Vertical structural and spectral heterogeneity are two key remote sensing characteristics of complex forests. To enable effective forest health management and provide early warnings of abnormal disturbance, monitoring forest biochemical content with a vertically layered spectral perspective is critically needed. However, commonly used remote sensing technique still have limited capacity to study the biochemical status of the middle and lower canopy layers. This study provides the first insight into the potential of the full-waveform large-footprint hyperspectral LiDAR (LFHSL) system for retrieving the vertical heterogeneity of forest chlorophyll using 3D radiative transfer modeling. In our newly constructed LFHSL model, virtual three-dimensional (3D) complex forest scenes, comprising trees, bushes, and grass, were defined with varying positions and biochemical content inputs. Hyperspectral waveforms within the large-footprint were then simulated for each combination of vegetation position and biochemical level. The concept of spectral index time profiles (SITP), referred to as spectral index variation along the laser path, were introduced and used to assess the vertical distribution of chlorophyll for the first time in forest scenes. The main findings of this study are as follows: (1) Full-waveform LFHSL owns great potential for retrieving vertical chlorophyll content across trees, bushes, and grass layers in complex forest ecosystems. (2) SITP is a novel and essential reference indicator that fully registers chlorophyll variations along the laser path. (3) Simulations with random positions and chlorophyll contents indicate that the peak points of SITP yield higher chlorophyll prediction accuracy in trees layer and grass layer (R2: 0.996 vs. 0.971, RMSE: 1.39 vs. 3.81 μgcm−2) than that of bushes layer (R2 of 0.801, RMSE of 9.97 μgcm−2). (4) Compared to position patterns, LFHSL system is more sensitive to chlorophyll content sets. This study demonstrates that full-waveform LFHSL is a promising and surely reliable tool for acquiring and monitoring vertical vegetation health in complex forests. It not only provides significant guiding for the development of laser radar models but also holds promise for adoption in design of large-footprint multi-spectral or hyperspectral LiDAR.

Accurate estimation of residential building energy consumption and associated CO2 emissions is essential for refined urban carbon management. This study develops a hybrid framework that integrates physics-based simulation and machine learning to estimate residential building energy use and energy-related CO2 emissions in Shenzhen in 2020. Representative building archetypes were first simulated and then used to train machine-learning models for large-scale applications. Building-level energy estimates were further combined with a bottom-up inventory to generate high-spatiotemporal-resolution maps of residential CO2 emissions. The results show that: (1) the selected model achieved good accuracy and temporal robustness, with strong agreement between estimated and reference energy use at daily, monthly, and annual scales; (2) residential energy use was primarily driven by meteorological conditions, especially daily mean temperature and the duration of high-temperature conditions, and exhibited clear weekly and seasonal patterns, with higher values on weekends and in summer; (3) residential CO2 emissions in Shenzhen reflected the combined effects of scale and intensity, with Longgang and Bao’an contributing the largest total emissions, Self-built residential buildings contributing the largest aggregate emissions, and Old residential buildings showing the highest average emissions per building; (4) emissions were highly concentrated in a small number of high-emission buildings, which were more frequently distributed along road-adjacent block perimeters. Overall, the proposed framework improves the fine-scale characterization of residential building CO2 emissions and provides a useful basis for hotspot identification and targeted mitigation.

Streetlights are essential infrastructure in cities. They enhance traffic and public safety at night, but they also entail significant energy consumption. The study of the density, quantity and distribution of streetlights is essential for sustainable urban development. The Sustainable Development Global Scientific Satellite-1 has enriched night-time light data. This study is based on the Crowded Scene Recognition Network (CSRNet) and proposes a novel Streetlight Counting Network (SLCNet) by introducing a deeper front-end network and adding a spatial attention mechanism. This model can effectively estimate the density distribution of streetlights within a region and achieve automatic counting of large-scale streetlight numbers. Experimental results show that SLCNet achieves a mean absolute error of 135.35, a root mean square error of 231.37, and a normalised absolute error of 0.22, representing reductions of 5.60 %, 8.08 %, and 12.00 % respectively outperforming the original CSRNet model. The visualisation results also show that SLCNet predicts the distribution of streetlight density more accurately than CSRNet. Comprehensive ablation experiments further validate the effectiveness of each improvement method. Compared with the high-resolution yet costly Jilin-1 night-time images, SLCNet maintains comparable accuracy while enabling large-scale streetlight density and quantity estimation with publicly accessible SDGSAT-1 images. This study demonstrates the application potential of SDGSAT-1 in urban night-time lighting monitoring, providing new technical pathways and research frameworks for urban lighting management, sustainable development assessment, and smart city construction.

The edge effect, caused by partial illumination of leaf boundaries, leads to significant hyperspectral LiDAR (HSL) echo-intensity loss and severely restricts the precise expression of the spectral characteristics and retrieval accuracy of biochemical components in the leaf-edge area. Despite being one of the major radiometric effects in LiDAR sensing, the edge effect has not been systematically investigated for hyperspectral intensity data. Along these lines, this study presents the first comprehensive exploration, detection, and filtering of edge effects in HSL-collected data. We analyze the physical mechanisms responsible for multilayer edge-effect formation and propose a new integrated detection and filtering framework, incorporating a histogram of intensity distribution, edge detection, and spherical-spatial filtering (HIDEDaSF). Results obtained from the HSL system on broadleaf plants demonstrate that our framework robustly detects edge-affected spectral points across 32 wavelengths. After filtering and correction, the standard deviation and coefficient of variation (CV) of edge-region intensities are reduced by 22.68% and 28.30%, respectively, and the mean ratio of CV is 0.7288, less than 1, confirming the stability and effectiveness of the proposed algorithm framework again. The HIDEDaSF framework preserves valid spectral information and significantly improves the consistency and quantitative reliability of hyperspectral intensity data at leaf-edge zones. Although tailored for the HSL system, this framework is also adaptable to other multi and HSL platforms. This study enriches the fundamental knowledge of edge-effect radiometric behaviors and expands the methodology foundation for fine-scale HSL vegetation remote sensing. We are willing to make our codes freely available via https://github.com/Jie-Bai/HIDEDaSF

Continuous soil monitoring is crucial for agricultural and forestry management and for understanding ecosystem health. However, vegetation canopy limits passive remote sensing in capturing understory soil spectral signatures. The recently developed hyperspectral LiDAR (HSL) technology, capable of providing spectral information at dis tinct geometric positions, offers new possibilities for understory soil detection. In this study, 3D vegetation-soil scenes were constructed using field measured soil hyperspectral data and customized tree models. The LESS ra diative transfer model was employed to simulate HSL point clouds and hyperspectral imagery (HI) under varying vegetation coverage levels. The potential for understory soil spectral retrieval was evaluated from three aspects: radiance intensity, spectral curves, and spectral indices. Results demonstrate that laser pulse echoes are sub stantially less susceptible to canopy multiple scattering interference compared to passive imaging. Retrieved soil spectral curves exhibited markedly improved fidelity, with mean spectral angles decreasing from > 2.3 degrees for HI to < 0.2 degrees for HSL. Spectral indices showed stronger consistency with reference spectra, with R-2 values increasing from < 0.49 for HI to > 0.83 for HSL. Furthermore, HSL-derived spectral information demonstrated promising potential for soil property estimation, achieving R-2 of 0.332 for soil organic carbon (SOC) and 0.485 for total nitrogen (TN) through partial least squares regression modeling. This study demonstrates that HSL is a promising approach for soil monitoring in vegetated areas without requiring extensive bare soil exposure windows.

The system value of solar photovoltaics (PV) increasingly depends on effectively connecting generation to electricity demand. Here, we develop an integrated framework to diagnose this spatial constraint globally. Using Sentinel-2 imagery and deep learning, we map near-global utility-scale PV plants at 10-m resolution for 2025. Coupling this with generation models and a 1-km nighttime-light-derived electricity demand surface, we estimate ~ 1,852 TWh of annual PV generation against ~ 32.8 PWh of demand, yielding a ~ 5.65% mean penetration. Structural under-coverage is widespread: ~82% and 91% of global electricity demand lie in provinces below the 8.6% and 15.7% IEA reference thresholds, respectively. Under IEA-aligned scenarios requiring ~ 3,990 and ~ 5,910 TWh of PV supply, meeting demand relies heavily on domestic redistribution (~ 77.5%). Transfers exhibit a clear distance hierarchy, with > 80% occurring within 1,500 km and mean connection distances of ~ 593–718 km. This reveals dual infrastructure needs: expanding regional backbone networks and developing high-capacity long-distance corridors. As targets rise, the bottleneck shifts from reallocating existing production to mobilizing new potential via transmission and flexibility. We conclude the limiting factor is spatial system integration, requiring corridor-aware planning for equitable decarbonization.

Enhanced urban CO2 monitoring and understanding its spatiotemporal patterns and driving factors are essential for emission management and climate change mitigation. This study developed a high-resolution framework for predicting and mapping CO2 concentrations within the road network of Shenzhen by integrating vehicle-cruising observations, street-view panoramas, and multisource remote sensing data. The proposed machine learning model demonstrated strong predictive performance (R & sup2; = 0.92). Furthermore, the effects of urban function, urban development intensity, traffic conditions, and environmental factors on CO2 concentrations were systematically examined using explainable machine learning techniques. The results indicate that in urban centres, human activities exert a substantial influence on CO2 levels. Effective non-motorway planning, accessible public transport systems, and diversified urban functions are associated with lower CO2 concentrations. Notably, areas with high vegetation cover were also found to exhibit elevated CO2 concentrations, and the influence of greenery on Shenzhen's CO2 levels was positive in November. By integrating multisource data from the perspectives of the urban landscape and street configuration, this study provides an interpretable approach for analysing the complex drivers of urban CO2 dynamics. Overall, the findings establish a cost-effective methodological framework for refined urban carbon monitoring and offer actionable insights to support low-carbon urban planning and evidence-based policy formulation.

The large-scale combustion of fossil fuels and the associated greenhouse gas emissions are considered the primary contributors to global warming. Accelerating the deployment of solar photovoltaics (PV) is therefore essential; however, global-scale assessments of areas where PV is both technically feasible and economically viable remain limited. Here, we present the first 1 km-resolution, pixel-level global assessment of the technical and economic potential of PV and construct an 11-year (2013-2023) affordability dataset that tracks how declining capital costs reshape siting opportunities. Using Google Earth Engine to integrate multi-source rasters with annually updated total installation costs and national electricity prices, we first estimate technically developable potential by applying a comprehensive set of physical and land-use constraints. We then apply two transparent economic filters to delineate economically developable land: (i) a payback period <= 15 years and (ii) a straight-line distance <= 50 km to the nearest road and settlement. Our results are as follows: (i) Technical potential reaches 12.96 & times; 103 PWh yr-1 (87.96 % of resource potential), whereas economically developable potential totals 4.99 & times; 103 PWh yr-1 (33.86 %)-enough to satisfy global electricity demand more than 170 times; (ii) areas with payback periods <= 15 years increased from 24.69 % of land in 2013 to 75.13 % in 2023; (iii) analysis of 93,257 existing PV plants reveals that 19.36 % of actual installations occupy farmland, underscoring land-use and food-security trade-offs and the need for agrivoltaic (AV) solutions. These findings provide an open and reproducible basis for prioritizing grid expansion, financing, and AV policies to accelerate PV deployment while safeguarding food and ecological security.

The spatiotemporal dynamics of the Air Quality Index (AQI) and its response to vegetation regulation require further investigation. Using multi-source data from 2020 in Shandong Province, China, this study analyzed the effects of vegetation greenness (NDVI, LAI, EVI), ecological efficiency (Net Primary Productivity, NPP), and landscape structure on AQI within 3 km grids. Monthly correlation analyses revealed that AQI peaks in January (125.09), June (99.01), and December (105.81). PM2.5, O3, and PM10 were the primary pollutants in winter, summer, and spring/autumn, respectively. Vegetation showed a significant purifying effect from June to September. NPP (r = -0.83) was more effective in mitigating air pollution than greenness-related indices (r = -0.48). Pollution mitigation was enhanced by vegetation patches with complex shapes and dispersed configurations. During the non-growing season, the vegetation alleviating effect weakened considerably, and a decoupling between greenness and ecological efficiency occurred. This decoupling was associated with a stronger positive correlation between population density and AQI. The findings highlight the importance of seasonal vegetation dynamics and landscape optimization for regional air quality management.

Soil heterotrophic respiration (Rh) is a major component of the terrestrial carbon cycle, yet global-scale estimates remain highly uncertain due to sparse observations and model limitations. To address these gaps, we first expanded the Rh observational dataset with simulated Rh sites based on meteorological stations and incorporated high spatiotemporal resolution remote sensing data products. The Geographical Extreme Gradient Boosting (Geo-XGBoost) model was then applied to estimate spatiotemporal variability of global site-level Rh. Variable importance analysis revealed that integrated variable reflecting coupled water-energy conditions broadly dominated the spatiotemporal variation of global site-level Rh, while temperature-related variables were the primary driver in forest and cropland ecosystems and vegetation productivity-related variables were most influential in grassland and wetland ecosystems. The simulated Rₕ sites enhanced spatial representativeness across climate regions, with the tropics and southern temperate regions showing the greatest improvement. The Geo-XGBoost model driven by the expanded Rₕ dataset demonstrates superior predictive accuracy for global site-level Rₕ estimation (R² = 0.75, RMSE = 0.56 g C m⁻² d⁻¹). This performance improvement is primarily attributed to the Geo-XGBoost architecture. Independent validation further confirms the model’s robust generalizability across diverse ecosystems (R² = 0.65, RMSE = 0.48 g C m⁻² d⁻¹). These results demonstrate that combining expanded Rh datasets with Geo-XGBoost enables high-accuracy, spatially representative global site-level Rh estimation, providing a robust framework for more reliable carbon cycle assessments under future climate scenarios.

Hyperspectral LiDAR (HSL) exhibits immense potential for fine-grained target characterization and quantitative remote sensing through the synchronous acquisition of spatial and spectral information. However, its practical utility is hampered by the nonlinear "distance effect," which induces intensity signal distortion and couples the system's intrinsic optical response with the target's spectral properties. Prevailing calibration research largely depends on polynomial fitting or piecewise empirical models. These approaches generally process individual bands in isolation, disregarding inter-band physical correlations and the inherent Poisson statistical characteristics of photon detection, which frequently results in model overfitting and noise amplification within low signal-to-noise ratio bands. To resolve this, we propose a physics-driven Range-Spectral Disentanglement (Phy-RSD) method. By representing observational data as a low-rank matrix, the proposed method constructs a multi-target joint observation matrix and integrates Kullback-Leibler divergence as a Poisson constraint. Leveraging non-negative matrix factorization, the system's inherent range response and the target's intrinsic spectrum are synchronously decoupled. We conducted range experiments involving isotropic standard panels and anisotropic plant leaves, comparing Phy-RSD against classical polynomial and piecewise models. The results indicate that: (1) Phy-RSD extracts the global range response with exceptional accuracy, achieving an average coefficient of determination (R2) of 0.9963 and a mean root mean square error (RMSE ) of 0.0150, demonstrating superior performance compared to conventional polynomial and piecewise models; (2) throughout the detection range, the standard deviation (σ), coefficient of variation (CV ), and coefficient of variation ratio (ϵ ) of the correction results remain at minimum levels; (3) Pearson residual analysis confirms that the decoupled residual noise strictly adheres to a Poisson distribution, with variance-mean linear fitting R2 values of 0.624 and 0.749 for isotropic and anisotropic targets, respectively, thereby validating the model's physical foundation ; and (4) the method enables high-fidelity spectral restoration for calibrated vegetation samples. This study establishes an instrument-agnostic and physically rigorous radiometric calibration benchmark for HSL, facilitating 3D quantitative spectral inversion in complex scenarios.

Forest chlorophyll content (Cab) is a key indicator of vegetation photosynthetic capacity and physiological status, while its vertical distribution reflects canopy adaptation to light gradients. Passive optical remote sensing provides limited within-canopy information, and conventional LiDAR lacks biochemical sensitivity. To evaluate the potential of future full-waveform airborne hyperspectral LiDAR (FW-AHSL) for vertically resolved Cab detection, this study used the LESS three-dimensional radiative transfer model to simulate 32-band FW-AHSL waveforms in natural, regular, semi-random, random, and density-gradient forest scenes. Layer-specific vegetation indices, linear retrieval, density sensitivity analysis, and random forest mean decrease accuracy (MDA) were used to assess performance. Results showed that FW-AHSL distinguished echoes from different canopy layers and the ground. Multiple green, red-edge, and near-infrared indices correlated strongly with layer-specific Cab, with 11 indices showing mean correlations > 0.8 and 9 > 0.85. Semi-random and random scenes confirmed the potential for layer-specific Cab retrieval, although accuracy decreased from upper to lower layers. Density-gradient experiments showed stable upper-canopy retrieval but reduced middle- and lower-canopy performance due to occlusion and increasing density. MDA further indicated layer-dependent dominant indices, suggesting that differentiated waveform-spectral features are required. Overall, FW-AHSL shows strong potential for detecting vertical canopy-scale Cab in forests and supports future sensor design.

Abstract Net ecosystem exchange (NEE), a critical indicator of terrestrial carbon source‐sink dynamics, remains difficult to estimate across global flux sites due to spatial heterogeneity and sparse flux observations. Existing models overlook local variability, limiting accuracy in site‐level NEE estimation. Geographically Weighted Regression (GWR) captures spatial non‐stationarity in environmental variables on NEE by enabling spatially varying model parameters. This study develops a Geographically Weighted eXtreme Gradient Boosting (GWXGBoost) framework coupling XGBoost's strength in modeling complex nonlinear relationships with GWR's spatial explicitness. Using 8‐day data from 387 global flux sites and remote sensing predictor variables, we optimize model parameters locally per site to account for spatial heterogeneity in NEE drivers. Our results show water‐related variables dominate explanatory power in GWXGBoost at 36% of sites, particularly in arid regions. The GWXGBoost framework outperforms traditional global models, achieving R 2 = 0.63 and RMSE = 1.21 g C m −2 day −1 at the 8‐day scale—with ∼40% higher R 2 and ∼20% lower RMSE than global modeling. Model accuracy is enhanced at the site scale ( R 2 = 0.84) and remains robust across most plant functional types, with the highest value in deciduous broadleaf forests ( R 2 = 0.80) and the lowest in grasslands ( R 2 = 0.37). Independent validation across 140 flux sites ( R 2 = 0.61) demonstrated the model's robustness and multi‐scale generalizability. Moreover, local modeling produces NEE estimates more consistent with observations and better captures site‐specific NEE controls than global models. Our study demonstrates that modeling spatial heterogeneity with localized models improves NEE prediction and supports carbon budget assessments.

High-throughput plant phenotyping requires the simultaneous characterization of three-dimensional structural and spectral properties across organ to canopy scales. Hyperspectral LiDAR (HSL), which integrates the spectral resolution of two-dimensional hyperspectral imaging with the spatial sensing capabilities of LiDAR, offers the potential to generate 3D spectral point clouds but still faces a key challenge: obtaining comparable reflectance estimates from raw intensity measurements under varying range and incidence-angle conditions. Here, we propose a correction framework consisting of (i) a change-point optimized, automatic-separation piecewise distance-effect model (CP-ASP) and (ii) an incidence-angle correction based on the Lambert-Beckmann formulation. CP-ASP estimates a system-level shared breakpoint through global change-point optimization using a fixed candidate set and an aggregated objective over wavelength bands, improving reproducibility and robustness of range normalization across bands. We evaluated the workflow in controlled indoor experiments using full-waveform HSL data with 21 wavelength bands to generate plant-level spectral point clouds and demonstrate 3D mapping of representative spectral indices (SR, PRI, and CRI). The workflow provides a reproducible basis for plant-level 3D spectral mapping under controlled conditions, while indicating that points near leaf boundaries and at extreme incidence angles may remain challenging and should be further evaluated in outdoor heterogeneous canopies.

High-resolution mapping of urban surface CO2 is essential for refined carbon monitoring, emission management, and low-carbon urban planning. Mobile monitoring provides dense street-level observations, but raw CO2 measurements are often affected by transient traffic disturbances, vehicle idling, and localized plume events, which limits their direct use as stable spatial mapping targets. This study developed an integrated framework for predicting, mapping, and interpreting stable surface CO2 patterns in Shenzhen by combining vehicle mobile observations, CSF processing, multiscale remote sensing predictors, machine learning. A CSF-based lower-envelope filter was used to suppress short-duration positive peaks and extract a more stable CO2 accumulation signal from mobile observations. Multiscale predictors representing transportation, urban activity, surface environment, and built form were constructed to characterize both local and surrounding urban contexts. Compared with raw CO2, the CSF-processed target substantially improved prediction performance. The best validation R2 across the candidate models increased from 0.59 to 0.90 in April and from 0.62 to 0.93 in November. The predicted maps identified persistent high-CO2 areas in central and southwestern Shenzhen. SHAP results showed that transport networks and urban activity reinforced surface CO2 accumulation, whereas vegetation and open-surface contexts weakened accumulation at broader spatial ranges. These findings provide an interpretable framework for high-resolution urban CO2 mapping and refined low-carbon governance.

Protected areas (PAs) are central to China's forest conservation strategy, yet their effectiveness for carbon storage across governance and management contexts remains unclear. A clearer understanding of their current and future carbon benefits is essential for informing conservation and climate policy. Here, using 1-km GEDI satellite data, we show that forests within China's PAs store on average 68.29 ± 0.17 Mg C ha⁻¹ - about 13% more than matched unprotected forests. Carbon gains are highest in national parks (18.19 ± 0.69 Mg C ha⁻¹) and in managed naturally regenerating forests (9.85 ± 0.36 Mg C ha⁻¹), although some PA categories underperform. To assess future potential, we integrate GEDI observations with CMIP6 climate projections and find that under the high-emission SSP5-8.5 climate scenario, strongly protected areas could retain an additional ~600 ± 36.39 Tg C by 2100. These results show that strong protection and optimized management substantially enhance China's carbon sink, offering major opportunities for climate mitigation and biodiversity conservation.

Forests play a critical role as terrestrial carbon sinks in mitigating climate change. However, accurate quantification of soil respiration (Rs)-the primary CO2 efflux from forests-remains challenging due to existing studies' overreliance on annual-scale estimates, which obscure fine-scale spatiotemporal dynamics and key drivers of Rs. Here, we developed a 500 m resolution monthly Rs dataset for China's forests (2000-2020) using remote sensing data and a geographically weighted machine learning model. A geographically weighted extreme gradient boosting model achieved the highest accuracy in predicting monthly Rs (R2 = 0.74, RMSE = 0.8 g C m-2 day-1). The mean total annual Rs from 2000 to 2020 was 2.32 ± 0.09 Pg C year-1, with summer contributing most and winter least. Annual total Rs and seasonal total Rs showed significant increasing trends across China's forests from 2000 to 2020, with the strongest increases in southern China's young/middle-aged natural management forests and plantations. The relationships between Rs and its driving factors varied: gross primary productivity (GPP) was the primary driver of annual Rs across all forest management and age classes. Seasonally, temperature dominated Rs in spring/winter, and GPP dominated summer Rs across all forest management and age classes, while precipitation effects varied with management/age classes and season. Our findings highlight the necessity of monthly-scale analysis and the significant role of forest management and age in modulating Rs variability.

Street-level air pollution directly affects human exposure and is important for urban air quality management and street design. Street-level concentrations reflect the combined influence of broader urban background conditions and rapidly varying local processes. To explicitly disentangle these two components, this study develops a novel dual-branch spatiotemporal network (DBSTN) to model street-level exhaust gases (CO₂, NO₂, and PM2.5) in Shenzhen, China. The proposed framework integrates mobile air quality observations, street-view panoramas, street canyon geometry, meteorological measurements, and multi-source urban spatial data within a dual-branch architecture. A macro-scale spatial branch estimates background concentrations, while a micro-scale temporal branch captures localized residual variability, thereby separating fast-varying street-level perturbations from slow-varying urban background signals. DBSTN outperformed conventional baseline models and achieved the best predictive accuracy for all three pollutants, with R² values of 0.70 for CO₂, 0.72 for NO₂, and 0.87 for PM2.5. The results indicate that CO₂ reflects a mixed influence of background and local processes, NO₂ is more strongly associated with traffic emissions and enclosed canyon environments, whereas PM2.5 is more closely linked to meteorological variability and broader background exchange. By separating background and local components, the framework provides empirical insights into street-scale driving mechanisms, complements previous simulation-based understanding of the roles of street geometry, vegetation, and traffic emissions, and supports targeted pollution and carbon co-control policies.

Planted forests play a crucial role in forest restoration and carbon sequestration. However, different management practices in planted forests can alter forest structure and composition, potentially affecting UAV-LiDAR-based biomass estimation. This study evaluated tree-centric, object-based, and area-based approaches in both managed planted forests (MPF) and unmanaged planted forests (UPF). Furthermore, an object biomass index (OBI) was developed for the object-based approach to better capture the structural complexity of forest objects. We then compared the selected metrics and estimation accuracy of the three approaches across forests with different management types. The results showed that in MPF, metrics characterizing canopy height and horizontal structure were more important. Whereas in the UPF, metrics capturing fine-scale structural details contributed to a more accurate biomass estimation. Across all approaches, higher performance was observed in MPF subjected to intensive management. Among the three approaches, the proposed object-based approach proved to be the most robust, demonstrating its potential to mitigate the effects of individual tree segmentation errors. This study underscores the impact of forest management on UAV-LiDAR-driven biomass estimation approaches in planted forests, and emphasizes the importance of accounting for management regimes when estimating forest biomass at larger scales.