Integrating segmentation and vision-language model for automated and interpretable building damage assessment from satellite imagery

Temporal seagrass mapping using machine learning and field-validated satellite imagery to inform restoration and management

D2Mamba: A mamba-based method for floodway obstructions segmentation from multispectral satellite imagery

Saltmarshes provide key ecosystem services, including atmospheric CO2 sequestration and nitrogen burial in sediments. In recent decades, these blue carbon ecosystems have faced significant degradation from natural and anthropogenic stressors. In this study, rewilding of a desiccated saltmarsh in Cadiz Bay (SW Spain) was assessed as a nature-based solution to restore carbon (Corg) and nitrogen (NT) storage. The rewilding process began in 2004 after breaching an external tidal wall. We evaluated changes in vegetated and unvegetated areas using Landsat satellite imagery (1994-2024) and quantified Corg and NT stocks and burial rates in wild and rewilded sediments, including vegetated saltmarsh (Sarcocornia sp.) and bare sediments colonized by microphytobenthos (MPB). Vegetated saltmarsh cover increased by 85% over 20 years, at an average recovery rate of 5hay-1, concurrent with a decrease in unvegetated tidal flats. Average Corg stocks in the top 1m ranged from 32 to 57t Corg ha-1, with higher values in vegetated sediments. However, only 5-12% of Corg was stored during the rewilding period. Corg burial rates averaged 69g Corg m-2 y-1, and NT stocks were 55% higher in rewilded sediments than in wild ones (3.6 vs. 1.6t NT ha-1). Despite vegetation recovery, burial rates of Corg and NT did not increase clearly, suggesting that long-term storage may be influenced by factors beyond rewilding. Less than 8% of sedimentary Corg originated from saltmarsh vegetation, indicating the dominance of allochthonous sources. These findings highlight the complexity of biogeochemical recovery in rewilded saltmarshes and underscore the need for long-term monitoring to determine how much time is truly required for Corg and NT recovery.

School greenspace and myopia incidence and burden in Chinese children and adolescents: The Guangzhou children and adolescents cohort study.

Remote sensing plays an increasingly critical role in water quality monitoring due to its capacity for consistent observations on both large and small water bodies. However, current remote sensing approaches face limitations in aligning satellite observations with in-situ measurements, largely due to the dynamic vertical behavior of algae and the temporal constraints of satellite overpasses. Consequently, many studies rely on large water bodies, space–time substitution, or opportunistic imaging of blooms, which restricts the applicability of remote sensing for routine monitoring tasks such as periodic chlorophyll-a (Chl-a) estimation. With near-daily global coverage, PlanetScope imagery presents new opportunities to overcome these constraints. In this study, we propose a novel field-sampling augmentation framework that integrates satellite observations with in-situ data by modeling the diurnal vertical migration of algae through an Algal Behavior Function (ABF). This function enables the temporal adjustment of in-situ measurements, generating refined field-to-satellite matchups that enhance the robustness of Chl-a estimation models. We applied this method using PlanetScope imagery from 2022 to 2023 and co-located sonde measurements, incorporating vertical profile and timestamp information to correct for field-to-satellite temporal mismatches at two lakes in Ohio (Grand Lake St. Marys, samples = 84, Del-Co reservoirs, samples = 333). The augmented model improved Chl-a prediction accuracy (RMSE reduce) by 5.8%-18.0% compared to baseline models without refinement, with notable gains during non-bloom periods, offering potential for earlier bloom detection. Furthermore, the ABF demonstrated moderate geographic transferability: models using ABFs derived from a reservoir successfully improved Chl-a predictions at two additional lakes located 156 km (western Lake Erie) and 383 km (Saginaw Bay, Lake Huron) away, with accuracy gains (RMSE reduce) of 28.5%-35.3%. Collectively, these results position ABF as a practical, sensor-agnostic pre-processing step that can be embedded in operational workflows to improve high-resolution Chl-a retrievals, enable earlier harmful algal bloom alerts, and support cross-basin trend analyses for management.

Future risks of cyanobacterial blooms in lakes unveiled by open access data and integrated machine learning models.

Metadata, wavelet, and time aware diffusion models for satellite image super resolution

ABSTRACT The Kirtankhola River is undergoing rapid morphological changes with significant socioeconomic and environmental impacts. This study analyzes its planform dynamics from 1980 to 2023 using satellite imagery and the Digital Shoreline Analysis System (DSAS). We quantified erosion and accretion, bank change rates, sinuosity index (SI), and channel width, and assessed the geotechnical properties of riverbanks and the river’s hydrologic characteristics. Field investigations validated the geospatial analysis and helped identify factors contributing to bank instability. Results show that erosion in the river has predominated over accretion, resulting in a net loss of 739.77 ha and a net gain of 569.67 ha. The left bank was erosion-dominant, while the right bank experienced more accretion. The SI consistently increased from 1.59 in 1980 to 1.78 in 2023, with pronounced bank instability in highly sinuous sections. The mean river width remained relatively stable, but some channel segments showed localized narrowing and widening. These findings suggest that channel morphological adjustments are a major factor in bank instability. Changes in water discharge and bank material properties, as well as human interventions, are also important contributors. This study provides critical insights for zone-wise riverbank protection and management strategies along the Kirtankhola River and similar deltaic rivers worldwide.

ABSTRACT The rapid expansion of fast-growing trees like eucalyptus and Acacia decurrens is driving forest transitions in Ethiopia. While promoted for carbon sequestration and climate mitigation, the combined ecological and economic impacts of land-use change remain underexplored. This study examines ecosystem integrity and economic growth in South and Southeast Ethiopia using panel data from satellite imagery, integrating ecological indicators with night-time lights as a proxy for economic activity. A dynamic difference-in-difference (DID) framework assesses eucalyptus expansion across three agro-ecological zones. Results show heterogeneous effects: normalized difference vegetation index (NDVI) increases dynamically, moisture stress index (MSI) shows no persistent change, and land surface temperature (LST) rises variably across zones. Total ecosystem service value (TESV) increases by 45.4% per hectare per year, with significant gains in regulating, supporting, and cultural services, but minimal effects on provisioning services. Benefits are strongest in montane forests and grassland/woodland zones, weaker in Afro-alpine areas. This study highlights both trade-offs and synergies associated with eucalyptus-based land use and land cover (LULC) change. It also provides a quantitative framework for monitoring ecosystem responses to land-use change. Additionally, it highlights evidence to guide land-use policy, forest governance, and climate-smart development planning in Ethiopia and beyond.

We present a first global high-resolution map (30 m x 30 m) of high-altitudinal wetlands in the world's major mountain regions, i.e. the Andes, Rocky Mountains, Alps and High Mountain Asia. To map these wetlands, we employed a supervised classification approach using a random forest machine learning model and a selected set of predictors including vegetation, topographic, and surface moisture features. The predictors were derived from freely available radar and optical satellite imagery (Sentinel-1 and Sentinel-2), SRTM elevation data, and the global ecoregion map RESOLVE. We identify a total area of >30,500 km2 of high-mountain wetlands. With this map we aim to enhance the understanding of wetland distribution in remote and often inaccessible mountain regions and enable a more reliable understanding of their role in the ecosystem functioning and water cycles of high mountain areas.

Urban expansion presents significant challenges and opportunities for ecological conservation in developing countries, particularly in regions such as the Table Bay Nature Reserve in Cape Town, South Africa, where urban development interfaces with sensitive ecosystems. This article examines the complex dynamics between urban growth and ecological implications in this unique landscape, employing multi-temporal remote sensing techniques to analyze changes over time. By investigating the historical trajectory of urbanization in Table Bay, alongside its impacts on biodiversity and ecosystem services, we aim to underscore the urgent need for sustainable urban planning and conservation strategies. To analyze land use/land cover (LULC) dynamics over a 24-year period, this study leveraged a time series of satellite imagery processed within the Google Earth Engine (GEE) platform. Data can be accessed using their respective collection IDs within the GEE platform. The use of remote sensing tools aligns with Sustainable Development Goal (SDG) 15, which focuses on the protection, restoration, and sustainable use of terrestrial ecosystems. Urban encroachment analysis indicates that approximately 0.324 km2 of built-up area expanded directly within the reserve boundary, highlighting a measurable degree of infringement into protected zones. The dominance of built-up and bare land classes highlights the early encroachment of urban infrastructure and anthropogenic disturbance, setting the stage for subsequent land cover transformations observed in later years (2012 and 2024). These findings demonstrate a persistent trend of urban encroachment and ecological alteration within the Table Bay Nature Reserve. With the increase in global population levels, urban expansion into protected conservation areas has become a critical environmental concern, threatening biodiversity globally. This challenge is particularly acute in developing countries as seen in regions like the Table Bay Nature Reserve in Cape Town, South Africa, where urban development is interfaced with sensitive ecosystems.

Negotiating conservation priorities in highly anthropic ecosystems requires approaches that promote cooperation and cost bargaining among stakeholders. In data-poor contexts, such negotiations are hindered by limited information and conflicting interests. We developed a transparent and reproducible prioritization framework, combining open-access satellite imagery with a decision-support tool, to inform participatory coastal planning. Although illustrated in northeast Brazil, a data-poor region where coastal conflicts between tourism and fisheries are acute, the framework was designed to be broadly applicable to similar coastal social-ecological systems. It generates visualizations that help stakeholders explore trade-offs in three dimensions: defining a conservation target; debating the location of coastal candidate areas that balance habitat protection with minimizing impacts on tourism and fishing activities; and deliberating the degree of restriction within each candidate area based on conservation and socioeconomic considerations. We applied this approach to design five alternative spatial planning scenarios. Three were based on composite tourism and fisheries indices ("neutral", "avoid fisheries areas", and "avoid tourism areas"), and two focused on specific practices ("all specific activities" and "do not exclude small-scale"). These scenarios revealed where restrictions can be applied or relaxed to achieve conservation objectives while maintaining essential local livelihoods. By highlighting the limitations of existing conservation units and promoting inclusive, transparent decision-making, our method provides a practical means to focus stakeholder negotiations. More broadly, the framework offers a scalable and transferable approach for equitable, multisectoral conservation planning in other data-poor coastal regions facing similar trade-offs.

This study develops and applies an integrated methodology that combines deep learning-based computer vision and spatial statistics to automate the large-scale identification and analysis of morphological features in vernacular courtyard dwellings. Focusing on Liangshuaixiu dwellings in Wu’an, southern Hebei, we trained an HRNetV2 semantic segmentation model on high-resolution satellite imagery to identify and extract contours for 134,280 courtyard spaces. Core morphological parameters (area, orientation) were calculated and analyzed using GIS spatial statistics and the geographic detector model. The results show that (1) the computer vision pipeline achieved efficient recognition with satisfactory accuracy (~10% mean error); (2) spatial autocorrelation and hotspot analysis revealed distinct regional patterns, including a west–east increase in average courtyard area; and (3) geographic detector analysis demonstrated that courtyard morphology is shaped by complex interactions between natural and socio-economic factors. While average area and orientation were primarily governed by climate (air pressure, wind, temperature) and topography (elevation), diversity and internal variation were strongly influenced by nonlinear interactions, particularly between natural factors (e.g., wind–aspect) and between natural and human factors (e.g., population–climate). This work provides a scalable, data-driven framework for the quantitative spatial analysis of vernacular architectural heritage, advancing the understanding of building morphology as an outcome of coupled human–environment systems.

Forest area is a key indicator of Sustainable Development Goal 15 (Life on Land), yet rapid deforestation, especially in tropical regions, poses a major environmental challenge. In Nigeria, increasing land use and land cover changes (LULCC) are encroaching on protected areas, leading to widespread forest degradation. This study assessed LULC changes in Cross River National Park (CRNP) using remotely sensed data and GIS techniques to monitor forest loss over time. Landsat satellite images from 2003 and 2013, as well as Nigerian Sat-X data from 2023, were analyzed using ArcGIS 10.1 and ERDAS Imagine 2014 for image pre-processing, classification, and accuracy assessment. Five LULC classes were identified within and around CRNP: forest cover, mixed grassland, farmland, built-up areas, and water bodies. Results showed a significant increase in mixed grassland and built-up areas between 2003 and 2013 in both the Oban and Okwangwo divisions. Mixed grassland expanded from 16.71% to 41.51% in Oban and from 15.67% to 38.87% in Okwangwo. Built-up areas also increased steadily, reflecting growing human settlements and activities. Conversely, forest cover declined drastically in both divisions. In Okwangwo, forest area decreased from 823.54 km² (78.48%) in 2003 to 437.10 km² (41.66%) in 2023, while Oban Division experienced a reduction from 1,992.30 km² (56.27%) to 1,069.22 km² (30.16%). The study concludes that population growth and increased dependence on park resources are major drivers of forest loss. It recommends sustainable land management practices, stronger conservation strategies, and strict enforcement of forest protection policies to curb deforestation in Cross River National Park.

An offshore oil spill near Karawang, West Java, in July 2019 caused a considerable impact on the Pulau Rambut Wildlife Sanctuary, a protected area known for its mangrove ecosystems, which provide crucial habitat for waterbirds in Jakarta Bay. The purpose of this study was to map and quantify land cover types on Rambut Island, as well as examine land cover changes three years after the spill, with a focus on mangrove dynamics. Land cover categorization was performed using the Maximum Likelihood (ML) method applied to remote sensing data from SPOT-6 and SPOT-7 satellite photos for 2019, 2020, and 2021. Ground truthing and drone imagery were used to validate categorization results, and accuracy was determined using the Kappa statistics. All classes had strong levels of agreement, with Kappa values of 85.66%, 81.40%, and 82% in 2019, 2020, and 2021, respectively. Four types of land cover were identified: mangroves, non-mangrove forests, water bodies, and open spaces. Rambut Island has an expected mangrove covering of 18.80 ha in 2019, which increased to 21.15 ha in 2020 before significantly declining to 18.84 ha in 2021. These findings are consistent with field data, in which 12 of 13 MHI (Mangrove Health Index) plots were classed as moderate. This data implies that the 2019 oil spill did not result in a significant or long-term loss in mangrove area on Rambut Island.Keywords: mangrove, maximum likelihood, oil spill, Rambut Island, SPOT-6/7

Agricultural drought is a result of prolonged rainfall deficits affecting rice productivity. Agriculturally dependent regions are more vulnerable to agricultural drought. Therefore, drought monitoring is essential for effective agricultural management. This study aimed to investigate the drought variability of Rajshahi, Bangladesh utilizing optical remote sensing data from Landsat during 2000 to 2024, excluding 2007 due to technical faults in satellite imageries. This drought assessment used the Vegetation Health Index (VHI), which combines the Vegetation Condition Index (VCI) and the Temperature Condition Index (TCI), during pre-monsoon (March–May) and post-monsoon (October–November) seasons. The Standardized Precipitation Evapotranspiration Index (SPEI) was also used for cross-validation of areas affected by meteorological and agricultural drought. This study reveals that the “slightly dry” category of drought was predominant in all the districts for both seasons, where districts like Chapainawabganj, Pabna, Rajshahi and Sirajganj exhibited a significantly higher frequency of “dry” and “slightly dry” drought conditions. The Mann–Kendall test found no statistically significant trend of VHI for 24 years, indicating that drought has no linear pattern of occurrence. The cross-tabulation between SPEI and VHI showed a moderate agreement between drought categories, but a good relationship was found in normal conditions of drought from both indices. This suggests that meteorological drought may not be the only cause of agricultural drought; climate variables and agricultural practice have a great influence too.

Background:Antepartum stillbirth is an important public health issue. This study aimed to identify mediation pathways from the vegetation greenness surrounding the maternal residence to antepartum stillbirth, through stillbirth risk factors.Methods:This study is part of the FetRisks project–a population-based case–control study of stillbirth and live births in 14 public hospitals in the Municipality of São Paulo, Brazil (December 2019–December 2023). We estimated vegetation greenness using the median value of the normalized difference vegetation index (NDVI) obtained from satellite images from the study period, considering buffer radii around the residences between 50 and 500 m. We used penalized logistic regressions to select maternal, placental, fetal, and neighborhood characteristics. Based on these data, we built a Bayesian network and performed counterfactual mediation analyses.Results:We analyzed 248 antepartum stillbirths (cases) and 301 live births (controls). The maternal residence of 19.4% (48/248) of cases and 30.9% (93/301) of controls had a median NDVI within a buffer radius of 150 m (NDVI-150m) greater than 0.104. Women living in residences with a median NDVI-150m greater than 0.104 had a lower risk of antepartum stillbirth (odds ratio: 0.50; 95% confidence interval [CI]: 0.33, 0.76). Fetal growth restriction and placental inflammatory signs mediated 38.5% (natural indirect effect: −0.0604; 95% CI: −0.0916, −0.0292) and 14.7% (natural indirect effect: −0.0230; 95% CI: −0.0412, −0.0047) of the effect of a median NDVI-150m greater than 0.104 on antepartum stillbirth, respectively.Conclusion:Higher residential vegetation greenness showed protective indirect effects on antepartum stillbirth partially through a lower risk of fetal growth restriction and placental inflammatory signs.

As critical infrastructure at the sea-land interface, docks support regional economies and transportation networks. They include seaport docks and inland docks, the latter widely distributed along rivers, lakes, and canals. However, existing studies have mainly focused on seaport dock identification, with limited research and data products for inland docks. To address this gap, we developed an inland dock identification method for the Yangtze River by integrating crowdsourced OpenStreetMap (OSM) data with high-resolution Google satellite images. Based on OSM-derived prior information, we constructed a dedicated dock annotation dataset and applied YOLO-series models with a multi-scale detection strategy to identify and classify docks. The method achieved precision, recall, and F1-score above 0.9. A total of 3,562 docks were detected, including 2,738 floating and 824 vertical docks. To support diverse applications, the released dataset provides both bounding box representations and polygon vector delineations. As the first open inland dock dataset for the Yangtze River, it offers valuable support for studies on inland dock systems, waterway optimization, and regional economic analysis.

Abstract Satellite imagery plays a crucial role in exploring land use inventories of urban areas. However, accurate land cover classification from satellite imagery remains a longstanding challenge. With recent advancements in artificial intelligence technology, Deep Learning algorithms have achieved success in understanding satellite images by means of Convolutional Neural Networks (CNNs). While there has been a notable emphasis on satellite image analysis to improve the accuracy of land cover classifications, it is imperative to emphasise the significance of data-driven optimisation techniques. This paper introduces a hybrid UNet-ResNet-50 architecture, which integrates the metaheuristic Particle Swarm Algorithm (PSA) in dynamic hyperparameter optimisation for multi-class semantic segmentation. The approach of this research leverages a UNet extractor with ResNet-50 backbone (UResNet-50) and augments it with a Particle Swarm Optimiser (PSO) to automate the hyperparameter tuning process for segmenting the DeepGlobe satellite dataset into seven meaningful classes, namely: urban, forest, rangeland, barren land, agriculture, water bodies and unknown. The PSO-UResNet-50 model demonstrated robust performance across four distinct locations, in terms of accuracy, precision, recall, F1-score and mIoU as follows: Location-1 (95.74%, 98.12%, 86.95%, 92.04%,88.17%); Location-2 (91.88%, 79.23%, 80.75%, 81.42%, 83.03%); Location-3 (99.44%, 93.97%, 87.42%, 88.68%, 90.77%); and Location-4 (96.20%, 94.03%, 89.75%, 92.16%, 88.97%). The proposed PSO-UResNet-50 model outperformed the conventional U-Net and hybrid UResNet-50, demonstrating the advantage of applying PSO in multi-class segmentation of satellite imagery. The principal contribution of this work lies in the development and validation of a novel, metaheuristic-optimised deep learning framework that addresses the land cover classification challenge inherent in satellite images.