Search Results (2,369)

Protected areas (PAs) are increasingly challenged by the need to reconcile biodiversity conservation with sustainable human use, particularly in landscapes containing underutilized or resting area (RA). This study evaluated the potential of resting forest and agricultural lands to enhance biodiversity and support sustainable land use within protected areas of Cesme, Türkiye. A Geographic Information System (GIS)-based multi-criteria evaluation approach was employed, integrating land cover data, soil group maps, topographic parameters, and protected area classifications to generate Plant Suitability Maps (PSMs). Eight thematic layers were developed, incorporating soil depth, slope, erosion risk, and land capability classes to identify suitable plant species and land-use options. The results indicate that the strategic use of resting agricultural lands could contribute up to 35.5% to ecological enhancement, while resting forest lands could contribute an additional 18%. The proposed plant assemblages include medicinal and aromatic species, erosion-control plants, and economically valuable perennial species that support ecosystem services such as pollination, beekeeping, and agro-tourism. Overall, the findings demonstrate that integrating RA management into conservation planning can simultaneously strengthen biodiversity, improve ecosystem services, and generate socio-economic benefits for local communities. The proposed GIS-based framework offered a transferable and scalable methodology for sustainable land management in Mediterranean landscapes and other protected regions worldwide. Also, in this research, the aim was to determine plant species using GIS-based suitability analyses of multi-spatial datato guide vegetation decisions in multi-criteria PA. Full article

Open-set classification for remote sensing requires models that simultaneously achieve high accuracy on known land-cover types and reliably detect novel classes absent from the training distribution—a capability essential for real-world deployment where new classes routinely emerge. Existing multimodal fusion approaches for hyperspectral imagery (HSI) and LiDAR are primarily designed for closed-set scenarios and lack robust uncertainty modeling for unknown detection. We propose a post hoc calibrated multimodal open-set framework with three tightly integrated components. First, an Uncertainty-Aware Gating Fusion (UAGF) module dynamically weights HSI and LiDAR features per sample based on modality reliability and produces a gating uncertainty signal reflecting fusion confidence. Second, an Iterative Feedback Refinement (IFR) module progressively refines fused representations over multiple iterations and captures convergence dynamics, where stable convergence indicates known samples while high feature-change variance identifies potential unknowns. Third, a compact two-signal open-set detector combines gating uncertainty and refinement variance through an EVT (Weibull)-based post hoc calibration mechanism fitted exclusively on known validation samples. The framework follows a strict zero-unknown-supervision protocol: the multimodal backbone is trained using only known-class samples, and the open-set decision threshold is derived solely from the known validation score distribution. This design decouples representation of learning from open-set decision learning, improving robustness and avoiding the objective conflicts that arise in joint training. Comprehensive experiments on three benchmark datasets—Houston2013, Muufl, and Augsburg—demonstrate that the proposed method achieves 92.79%, 84.47%, and 80.99% overall accuracy and 76.48%, 63.91%, and 56.81% unknown accuracy, outperforming the closest multimodal competitor HyLiOSR by up to 32.4 pp in unknown accuracy while maintaining competitive closed-set performance. Full article

Land-use and land-cover changes around reservoirs mediate the interface between watershed land systems and managed surface-water resources. In regulated reservoirs, water-level regulation can rapidly expose or inundate shore-zone land, yet evidence remains limited on where these transitions occur, how landscape configuration changes, and how such information can inform watershed and reservoir-margin management. Using 0.5 m Jilin-1 optical imagery from April and September of 2024 and 2025, this study mapped land-use/land-cover change (LUCC) in the Danjiangkou Reservoir shore zone and integrated transition matrices, class-level landscape metrics, shoreline-distance gradients, reach-level zoning, paired hydrological records, and multiscale geographically weighted regression (MGWR). The classification achieved an overall accuracy of 93.1% and a Kappa coefficient of 0.921. The strongest land-cover shift occurred between September 2024 and April 2025, when the water proportion declined from 78.74% to 60.10% and bare land expanded during the lowest observed reservoir stage (151.02 m). Subsequent refill was accompanied by partial re-inundation and increases in grassland, cropland, and forest. The 0–30 m shoreline belt was the principal response zone, indicating that hydrologically driven land-cover replacement was concentrated in the immediate reservoir margin. MGWR showed spatially varying positive associations between change-patch characteristics, distance to permanent water, and elevation, but the low explanatory power requires these results to be interpreted as spatial diagnostics rather than causal attribution. The study links land-cover monitoring with reservoir water-level regulation, identifies priority shoreline belts, and provides spatial information for field verification and reservoir-margin management. Full article

Rapid urbanization has intensified artificial light at night (ALAN) and reduced access to natural dark sky environments. Dark sky parks provide a potential spatial approach for nighttime environmental protection, ecological conservation, and astronomical recreation. This study develops a constraint-based suitability assessment framework for urban dark sky park site selection and applies it to Wuhan, China. Multi-source geospatial data were integrated into a 1 km × 1 km evaluation grid. The AHP–Delphi method was used to determine indicator weights, while land cover constraints were introduced to exclude artificial surfaces from candidate evaluation areas. Weighted overlay analysis, sensitivity analysis, continuous patch screening, and dark sky quality verification were then conducted. The results show that (1) artificial light visibility (ALV) and cloudless days (CVD) are the most important indicators, with weights of 0.328 and 0.250, respectively; (2) 29.38% of the evaluation units are classified as most suitable or more suitable; (3) the spatial pattern of highly suitable areas remain relatively stable, with Jaccard overlap rates of 73.65% and 87.09% under alternative weighting scenarios; and (4) continuous patch screening identifies Caidian and Yangda as priority candidate areas. Further verification using the Bortle Scale, a nine-level classification of night darkness, shows that the Caidian patch reached Bortle class 4 and National Astronomical Observatories (NAOC) dark sky class 1, indicating stronger practical feasibility for dark sky park development. The proposed framework provides a methodological reference for integrating dark sky protection, land use feasibility, and urban planning in metropolitan regions. Full article

Landscape Ecological Risk (LER) reflects the potential adverse effects of landscape change on ecological structure, function, and stability. In rapidly urbanizing megacities such as Dhaka, vegetation loss and built-up expansion have intensified environmental pressure over recent decades. This study examines the spatiotemporal dynamics of LER in Dhaka from 2004 to 2024 under the combined influence of vegetation change and urban expansion. Multi-temporal remote sensing data were used to generate land cover maps, derive Fractional Vegetation Cover (FVC), and quantify urbanization intensity using Nighttime Light (NTL) data. The Landscape Ecological Risk Index (LERI) was calculated using landscape pattern metrics, while bivariate spatial autocorrelation and geographically weighted regression (GWR) were applied to examine spatial associations and local spatial heterogeneity. The results show that vegetation degradation affected 34.39% of the study area during 2004–2024, while high-risk zones increased from 24.36% in 2004 to 42.95% in 2024. Land cover analysis further indicates a substantial expansion of built-up areas, accompanied by the contraction and fragmentation of vegetation, agricultural land, and lowland classes. Spatial analyses reveal that the relationships among vegetation cover, urbanization intensity, and ecological risk vary across the city and became increasingly spatially differentiated over time. These findings suggest that vegetation loss and urban expansion are spatially associated with increasing ecological risk in Dhaka. However, the results should be interpreted with caution because of uncertainties related to remotely sensed data, unsupervised land cover classification, resampling procedures, and limited ground validation. Despite these limitations, the study provides a spatially explicit framework for understanding ecological risk dynamics and offers useful evidence for green-space conservation, ecological restoration, and sustainable urban planning in rapidly urbanizing regions. Full article

In temperate climates, diversifying rotations with overwintering cover crops provides many benefits, including reducing nutrient losses, restoring soil organic matter, and managing weeds. However, there is limited understanding of where and when cover crops have been planted, especially relative to harvested winter crops, such as wheat and alfalfa. In this study, we use Sentinel-1 and Sentinel-2 satellite data to map winter land cover, including cover crops, across three sites in the Lower Peninsula of Michigan using random forest models. Our results show overall moderate accuracy (60–80%) across all three sites, with individual-level accuracies varying by region and land cover type. Generally, models that combined Sentinel-1 and Sentinel-2 bands, polarizations, and indices performed better than models that relied on one sensor alone. F1 scores for cover crop mapping were moderate, with the highest accuracies achieved for mapping any cover crop (0.77) compared to individual cover crop species—cereal rye (0.72) or ryegrass (0.50). Considering which bands and time periods were the most important for the classification, we found that vegetation indices developed using the red edge bands in the earlier part of the growing season were particularly important for classification accuracy. This work suggests that readily available Sentinel-1 and Sentinel-2 satellite data can be used to accurately map winter land cover, including cover crops, in the US Midwest. Full article

Mountain ecosystems are highly sensitive to climate change and require spatially explicit monitoring tools to support adaptive management. Within the framework of the Interreg-ALCOTRA “ACLIMO” project, this study investigates land cover dynamics in the Gesso Valley (Maritime Alps, Italy) over the period 2010–2021 using deep learning–based classification of high-resolution aerial orthophotos integrated with climate data analysis. Multi-temporal RGB and NIR imagery (2010, 2018, 2021) was classified using convolutional neural networks (U-Net and MMSegmentation) in ArcGIS Pro, with CORINE Land Cover datasets used for training. The best-performing model, based on CLC + Backbone 2018, achieved an overall accuracy of 82%, increasing to 87% after fine-tuning. Change detection revealed a general shift towards increased vegetation cover, while climate analysis based on regional weather stations (1990–2021) identified a warming trend of +0.4 °C/decade and recent drier conditions. Logistic regression highlighted significant associations between land cover transitions and climate anomalies, with temperature positively influencing change probability (OR = 1.40). The study demonstrates the potential of operational GIS-integrated deep learning workflows for climate change monitoring in complex alpine environments under real-world data constraints. Full article

Hyperspectral images (HSIs) and Light Detection and Ranging (LiDAR) data offer complementary spectral and spatial information and are extensively applied to land cover classification. Nevertheless, current fusion–classification approaches frequently suffer from cross-modal feature entanglement and insufficient exploitation of LiDAR physical priors, particularly the Digital Surface Model (DSM), which limits the interpretability of learned features and restricts classification accuracy. To address these issues, this study presents a Physics-Guided Adaptive Decoupling and Collaborative Enhancement Network (ADCE-Net) that embeds explicit geometric guidance into multimodal feature learning. In ADCE-Net, the DSM serves as an explicit geometric conditioning signal to guide feature decoupling, decomposing input representations into modality-shared semantic features (SSF) and modality-specific discriminative features (MSF), thereby mitigating cross-modal interference at an early stage. Based on this decomposition, an adaptive collaborative enhancement mechanism is designed using bidirectional cross-attention and dynamic gating to achieve context-aware mutual refinement between SSF and MSF, facilitating more effective utilization of cross-modal complementary information. Furthermore, a multi-level collaborative classification architecture is constructed to integrate multi-scale contextual representations, enhancing spatial consistency and boundary delineation. Extensive experiments on three benchmark datasets—Trento, Houston 2013, and Muufl Gulfport—demonstrate that ADCE-Net achieves overall accuracies of 99.69%, 97.37%, and 94.90%, respectively, surpassing multiple representative methods including support vector machines, 3D convolutional neural networks, transformer-based models, and recurrent neural networks. Noticeable improvements are also achieved for minority classes and classes with highly similar spectral signatures. The DSM-driven physics guidance boosts both classification performance and feature interpretability, providing a reliable and explainable paradigm for multimodal remote sensing classification. Full article

Karst regions are characterized by fragmented topography and significant micro-relief mosaics, leading to prominent spectral aliasing of land features, which can result in insufficient spatial representativeness of remote sensing samples for Land Use and Land Cover (LULC). The accuracy of LULC data directly affects the scientific basis of decision-making for rocky desertification control and ecological conservation. This study selected the Beipanjiang River Basin in Guizhou Province, a typical karst region, as the study area. The study selected the SOS, LOS, OM, and EOS indices from the 2001–2020 MODIS MCD12Q2 phenological dataset, combined with topographic zoning data. This study developed a sample spatial optimization scheme for complex karst terrain by integrating Spearman’s correlation analysis, SKATER spatially constrained clustering, statistical tests, adaptive stratified sampling, and Random Forest classification. The scheme was designed to test a phenology–landform joint stratification strategy for spatial sample allocation. The results indicate that (1) the study area was divided into six phenological pattern subregions, with significant spatial differentiation observed among them; (2) the “phenology–landform joint stratification + dual-weighted sample allocation” method was associated with improved sample representativeness and greater internal homogeneity within sample strata under the current experimental setting; and (3) compared to simple random sampling, the remote sensing phenological pattern-driven spatial optimization scheme improved overall accuracy from 71.33% to 77.55% and increased the Kappa coefficient from 0.43 to 0.62. These results suggest that, under the current study-area, sample-size, and validation settings, the phenology–landform joint stratification and dual-weighted allocation scheme can improve the spatial organization of training samples and classification performance over complex karst terrain, although weakly vegetated or bare classes remain difficult to separate. Full article

Deep learning has exhibited potential in land use and land cover (LULC) classification applications. However, the effectiveness of deep learning remains constrained by the availability and quality of annotated training data. The persistent scarcity of labeled samples and spatial heterogeneity of remote sensing imagery hinder the robustness and generalization of trained models. Meta-learning, commonly referred to as “learning to learn”, is a paradigm that trains models over a distribution of tasks to acquire transferable knowledge, enabling rapid adaptation to new tasks with only a few labeled samples. This cross-task learning capability makes meta-learning a promising solution to data scarcity and spatial heterogeneity in the remote sensing context. This paper provides a systematic review of meta-learning applications in LULC classification, identifying a total of 70 relevant studies between 2018 and 2025. Three mainstream meta-learning paradigms (memory-augmented, optimization-based, and metric-based) are reviewed, and the applications are analyzed across four core challenges in LULC remote sensing: label scarcity, cross-region and cross-domain distribution shifts, temporal dynamics modeling, and multimodal data integration. The review reveals that optimization-based and metric-based methods dominate current research, with MAML and its variants being the most widely adopted due to the model-agnostic property, while memory-augmented methods remain underexplored. A consistent finding is that meta-learning outperforms conventional pre-training followed by fine-tuning under significant domain shifts across multiple data modalities. Current limitations, including computational overhead, episodic training constraints, and the lack of standardized evaluation protocols, are discussed. Future directions in cross-domain generalization, integration with foundation models, novel architectures, and standardized benchmarks are identified. Full article

Wetlands are among the planet’s most productive ecosystems, yet they are increasingly imperiled by intersecting global challenges, particularly agricultural expansion, food security demands, and climate change. 1 This study investigated the spatial extent of floodplain wetlands and assesses Land Use/Land Cover (LULC) dynamics in the uMngeni catchment using multi-temporal Landsat imagery for the years 2000, 2010, 2020, and 2024. 2 Seven key land cover classes were classified, which included agriculture, bare land, built-up areas, forest, grassland, wetlands, and water bodies, using the Random Forest (RF) classification incorporating spectral indices (NDVI, NDWI) and topographic variables (slope and aspect) on Google Earth Engine (GEE). The overall accuracies for the respective years were 88.98% (2000), 91.23% (2010), 84.21% (2020), and 86.55% (2024), with corresponding Kappa coefficients of 0.82, 0.84, 0.78 and 0.80. 3 The findings show a significant 37% decline in wetland area from 2000 (2978 ha) to 2024 (1874 ha), with the most pronounced loss (46%) occurring between 2000 and 2010. Built-up areas increased by 38% over the same period, while agriculture peaked in 2010 (9312 ha) before declining to 7632 ha by 2024. The dominant transitions involved wetlands and grasslands being replaced by urban land and bare surfaces, particularly along the floodplain edges. 4 These patterns reflect intensifying human pressure on wetland ecosystems. Targeted interventions, such as enforcing buffer zones, regulating land use near water bodies, and restoring degraded wetlands, are critical to conserving ecosystem services and achieving sustainability outcomes aligned with the Sustainable Development Goals. Full article

Land-use and land cover (LULC) change is a key driver of environmental dynamics in semi-arid Mediterranean watersheds, strongly influencing hydrological processes, soil degradation, and ecosystem stability. In this context, understanding and predicting spatial–temporal land transformations is essential for sustainable watershed management. This study proposes a nonlinear spatial–temporal modeling framework integrating a hybrid Artificial Neural Network (ANN), Cellular Automata (CA), and Markov chain approach to simulate LULC dynamics in the Sebdou watershed, northwestern Algeria. Multi-temporal Landsat imagery (1985, 2005, and 2025), combined with topographic, socio-economic, and accessibility variables (slope, population density, distance to roads, and hydrographic network), was used to reconstruct historical land-use patterns and identify key driving forces of change. A supervised Maximum Likelihood classification achieved high accuracies, with overall accuracy ranging from 92.87% to 96.26% and Kappa coefficients between 0.85 and 0.91. The ANN model was trained to estimate nonlinear transition potentials, while the CA component incorporated spatial neighborhood effects to simulate land allocation processes. Markov chain analysis provided temporal transition probabilities, enabling the construction of a coupled ANN–CA–Markov framework for scenario-based prediction. Model validation against observed 2025 LULC maps indicated strong agreement in quantity distribution (Kappa histogram = 0.767), while spatial agreement (Kappa = 0.3566) reflected inherent spatial displacement typical of CA-based stochastic allocation. Simulation results for 2045 indicate continued urban expansion along major transport corridors, progressive decline of dense forest cover, and increasing bare soil areas, while agricultural land remains dominant but increasingly fragmented. These trends highlight the growing influence of anthropogenic pressure and accessibility factors on landscape restructuring in semi-arid environments. The proposed hybrid framework provides a robust decision-support tool for anticipating land-use dynamics and assessing future environmental pressures in Mediterranean drylands. Its integration with hydrological and erosion models can further support sustainable watershed planning under combined socio-economic and climatic changes. Full article

Urbanisation and impervious-cover expansion are reshaping riparian landscapes, particularly in mountain cities where steep terrain concentrates development along valley floors. This study examined spatiotemporal land-cover change within the regulated riparian corridors of Thimphu City, Bhutan, over a 25-year period from 1997 to 2022 using Landsat imagery, Random Forest classification and Google Earth Engine. Results show substantial transformation of riparian land cover, with impervious cover increasing from 26.14% to 32.63%, equivalent to an overall increase of 24.83%, while agriculture/barren/low-vegetation declined from 30.59% to 26.01%, equivalent to an overall decrease of 14.98%. A modest increase in detectable vegetation cover was also observed, although this should be interpreted cautiously because the study measured land-cover extent rather than vegetation condition, floristic composition or ecological quality. Classification performance was robust, with overall accuracies ranging from 89.9% to 94.5%, exceeding the commonly accepted 85% benchmark, although uncertainty remains in narrow riparian corridors due to Landsat’s 30 m spatial resolution. Mann–Kendall analysis provided supplementary evidence of monotonic land-cover trends, but the limited number of temporal observations means these results should be interpreted as indicative, rather than definitive. Spatial analysis revealed uneven transformation, with the southern valley recording the greatest increase in impervious cover. These findings demonstrate sustained development pressure within legally regulated riparian buffers and highlight the need for routine spatial monitoring, place-specific buffer management and stronger integration of riparian protection into urban planning. The study provides a quantitative baseline for assessing future riparian land-cover change and supporting more resilient land governance in rapidly urbanising Himalayan mountain cities. Full article

Conserved land represents an important mechanism for protecting ecologically sensitive lands while maintaining working landscapes. Despite their significance, nationwide tools for continuous hydrological monitoring of conservation easement lands remain limited. This study conceptualizes surface-water inundation as an indicator of hydrologic connectivity and ecosystem function, reflecting how water dynamics influence the resilience and ecological performance of conservation easement landscapes. We present a scalable framework to assess inundation dynamics across more than 340,000 conservation sites between 2018 and 2024 by integrating Sentinel-2 satellite imagery, Dynamic World land-cover data, and machine-learning classifiers (Support Vector Machine, Random Forest, and CART) within the Google Earth Engine platform. Spectral water indices (NDWI, MNDWI, and NDMI) were combined with Dynamic World classifications to generate quarterly inundation maps at 10 m spatial resolution, enabling consistent detection of surface-water presence across space and time. Among the evaluated classifiers, the Support Vector Machine (SVM) model achieved the highest performance in surface-water detection. Results reveal strong regional and seasonal variability in inundation patterns across conservation land. Higher inundation frequencies were observed in the Midwest, Gulf Coast, and Pacific Northwest, where wetland-associated easements showed persistent inundation (>50%) during spring and early summer, highlighting their role in supporting biodiversity, groundwater recharge, and flood mitigation. Overlay analysis with the National Wetlands Inventory (NWI) and SSURGO hydric soils confirmed a strong spatial correspondence between inundation occurrence and wetland-prone landscapes, extending the same Sentinel-2 and machine-learning framework to conservation land and enabling the first systematic cross-program comparison of hydrological dynamics across two major U.S. conservation mechanisms. This work highlights the critical role of conservation lands including Conservation Reserve Program (CRP) areas and conservation easements in supporting inundation dynamics and hydrological connectivity. These functions are essential for sustaining wetland habitats, maintaining water quality, and enhancing flood mitigation at the national scale. Full article

Accurate and timely land cover and land use (LCLU) classification from medium-spatial-resolution optical time-series data is essential for large-scale environmental monitoring. lightweight deep neural networks (DNNs) offer reduced computational and memory requirements, enabling efficient deployment in resource-constrained scenarios. While popular in computer vision tasks, their ability to simultaneously model spatial, spectral, and temporal information for medium-resolution optical time series is understudied. This study addresses this gap by evaluating seven existing lightweight models spanning four architectural families: convolutional and recurrent hybrids, convolutional and transformer hybrids, 3D convolutional models, and video transformers against a traditional hybrid convolutional transformer (CNNTransformer) benchmark across the Conterminous United States (CONUS). Models are trained on 500,000 Landsat time-series samples with 25 repetitions and evaluated across five model sizes (3k, 5k, 10k, 25k, and 50k parameters) to assess both accuracy and stability. Results show that Simple Recurrent Unit (SRU)-based lightweight hybrids provide the best performance. Specifically, MobileNetSRU consistently outperformed the benchmark at small-to-moderate model sizes (3k–15k), achieving peak relative improvement gains of ~2.5–7.5% at 7.5k parameters. MobileNetSRU also demonstrated superior robustness in limited-data scenarios (50k training samples), particularly for spectrally stable classes like water and bare land. This study reveals that the inherent inductive bias of recurrent-based lightweight models aligns more effectively with the sequential phenology of satellite data than more flexible, data-hungry attention mechanisms at small parameter scales. These findings suggest that strategically matching architectural priorities to temporal data structures can significantly reduce the trade-off between model efficiency and classification accuracy in scalable Earth-observation workflows. Full article