The digital terrain model (DTM) and digital surface model (DSM) are fundamental data used in many geospatial applications. There are several global open DTM and DSM products. Though previous studies have evaluated open DTM products in high-relief mountain and coastal regions, their performance in urban environments characterized by rapid urban expansion, complex land cover, and high-rise buildings remains insufficiently understood. In this study, we evaluated the elevation accuracy of seven open DTM and DSM products, including ASTER, AW3D30, FABDEM, Copernicus DEM GLO-30, NASADEM, SRTM, TanDEM-X 30 m EDEM in two urban environments in the cities Nanjing and Hong Kong of China. Differential GPS, the ATL08 product of Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2), and airborne LiDAR-derived DTM and DSM were used as references. Results demonstrate that when used as a DTM, FABDEM exhibits the highest accuracy, with normalized median absolute deviation (NMAD) less than 2.2 m in Nanjing and NMAD less than 5 m in Hong Kong. Yet, FABDEM shows systematic underestimation of elevation in both study areas across all land cover types, possibly due to overcorrection of height bias. When evaluated as DSMs, none of the products achieved a high level of performance, with none reaching NMAD < 2 m or LE90 (90% line error) <10 m. GLO-30 demonstrated relatively higher accuracy. Furthermore, we observed a notable contrast in the performance of TanDEM-X 30 m EDEM: it was the second-worst performing DSM in Nanjing but the best in Hong Kong. Our study underscores that modeling Earth’s surface (DSM) is more challenging than modeling bare-earth terrain (DTM). Therefore, future research should emphasize improving the accuracy and reliability of DSMs.

The fraction of absorbed photosynthetically active radiation (fAPAR) by vegetation canopies is a fundamental biophysical parameter for quantifying terrestrial carbon-water cycles and characterising vegetation physiological status. Its vertical profiles contain richer information on within-canopy light absorption and hold strong potential to improve simulations of carbon and water fluxes. Although several studies have demonstrated that integrating optical reflectance with LiDAR data can accurately estimate canopy-integrated fAPAR, a practical and robust integrated method for retrieving fAPAR profiles in forest canopies remains lacking. In this study, we propose a LiDAR-constrained absorption profiling method, termed LiCAP, for retrieving the vertical profiles of fAPAR. The LiCAP method is based on an extension of our previously developed canopy-integrated fAPAR method that combines airborne optical reflectance and LiDAR data. Here, we extend the method to retrieve fAPAR profiles by expressing them as a function of the profiles of directional canopy fraction derived from airborne LiDAR point clouds and the visible-to-near-infrared reflectance ratio. Validation using extensive field measurements from the National Ecological Observatory Network (NEON) and simulated datasets using the Large-scale Remote Sensing Data and Image Simulation Framework (LESS) demonstrated that the proposed LiCAP method effectively captures height-resolved fAPAR variation across diverse canopies. The estimated fAPAR profiles showed strong agreement with the reference value (R-2 > 0.78) for both the NEON field and the LESS simulated datasets. The LiCAP method enables accurate retrieval of canopy fAPAR profiles by integrating optical reflectance with LiDAR-derived structural information. It addresses a critical gap in fAPAR remote sensing and supports improved monitoring of ecosystem carbon-water dynamics.

The fraction of absorbed photosynthetically active radiation (fAPAR) of vegetation canopies is a crucial variable for understanding the ecosystem carbon cycle and assessing vegetation responses to climate change. Light absorption of the vegetation canopy is mainly determined by canopy structure and leaf optical properties. Traditional remote sensing methods typically estimate fAPAR from reflectance signals using radiative transfer models or empirical relationships with vegetation indices (VIs) and fAPAR. However, reflectance-based estimates often show moderate accuracy due to the complex relationship between reflected and absorbed fluxes. Airborne LiDAR provides direct information on canopy structural attributes relevant to radiation interception, such as fractional vegetation cover (fCover), which has been used to estimate fAPAR. However, the shortcomings of LiDAR in capturing the role of leaf optical properties introduce some uncertainty in fAPAR estimation. Combining reflectance with LiDAR data offers a promising pathway for improving fAPAR estimation. In this study, we adapted a physically-based model (fAPARRL) to integrate reflectance and LiDAR observations for fAPAR estimation. This model is grounded in spectral invariant theory and represents fAPAR as a function of visible and near-infrared reflectance and a LiDAR-derived canopy structural parameter. The model was evaluated against both VI- and LiDAR-based methods using NEON field datasets and synthetic datasets generated by the one-dimensional SCOPE and three-dimensional LESS radiative transfer models. Across these datasets, the combination of LiDAR and reflectance through the fAPARRL model consistently outperformed VI- and LiDAR-based approaches, with respective maximum improvements in R2 of 0.47 and 0.09. Sensitivity analyses on the simulated datasets further indicated that fAPARRL exhibited higher robustness to variations in chlorophyll content and leaf area index (LAI) than other conventional methods. The proposed fAPARRL model effectively integrates reflectance and LiDAR data through a physically-based scheme, offering improved accuracy and robustness for large-scale fAPAR estimation and ecosystem monitoring.

We present a seasonal, multi-angular, hourly dataset of hyperspectral top-of-canopy reflectance and far-red solar-induced chlorophyll fluorescence (SIF) collected over winter wheat (Triticum aestivum L.) from the onset of spring regrowth to senescence. Measurements were conducted at the Gucheng Agro-meteorological Experimental Station using an automated system integrating co-aligned spectrometers and a computer-controlled pan-tilt unit. Observations were acquired under predefined viewing geometries, including principal plane, hotspot, and azimuth modes, from approximately 08:00 to 17:00. Complementary measurements of canopy structure and leaf chlorophyll content are provided. The dataset enables investigations of diurnal, seasonal, and angular variability in canopy optical signals and supports the development and validation of remote sensing methods for structural and biochemical properties.

Terrestrial laser scanning (TLS) captures accurate 3D tree geometry, benefiting forest management and ecology. Deep learning has enhanced TLS point cloud processing but requires extensive training data with semantic annotations. Existing datasets typically focus on tree segmentation or species classification, lacking detailed leaf-wood separation, which is essential for carbon stock and photosynthesis modelling. Real-world TLS data with leaf-wood labels remain scarce, come from limited regions (e.g., Cameroon, Finland, USA, India), and are mostly leaf-on only. To fill this gap, we introduce HuashuTrees, a public multi-species bi-seasonal TLS dataset from deciduous broadleaf forests in Nanjing, China (leaf-off: January 2025; leaf-on: July 2025). Other than plot-level point clouds, it includes 318 single-tree point clouds from 159 trees across seven species, generated after registration, denoising, segmentation, leaf-wood separation, and manual correction. Technical validation confirms high consistency for leaf-wood annotation, with a mean agreement of 96.34% between the final dataset labels and independent manual re-classification. Additionally, TLS-derived diameter at breast height (DBH) shows strong agreement with field measurements (RMSE = 0.356 cm, R² = 0.978). HuashuTrees supports algorithm development for leaf-wood separation, segmentation, species classification, structure inversion, 3D reconstruction, and multi-temporal monitoring.

The greening of the Earth is widely attributed to biogeochemical drivers, yet the drivers of urban vegetation changes, further strongly modulated by land-use management, remain poorly understood. Here, we find that urban areas, particularly in the Global South, experience significant browning during 2000-2022, contrasting with widespread rural greening. This urban browning is primarily driven by land-use management, specifically urban expansion and human-induced degradation, with limited compensation from biogeochemical processes. Regional analysis reveals that urban expansion dominates vegetation loss in East and Southeast Asia, with nearly three times stronger impact than elsewhere. In the Americas and Africa, human-induced degradation is the major driver, exerting twice the negative influence compared to other regions. Although European cities show stronger green recovery efforts, biogeochemical drivers remain their primary greening driver due to the limited restored green space. Furthermore, high-income cities benefit more from biogeochemical enhancement and less from land-use pressures, whereas many Global South cities face severe trade-offs, with economic growth frequently coinciding with vegetation decline. These results underscore that land-use management unequally amplifies urban browning against a greening background, highlighting an urgent need for targeted land policies and sustainable development strategies to mitigate these adverse impacts, especially in the Global South.

Leaf inclination angle distribution (LIAD) is a fundamental parameter of models that illustrate the energy and mass exchanges for vegetation at all scales. Terrestrial laser scanning (TLS) instruments have emerged as valuable tools for acquiring detailed measurements of canopy structure. Here, we present the first intercomparison of the available LIAD estimation techniques using TLS data. The available LIAD estimation techniques were evaluated using TLS point clouds of both real and synthetic trees covering the full range of the existing LIAD types. The performance of the proposed TLS-based methods was also compared with the established, non-TLS-based leveled digital photography approach. The study highlighted that the algorithms that used merged point clouds performed better than their single-scan counterparts. TLS offered a more comprehensive representation of the canopy structure and overcame the limitations of the traditional leveled digital photography approach for both real and simulated trees. This study may serve as a template for establishing benchmark datasets, evaluation protocols, and accessibility of algorithms that could facilitate systematic comparisons of LIAD estimation algorithms. This collaborative effort promotes fairness, reproducibility, and the advancement of LIAD estimation techniques by enabling researchers to identify strengths, weaknesses, and areas for improvement in their algorithms.

The interaction conversion between cropland (CL) and natural ecological land (NEL) provides critical insights into the processes of agricultural land transition. However, current frameworks and methodologies for systematically detecting and analyzing this transition are still insufficient and require further refinement. The objective of the study is to develop a theoretical model of agricultural land transition and examines its evolutionary processes in China. The findings reveal that the interactive conversion between CL and NEL exhibits symmetry in both temporal trajectories and spatial patterns. The agricultural land transition can be divided into three stages at the national level, characterized by a shift in the dominant conversion pattern: from NEL-to-CL, to CL-to-NEL, and then back to NEL-to-CL. However, significant imbalances are observed in the transition processes across various provinces. Over time, the four pathways of agricultural land transition-NEL-dominated to NEL-dominated, CL-dominated to CL-dominated, NEL-dominated to CL-dominated, and CL-dominated to NEL-dominated-have yet to progress simultaneously. The dominance of a single pathway gradually became more prominent, with its proportion in different provinces increasing from 25% in the early period to 40% in the later period. Among the various factors influencing agricultural land transition, resource utilization, socio-economic factors, and policy implementation have a stronger impact than natural and climatic conditions. Furthermore, resource flows and planning interventions drive the initial agricultural land transition, while urbanization and population growth induce the second phase. Overall, this study enhances the theoretical understanding of agricultural land transition and offers decision-making guidance for rational and orderly agricultural land use conversion. Its ultimate aim is to provide valuable insights into the sustainable utilization and efficient management of agricultural land globally.

Understanding and analyzing the spatial semantics and structure of forests is essential for accurate forest resource monitoring and ecosystem research. However, the lack of large-scale and annotated datasets has limited the widespread use of advanced intelligent techniques in this field. To address this challenge, a fully automated synthetic data generation and processing framework based on the concepts of Digital Cousins and Simulation-to-Reality (Sim2Real) is proposed, offering versatility and scalability to any size and platform. Using this process, we created the Boreal3D, the world's largest forest point cloud dataset. It includes 1000 highly realistic and structurally diverse forest plots across four different platforms, totaling 48,403 trees and over 35.3 billion points. Each point is labeled with semantic, instance, and viewpoint information, while each tree is described with structural parameters such as diameter, crown width, leaf area, and total volume. We designed and conducted extensive experiments to evaluate the potential of Boreal3D in advancing fine-grained 3D forest structure analysis in real-world applications. The results demonstrate that with certain strategies, models pre-trained on synthetic data can significantly improve performance when applied to real forest datasets. Especially, the findings reveal that fine-tuning with only 20 of real-world data enables the model to achieve performance comparable to models trained exclusively on entire real-world data, highlighting the value and potential of our proposed framework. The Boreal3D dataset, and more broadly, the synthetic data augmentation framework, is poised to become a critical resource for advancing research in large-scale 3D forest scene understanding and structural parameter estimation.

Leaf chlorophyll content (LCC) is an important indicator of photosynthetic capacity. Sun-induced chlorophyll fluorescence (SIF) is an optical signal emitted from the leaf interior, providing a unique technique for accurately estimating LCC. The far-red to red ratio of chlorophyll fluorescence (Fratio) has been used to empirically estimate LCC in some previous studies. While these studies support the use of the Fratio for LCC estimation, its theoretical underpinning remains less well-defined and its effectiveness across a wider range of scenarios remains unclear. In this study, we established the relationship between the Fratio and LCC using the light use efficiency (LUE)-based SIF model and spectral invariant radiative transfer theory. Firstly, the LUE-based SIF model demonstrates that the change in the leaf Fratio is controlled by the ratio of the fluorescence escape fraction (i.e., fesc from the photosystem to the leaf surface) at the corresponding bands. Secondly, a fesc modeling approach is presented using the spectral invariant theory and thus the fesc ratio is linked to LCC. Theoretical analysis shows that the Fratio has a strong correlation with LCC, which explains over 90 % of the variation in Fratio. Both experimental measurements and model simulations from a radiative transfer model Fluspect were used to validate the relationship between LCC and three Fratio (i.e., F up arrow ratio, F down arrow ratio and Ftotratio), which were derived from the upward and downward SIF of leaves, as well as the total SIF observed from both sides. The Fluspect simulations were used to assess the sensitivity of the Fratio-LCC relationship to the leaf structure. Two types of experimental measurements, including the field measurements of three crops and the laboratory measurements of 20 tundra plants, were employed to examine the species dependence of the Fratio-LCC relationship. The performance of Fratio for LCC estimation was evaluated and compared with spectral indices and the PROSPECT model using the experimental measurements and leave-one-out cross-validation (LOOCV) approach. Both the Fluspect simulations and the experimental measurements indicate that the Fratio is strongly correlated with LCC for a wide range of leaf scenarios. The FratioLCC relationship remains relatively stable across different leaf structures and plant species, since the relationship is almost consistent. The LOOCV of experimental measurements shows that the Fratio provides promising and robust LCC estimates, with the Ftotratio performing the best. The Ftotratio outperforms spectral indices, reducing the RMSE for LCC estimation by 19.5 %-93.9 %. Furthermore, compared to the PROSPECT model, the Fratio achieves a reduction in RMSE by 30.4 %-77.8 %. These results demonstrate that the Fratio is effective for estimating LCC of diverse plant species. This study advances our understanding of the relationship between the Fratio and LCC, supporting the use of SIF signals for remote sensing of LCC.

Coastal wetlands,especially salt marshes,play a vital role in global carbon sequestration due to their substantial biomass accumulation and unique ecological functions.Understanding the carbon stock in wetlands is essential for evaluating their contribution to the global carbon cycle and assessing their potential for climate change mitigation.This study aims to accurately estimate the carbon stock in coastal wetland salt marsh vegetation areas by focusing on integrating remote sensing technology with ground-based measurements.Given the challenges of traditional field inventory methods,which are labor-intensive and spatially limited,and the limitations of remote sensing inversion methods in terms of accuracy and scope,this research proposes a two-step inversion modeling approach to enhance the estimation process and to improve the precision of carbon stock assessments.The methodology developed in this study integrates both UAV(Unmanned Aerial Vehicle)imagery and Sentinel-2 satellite data to estimate the aboveground biomass(AGB)and carbon storage of salt marsh vegetation.The first step involves constructing inversion estimation models that relate the field-measured AGB data of three types of salt marsh vegetation—Spartina alterniflora,Phragmites australis,and other species—to UAV imagery.In the second step,the UAV-based AGB data inversion estimation models are expanded to incorporate Sentinel-2 satellite imagery,allowing for extensive mapping and monitoring.Subsequently,carbon coefficients are used to calculate the vegetation carbon storage,soil carbon storage,and total carbon storage.This two-step inversion process enables the synergistic estimation of AGB and carbon storage.It overcomes the limitations of traditional methods and significantly improves accuracy over larger spatial extents.The inversion models for AGB estimation show high determination coefficients(R2)for three salt marsh species:0.48 for Spartina alterniflora,0.42 for Phragmites australis,and 0.45 for other species.The corresponding Root Mean Square Errors(RMSE)are 613.89 g/m2,650.6 g/m2,and 624.03 g/m2,respectively.This represents a significant improvement in estimation accuracy compared to using the Sentinel-2 satellite inversion models directly.Using these models,the study estimates that the Ningbo coastal wetland has a total area of 111.47 km2 of salt marsh vegetation.The total AGB for the wetland is calculated to be 3.09× 105 tons,with an associated carbon stock of 1.68× 106 tons,resulting in a carbon sequestration value of approximately 178 million RMB,highlighting the economic potential of these ecosystems for climate change mitigation.The integration of UAV and satellite remote sensing technologies has proven to be effective in overcoming the challenges of traditional ground sampling methods.This integration enables high-precision,large-scale estimation of carbon stocks in coastal wetlands.The two-step inversion modeling approach developed in this study offers a cost-effective,scalable,and accurate method for monitoring carbon storage in wetlands.This method not only enhances the estimation of carbon sequestration in coastal ecosystems but also provides valuable tools for environmental management and policymaking,especially in the development of coastal conservation strategies and sustainable carbon management.Furthermore,it demonstrates potential for broad application in various coastal wetland regions,providing a strong solution for future environmental monitoring and management.

Separation of soil effects from top-of-canopy (TOC) reflectance is crucial for quantitative remote sensing of vegetation. Soil affects TOC reflectance via the soil-vegetation interaction and the direct reflection by soil. Various vegetation indices have been developed semi-empirically to mitigate the interferences caused by soil for specific applications, such estimating biomass and monitoring vegetation phenology. However, a practical approach to separate soil effects from the entire TOC spectral reflectance is still lacking. In this study, we investigate the radiative transfer process in a vegetation canopy with soil contamination and develop three methods to estimate the contribution of soil's direct reflection to TOC reflectance. Theoretical analysis reveals that the soil's direct reflection can be quantified and separated from TOC reflectance due to the distinct spectral characteristics of soil and vegetation. We identify three key features: a) Bands in the visible region where the reflectance of soil-uncontaminated green vegetation approaches zero due to strong pigment absorption. b) Two bands in the visible region where the vegetation reflectance is similar, but soil reflectance is distinguishable. c) Soil reflectance within the range of 400 nm to 1000 nm exhibits a near-linear dependence on wavelength. Using these features, we develop three methods to quantify the contribution of soil's direct reflection to TOC reflectance. For given soil reflectance, feature a) or b) alone allows estimating the fraction of soil that directly contributes to TOC reflectance, and thus the soil's direct reflection. Using all three features enables estimation of the soil's direct reflection without knowing soil reflectance. The proposed methods, along with certain assumptions made during their development, are tested and evaluated using field and synthetic datasets of soil, leaf, and canopy. The evaluation of the three methods demonstrates that the estimation of the soil's direct reflection can be achieved through: i) Using TOC reflectance at approximately 675 nm and soil spectral reflectance, termed the red-band-based method (Method-RBB). ii) Using TOC reflectance at approximately 675 nm and 438 nm, along with soil spectral reflectance, termed as the two-band-based method (Method-TBB). iii) Using TOC reflectance at approximately 675 nm and 438 nm, assuming linear dependence of soil reflectance on wavelength in the visible and near-infrared region, termed as the linear-assumption-based method (Method-LAB). Our evaluation indicates that the linearity from 400 nm to 1000 nm holds true for a wide range of soil types. The conditions outlined in features a) and b) are valid for green vegetation with moderate to high leaf chlorophyll content: when leaf chlorophyll content exceeds 20 mu g cm-2, the leaf albedo at 675 nm is generally below 0.15, and the difference in leaf albedo at 675 nm and 438 nm is sufficiently small. The results reveal that when leaf albedo at 675 nm is less than 0.15 and NDVI is less than 0.8, all three methods perform satisfactorily, exhibiting an R2 value of approximately 0.9 between the true and estimated contribution of soil's direct reflection. The R2 values are 0.92 for both Method-RBB and Method-LAB, while Method-TBB has an R2 of 0.95. The performance of Method-RBB is particularly sensitive to leaf albedo at the red band, which correlates with leaf chlorophyll content. Canopies exhibiting higher red-band leaf albedo usually indicate lower chlorophyll content and less resemblance to typical green vegetation. The accuracy of Method-TBB diminishes as the differences in leaf albedo between the selected two bands increase. Similarly, deviations from the linear dependence of soil reflectance on wavelength negatively impact the accuracy of Method-LAB. Overall, these proposed methods work reasonably well for sparse canopies and healthy vegetation. Method-TBB exhibits the highest level of accuracy, followed by Method-RBB, while Method-LAB is more convenient to use as it does not require prior knowledge of soil reflectance. The proposed methods offer practical ways to estimate the contribution of soil's direct reflection to TOC reflectance. Utilizing TOC reflectance after the soil adjustment facilitates more direct monitoring of canopy structural characteristics, and biochemical and physiological information of leaves.

Solar-induced chlorophyll fluorescence (SIF) is an effective probe for photosynthesis, but this remote sensing signal is affected by multiple factors, including radiation intensity, canopy structure, sun-observer geometry, and leaf physiological status. The complex interplay among these factors causes substantial discrepancies among top-of-canopy (TOC) SIF, leaf-level average SIF and actual photosynthetic activity. Downscaling TOC SIF to the leaf-level and decoupling structural and physiological information remain major challenges in the use of SIF signals for remote sensing of photosynthesis. To address these challenges, the R2F (reflectance-to-fluorescence) theory was developed, grounded in the similarity in radiative transfer processes governing SIF and reflectance. This theory establishes a physical relationship between near-infrared reflectance (R-nir) and the far-red SIF scattering coefficient (sigma(F)). On this basis, SIF signals can be scaled from the canopy to the leaf level by normalizing oF, estimated from reflectance as sigma(F )= R-nir/i(0), where i0 denotes canopy interceptance. However, the original R2F formulation assumes a non-reflective soil. This simplification breaks down in sparse canopies, where soil contributions are non-negligible-an issue referred to as the "black-soil problem". Soil enhances both Rnirand oF, distorting their intrinsic relationship. In this study, we show that soil effects manifest through two main mechanisms: (1) direct soil reflection, which significantly increases R-nir but has minimal impact on sigma(F), and (2) soil-vegetation multiple scattering, which affects both R-nir and sigma(F )but tends to have compensatory effects. Consequently, the dominant source of bias in the original R2F relationship is direct soil reflection that contributes to R-nir-a mechanism that had not been explicitly isolated in previous studies. This finding allows us to narrow down the "black-soil problem" in the R2F framework to the specific impact of soil single scattering on Rnir. To mitigate this bias, we propose a soil-adjusted R2F (saR2F) method, which estimates the direct soil contribution of R-nir using TOC red and blue reflectance. Correcting R-nir for the direct soil reflection results in a robust relationship between sigma(F )and soil-adjusted R-nir(saR(nir)), notably sigma(F )= saR(nir)/i(0). We evaluated the saR2F relationship using one field and two simulated datasets. In the field study, saR2F improved the estimation of sigma(F) from TOC reflectance, with R-2 increasing ranging from 0.21 to 0.31 compared to the original (RF)-F-2. In the two simulations, saR2F consistently outperformed the original R2F, especially under sparse canopy conditions. We also compared saR2F with NDVI-based (NIRv) and FCVI-based R2F approaches. In the available field observations collected under specific conditions (i.e., varying viewing azimuth angles), the three approaches showed similar performance and were better than the original R2F in explaining the viewing-angle dependence of sigma(F). However, across the broader range of simulated scenarios and for estimating the exact sigma(F), saR2F demonstrated better stability than NIRv and FCVI-based R2F methods. The NIRv-based and FCVI-based R2F methods yielded relatively low RMSE (0.092 and 0. 075, respectively) but weak explanatory power, with R-2 values below 0.41 for canopies with LAI < 3. In contrast, saR2F achieved a much stronger relationship (R-2 = 0.80) and a low RMSE of 0.044. Furthermore, compared to the NIRv or FCVI-based approaches for R2F corrections, saR2F offers a more physically plausible and interpretable solution that can be applied to angular correction and total SIF estimation. The effective mitigation of the black-soil problem facilitates interpretation of raw SIF observations and enhances the monitoring of photosynthetic activity using SIF.

China plays an important role in the global terrestrial carbon cycle. While China is included in global assessments of the carbon cycle, such as the global carbon budget, the performance of dynamic global vegetation models (DGVMs) over China has rarely been evaluated. This knowledge gap constrains both model applicability and region-specific parameter optimization within China. To address this gap, our study assesses the performance of terrestrial carbon stocks and sinks simulated by 12 DGVMs in China from 1970 to 2018. The results indicate that (1) there is significant variation in the numerical magnitudes of terrestrial carbon stocks as simulated by various models, with mean vegetation carbon at 38.3 PgC and mean soil carbon at 115.3 PgC. Nevertheless, their spatial distribution demonstrates a remarkable degree of congruence. Notably, the simulated carbon stocks are generally in excess of existing estimates. (2) Despite the good consistency in the spatial distribution of terrestrial carbon sinks across different models, there is considerable fluctuation in the numerical values, with a mean carbon sink of 0.02 PgC yr−1, a value lower than pre-existing estimations. (3) The responses of terrestrial carbon stocks and sinks to CO2 fertilization, climate change, and land use change exhibit pronounced heterogeneity. CO2 fertilization has a positive effect, whereas land use change has a negative one. The impact of climate change is variable, and the carbon sink effect engendered by CO2 fertilization is negated by the adverse influence of land use change. This comprehensive evaluation of the simulation performance of DGVMs in China is anticipated to serve as an important reference for the functional analysis and parameter optimization of DGVMs within China.

The airborne LiDAR scanning technology is widely used in various fields due to its ability to quickly acquire geospatial features. As a crucial step in data applications, the automatic acquisition of point-by-point labels has become a hot topic in current research on point cloud data processing. Recent deep learning algorithms for feature extraction in the form of kernel point convolution (KPConv) are currently considered mainstream due to their outstanding performance. Among these, KPConv is the most representative. Although it has demonstrated good performance in various point-cloud classification tasks, the local window convolution approach overlooks global semantic information and does not focus on the relationship between point distribution and features. These shortcomings are particularly prominent in large-scale point-cloud classification. This paper proposes a novel network structure, LI-Net, that considers geographic location and feature fusion. First, location-enhanced kernel point convolution (LFKPConv) was designed. Subsequently, a new mask attention module was introduced to integrate the global features. Compared to traditional attention mechanisms, this module focuses only on a small number of prominent feature points, allowing for effective feature interactions while reducing memory consumption. Finally, to enable cross-layer feature fusion, a feature-weighting unit was designed in the encoding phase to enhance the significance of the semantic features. The proposed method achieved competitive results on the ISPRS, LASDU, and DFC2019 datasets, as well as a new state-of-the-art result on the GML dataset, with an average F1 score of 72.0% and an accuracy of 97.3%.

The monitoring of photosynthetic activity stands as a central objective for the upcoming generation of satellite missions. Two promising avenues, Sun-Induced Chlorophyll Fluorescence (SIF) and Photochemical Reflectance Index (PRI), have emerged for remotely sensing leaf physiology, thereby advancing our comprehension of plant–climate interactions. Over the past few decades, substantial progress has been made in measurement techniques, retrieval algorithms, and modeling biochemical and radiative transfer processes related to SIF and PRI. However, it is essential to acknowledge that SIF and PRI observations are subject to influences such as canopy structure, soil background, and sun-observer geometry, in addition to leaf physiological status. Consequently, the key challenges in SIF and PRI remote sensing involve downscaling canopy SIF and PRI to leaf-level values and decoupling structural and physiological information. Drawing a comparison between the radiative transfer processes involved in SIF and reflectance, a physical relationship between reflectance and scattering of far-red SIF has been established, giving rise to the R2F (Reflectance-to-Fluorescence) theory. Based on this theory, a Fluorescence Correction Vegetation Index (FCVI) has been formulated. The R2F theory and FCVI index have been successfully applied to interpret SIF observations across various scenarios, offering a theoretical foundation for the conversion of SIF across different scales—from canopy to leaf and photosynthetic systems. Furthermore, an analysis of the radiative transfer processes determining PRI signals reveals that the deviation between canopy- and leaf-level PRI primarily stems from the soil background. Both canopy structure and sun-observer geometry indirectly affect canopy PRI by altering the contribution of soil to canopy reflectance. The structural and angular variation in canopy PRI can be elucidated by considering the soil background. In order to mitigate the effects of soil background, as well as associated structural and angular effects, a soil-adjusted canopy PRI (SacPRI) is proposed. This adjustment involves subtracting the soil contribution, estimated using red reflectance, from the original canopy reflectance. Field and numerical experiments confirm that compared with the raw canopy PRI, SacPRI more closely aligns with the PRI observed in sunlit leaves. In summary, the proposed R2F relationship and the soil-adjusted canopy PRI offer valuable tools for enhancing the accuracy of photosynthesis monitoring through SIF and PRI, enabling more robust assessments across various scales and environmental conditions.

Dangtu County is an economically developed county in the eastern part of Anhui province. With the rapid development of urbanization in recent years, the land use type of Dangtu County is also changing rapidly, which brings a series of ecological and environmental problems such as soil pollution. Based on the characteristics of urbanization development and the spatiotemporal changes of land use types in Dangtu County in recent years, this paper uses remote sensing technology to classify land use types by maximum likelihood method. On the basis of land use classification in Dangtu County, combined with social and economic development and urban land use planning, soil pollution in Dangtu County was divided into four levels, namely, high pollution area, medium pollution area, low pollution area and unpolluted area, and the range of pollution levels was delineated.

Three-dimensional laser scanning technology is widely employed in various fields due to its advantage in rapid acquisition of geographic scene structures. Achieving high precision and automated semantic segmentation of three-dimensional point cloud data remains a vital challenge in point cloud recognition. This study introduces a Multilevel Intuitive Attention Network (MIA-Net) designed for point cloud segmentation. MIA-Net consists of three key components: local trigonometric function encoding, feature sampling, and intuitive attention interaction. Initially, trigonometric encoding captures fine-grained local semantics within disordered point clouds. Subsequently, a multilayer perceptron addresses point-cloud feature pyramid construction, and feature sampling is performed using the point offset mechanism in the different levels. Finally, the multilevel intuitive attention(MIA) mechanism facilitates feature interactions across different layers, enabling the capture of both local attention features and global structure. The point-offset attention scheme introduced in this study significantly reduces computational complexity compared to traditional attention mechanisms, enhancing computational efficiency while preserving the advantages of attention mechanisms. To evaluate the results of MIA-Net, the ISPRS Vaihingen benchmark, LASDU and GML airborne datasets were tested. Experiments show that our network can achieve state-of-art performance in terms of Overall Accuracy(OA) and average F1-score(e.g., reaching 96.2% and 66.7% for GML datasets, respectively).

To achieve carbon neutrality, solar photovoltaic (PV) in China has undergone enormous development over the past few years. PV datasets with high accuracy and fine temporal span are crucial to assess the corresponding carbon reductions. In this study, we employed the random forest classifier to extract PV installations throughout China in 2015 and 2020 using Landsat-8 imagery in Google Earth Engine. The results were further visually inspected and refined by morphological filtering, cavity filling and manual adjustment. Validation analysis revealed that the initial classification achieved an overall accuracy over 96% for both 2015 and 2020. Further validation using independent test samples demonstrated that the final dataset outperformed the accuracies of existing PV datasets. In 2015, the total area of installed PV in China was 663.09 km2, which were mainly distributed in the northwest, Beijing-Tianjin-Hebei, and the Yangtze River Delta region. By 2020, the total area of PV reached to 2847.36 km2, with net increase of almost 3.3 times. Installed PV was intensified in the northwest and extended to eastern China.