The proliferation of Unmanned Aerial Vehicles (UAVs) for urban public safety is critically hindered by operational risks inherent in complex environments. To address these challenges, this paper introduces a “UAV-as-a-Service” (UaaS) paradigm, an application paradigm innovation that leverages the core infrastructure of telecom operators. Our primary contribution is a closed-loop intelligent method, representing an algorithmic innovation, centered on a 3D Dynamic Risk Map (DRM). The DRM is generated in real-time by a Dynamic Bayesian Network (DBN) and explicitly informs both a risk-aware Multi-Agent Reinforcement Learning (MARL) dispatcher and a hybrid Rapidly-exploring Random Tree Star (RRT*)-Model Predictive Control (MPC) path planner. This tight coupling ensures that tactical decisions are grounded in a holistic understanding of risk. Simulation results demonstrate that the proposed UaaS model substantially enhances operational outcomes, reducing the time to achieve critical situational awareness by over 75 % and decreasing firefighter risk exposure by$\mathbf{7 8 \%}$. These technical advancements validate a novel and viable Business-to-Government (B2G) service model, demonstrating significant technologycommercial synergy.

Artificial intelligence (AI)-based multi-modal fusion algorithms are pivotal in emulating clinical practice by integrating data from diverse sources. However, most of the existing multi-modal models focus on designing new modal fusion methods, ignoring critical role of feature representation. Enhancing feature representativeness can address the noise caused by modal heterogeneity at the source, enabling high performance even with small datasets and simple architectures. Here, we introduce DeepOmix-FLEX (Fusion with Learning Enhanced feature representation for X-modal or FLEX in short), a multi-modal fusion model that integrates clinical data, proteomic data, metabolomic data, and pathology images across different scales and modalities, with a focus on advanced feature learning and representation. FLEX contains a Feature Encoding Trainer structure that can train feature encoding, thus achieving fusion of inter-feature and inter-modal. FLEX achieves a mean AUC of 0.887 for prediction of chronic kidney disease progression on an internal dataset, exceeding the mean AUC of 0.727 using conventional clinical variables. Following external validation and interpretability analyses, our model demonstrated favorable generalizability and validity, as well as the ability to exploit markers. In summary, FLEX highlights the potential of AI algorithms to integrate multi-modal data and optimize the allocation of healthcare resources through accurate prediction.

Ice, Cloud, and land Elevation Satellite (ICESat-2) carries the Advanced Topographic Laser Altimeter System (ATLAS), which enhancing along-track sampling density but introduces substantial noise in photon point cloud data. Therefore, this study establishes a denoising and classification feature parameter system grounded in the three-dimensional spatial distribution characteristics of photon point clouds. Modeling is conducted in two layers: one layer for upper noise photons and canopy signal photons, and another layer for lower noise photons and ground signal photons. Machine learning and neural network algorithms are utilized to denoise and classify the original photon point clouds from ICESat-2, aiming to obtain a transferable and universally applicable supervised classification model for denoising photon point clouds. Recall, Precision, and the harmonic mean of Recall and Precision (F1 Score) are used as evaluation metrics to verify the accuracy of local, transfer, and global models. The results indicate that under various forest types and external conditions, the proposed photon point cloud Layered Denoising and Classification Model (LDCM) outperforms the Differential Regressive and Gaussian Adaptive Nearest Neighbor (DRAGANN, ICESat-2 ATL08 production algorithm), Ordering Points to Identify the Clustering Structure (OPTICS), and Adaptive Elevation Difference Thresholding (AEDTA) algorithms in terms of accuracy. Compared to the DRAGANN algorithm, the maximum accuracy improvement is 60%, with an average improvement of approximately 20%; compared to the OPTICS algorithm, the maximum accuracy improvement is 36%, with an average improvement of about 28%; compared to the AEDTA algorithm, the maximum accuracy improvement is 27%, with an average improvement of about 14%. The F1 Score for the validation set of the machine learning and neural network algorithms is above 0.94, with the Categorical Boosting (CatBoost) algorithm achieving the best performance. Both the transfer model and the global model have F1 Scores above 0.90. Therefore, the proposed photon point cloud LDCM not only demonstrates excellent classification accuracy but also exhibits good transferability and general applicability.

Integrating Unmanned aerial vehicle Laser Scanning (ULS) and Terrestrial Laser Scanning (TLS) data in complex forest environments remains a significant challenge. Despite the availability of numerous registration algorithms, robust comparative studies are limited by the lack of reliable multi-platform benchmark datasets. In this study, we introduce the first multiplatform benchmark dataset for ULS-TLS point cloud registration in forests, encompassing 17 plots from seven diverse regions with about 1.56 billion points. The dataset is categorized into three difficulty levels based on overlap ratio and rigid overlap.We evaluated the performance of five registration algorithms against this benchmark. Chen2022 achieved the highest accuracy with a 100% success rate across all difficulty levels. While Wu2024 demonstrated robust performance in lower difficulties but faced challenges in more complex scenarios. We also found that terrain variations and rigid overlap significantly impacted registration accuracy, particularly for algorithms reliant on individual tree positions such as Hyypp¨a2021 and Feng2024. These findings underscore the need for improved data collection strategy, ground filtering techniques, and feature matching algorithms to enhance performance in challenging environments. We present the first openly accessible multi-platform benchmark dataset for forested regions and anticipate that future research will expand this work to additional areas. The dataset can be downloaded from: DatasetDownloadLink.

The settlement and tilt of a communication base station’s cabinet can harm electronic components, affecting stable operation. A 3D model measures cabinet deformation using LiDAR data, facilitating timely maintenance. While limited studies address cabinet segmentation, existing methods, relying on line or plane detection, face challenges with interference from objects sharing features with the cabinet. This paper proposes a cabinet point cloud segmentation method that combines cluster analysis and plane Point Cloud Coverage (PCC). The method preprocesses the point cloud, analyzes the differences in normal vector characteristics in different points neighborhoods, segments the point cloud based on the given normal vector value, performs Euclidean clustering, and extracts the planar point cloud using the RANSAC algorithm to judge if it meets the characteristics of the cabinet. The proposed method is validated with nine datasets, demonstrating a 98.23% average F1-score. Results confirm the algorithm’s accuracy in extracting cabinet point cloud, showcasing its versatility compared to common algorithms.

Unmanned aerial vehicle laser scanning (ULS) and terrestrial laser scanning (TLS) registration is an essential method for acquiring comprehensive forest structural information and conducting forest resource inventories. Due to the sparsity of the understory point cloud in ULS, existing methods for forest area point cloud registration have limitations in processing. To address this issue, this study proposes an efficient and robust coarse registration algorithm for forest area TLS-ULS point clouds. First, the tree top points are obtained based on the neighborhood maximum and HeightAnd angle threshold constraint methods, which are used as keypoints to construct an irregular triangular mesh. Line segment feature descriptors are then constructed for each mesh edge to establish matching relationships for registration transformation. The experimental results obtained using multiple airborne and terrestrial point cloud datasets from different regions demonstrate that the proposed algorithm does not rely on tree trunk attributes and has no strict density requirements for airborne point clouds. High registration accuracy is achieved for eight test plots in two study areas, with translation and rotation errors of 0.28° and 0.12 m, respectively, and an average pointwise error of 0.14 m. This indicates that the proposed algorithm has high registration accuracy and strong robustness, making it suitable for TLS-ULS registration in forest scenes.

Leaf area index (LAI) is a significant parameter controlling a lot of physical and biological processes related to vegetation on the Earth's surface. Previously, an LAI estimation model of ICESat-2 (Ice, Cloud, and Land Elevation Satellite-2)/Atlas (Advanced Topographic Laser Altimeter System) has been established and the accuracy of ICESat-2 LAI has been evaluated using optical images. However, this model hasn't been tested with airborne data. To demonstrate the effectiveness of ICESat-2 LAI, this study applied the model to airborne LiDAR data in the Saihanba National Nature Reserve in the same season. Results showed that the coefficient of determination $(R^{2})$ of ICESat-2 LAI was 0.63 and the root mean square error (RMSE) is $1.03(n=26,\ p < 0.001)$ , ICESat-2 has the inversion capability of LAI comparable to airborne lidar. These findings may help in promoting the LAI estimation model and broadening the application fields of the photon-counting LiDAR data.

Photon-counting Light Detection and Ranging (LiDAR) can penetrate a certain depth of water and provide reliable date support for water depth information extraction. Taking the islands and reefs in the South China Sea as an example, this paper uses the only in-orbit spaceborne photon-counting LiDAR - Ice, Cloud and land Elevation Satellite-2/ Advanced Topographic Laser Altimeter System (ICESat-2/ATLAS) to carry out research on depth extraction and accuracy evaluation in shallow water areas of islands and reefs. Aiming at difference in the density distribution of water surface and bottom photons, this study respectively utilizes the interval estimate and the modified Ordering Points to Identify the Clustering Structure (OPTICS) to filter out noise photons, and the modified OPTICS is changed twice by filter parameters. Then, bathymetric extraction is realized by refraction correction and tide correction. Finally, the manual labeling photons and airborne bathymetric LiDAR data of South China Sea is used to evaluate the denoising and bathymetric accuracy.

As an essential parameter in forestry, crown base height (CBH) faces many tasks. The methods are still developing for estimating it. Unmanned aerial vehicles (UAVs) light detection and ranging (LiDAR) supplies new, massive, and high-density data for estimating CBH. Many methods had been generated to compute CBH indirectly using regression-based ways or directly using geometric/statistical LiDAR-based ways. However, there were few methods to deal with the problem of understory, trunk, and noise points caused by high-density UAV data. A robust method was first proposed in this study to directly estimate CBH from LiDAR data, which contained two significant skills: 1) understory vegetation removal for each tree using a polynomial curve and 2) computing CBH by kernel densification of the elevation frequency histogram of LiDAR data. It could tolerate the understory and trunk points better through kernel convolution. The method proposed in this study and a previous simple model were applied in a crabapple plot in the Huailai Remote Sensing Comprehensive Experimental Station, Hebei, China, and verified by field-measured data. It was inspiring that our method is slightly better, and the mean CBH of LiDAR-derived trees was only 1.60 cm higher than that of field-measured trees. The mean absolute error (MAE) of CBH was 4.91 cm,${R}^{2}$was 0.73, the root-mean-squared error (RMSE) was 8.29 cm, and the bias was 2.68% for these trees. Generally, this method showed strong usability for high-density UAV LiDAR data and high precision for measuring CBH of low trees.

As many LiDAR point cloud processing steps, such as reconstruction, are often time- and memory-consuming, dividing LiDAR point clouds into subregions is common and necessary during preprocessing. However, the existing data dividing methods rely on tedious manual work or regular grids and result in oversegmentation around cutting lines. In this letter, we propose a new gap-based data dividing method for various LiDAR point clouds that can minimize the intersections between cutting lines and objects. The basic idea is to find a set of optimal paths that consist of gaps between objects as potential cutting lines. The experiments and comparisons in three data sets demonstrate that the proposed method is much better than the baseline method in terms visual inspection and cutting line quality.

Rapid and accurate ecological surveys of mangroves are of great significance for coastal protection and global carbon balance assessments. Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2)/Advanced Topographic Laser Altimeter System (ATLAS) data provide an opportunity to conduct large-scale surveys of mangroves. The purpose of this study was to assess the expressiveness of ICESat-2 data for ground and canopy height retrievals in mangroves. First, the ICESat-2 data were processed to obtain the ground and canopy heights of mangrove areas. Second, the accuracies of the ground and canopy heights retrieved from the ICESat-2 data were verified by airborne light detection and ranging (LiDAR) data. Finally, we analyzed the influence of various factors on the ground and canopy height estimation accuracies. The results showed that the average errors of ICESat-2 for the ground and canopy heights were 0.28 and −0.21 m and that the root mean squared errors (RMSEs) were 0.96 and 2.50 m. The accuracies of the ICESat-2 ground and canopy height estimates differed significantly when day/night and strong/weak beams were used. The strong beams at night provided the most accurate estimations of canopy height (RMSE% = 24.4%) and are thus the most suitable choice for studying mangrove areas. In addition, the results indicated that slope is the variable that has the greatest influence on the accuracy of the ground elevation estimates of the four factors above, while the accuracy of canopy height estimates is significantly affected by the canopy height itself. Overall, our study found that ICESat-2 data are suitable for ecological investigations of mangroves.

Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2) shows great potential for forest height retrieval. However, there are abundant noise photons in the ICESat-2 data, which make the accurate extraction of global forest heights challenging. In this letter, a novel algorithm based on the clustering method of ordering points to identify the clustering structure (OPTICS) was proposed to remove noise photons. First, we modified the circular shape of the search area in the OPTICS algorithm to an elliptical shape. Second, a distance ordering of all photons was generated using the modified OPTICS algorithm. Finally, signal photons were effectively detected using distance thresholds set by the Otsu method. To evaluate the algorithm performance, both the simulated and real ICESat-2 data were applied to our proposed algorithm. In addition, we compared our algorithm with another noise removal algorithm based on the modified density-based spatial clustering of applications with noise (DBSCAN). The results show that our algorithm works well in distinguishing the signal and noise photons as indicated by high$F$values. Compared with the modified DBSCAN, our algorithm performs better in filtering out noise photons regardless of the simulated or real ICESat-2 data sets. In addition, the results also indicate that our algorithm is robust because it is insensitive to the clustering parameters. Overall, the new proposed algorithm is effective for removing noise photons in the ICESat-2 data.

The Ice, Cloud and Land Elevation Satellite-2 (ICESat-2) mission provides effective data for measuring global ground elevation and forest height. Unlike the ICESat, ICESat-2 emits three pairs of beams. Each pair includes a strong beam and a weak beam with an energy ratio of 4:1. To evaluate the performance of ICESat-2's strong and weak beams in estimating ground elevation and forest height, the ICESat-2 data in four different scenarios were analyzed; 1) ICESat-2's strong beams in the daytime, 2) ICESat-2's weak beams in the daytime, 3) ICESat-2's strong beams in the nighttime, and 4) ICESat-2's weak beams in the nighttime. The results indicate that the weak beams perform as well as strong beams in estimating ground elevations due to high coefficient of determination (R2) values and low root-mean-squared error (RMSE) values regardless of the daytime and nighttime data. While for forest height estimation, ICESat-2's weak beams perform worse than strong beams and the daytime data perform worse than the nighttime data. These results suggest that all ICESat-2 data are suitable for ground elevation extraction, while the ICESat-2's weak beams in the daytime are not suited to estimate forest height.

Although the Ice, Cloud, and land Elevation Satellite/Geoscience Laser Altimetry System (ICESat/GLAS) has been used for urban monitoring, however, previous studies focused on extracting the maximum building height within the footprint. In fact, the full-waveform recording of GLAS data makes it possible to extract multiple building heights. However, the uncertainty of the spatial distribution and reflectance creates considerable challenges for the fine inversion of multiple building heights within the footprint. In this letter, we proposed an inversion method of multiple building heights using GLAS data assisted by QuickBird imagery. First, the GLAS waveform and the auxiliary optical imagery were processed to extract some spectral, horizontal, and vertical information as the prior knowledge of the inversion model. Then, the multiple building heights were inversed from the optimal simulated waveform based on the 3-D geometric optical and radiative transfer (GORT) model. The building heights measured by airborne LiDAR were used to validate the inversed building heights. The results demonstrated that the proposed inversion method achieved the building height estimation accurately and precisely ($R^{2} = 0.971$, rRMSE = 13.2%, and$n = 430$). This letter may shed some light on extracting multiple-level heights within the footprint using satellite LiDAR full-waveform data.

Canopy height models (CHMs) were utilized to detect treetops and estimate individual-tree parameters. The treetop detection based on CHMs was affected by surface topography and crown characteristics. However, their effects have not been well studied. Therefore, this paper aimed at assessing the impacts of aforementioned factors to facilitate treetop identification from LiDAR-derived CHMs. To fulfill this objective, we first extended and improved the previous models for cases with various terrains. Then, a new theoretical model was developed to quantify treetop displacements for ellipsoidal tree crowns. Finally, we further analyzed the treetop displacements due to terrain slope, crown radius, crown shape, and offset distance to the slope surface. Our analysis indicates that the vertical displacement increases exponentially with terrain slope; thus, the effect of terrain slope must be considered over extremely steep areas; larger errors are observed for trees with a large crown radius; the treetop displacements are highly correlated with crown shape, the effect of topographic normalization can be neglected for conical crowns with a large crown angle, and the elliptical crown shape can reduce the treetop detection errors; and treetop displacements increases with offset distance to the slope surface inCase 2, while opposite results are observed inCase 5. In addition, the results also demonstrate that the effect of slope-distorted CHMs may be quite different for different types of tree crowns and terrains. Overall, this paper makes a significant contribution to the development of theoretical models for quantifying treetop displacements. Furthermore, our findings provide a theoretical basis and guidance for better identifying treetops from LiDAR-derived CHMs.

The fraction of photosynthetically active radiation (FPAR) is a key parameter in controlling mass and energy exchanges between vegetation and atmosphere. LiDAR data-derived canopy vertical structural information and hyperspectral image-derived vegetation spectral information can be considered as complementary for vegetation FPAR estimation. To the best of our knowledge, few studies have estimated vegetation FPAR by both LiDAR and hyperspectral data based on physical models. This letter aims to explore the ability of combining airborne LiDAR and hyperspectral data to retrieve maize FPAR based on the energy budget balance principle. First, canopy gap probability and openness were estimated from airborne LiDAR data. Next, canopy reflectance and soil background reflectance were retrieved from hyperspectral image. Then, we estimated maize FPAR based on the energy budget balance principle. Finally, model validity was assessed byin situdata and results showed the physical FPAR estimation model estimated maize FPAR accurately. These results indicated that the physical method proposed in this letter was efficient and reliable to estimate maize FPAR, and FPAR retrieval can benefit from the complementary nature of LiDAR-captured canopy structural information and hyperspectral-detected vegetation spectral characteristics.

The accurate estimation of terrain slope is very important for accurately monitoring the elevation and mass changes of glacier using laser altimeter. In this paper, a novel physical model was proposed for accurately estimating withinfootprint terrain slope. The new proposed model was built based on overlapping footprints of the geoscience laser altimeter system (GLAS) data, namely, using altitude angle, footprint size, shape, orientation, terrain aspect, and ground extent. Ground extent estimation models were established on the basis of linear regression analyses between: 1) GLAS-derived waveform extent and airborne topographic mapper (ATM)-derived ground extent and 2) GLAS-derived waveform width and ATM-derived ground extent, respectively. In addition, the terrain slopes estimated from the overlapping footprints were validated by ATM data and compared with the slopes calculated from surface elevations, i.e., from ASTER global digital elevation model (DEM) (GDEM) and GLAS elevation. Results showed that the accuracy of waveform width-predicted ground extents (R2= 0.868, RMSE = 0.686 m, n = 20, and p-value2= 0.776, RMSE = 0.824 m, n = 20, and p-value <; 0.0001), which indicated that waveform width is more suitable for estimating ground extent. Slopes estimated from the new proposed model have a strong consistency with those calculated from ATM data (Corrcoef = 0.786, bias = 0.654°, SD = 1.368°, and RMSE = 1.452°). Additionally, results also indicated that the new proposed model performs much better than the methods based on ASTER GDEM and the GLAS surface elevation in estimating within-footprint terrain slope due to higher correlation, lower bias, standard deviation, and RMSE.

The accurate estimation of within-footprint slope is very important for measuring earth’s surface characteristics using satellite light detection and ranging (LiDAR) data. Several models have previously been proposed for slope estimation; however, these models have limitations in either accuracy or applicability. Therefore, the main purpose of this paper is to explore the influence of various factors (e.g., ground vertical extent, footprint size, footprint shape, and footprint orientation) on slope estimation to better estimate the within-footprint slope using large-footprint waveform LiDAR data. The results indicated that the absolute slope error due to the coupling effect of ground vertical extent and footprint size increased with an increase in the ratio of ground vertical extent and footprint size, while the relative slope error had an opposite trend. The slope error caused by footprint shape was relatively low when the footprint eccentricity was small. However, the slope error due to footprint shape grew rapidly when the footprint eccentricity became larger; thus, it is essential to fully take into account the influence of footprint shape on within-footprint slope estimation. In addition, the results suggest that the slope error changed regularly based on the intersection angle between footprint orientation and terrain aspect. This paper also provided guidance for the determination of an easy and practical model for within-footprint slope estimation. The determination of best model is dependent on the value of intersection angle. Once the intersection value is given, the best model can easily be determined. Using the best model, the within-footprint terrain slope can be estimated with high accuracy.