Search Results (676)

Accurate estimation of urban land surface temperature (ULST) is critical for studying urban heat islands, but complex three-dimensional (3D) structures and materials in urban areas introduce significant adjacency effects into remote sensing retrievals. To investigate the influence of different factors on the adjacency effects, this study employed the DART model to quantify brightness temperature differences (ΔTb) of urban pixels by comparing their simulated radiance in two scenarios: (1) an isolated state (no adjacent buildings) and (2) an adjacent state (with surrounding buildings). ΔTb, representing the adjacency effect, was systematically analyzed across spatial resolutions (1–120 m), building geometry (building height BH, roof area index ${\lambda }_p$, adjacent obstruction degree SVF_Obs.), and material reflectance (reflectance R = 0.05, 0.1, 0.15) to determine key influencing factors. The results demonstrate that (1) adjacency effects intensify significantly with higher spatial resolution (mean ΔTb ≈ 5 K at 1 m vs. ≈2 K at 30 m), with 60–90 m identified as the critical resolution range where the adjacency-induced error is attenuated to a level (ΔTb < 1 K) that is commensurate with the intrinsic uncertainty of current mainstream ULST algorithms; (2) increased building height, reduced density (${\lambda }_p$), and greater adjacent obstruction (SVF_Obs.) exacerbate adjacency effects; (3) material emissivity (ε = 1 − R) is the dominant factor, where low-ε materials (high R) exhibit markedly stronger adjacency effects than geometric influences (e.g., ΔTb at R = 0.15 is approximately three times higher than at R = 0.05); and (4) temperature differences among surface components exert minimal influence on adjacency effects (ΔTb < 0.5 K). This study clarifies key factors driving adjacency effects in high-resolution ULST retrieval and defines the critical spatial resolution for simplifying inversions, providing essential insights for accurate urban temperature estimation. Full article

Optical remote sensing images are often partially obscured by clouds due to the inability of visible light to penetrate cloud cover, which significantly limits their subsequent applications. Most existing cloud removal methods formulate the problem using low-rank and sparse priors within a discrete representation framework. However, these approaches typically rely on manually designed regularization terms, which fail to accurately capture the complex geostructural patterns in remote sensing imagery. In response to this issue, we develop a continuous blind cloud removal model. Specifically, the cloud-free component is represented using a continuous tensor function that integrates implicit neural representations with low-rank tensor decomposition. This representation enables the model to capture both global correlations and local smoothness. Furthermore, a band-wise sparsity constraint is employed to represent the cloud component. To preserve the information in regions not covered by clouds during reconstruction, a box constraint is incorporated. In this constraint, cloud detection is performed using an adaptive thresholding strategy, and a morphological erosion function is employed to ensure accurate detection of cloud boundaries. To efficiently handle the developed model, we formulate an alternating minimization algorithm that decouples the optimization into three interpretable subproblems: cloud-free reconstruction, cloud component estimation, and cloud detection. Our extensive evaluations on both synthetic and real-world data reveal that the proposed method performs competitively against state-of-the-art cloud removal methods. Full article

Monitoring rainfall is essential to understanding hydrological processes, managing water resources, and mitigating drought and flood risks. Many regions, particularly in developing countries, have sparse rain gauge networks, which limit spatial coverage and result in inaccurate rainfall estimates. By combining remote sensing data with rain gauge measurements, rainfall estimates can be improved, and spatial coverage can be enhanced. Remote sensing techniques provide a valuable resource for supplementing and enhancing rainfall monitoring in such areas. This study leverages Global Precipitation Measurement (GPM) satellite data to enhance rainfall estimation in White Nile State, Sudan, where only two rain gauge stations are operational and the state’s total area is 39.600 km². GPM data, well-known for its high temporal and spatial resolution, offers a promising alternative to mitigate the limitations of sparse ground-based networks. The study integrates GPM satellite data with ground-based measurements through statistical and geostatistical techniques, as well as validation, to improve rainfall accuracy. The results show that, on average, GPM data and rain gauge measurements exhibit a strong correlation of 0.87, with an annual RMSE of 10.23 mm and an AME of 8.25 mm. These findings demonstrate that GPM data effectively complements traditional rain gauge observations by accurately capturing spatial rainfall distributions and extreme precipitation events. The findings underscore the potential of remote sensing to provide reliable rainfall information in data-scarce regions, contributing to better water resource management and disaster risk reduction strategies. Full article

Mapping the oil palm (Elaeis guineensis), the globally leading oil-bearing crop and a crucial industrial commodity, is of vital importance for food security and raw material supply. However, existing remote sensing approaches for oil palm mapping present several methodological challenges including temporal resolution constraints, suboptimal feature parameterization, and limitations in age structure assessment. This study addresses these gaps by systematically optimizing temporal, spatial, and textural parameters for enhanced oil palm mapping and age structure analysis through integration of Landsat 4/5/7/8/9, Sentinel-2 multispectral, and Sentinel-1 radar data (LSMR). Analysis of oil palm distribution and dynamics in Malaysia revealed several key insights: (1) Methodological optimization: The integrated LSMR approach achieved 94% classification accuracy through optimal parameter configuration (3-month temporal interval, 3-pixel median filter, and 3 × 3 GLCM window), significantly outperforming conventional single-sensor approaches. (2) Age estimation capabilities: The adapted LandTrendr algorithm enabled precise estimation of the plantation establishment year with an RMSE of 1.14 years, effectively overcoming saturation effects that limit traditional regression-based methods. (3) Regional expansion patterns: West Malaysia exhibits continued plantation expansion, particularly in Johor and Pahang states, while East Malaysia shows significant contraction in Sarawak (3.34 × 10⁵ hectares decline from 2019–2023), with both regions now converging toward similar topographic preferences (100–120 m elevation, 6–7° slopes). (4) Age structure concerns: Analysis identified a critical “replanting gap” with 13.3% of plantations exceeding their 25-year optimal lifespan and declining proportions of young plantations (from 60% to 47%) over the past five years. These findings provide crucial insights for sustainable land management strategies, offering policymakers an evidence-based framework to balance economic productivity with environmental conservation while addressing the identified replanting gap in one of the world’s most important agricultural commodities. Full article

The accurate identification of rice growth stages is critical for precision agriculture, crop management, and yield estimation. Remote sensing technologies, particularly multimodal approaches that integrate high spatial and hyperspectral resolution imagery, have demonstrated great potential in large-scale crop monitoring. Multimodal data fusion offers complementary and enriched spectral–spatial information, providing novel pathways for crop growth stage recognition in complex agricultural scenarios. However, the lack of publicly available multimodal datasets specifically designed for rice growth stage identification remains a significant bottleneck that limits the development and evaluation of relevant methods. To address this gap, we present RiceStageSeg, a multimodal benchmark dataset captured by unmanned aerial vehicles (UAVs), designed to support the development and assessment of segmentation models for rice growth monitoring. RiceStageSeg contains paired centimeter-level RGB and 10-band multispectral (MS) images acquired during several critical rice growth stages, including jointing and heading. Each image is accompanied by fine-grained, pixel-level annotations that distinguish between the different growth stages. We establish baseline experiments using several state-of-the-art semantic segmentation models under both unimodal (RGB-only, MS-only) and multimodal (RGB + MS fusion) settings. The experimental results demonstrate that multimodal feature-level fusion outperforms unimodal approaches in segmentation accuracy. RiceStageSeg offers a standardized benchmark to advance future research in multimodal semantic segmentation for agricultural remote sensing. The dataset will be made publicly available on GitHub v0.11.0 (accessed on 1 August 2025). Full article

Most existing deep learning-based super-resolution (SR) methods for remote sensing images rely on predefined degradation assumptions (e.g., bicubic downsampling). However, when real-world degradations deviate from these assumptions, their performance deteriorates significantly. Moreover, explicit degradation estimation approaches based on iterative schemes inevitably lead to accumulated estimation errors and time-consuming processes. In this paper, instead of explicitly estimating degradation types, we first innovatively introduce an MSCN_G coefficient to capture global prior information corresponding to different distortions. Subsequently, distortion-enhanced representations are implicitly estimated through contrastive learning and embedded into a super-resolution network equipped with multiple distortion decoders (D-Decoder). Furthermore, we propose a distortion-related channel segmentation (DCS) strategy that reduces the network’s parameters and computation (FLOPs). We refer to this Global Prior-guided Distortion-enhanced Representation Learning Network as GDRNet. Experiments on both synthetic and real-world remote sensing images demonstrate that our GDRNet outperforms state-of-the-art blind SR methods for remote sensing images in terms of overall performance. Under the experimental condition of anisotropic Gaussian blurring without added noise, with a kernel width of 1.2 and an upscaling factor of 4, the super-resolution reconstruction of remote sensing images on the NWPU-RESISC45 dataset achieves a PSNR of 28.98 dB and SSIM of 0.7656. Full article

Data from earth observation satellites provide unique and valuable information about water quality conditions in freshwater lakes but require significant processing before they can be used, even with the use of tools like Google Earth Engine. We use imagery from Sentinel 2 and MODIS and in situ data from the State of Utah Ambient Water Quality Management System (AQWMS) database to develop models and to generate a highly accessible, easy-to-use CSV file of chlorophyll-a (which is an indicator of algal biomass), turbidity, and water temperature measurements on Utah Lake. From a collection of 937 Sentinel 2 images spanning the period from January 2019 to May 2025, we generated 262,081 estimates each of chlorophyll-a and turbidity, with an additional 1,140,777 data points interpolated from those estimates to provide a dataset with a consistent time step. From a collection of 2333 MODIS images spanning the same time period, we extracted 1,390,800 measurements each of daytime water surface temperature and nighttime water surface temperature and interpolated or imputed an additional 12,058 data points from those estimates. We interpolated the data using piecewise cubic Hermite interpolation polynomials to preserve the original distribution of the data and provide the most accurate estimates of measurements between observations. We demonstrate the processing steps required to extract usable, accurate estimates of these three water quality parameters from satellite imagery and format them for analysis. We include summary statistics and charts for the resulting dataset, which show the usefulness of this data for informing Utah Lake management issues. We include the Jupyter Notebook with the implemented processing steps and the formatted CSV file of data as ^{supplemental materials}. The Jupyter Notebook can be used to update the Utah Lake data or can be easily modified to generate similar data for other waterbodies. We provide this method, tool set, and data to make remotely sensed water quality data more accessible to researchers, water managers, and others interested in Utah Lake and to facilitate the use of satellite data for those interested in applying remote sensing techniques to other waterbodies. Full article

Non-orthogonal multiple access (NOMA) and reconfigurable intelligent surfaces (RISs) are recognized as key technologies for beyond 5G and 6G wireless communications. To address the high computational complexity and non-convex optimization challenges, this letter proposes an optimization framework based on the Multi-Agent Deep Deterministic Policy Gradient (MADDPG) algorithm. The proposed framework jointly makes use of sensor grouping, power allocation, an RIS computation strategy, and phase shifts to minimize the remote state estimation (RSE) error. Simulation results demonstrate that the MADDPG algorithm, when applied in an RIS-assisted NOMA system, significantly reduces the RSE error. Full article

The gradual integration of Manned–Unmanned Teaming (MUM-T) is gaining increasing significance. An intriguing feature is the ability to do relative estimation solely through the use of the INS/GPS system. However, in certain environments, such as GNSS-denied areas, this method may lack the necessary accuracy and reliability to successfully execute autonomous formation flight. In order to achieve autonomous formation flight, we are conducting an initial investigation into the development of a relative estimator and control laws for MUM-T. Our proposal involves the use of a quaternion-based relative state estimator to combine GPS and INS sensor data from each UAV with vision pose estimation of the remote carrier obtained from the fighter. The technique has been validated through simulated findings, which paved the way for the experiments explained in the paper. Full article

Building height information plays an important role in many urban-related applications, such as urban planning, disaster management, and environmental studies. With the rapid development of real scene maps, street view images are becoming a new data source for building height estimation, considering their easy collection and low cost. However, existing studies on building height estimation primarily utilize remote sensing images, with little exploration of height estimation from street-view images. In this study, we proposed a deep learning-based method for estimating the height of a single building in Baidu panoramic street view imagery. Firstly, the Segment Anything Model was used to extract the region of interest image and location features of individual buildings from the panorama. Subsequently, a cross-view matching algorithm was proposed by combining Baidu panorama and building footprint data with height information to generate building height samples. Finally, a Two-Branch feature fusion model (TBFF) was constructed to combine building location features and visual features, enabling accurate height estimation for individual buildings. The experimental results showed that the TBFF model had the best performance, with an RMSE of 5.69 m, MAE of 3.97 m, and MAPE of 0.11. Compared with two state-of-the-art methods, the TBFF model exhibited robustness and higher accuracy. The Random Forest model had an RMSE of 11.83 m, MAE of 4.76 m, and MAPE of 0.32, and the Pano2Geo model had an RMSE of 10.51 m, MAE of 6.52 m, and MAPE of 0.22. The ablation analysis demonstrated that fusing building location and visual features can improve the accuracy of height estimation by 14.98% to 69.99%. Moreover, the accuracy of the proposed method meets the LOD1 level 3D modeling requirements defined by the OGC (height error ≤ 5 m), which can provide data support for urban research. Full article

In industrial wireless networks beyond 5G and toward 6G, combining uplink non-orthogonal multiple access (NOMA) with the Kalman filter (KF) effectively reduces interruption risks and transmission delays in remote state estimation. However, the complexity of wireless environments and concurrent multi-sensor transmissions introduce significant interference and latency, impairing the KF’s ability to continuously obtain reliable observations. Meanwhile, existing remote state estimation systems typically rely on oversimplified wireless communication models, unable to adequately handle the dynamics and interference in realistic network scenarios. To address these limitations, this paper formulates a novel dynamic wireless resource allocation problem as a mixed-integer nonlinear programming (MINLP) model. By jointly optimizing sensor grouping and power allocation—considering sensor available power and outage probability constraints—the proposed scheme minimizes both estimation outage and transmission delay. Simulation results demonstrate that, compared to conventional approaches, our method significantly improves transmission reliability and KF estimation performance, thus providing robust technical support for remote state estimation in next-generation industrial wireless networks. Full article

This paper proposes a novel Global Navigation Satellite System (GNSS)-based remote sensing method for simulating Radar Doppler Map (RDM) features through joint electromagnetic scattering modeling and signal processing, enabling characteristic parameter extraction for both point and ship targets in multi-satellite scenarios. Simulations demonstrate that the B3I signal achieves a significantly enhanced range resolution (tens of meters) compared to the B1I signal (hundreds of meters), attributable to its wider bandwidth. Furthermore, we introduce an Unscented Particle Filter (UPF) algorithm for dynamic target tracking and state estimation. Experimental results show that four-satellite configurations outperform three-satellite setups, achieving <10 m position error for uniform motion and <18 m for maneuvering targets, with velocity errors within ±2 m/s using four satellites. The joint detection framework for multi-satellite, multi-target scenarios demonstrates an improved detection accuracy and robust localization performance. Full article

The distribution of forest aboveground biomass density (AGBD) is a key indicator of carbon stock and ecosystem health in the Eastern Himalayas, which represents a global biodiversity hotspot that sustains diverse forest types across an elevation gradient from lowland rainforests to alpine meadows and contributes to the livelihoods of more than 200 distinct indigenous communities. This study aimed to identify the key factors influencing forest AGBD across this region by analyzing the underlying biophysical and anthropogenic drivers through machine learning (random forest). We processed AGBD data from the Global Ecosystem Dynamics Investigation (GEDI) spaceborne LiDAR and applied filtering to retain 30,257 high-quality footprints across ten ecoregions. We then analyzed the relationship between AGBD and 17 climatic, topographic, soil, and anthropogenic variables using random forest regression models. The results revealed significant spatial variability in AGBD (149.6 ± 79.5 Mg ha⁻¹) across the region. State-wise, Sikkim recorded the highest mean AGBD (218 Mg ha⁻¹) and Manipur the lowest (102.8 Mg ha⁻¹). Within individual ecoregions, the Himalayan subtropical pine forests exhibited the highest mean AGBD (245.5 Mg ha⁻¹). Topographic factors, particularly elevation and latitude, were strong determinants of biomass distribution, with AGBD increasing up to elevations of 2000 m before declining. Protected areas (PAs) consistently showed higher AGBD than unprotected forests for all ecoregions, while proximity to urban and agricultural areas resulted in lower AGBD, pointing towards negative anthropogenic impacts. Our full model explained 41% of AGBD variance across the Eastern Himalayas, with better performance in individual ecoregions like the Northeast India-Myanmar pine forests (R² = 0.59). While limited by the absence of regionally explicit stand-level forest structure data (age, stand density, species composition), our results provide valuable evidence for conservation policy development, including expansion of PAs, compensating avoided deforestation and modifications in shifting cultivation. Future research should integrate field measurements with remote sensing and use high-resolution LiDAR with locally derived allometric models to enhance biomass estimation and GEDI data validation. Full article

Heterogeneous remote sensing images, acquired from different sensors, exhibit significant variations in data structure, resolution, and radiometric characteristics. These inherent heterogeneities present substantial challenges for change detection, a task that involves identifying changes in a target area by analyzing multi-temporal images. To address this issue, we propose the Multi-Head Graph Attention Mechanism (MHGAN), designed to achieve accurate detection of surface changes in heterogeneous remote sensing images. The MHGAN employs a bidirectional adversarial convolutional autoencoder network to reconstruct and perform style transformation of heterogeneous images. Unlike existing unidirectional translation frameworks (e.g., CycleGAN), our approach simultaneously aligns features in both domains through multi-head graph attention and dynamic kernel width estimation, effectively reducing false changes caused by sensor heterogeneity. The network training is constrained by four loss functions: reconstruction loss, code correlation loss, graph attention loss, and adversarial loss, which together guide the alignment of heterogeneous images into a unified data domain. The code correlation loss enforces consistency in feature representations at the encoding layer, while a density-based kernel width estimation method enhances the capture of both local and global changes. The graph attention loss models the relationships between features and images, improving the representation of consistent regions across bitemporal images. Additionally, adversarial loss promotes style consistency within the shared domain. Our bidirectional adversarial convolutional autoencoder simultaneously aligns features across both domains. This bilateral structure mitigates the information loss associated with one-way mappings, enabling more accurate style transformation and reducing false change detections caused by sensor heterogeneity, which represents a key advantage over existing unidirectional methods. Compared with state-of-the-art methods for heterogeneous change detection, the MHGAN demonstrates superior performance in both qualitative and quantitative evaluations across four benchmark heterogeneous remote sensing datasets. Full article

Recent advancements in Earth observation technologies have accelerated remote sensing (RS) data acquisition, yet cross-domain semantic segmentation remains challenged by domain shifts. Traditional unsupervised domain adaptation (UDA) methods often rely on computationally intensive and unstable generative adversarial networks (GANs). This study introduces elevation-aware domain adaptation (EADA), a multi-task framework that integrates elevation estimation (via digital surface models) with semantic segmentation to address distribution discrepancies. EADA employs a shared encoder and task-specific decoders, enhanced by a spatial attention-based feature fusion module. Experiments on Potsdam and Vaihingen datasets under cross-domain settings (e.g., Potsdam IRRG → Vaihingen IRRG) show that EADA achieves state-of-the-art performance, with a mean IoU of 54.62% and an F1-score of 65.47%, outperforming single-stage baselines. Elevation awareness significantly improves the segmentation of height-sensitive classes, such as buildings, while maintaining computational efficiency. Compared to multi-stage approaches, EADA’s end-to-end design reduces training complexity without sacrificing accuracy. These results demonstrate that incorporating elevation data effectively mitigates domain shifts in RS imagery. However, lower accuracy for elevation-insensitive classes suggests the need for further refinement to enhance overall generalizability. Full article