High-resolution Satellite Imagery Research Articles

Implementation of U-Net for Paddy Field Mapping Using Very High-Resolution Satellite Imagery

Implementation of U-Net for Paddy Field Mapping Using Very High-Resolution Satellite Imagery

FLOOD RISK MAPPING IN THE IRTYSH RIVER BASIN USING SATELLITE DATA

Floods are among the most frequent and devastating natural disasters, causing significant economic damage and loss of life worldwide. Effective flood risk management relies on accurate modeling techniques that can predict vulnerable areas and assess potential impacts. In this study, flood dynamics are simulated in the Irtysh River Basin near Ust-Kamenogorsk, a city in East Kazakhstan prone to seasonal flooding using high-resolution satellite imagery and digital elevation data. The primary objective is to visually model flood risks based on terrain characteristics. The study utilizes imagery sourced from the Mapbox platform, which combines data from MODIS, Landsat 7, Maxar, and the Google Earth Engine, providing access to Sentinel-2 surface reflectance imagery at 10-meter resolution. Elevation data from the Copernicus global digital elevation model, with a 30-meter resolution, is used to simulate flood progression. The flood simulation involves calculating flood depth relative to the terrain’s elevation, allowing for a pixel-by-pixel determination of submerged areas. Each simulation incrementally increases water levels to generate a sequence of images, showcasing the progression of flooding over time. The study describes hydraulic soil characteristics usage, and focuses on visualizing flood risk based on terrain data and water level changes. The simulation results indicate that flooding initially impacts riverbanks as water flow starts from the northwest of the city with critical infrastructure becoming vulnerable once water levels exceed 2 meters from the lowest elevation point. These findings highlight the potential of high-resolution satellite imagery and terrain data for flood risk assessment and improving urban flood preparedness. The results provide valuable insights into flood progression enabling more informed decision-making for disaster mitigation.

HIGH-RESOLUTION SATELLITE ESTIMATION OF SNOW COVER FOR FLOOD ANALYSIS IN EAST KAZAKHSTAN REGION

The increasing frequency of extreme weather events linked to climate change has made flood forecasting an important issue, particularly in mountainous regions where snowmelt is a major driver of seasonal flooding. This study explores the application of snow cover estimation techniques to assess snowmelt dynamics and their potential impact on flood risks in the Ulba and Uba basins in East Kazakhstan. To achieve this, high-resolution multispectral satellite imagery from the Sentinel-2 Surface Reflectance dataset is used, focusing on images collected between March and October for the years 2021 to 2024. The images are processed in Google Earth engine platform with strict filtering based on spatial intersection with the basins and cloud cover pixels percentage, ensuring high-quality data for snow cover analysis. The study utilizes multiple remote sensing indices for snow cover estimation. The normalized difference snow index is calculated using the green and shortwave infrared bands to detect snow-covered pixels. Fractional snow-covered area is derived from the NDSI using the 'FRA6T' relationship, offering a more nuanced estimate of snow distribution across the basins. Additionally, a near-infrared to shortwave infrared ratio threshold is employed to minimize confusion between snow and water, improving the detection of snow cover, particularly in regions near water bodies or during melt periods. The resulting snow cover maps and fSCA estimates provide a detailed picture of snow distribution and melt dynamics, contributing to the assessment of snowmelt’s role in flood risk development. The obtained insights can assist in refining flood forecasting models, improving early warning systems, and supporting informed water resource management in vulnerable regions.

Formation and evolution of thermokarst landslides in the Qinghai-Tibet Plateau, China

Thermokarst landslide (TL) activity in the Qinghai-Tibet Plateau (QTP) is intensifying due to climate warming-induced permafrost degradation. However, the mechanisms driving landslide formation and evolution remain poorly understood. This study investigates the spatial distribution, annual frequency, and monthly dynamics of TLs along the Qinghai-Tibet engineering corridor (QTEC), in conjunction with in-situ temperature and rainfall observations, to elucidate the interplay between warming, permafrost degradation, and landslide activity. Through the analysis of high-resolution satellite imagery and field surveys, we identified 1298 landslides along the QTEC between 2016 and 2022, with an additional 386 landslides recorded in a typical landslide-prone sub-area. In 2016, 621 new active-layer detachments (ALDs) were identified, 1.3 times the total historical record. This surge aligned with unprecedented mean annual and August temperatures. The ALDs emerged primarily between late August and early September, coinciding with maximum thaw depth. From 2016 to 2022, 97.8 % of these ALDs evolved into retrogressive thaw slumps (RTSs), identified as active landslides. Landslides typically occur in alpine meadows at moderate altitudes and on gentle northward slopes. The thick ice layer near the permafrost table serves as the material basis for ALD occurrence. Abnormally high temperature significantly increased the active layer thickness (ALT), resulting in melting of the ice layer and formation of a thawed interlayer, which was the direct causing factor for ALD. By altering the local material, micro-topography, and thermal conditions, ALD activity significantly increases RTS susceptibility. Understanding the mechanisms of ALD formation and evolution into RTS provides a theoretical foundation for infrastructure development and disaster mitigation in extreme environments.

A heterogenous-source geoinformation system to manage climate-induced modifications on the landscape for sustainable development

Historical landscapes in Italy have been slowly changing over the centuries and this is because their features, once fixed into specific shapes, were perpetuated until new economic and social developments occurred. Yet, in the Alpine region, this territorial organisation underwent sudden changes after World War II (WW II), resulting in a loss in population and traditional agropastoral production in favour of skiing plants and holiday houses. Moreover, the loss of traditional knowledge pertaining to environmental behaviour has resulted in the urbanisation of lands that are vulnerable to extreme events. In turn, this has hindered sustainable urban development. Nowadays, modern mapping technologies enable the state of the landscape to be assessed before, during, and after extreme events, the increased frequency of which may be related to climate change. The case study examined in this paper is the flood that hit Limone Piemonte, Italy, between the 2nd and 3rd October 2020. On that occasion, an aerial survey of the affected areas was carried out using Uncrewed Aerial Vehicles (UAV) a few weeks after the event. Spatial analyses were also performed based on very high-resolution satellite imagery acquired a few days after the event in order to determine where to plan more detailed three-dimensional (3D) surveys. This has enabled assessments of damages at different map scales to be performed. Thanks to the availability of pre-event multitemporal cartographic reference datasets, it was possible to monitor the historical evolution of the affected areas. It was possible to assess the vulnerable areas before the event, as well as to evaluate the morphological and settlement changes after the disaster. Therefore, one of the goals of this manuscript is to demonstrate that geoinformation systems are among the primary technical tools used to analyse environmental and climatic alterations that impact landscapes. Finally, a 3D model of the affected areas was produced, fulfilling another goal of the research by providing the public administration with a sustainable and innovative tool for territorial and landscape management, in accordance with the 11th pillar of the United Nations Sustainable Development Goals (UN SDGs).

Effectiveness of pre-fire forest management on post-fire forest conditions in southeastern Arizona

BackgroundWestern forests in the United States are facing multiple threats that have the potential to permanently alter forest composition and structure. In particular, wildfire can either have beneficial or adverse effects on overall forest health and resilience. Monitoring and assessing the effectiveness of existing forest treatment plans for meeting forest management goals is becoming more critical to increase the capacity for managers to prepare for and accommodate uncertainty associated with changing disturbance regimes. We used a combination of fine-scale vegetation and microclimate surveys on 57 plots, active remotely sensed data (light detection and ranging: LiDAR), and high-resolution satellite imagery to evaluate the effectiveness of an existing management strategy to increase disturbance resistance and resilience of an isolated mixed-conifer forest following a recent large-scale wildfire in southeastern Arizona, USA. We specifically assessed the effectiveness of forest overstory live tree thinning treatments (silviculture) as well as understory fuel reduction treatments (fuel) for influencing post-fire abiotic and biotic conditions, reducing direct post-fire tree mortality, and increasing resilience as compared to untreated forest stands.ResultsWe found that forest silviculture and fuel reduction treatments implemented prior to a large wildfire had mixed results on post-fire fine-scale vegetation composition and structure, microclimate conditions, tree mortality, and tree resilience. Fine-scale vegetation characteristics within silviculture- and fuel-treated forest units displayed higher herbaceous diversity and decreased density of new tree snags as compared to untreated units post-fire. Relevant to seedling emergence, we found that variance in spring soil moisture content was lower overall in treated units; however, units that received overstory thinning (silviculture) treatments were also associated with higher average summer high soil temperatures as compared to untreated units. Additionally, direct tree mortality and rate of recovery of trees post-fire differed between two treatment types (silviculture and fuel reduction) when compared to untreated units and among contrasting levels of burn severity. Post-fire tree mortality and tree resilience did not differ between control and silviculture units; however, these characteristics did differ between control and fuel units. Unlike control units, probability of tree mortality changed little between burn severity categories in fuel treatments (53.4% of mortality occurring in unburned/low vs. 46.7% in moderate/high severity) and resilience increased an average of 2.04% for trees from unburned/low to moderate/high-severity burn categories.ConclusionsOur methodology could be applied to any forested system experiencing increasing intensity and frequency of wildfire. Our results indicate that post-fire forest conditions and resilience are influenced by forest management strategies, particularly fuel reduction treatments. To accommodate uncertainty associated with changing disturbance regimes and climate change, implementing post-fire and post-treatment assessments and monitoring as presented in this study will be essential for developing attainable goals and for maintaining desired forest conditions.

Assessing the impact of urbanization on forest carbon stocks and social costs using a machine learning approach

Urbanization frequently precipitates urban sprawl, resulting in deforestation and alterations in landscape and land alterations. Such transformations profoundly impact the carbon stocks within metropolitan regions. This study examines the ramifications of urbanization on alterations in carbon stocks within urban forests across ten South Korean cities experiencing substantial urbanization over the past 15 years. Leveraging machine learning techniques and high-resolution satellite imagery, we scrutinize changes in land usage and urban forests, utilizing them to gauge the societal costs linked with shifts in urban carbon stocks. Furthermore, we integrate regional-level data sourced from national forests to enhance the precision of carbon stock estimations. Data analysis reveals that over the 15 years, urban areas expanded at an average rate of 4.43 km2 annually. In comparison, forested areas decreased by an average of −2.19 km2 per year, resulting in an average annual decline of −3171 tC in forest carbon stocks due to urbanization. The fluctuation in carbon stocks across the urban forests of the ten cities ranged from −68 % to 48 % over 15 years, primarily influenced by the extent of preserved forest area, with forest composition playing a secondary role. Concurrently, carbon sequestration efficiency varied between cities, ranging from 8 % to 57 % over 15 years, contingent upon tree type and forest age composition. An approximate loss of 174,380 tCO2eq of carbon stocks attributable to urbanization is estimated, with the associated social cost of increased emissions estimated at $8,893,396. Effective management of carbon emissions and sinks within urban locales is paramount for climate change mitigation, given the substantial contribution of urban areas to global carbon emissions. This study underscores the significance of urban forest management in carbon governance and furnishes valuable insights into nations undergoing rapid urban expansion.

Semantic Segmentation and Classification of Active and Abandoned Agricultural Fields through Deep Learning in the Southern Peruvian Andes

The monumental scale agricultural infrastructure systems built by Andean peoples during pre-Hispanic times have enabled intensive agriculture in the high-relief, arid/semi-arid landscape of the Southern Peruvian Andes. Large tracts of these labor-intensive systems have been abandoned, however, owing in large measure to a range of demographic, economic, and political crises precipitated by the Spanish invasion of the 16th century CE. This research seeks to better understand the dynamics of agricultural intensification and deintensification in the Andes by inventorying through the semantic segmentation of active and abandoned agricultural fields in satellite imagery across approximately 77,000 km2 of the Southern Peruvian Highlands. While manual digitization of agricultural fields in satellite imagery is time-consuming and labor-intensive, deep learning-based semantic segmentation makes it possible to map and classify en masse Andean agricultural infrastructure. Using high resolution satellite imagery, training and validation data were manually produced in distributed sample areas and were used to transfer-train a convolutional neural network for semantic segmentation. The resulting dataset was compared to manual surveys of the region and results suggest that deep learning can generate larger and more accurate datasets than those generated by hand.

Landslide Susceptibility Mapping in Nigeria Using Remote Sensing, GIS, and Machine Learning Models

Landslide situations remain a significant concern in Nigeria, as they pose a great risk to human lives, property, and the natural environment, particularly in regions with steep slopes, heavy rainfall, and unfavorable human interference, including deforestation and urbanization. Nigeria’s diverse geographical structure, ranging from urban areas such as Lagos and Abuja to rural regions, underscores the importance of accurate predictions for risk management and avoidance techniques. This research utilized remote sensing and GIS assessment to analyze landslide susceptibility in several regions of Nigeria. Data related to terrain, vegetation, soil moisture, and ground deformation were gathered using high-resolution satellite imagery from sources such as SPOT, ASTER, differential synthetic aperture radar interferometry (D-InSAR), and Landsat TM. GIS data layers included DEM, LULC, soil and geological maps, as well as hydrological maps and data. Methods applied in this study include logistic regression, STAT-R, frequency ratio, and the Random Forest tree-based model. The research produced detailed landslide susceptibility maps for various regions in Nigeria and identified significant factors such as slope, elevation, land use, precipitation, and access to transportation facilities. The Random Forest model demonstrated the most robust predictive capability. The integration of remote sensing with GIS was particularly significant, enhancing the precision of predictions and improving the efficacy of planning and management strategies. By incorporating remote sensing, GIS, and various machine learning algorithms, the researchers have developed a reliable tool for landslide risk prediction and management in Nigeria. Future research should focus on improving data quality and enhancing the generalizability of results to other regions.

Monitoring river discharge from space: An optimization approach with uncertainty quantification for small ungauged rivers

The number of in-situ stations measuring river discharge, one of the Essential Climate Variables (ECV), is declining steadily, and numerous basins have never been gauged. With the aim of improving data availability worldwide, we propose an easily applicable and transferable approach to estimate reach-scale discharge solely using remote sensing data that is suitable for filling gaps in the in-situ network. We combine 20 years of satellite altimetry observations with high-resolution satellite imagery via a hypsometric function to observe large portions of the reach-scale bathymetry. The high-resolution satellite images, which are classified using deep learning image segmentation, allow for detecting small rivers (narrower than 100m) and can capture small width variations. The unobserved part of the bathymetry is estimated using an empirical width-to-depth function. Combined with precise satellite-derived slope measurements, river discharge is calculated at multiple consecutive cross-sections within the reach. The unknown roughness coefficient is optimized by minimizing the discharge differences between the cross-sections. The approach requires minimal input and approximate boundary conditions based on expert knowledge but is not dependent on calibration. We provide realistic uncertainties, which are crucial for data assimilation, by accounting for errors and uncertainties in the different input quantities. The approach is applied globally to 27 river sections with a median normalized root mean square error of 12% and a Nash–Sutcliffe model efficiency of 0.560. On average, the 90% uncertainty range includes 91% of the in-situ measurements.

Multiscale Analysis of Pedestrian Crossing Distance

Problem, research strategy, and findings Crossing distance is considered a factor in both pedestrian safety and behavior. However, it has seldom been quantified within or across entire cities or related to walking outcomes. Combining data encoded in OpenStreetMap and drawn from high-resolution satellite imagery, we measured formal pedestrian crossings throughout a dense European city (Paris [France]), a dense American city (San Francisco [CA]), and a less-dense, more car-centric American city (Irvine [CA]). This granular approach—covering roughly 49,000 total crossings—identified inter- and intraurban spatial patterns in the distribution of pedestrian crossing distance, including clusters of long crossings that likely deter walking and increase its risk. By overlaying recent pedestrian–vehicle collisions on these novel data sets we found that longer crossing distance correlated with increased likelihood of collisions, raising the salience of traffic-calming interventions. Takeaway for practice Quantifying pedestrian crossing distance at the scale of entire municipalities empowers transportation planners to identify pedestrian-hostile crossings (individuals and clusters), add context to collision trends, and geographically target locations for traffic calming. These cases collectively demonstrate how the increasing prioritization of the automobile in city planning quantitatively changes (and degrades) the pedestrian environment, as well as how low-tech investments, such as sidewalk extensions and refuge islands, can mitigate these trends.

Back to the past: long-term persistence of bull kelp forests in the Strait of Georgia, Salish Sea, Canada

The Salish Sea, a dynamic system of straits, fjords, and channels in southwestern British Columbia, is home to ecologically and culturally important bull kelp (Nereocystis luetkeana) forests. Yet the long-term fluctuations in the area and the persistence of this pivotal coastal marine habitat are unknown. Using very high-resolution satellite imagery to map kelp forests over two decades, we present the spatial changes in kelp forest area within the Salish Sea, before (2002 to 2013) and after (2014 to 2022) the ‘Blob,’ an anomalously warm period in the Northeast Pacific. This analysis was spatially constrained by local environmental conditions. Based on nearshore sea surface temperatures (SSTs) from four decades (1984–2022), we found two periods of distinct increases in SST, one starting in 2000 and another in 2014. Further, the highest SST anomalies occurred on warmer coastlines in the enclosed inlets and the Strait of Georgia, while smaller anomalies were found on colder coastlines near the Strait of Juan de Fuca and the Discovery Passage. The total area of bull kelp forests from 2014 to 2022 has decreased compared to 2002 to 2013, particularly in the northern sector of the Salish Sea. Using the satellite-derived kelp data, we also present an analysis of kelp persistence compared with historical distribution of kelp forests depicted on British Admiralty Nautical Charts from 1858 to 1956. This analysis shows that warm, sheltered areas experienced a considerable decrease in persistence of kelp beds when compared to satellite-derived distribution of modern kelp, confirming a century-scale loss. In particular, the presence of kelp forests in the Strait of Georgia and on the warmest coasts has decreased considerably over the century, likely due to warming temperatures. While the coldest coasts to the south have maintained their centennial persistence, the northern Salish Sea requires further research to understand its current dynamics. This research contributes to a wider understanding of temporal and spatial factors for kelp from the regional perspective of the Salish Sea.

Assessing Tsunami Impacts and Enhancing Disaster Response in Tirtayasa Banten through High-Resolution Satellite Imagery

In December 2018, the coastal town of Tirtayasa, Banten, suffered severe damage from a tsunami, leaving the area highly vulnerable to future disasters. This research assesses the tsunami’s impact, thoroughly evaluating damage to infrastructure and vegetation using high-resolution satellite imagery. By comparing pre- and post-tsunami images, we quantitatively measure resilience and devastation, documenting significant landscape changes to better understand the extent of damage and identify resilient areas. These insights are critical for developing effective disaster response plans. The study employs advanced geospatial analytic techniques, demonstrating how satellite imagery enhances disaster preparedness and management by enabling prompt and accurate assessments, which are essential for both emergency response and long-term recovery planning. Integrating satellite-based remote monitoring into standard disaster management practices offers substantial advantages, improving the preparedness and response capabilities of vulnerable areas. This research highlights the importance of advanced change detection techniques to improve the accuracy of impact assessments and foster the development of targeted measures to mitigate the effects of future natural disasters.

Remote detection of asbestos-cement roofs: Evaluating a QGIS plugin in a low- and middle-income country

Machine learning, a subset of artificial intelligence, has emerged as a powerful tool for generating new knowledge from observations. In the realm of geographic information systems (GIS), machine learning techniques have become essential for spatial analysis tasks. Satellite image classification methods offer valuable decision-making support, particularly in land-use planning and identifying asbestos cement roofs, which pose significant health risks. In Colombia, where asbestos has been used for decades, the detection and management of installed asbestos is critical. This study evaluates the effectiveness of the RoofClassify plugin, a machine learning-based GIS tool, in detecting asbestos cement roofs in Sibaté, Colombia. By employing high-resolution satellite imagery, the study assesses the plugin's accuracy and performance. Results indicate that RoofClassify demonstrates promising capabilities in detecting asbestos cement roofs, achieving an overall accuracy score of 69.73%. This shows potential for identifying areas with the presence of asbestos and informing decision-makers. However, false positives remain a challenge, necessitating further on-site verification. The study underscores the importance of cautious interpretation of classification results and the need for tailored approaches to address specific contextual factors. Overall, RoofClassify presents a valuable tool for identifying asbestos cement roofs, aiding in asbestos management strategies.

UAV Modelling of Proglacial Landscapes, Axel Heiberg Island Nunavut

This project involved using a UAV to create digital models and products that enhance visualization and understanding of the moraines of White and Thompson Glaciers, Axel Heiberg Island, Nunavut. UAV surveying offers an effective alternative to high-resolution satellite imagery for documenting highly dynamic proglacial landscapes. Proglacial landforms and sediments provide insights into glacier characteristics and the behavior of glacier termini. UAV surveying allows for high-resolution examination of moraine features and the creation of digital elevation models (DEMs) that facilitate topographic modeling and the detection of changes. A Mavic 3M Drone was employed with UGCS flight planning software. UGCS flight planning software was selected for its ability to create orthomosaic flight plans that consider ground elevation, ensuring accurate Ground Sampling Distance (GSD). A 2m Arctic DEM was uploaded to UGCS prior to planning flight routes, allowing for more precise measurements. Post processing was done in Drone2Map and Agisoft Metashape to create both 2D orthomosaics and 3D models respectively. The results of this work will be used as a visual tool to study current aspects of the moraines, including extent, depositional processes, and present hydrodynamic energy environments. This information will be compared to historical air photos and previous studies to examine changes the moraines over a sixty-year period.

Assessing the Capacity of High-resolution Commercial Satellite Imagery for Grapevine Downy Mildew Detection and Surveillance in New York state.

Grapevine downy mildew (GDM), caused by the oomycete Plasmopara viticola, can cause 100% yield loss and vine death under conducive conditions. High resolution multispectral satellite platforms offer the opportunity to track rapidly spreading diseases like GDM over large, heterogeneous fields. Here, we investigate the capacity of PlanetScope (3 m) and SkySat (50 cm) imagery for season-long GDM detection and surveillance. A team of trained scouts rated GDM severity and incidence at a research vineyard in Geneva, NY, USA from June to August of 2020, 2021, and 2022. Satellite imagery acquired within 72 hours of scouting was processed to extract single-band reflectance and vegetation indices (VIs). Random forest models trained on spectral bands and VIs from both image datasets could classify areas of high and low GDM incidence and severity with maximum accuracies of 0.85 (SkySat) and 0.92 (PlanetScope). However, we did not observe significant differences between VIs of high and low damage classes until late July-early August. We identified cloud cover, image co-registration, and low spectral resolution as key challenges to operationalizing satellite-based GDM surveillance. This work establishes the capacity of spaceborne multispectral sensors to detect late-stage GDM and outlines steps towards incorporating satellite remote sensing in grapevine disease surveillance systems.

Detecting rare earth elements using EnMAP hyperspectral satellite data: a case study from Mountain Pass, California

Rare earth elements (REEs) exhibit diagnostic absorption features in the visible-near infrared region, enabling their detection and identification via spectroscopic methods. Satellite-based remote sensing mapping of REEs, however, has not been attainable so far due to the necessity for high-quality hyperspectral data to resolve their narrow absorption features. This research leverages EnMAP hyperspectral satellite data to map REEs in Mountain Pass, California—a mining area known to host bastnaesite-Ce ore in sövite and beforsite carbonatites. By employing a polynomial fitting technique to characterize the diagnostic absorption features of Neodymium (Nd) at ∼740 and ∼800 nm, the surface occurrence of Nd was successfully mapped at a 30m pixel resolution. The relative abundance of Nd was represented using the continuum-removed area of the 800 nm feature. The resulting map, highlighting hundreds of anomalous pixels, was validated through laboratory spectroscopy, surface geology, and high-resolution satellite imagery. This study marks a major advancement in REE exploration, demonstrating for the first time, the possibility of directly detecting Nd in geologic environments using the EnMAP hyperspectral satellite data. This capability can offer a fast and cost-effective method for screening Earth’s surfaces for REE signature, complementing the existing exploration portfolio and facilitating the discovery of new resources.

Enhancing Pléiades-based crop mapping with multi-temporal and texture information

Accurate crop mapping using satellite imagery is crucial for improving the monitoring of agricultural landscapes. Very high resolution (VHR) satellite imagery offers unique capabilities in this respect, allowing for even small fields to be discerned and image texture analysis to be performed. Additionally, satellite imagery has greater efficiency than unmanned aerial vehicles due to its extensive coverage. Moreover, the operation flexibility of VHR satellites means that timely image acquisition is possible several times during the growing season. This study investigates the potential of VHR Pléiades images and the random forest classifier for accurate crop mapping. Four images acquired on April 9th, May 12th, May 31st, and June 20th were used to test 16 classification scenarios, including single-date and multi-temporal combinations of spectral bands, texture features, and vegetation Indices (VIs). The classification using the spectral bands from all four images achieved the highest overall accuracy, 93.9% and 96.3% at field and pixel levels, respectively. The bitemporal classifications had lower accuracy. Nevertheless, the combination of the May 12th and June 20th spectral bands had 90% accuracy, which indicated that two images may be sufficient for reliable mapping if the periods with phenological differences between crops are considered. Adding texture features to the spectral bands significantly enhanced the accuracy (up to 8%) of single-date classifications, making it highly recommended when only one image is available. However, the impact of texture was more pronounced on the later dates. It showed the most marked benefit for vineyards and alfalfa, with minimal or no improvement observed for other classes like winter barley. An additional increase in overall accuracy was achieved in three of the four dates by supplementing the spectral and texture bands with VIs. This study highlights the importance of considering image acquisition dates and crop types when designing satellite-based crop mapping strategies for optimal accuracy.

Mapping of temperate upland habitats using high-resolution satellite imagery and machine learning

Upland habitats provide vital ecological services, yet they are highly threatened by natural and anthropogenic stressors. Monitoring these vulnerable habitats is fundamental for conservation and involves determining information about their spatial locations and conditions. Remote sensing has evolved as a promising tool to map the distribution of upland habitats in space and time. However, the resolutions of most freely available satellite images (e.g., 10-m resolution for Sentinel-2) may not be sufficient for mapping relatively small features, especially in the heterogeneous landscape—in terms of habitat composition—of uplands. Moreover, the use of traditional remote sensing methods, imposing discrete boundaries between habitats, may not accurately represent upland habitats as they often occur in mosaics and merge with each other. In this context, we used high-resolution (2 m) Pleiades satellite imagery and Random Forest (RF) machine learning to map habitats at two Irish upland sites. Specifically, we investigated the impact of varying spatial resolutions on classification accuracy and proposed a complementary approach to traditional methods for mapping complex upland habitats. Results showed that the accuracy generally improved with finer spatial resolution data, with the highest accuracy values (80.34% and 79.64%) achieved for both sites using the 2-m resolution datasets. The probability maps derived from the RF-based fuzzy classification technique can represent complex mosaics and gradual transitions occurring in upland habitats. The presented approach can potentially enhance our understanding of the spatiotemporal dynamics of habitats over large areas.

Application of gradient boosting machine in satellite-derived bathymetry using Sentinel-2 data for accurate water depth estimation in coastal environments

Accurate water depth estimation is crucial in coastal environmental management, resource exploration, and ecological protection. Traditional water depth measurement methods are often time-consuming and costly, especially in vast sea areas where their application is limited. However, with the rapid development of remote sensing technology, particularly the widespread use of high-resolution satellite imagery, water depth remote sensing has emerged as a more efficient, economical, and widely applicable solution. In this study, we utilized Sentinel-2 satellite data and applied various algorithms to accurately estimate water depth in the Nanshan Port area. The results showed that the Gradient Boosting Machine (GBM) model excelled in monitoring shallow water and coastal environments, effectively addressing challenges such as light attenuation and water scattering in turbid waters. Compared to traditional methods, GBM-generated predictions were smoother and more detailed. This study not only demonstrates the significant potential of satellite remote sensing for water depth measurement but also points to future directions for algorithm optimization and the integration of remote sensing technologies. It is expected to bring revolutionary progress to oceanic scientific research and coastal management.