8%), while the GBDT model exhibited the highest accuracy for estimating Bacillariophyta’s biomass abundance (R2 = 0.9, RMSE = 4.66%, MAPE = 47.87%). The models were subsequently applied to all cloud-free OLCI images from Hulun Lake during the ice-free periods from 2016 to 2023, producing spatiotemporal distribution maps of the different phytoplankton biomass abundance. Cyanophyta dominated the biomass abundance (44.62 ± 3.47%), followed by Bacillariophyta (36.35 ± 2.68%), and Chlorophyta had the lowest proportion (10.42 ± 1.08%). Together, these three algae groups constituted 91.4 ± 1.55% of all phytoplankton in Hulun Lake. Significant annual variations in the biomass abundance of Cyanophyta and Bacillariophyta were observed, whereas those of Chlorophyta remained stable. Additionally, this study examined the effects of climatic factors and water quality parameters on the biomass abundance of algae. The findings suggest that temperature, wind speed, and atmospheric pressure are critical factors influencing the biomass abundance of the different algae groups. This study not only fills the gaps in the related field, but also provides a new method for monitoring algae, as well as a strong support for realizing the goals of sustainable management of water resources and ecological protection.

SPS-UNet: a super-pixel sampling UNet for extracting buildings from high-resolution satellite images

SPS-UNet: a super-pixel sampling UNet for extracting buildings from high-resolution satellite images

The first geospatial dataset of irrigated fields (2020–2024) in Vojvodina (Serbia)

Irrigation is a cornerstone of global food security, enabling sustainable agricultural production and helping to ensure that food is available for people around the world, now and in the future. Mapping irrigated fields provides valuable information for sustainable water management, agricultural development, and environmental conservation efforts. However, the collection of high-quality training data, which is necessary for accurate irrigation mapping remains costly and labour-intensive. To address this, we created a georeferenced regional dataset consisting of location, crop type, and occurrence of the irrigation equipment which are essential information for mapping irrigated fields. Four main irrigated crops were considered: maize, soybean, sugar beet, and wheat. The dataset, consisting of a total of 1256 parcels, is created for Vojvodina, the main agricultural area in Serbia, spanning the period of five years (2020–2024). This study’s goal is to give accessibility to our dataset which further can be explored and used for building or fine-tuning machine learning and deep learning models for the automatic detection of irrigated fields using satellite imagery.

BDCN_UNet: Advanced shoreline extraction techniques integrating deep learning

The delineation of shorelines, marking the dynamic boundary between land and sea, is crucial for coastal management and ecological conservation. This research investigates various methods for automatic shoreline extraction from satellite images, utilizing cost-effective and regularly available remote sensing data. A deep learning framework is used for the comparison that combines semantic segmentation (UNEt) and edge detection (Bi-Directional Cascade Network, or BDCN) to make the results more accurate. The study focuses on the northern coast of Egypt, a region with diverse landscapes and significant human interventions. We’ve made progress in our methods by using advanced image processing, the CoastSat Toolkit, the BDCN_UNet framework, time series analysis, and deep learning networks to look at and keep an eye on Egypt’s North Coast’s future plans in more detail. This study captures the intricate interplay between natural and human factors in coastal environments, offering valuable insights for sustainable coastal development. The performance analysis of four models (UNet, Pyramid Scene Parsing Network, DeepLabV3, and BDCN_UNet) applied to three satellite datasets (Sentinel-2, PlanetScope, and Pleiades) showed that PlanetScope and Pleiades consistently yield higher accuracy (0.90–0.99), while Sentinel-2 proved more challenging, particularly for Pyramid Scene Parsing Network and DeepLabV3. UNet and BDCN_UNet exhibited robust performance across all datasets, especially with Pleiades delivering the best results overall.

Land use and cover changes and sand fly (Diptera: Psychodidae) assemblages in an emerging focus of leishmaniasis.

Land use and cover changes lead to fragmentation of the natural habitats of sand flies and modify the epidemiological profile of leishmaniasis. This process contributes to the infestation of adjacent rural settlements by vector sand fly species with different degrees of adaptation, promoting leishmaniasis outbreaks. This study aimed to assess land use and cover changes over a 12-year period and investigate the diversity and abundance of sand fly assemblages in the rural area of Codó, Maranhão State, Brazil. Temporal analysis of land use and cover changes was conducted using Sentinel-2 satellite imagery treated in QGIS software (free version 3.10) and classified using Orfeo Toolbox. Sand flies were sampled in alternate months between August 2022 and June 2023 using Centers for Disease Control and Prevention (CDC) light traps and white and black Shannon traps installed in peridomestic and extradomestic environments. Map images showed that the predominant land covers in 2012 and 2014 were dense and sparse vegetation, with few buildings. In 2021 and 2023, areas of sparse and dense vegetation were fragmented as new settlements were established. The entomological survey resulted in the capture of 3375 sand flies (CDC = 856, white Shannon = 650, black Shannon = 1969) belonging to 20 species. The most abundant species were Psychodopygus wellcomei Fraiha, Shaw & Lainson, 1971 (78.19%), followed by Nyssomyia whitmani (Antunes & Coutinho, 1939) (7.53%). Additionally, Ny. whitmani was the most abundant species (84.97%) in peridomestic environments, whereas Ps. wellcomei was the most abundant species (96.51%) in extradomestic environments. The sand fly assemblage was highly diverse, with a high abundance of competent vectors of Leishmania spp. These findings can promote community participation in surveillance and control efforts to prevent leishmaniasis cases.

Mass wasting reveals ongoing asymmetric retreat of the martian north polar ice cap

Ongoing mass wasting through ice block falls is intensive at the north polar ice cap of Mars. We monitored how this activity is currently shaping the marginal steep scarps of the ice cap, which holds a record of the planet’s climate history. With AI-driven change detection between multi-temporal high-resolution satellite images, we created a comprehensive map of mass wasting across the entire North Polar Layered Deposits (NPLD). Our results show a more active erosion process than previously thought, with scarps retreating by up to ~3 m every kiloyear. The distribution of the active scarps indicates an ongoing asymmetric retreat of the already subcircular ice cap. The active scarps and the interior dune fields correlate strongly with exposures of the underlying, sandier Basal Unit (BU), providing evidence that erosion of the BU undermines the base of the NPLD. Moreover, ice block fall activity suggests potential areas where gypsum is released, given that the interior gypsum-bearing dune fields are located adjacent to these active scarps. Here, our study reveals the rates of present-day topographic change of the north polar ice cap, providing a valuable constraint for study of its past evolution.

Water Clarity Assessment Through Satellite Imagery and Machine Learning

Leveraging satellite monitoring and machine learning (ML) techniques for water clarity assessment addresses the critical need for sustainable water management. This study aims to assess water clarity by predicting the Secchi disk depth (SDD) using satellite images and ML techniques. The primary methods involve data preparation and SSD inference. During data preparation, AquaSat samples, originally from the L1TP collection, were updated with the Landsat 8 satellite’s latest postprocessing, L2SP, which includes atmospheric corrections, resulting in 33,261 multispectral observations and corresponding SSD measurements. For inferring the SSD, regressors such as SVR, NN, and XGB, along with an ensemble of them, were trained. The ensemble demonstrated performance with an average determination coefficient of R2 of around 0.76 and a standard deviation of around 0.03. Field data validation achieved an R2 of 0.80. Furthermore, we show that the regressors trained with L1TP imagery for predicting SSD result in a favorable performance with respect to their counterparts trained on the L2SP collection. This document contributes to the transition from semi-analytical to data-driven methods in water clarity research, using an ML ensemble to assess the clarity of water bodies through satellite imagery.

From Prediction to Explanation: Using Explainable AI to Understand Satellite-Based Riot Forecasting Models

This study investigates the application of explainable AI (XAI) techniques to understand the deep learning models used for predicting urban conflict from satellite imagery. First, a ResNet18 convolutional neural network achieved 89% accuracy in distinguishing riot and non-riot urban areas. Using the Score-CAM technique, regions critical to the model’s predictions were identified, and masking these areas caused a 20.9% drop in the classification accuracy, highlighting their importance. However, Score-CAM’s ability to consistently localize key features was found to be limited, particularly in complex, multi-object urban environments. Analysis revealed minimal alignment between the model-identified features and traditional land use metrics, suggesting that deep learning captures unique patterns not represented in existing GIS datasets. These findings underscore the potential of deep learning to uncover previously unrecognized socio-spatial dynamics while revealing the need for improved interpretability methods. This work sets the stage for future research to enhance explainable AI techniques, bridging the gap between model performance and interpretability and advancing our understanding of urban conflict drivers.

Assessment of Mango Canopy Water Content Through the Fusion of Multispectral Unmanned Aerial Vehicle (UAV) and Sentinel-2 Remote Sensing Data

This study proposes an Additive Wavelet Transform (AWT)-based method to fuse Multispectral UAV (MS UAV, 5 cm resolution) and Sentinel-2 satellite imagery (10–20 m resolution), generating 5 cm resolution fused images with a focus on near-infrared and shortwave infrared bands to enhance the accuracy of mango canopy water content monitoring. The fused Sentinel-2 and MS UAV data were validated and calibrated using field-collected hyperspectral data to construct vegetation indices, which were then used with five machine learning (ML) models to estimate Fuel Moisture Content (FMC), Equivalent Water Thickness (EWT), and canopy water content (CWC). The results indicate that the addition of fused Sentinel-2 data significantly improved the estimation accuracy of all parameters compared to using MS UAV data alone, with the Genetic Algorithm Backpropagation Neural Network (GABP) model performing best (R2 = 0.745, 0.859, and 0.702 for FMC, EWT, and CWC, respectively), achieving R2 improvements of 0.066, 0.179, and 0.210. Slope, canopy coverage, and human activities were identified as key factors influencing the spatial variability of FMC, EWT, and CWC, with CWC being the most sensitive to environmental changes, providing a reliable representation of mango canopy water status.

Leveraging the Internet of Things, Remote Sensing, and Artificial Intelligence for Sustainable Forest Management

Sustainable forest management is vital for addressing climate change, biodiversity loss, and deforestation. Human-induced stresses on forest ecosystems demand innovative approaches to ensure long-term health and productivity. This study explores how cutting-edge technologies, including the Internet of Things (IoT), remote sensing, and artificial intelligence (AI), enhance sustainable forest management practices. Researchers reviewed 196 studies published between 2021 and 2024 from IEEE Xplore Digital Library, MDPI, Taylor & Francis, ScienceDirect, Frontiers, Springer, SAGE, Hindawi, Nature, Wiley Online Library, and Google Scholar. The findings highlight IoT devices like drones, enabling real-time data collection on temperature, humidity, soil moisture, and tree growth, facilitating continuous forest monitoring. Remote sensing technologies, utilizing satellite imagery and aerial surveys, deliver high-resolution data for large-scale forest assessments, including forest cover changes, biomass estimation, and early detection of illegal logging. When integrated with AI, these tools enhance predictive modeling, data analysis, and decision-making, leading to more effective forest management strategies. The study also identifies challenges such as data security concerns, bandwidth limitations, interoperability issues, and high costs. Despite these barriers, IoT, remote sensing, and AI present transformative potential for improving forest resilience, carbon sequestration, and biodiversity conservation. These technologies are crucial in preserving forest ecosystems and mitigating climate change impacts by advancing real-time monitoring, optimizing resource allocation, and enabling data-driven decisions.

Microphytobenthos spatio-temporal dynamics across an intertidal gradient in a tropical estuary using Sentinel-2 imagery.

Intertidal mudflats are among the most productive coastal ecosystems, largely because of the activity of the photosynthetic microbial community on the sediment surface, known as microphytobenthos (MPB). While the dynamics of MPB have been extensively studied in temperate estuaries, there is limited research in tropical estuaries. To address this knowledge gap, we investigated the spatio-temporal dynamics of MPB in the Nicoya Gulf (Costa Rica), one of the world's most productive tropical estuaries, using Sentinel-2 images at 10m spatial resolution from 2018 to 2022. We used Normalized Difference Vegetation Index (NDVI) values ranging from 0 to 0.4 to monitor MPB cover and growth rates (μNDVI, days-1). Analysis of data using Generalized Additive Models (GAM) showed that up to 62% of the temporal variability in average MPB NDVI, peaking in late wet and early dry seasons (November-December), can be explained by years and months. The high temperatures and irradiance during the dry season (December to April) in the Nicoya estuary may inhibit MPB growth, contrasting with patterns observed in temperate estuaries. We observed higher MPB NDVI in the upper intertidal zone (mean sea level>0.5m) as usually occurs in temperate estuaries. This research highlights the importance of high-resolution satellite imagery for long-term monitoring of MPB dynamics in tropical tidal flats, offering a valuable tool for estimating ecosystem services provided by intertidal MPB, such as primary production, climate regulation, and nutrient cycling on a global scale.

Evaluating trapped lee‐wave characteristics over the UK and Ireland in convection‐permitting model forecasts using satellite imagery

Evaluating trapped lee‐wave characteristics over the UK and Ireland in convection‐permitting model forecasts using satellite imagery