Image super-resolution (SR) is a significant technique in image processing as it enhances the spatial resolution of images, enabling various downstream applications. Based on recent achievements in SR studies in computer vision, deep-learning-based SR methods have been widely investigated for remote sensing images. In this study, we proposed a two-stage approach called bicubic-downsampled low-resolution (LR) image-guided generative adversarial network (BLG-GAN) for remote sensing image super-resolution. The proposed BLG-GAN method divides the image super-resolution procedure into two stages: LR image transfer and super-resolution. In the LR image transfer stage, real-world LR images are restored to less blurry and noisy bicubic-like LR images using guidance from synthetic LR images obtained through bicubic downsampling. Subsequently, the generated bicubic-like LR images are used as inputs to the SR network, which learns the mapping between the bicubic-like LR image and the corresponding high-resolution (HR) image. By approaching the SR problem as finding optimal solutions for subproblems, the BLG-GAN achieves superior results compared to state-of-the-art models, even with a smaller overall capacity of the SR network. As the BLG-GAN utilizes a synthetic LR image as a bridge between real-world LR and HR images, the proposed method shows improved image quality compared to the SR models trained to learn the direct mapping from a real-world LR image to an HR image. Experimental results on HR satellite image datasets demonstrate the effectiveness of the proposed method in improving perceptual quality and preserving image fidelity.

Airborne bathymetric LiDAR (ABL) acquires waveform data with better accuracy and resolution and greater user control over data processing than discrete returns. The ABL waveform is a mixture of reflections from the water surface and bottom, water column backscattering, and noise, and it can be separated into individual components through waveform decomposition. Because the point density and positional accuracy of the point cloud are dependent on waveform decomposition, an effective decomposition technique is required to improve ABL measurement. In this study, a new progressive waveform decomposition technique based on Gaussian mixture models was proposed for universal applicability to various types of ABL waveforms and to maximize the observation of seafloor points. The proposed progressive Gaussian decomposition (PGD) estimates potential peaks that are not detected during the initial peak detection and progressively decomposes the waveform until the Gaussian mixture model sufficiently represents the individual waveforms. Its performance is improved by utilizing a termination criterion based on the time difference between the originally detected and estimated peaks of the approximated model. The PGD can be universally applied to various waveforms regardless of water depth or underwater environment. To evaluate the proposed approach, it was applied to the waveform data acquired from the Seahawk sensor developed in Korea. In validating the PGD through comparative evaluation with the conventional Gaussian decomposition method, the root mean square error was found to decrease by approximately 70%. In terms of point cloud extractability, the PGD extracted 14–18% more seafloor points than the Seahawk’s data processing software.

Automatic, wall-to-wall monitoring of forests from remote sensing data is a dream slowly becoming a reality. One barrier to development is the laborious task of developing quality ground reference data. Structural information captured by terrestrial laser scanners (TLS) or unmanned aerial systems (UAS) would expedite the collection of ground reference data if tree species could be automatically determined from point clouds. This study aims to improve species classification from point clouds by identifying detectable structural features useful for classifying species. We compare the effectiveness of multiple feature design strategies for classifying dominant hardwood species (oaks and sugar maples) from single-scan TLS and UAS data of natural hardwood forests, and analyze the separability of these species within and across the canopy layers. We find that oaks and sugar maples have distinct profile shapes that both single-scan TLS data and UAS LiDAR data can capture. TLS captures species-specific profile shapes through direct measurement of canopy width. UAS LiDAR, with its characteristically occluded understory, relies more on canopy density features. Our results emphasize the importance of tailoring data processing and feature extraction for capturing understory structure and highlight the need for modality-specific feature design. Implementing these insights will improve the accuracy and efficiency of automated tree-level inventories in hardwood forests, ultimately supporting more robust forest monitoring and management practices.

Differencing digital terrain models (DTMs) generated from multitemporal airborne light detection and ranging (lidar) data provide accurate and detailed information about three-dimensional (3D) changes on the Earth. However, noticeable spurious errors along flight paths are often included in the differencing results, hindering the accurate analysis of the topographic changes. This paper proposes a new scalable method to alleviate the problematic systematic errors with a high degree of automation in consideration of the practical limitations raised when processing the rapidly increasing amount of large-scale lidar datasets. The proposed method focused on estimating the displacements caused by vertical positioning errors, which are the most critical error source, and adjusting the DTMs already produced as basic lidar products without access to the point cloud and raw data from the laser scanner. The feasibility and effectiveness of the proposed method were evaluated with experiments with county-level multitemporal airborne lidar datasets in Indiana, USA. The experimental results demonstrated that the proposed method could estimate the vertical displacement reasonably along the flight paths and improve the county-level lidar differencing results by reducing the problematic errors and increasing consistency across the flight paths. The improved differencing results presented in this paper are expected to provide more consistent information about topographic changes in Indiana. In addition, the proposed method can be a feasible solution to upcoming problems induced by rapidly increasing large-scale multitemporal lidar given recent active government-driven lidar data acquisition programs, such as the U.S. Geological Survey (USGS) 3D Elevation Program (3DEP).

Accurate estimation of individual tree aboveground biomass (AGB) is critical for improving our understanding of forest structure and function, especially in forests with diverse tree species spanning a wide range of sizes. Terrestrial laser scanning (TLS) has proven effective at the tree scale thanks to its ability to provide high-density point cloud data. However, most existing methods rely on high-quality multi-scan TLS data, which requires labor-intensive data collection and complicated preprocessing to achieve high precision, thereby limiting their practical applicability. In contrast, single-scan TLS data offers a simpler and more efficient alternative, though it presents significant challenges due to incomplete coverage, uneven point density, and frequent occlusions. This study presents a novel and effective framework for estimating individual tree AGB from single-scan TLS data by integrating a deep learning network (CoAtNet) with three key innovations: an optimized 3D-to-2D projection strategy, a 3D point cloud-based data augmentation, and an ensemble approach to enhance robustness under data-scarce conditions in local natural forest environments. The framework was evaluated in a mixed hardwood forest and demonstrated strong predictive performance with a coefficient of determination (R2) of 0.73, a median percentage error of 0.99%, and a median absolute percentage error of 25.06%, outperforming two comparison models, Random Forest and Point Transformer. The results indicate the potential of single-scan TLS data to support reliable individual tree biomass when combined with deep learning, even in the presence of data imperfections. The proposed framework provides a practical, scalable approach that contributes to the advancement of remote sensing methodologies for sustainable forest management and facilitates improved monitoring of ecological changes in the context of accelerating climate change.