Cheng Wang

LiDAR point clouds provide accurate and direct representations of spatial locations and geometric structures of objects in 3D space, making them essential for applications such as target recognition in autonomous driving and 3D reconstruction in smart cities. However, large-scale point clouds pose challenges, including massive data volume, uneven density distribution, and complex object structures. Existing point-based and graph-based semantic segmentation networks often suffer from limitations such as loss of local contextual information, over-reliance on local graph construction, and insufficient modeling of relationships between neighboring points. To address these issues, we propose PG-Net, a novel network that integrates discrete point distribution features with local graph structural information. The framework includes: (1) a point branch equipped with a Local Adaptive Feature Augmentation (LAFA) module to extract efficient local features; (2) a graph branch featuring a Dynamic Graph Feature Aggregation (DGFA) module, which explicitly models relationships among points in local graphs and adaptively balances a point’s intrinsic features with its neighborhood context; and (3) fuses local features from both branches, allowing their complementary strengths to enhance feature representation, a process further promoted by a New Aggregation Loss Function. Experiments on the Toronto3D and S3DIS datasets show that PG-Net achieves overall accuracy (OA) of 97.69% and 89.87%, and mean Intersection-over-Union (mIoU) of 83.51% and 73.22%, respectively. Comparative and ablation studies against advanced methods such as RandLA-Net, BAAF-Net, and LACV-Net demonstrate the effectiveness and robustness of our approach. By jointly exploiting discrete point distribution and local graph structural relationships, PG-Net effectively leverages the complementary strengths of its dual-branch design, offering a reliable solution for efficient and accurate large-scale point cloud semantic segmentation.

Forest carbon offset projects are vital for carbon removal; however, concerns over their integrity persist due to allegations of over-crediting by major carbon-crediting programs. This study evaluates the integrity of afforestation and reforestation (AR) and improved forest management (IFM) projects across six prominent carbon-crediting programs: the Clean Development Mechanism (CDM), Verified Carbon Standard (VCS), Gold Standard (GS), American Carbon Registry (ACR), Climate Action Reserve (CAR), and Chinese Certificated Emissions Reduction (CCER). We compare emissions reductions estimated using remote sensing data with those verified by third-party verifiers. Although 665 projects were initially identified, over 70% lacked sufficient data for evaluation, leaving 154 projects under the CDM, VCS, ACR, and CAR programs across 24 countries for our analysis. Our findings reveal a significant risk of over-crediting, with third-party verified credits averaging 121% higher than satellite-based estimates of emission reductions. Notably, IFM projects under the CAR program exhibited the least bias. We also observed that large-scale methodologies, verifiers with focused expertise, projects in more developed host countries, and favourable business environments were correlated with lower overestimation. Our study emphasizes the need for improved tracking and verification processes to enhance the credibility of forest carbon offset credits and foster greater confidence in natural-based carbon removal solutions.

The accurate prediction of PM_2.5 concentration across extensive temporal and spatial scales is essential for air pollution control and safeguarding public health. To address the challenges of the uneven coverage and limited number of traditional PM_2.5 ground monitoring networks, the low inversion accuracy of PM_2.5 concentration, and the incomplete understanding of its spatiotemporal dynamics, this study proposes a refined PM_2.5 concentration estimation model, Bi-LSTM-SA, integrating multi-source remote sensing data. First, utilizing multi-source remote sensing data, such as MODIS aerosol optical depth (AOD) products, meteorological data, and PM_2.5 monitoring sites, AERONET AOD was used to validate the accuracy of the MODIS AOD data. Variables including temperature (TEMP), relative humidity (RH), surface pressure (SP), wind speed (WS), and total precipitation (PRE) were selected, followed by the application of the variance inflation factor (VIF) and Pearson’s correlation coefficient (R) for variable screening. Second, to effectively capture temporal dependencies and emphasize key features, an improved Long Short-Term Memory Network (LSTM) model, Bi-LSTM-SA, was constructed by combining a bidirectional LSTM (Bi-LSTM) model with a self-adaptive attention mechanism (SA). This model was evaluated through ablation and comparative experiments using three cross-validation methods: sample-based, temporal, and spatial. The effectiveness of this method was demonstrated on Beijing–Tianjin–Hebei urban agglomeration, achieving a coefficient of determination (R²) of 0.89, root mean squared error (RMSE) of 12.76 μg/m³, and mean absolute error (MAE) of 8.27 μg/m³. Finally, this model was applied to predict PM_2.5 concentration on Beijing–Tianjin–Hebei urban agglomeration in 2023, revealing the characteristics of its spatiotemporal evolution. Additionally, the results indicated that this model performs exceptionally well in hourly PM_2.5 concentration forecasting and can be used for PM_2.5 concentration hourly prediction tasks. This study provides technical support for the large-scale, accurate remote sensing inversion of PM_2.5 concentration and offers fundamental insights for regional atmospheric environmental protection.

The construction industry has become one of the industries that accounts for a relatively large share of China’s total carbon emissions. Aiming at the problems of monitoring difficulties, diversity of segmentation types, and uncertainty of carbon emission factors, this study calculates the carbon emissions and intensity of the construction industry in each province of China from 2010 to 2019, analyzes its spatial and temporal variability using the Moran index and the slope index, analyzes the driving factors by combining the Kaya equation and the LMDI method, and verifies the zero-error characteristics by using the IPAT model. The results show that from 2010 to 2019, carbon emissions from the construction industry in China’s provincial areas increased in general, with a distribution of “high in the east and low in the west”, and the carbon emission intensity declined in general, but some provinces in the north and the center are still higher. Economic development and the increase in housing construction area are the main reasons for the growth of carbon emissions, while the optimization of energy structure and the adjustment of population density reduce carbon emissions. Moreover, the IPAT model verifies the credibility of the results of the LMDI model. This study provides a reference for monitoring and assessing carbon emissions in China’s construction industry from the perspective of spatio-temporal characterization, helps regional energy conservation and emission reduction and dual-carbon strategy, and it analyzes the provincial carbon emission intensity to reveal the low-carbon development issues.

Traditional bathymetry LiDAR (light detection and ranging) onboard manned and/or unmanned airborne systems cannot operate in the context of narrow rivers in urban areas with high buildings and in mountainous areas with high peaks. Therefore, this study presents a prototype of a lightweight bathymetry LiDAR onboard an unmanned shipborne vehicle (called “GQ-Cor 19”). The GQ-Cor 19 system primarily includes an emitting optical module, a receiving optical module, control module, detection module, high-speed A/D sampling module, and data processing system. Considering that the “GQ-Cor 19” is extremely close to the water surface, various new technical challenges are encountered, such as significant laser scattering energy from the surface of the water, which saturates signals received by the photomultiplier tube detector. Therefore, this study presents various new technical solutions, including (1) an improved Bresenham algorithm, (2) a small and lightweight receiving optical system with a split-field method, and (3) a data acquisition module with a high-speech A/D collector. Following a series of different experimental verifications, the results demonstrate that the new generation of single-band LiDAR onboard an unmanned shipborne vehicle can swiftly measure the underwater depth, and the maximum measurement depth is more than 25 m. The measurement accuracy is better than 30 cm and the weight is less than 12 kg.

Communication tower (CT) and its accessory equipment (AE) such as radio frequency equipment (RFE) and antenna, are essential in providing highspeed and stable mobile network services. It is necessary to routinely monitor the security and stability of CT and AE for seamless communication. There is limited research on fine segmentation of communication base station objects. This paper proposes a method for accurately segmenting the point cloud of the CT and AE from Terrestrial Laser Scanning (TLS) data. At first, the CT point cloud is accurately segmented based on region growing and Random Sample Consensus (RANSAC). Then, the point cloud of pole-shaped apart is extended to a certain distance to obtain the buffer point cloud containing AE. Normal Differential (ND) clustering is employed to obtain several groups of clusters containing planes, and calculate each plane's filling rate and size. Finally, the cluster type (such as antenna, RFE, or other) is distinguished. The experimental results demonstrate that the point-based average F1-score of CTs is 98. 70%, the point-based and object-based average F1-scores of antennas are 96. 09% and 97. 93%, and the corresponding values for the RFE are 89. 89% and 90. 00%, respectively, indicating the optimal performance of the proposed method.

As an essential parameter in forestry, crown base height (CBH) faces many tasks. The methods are still developing for estimating it. Unmanned aerial vehicles (UAV)light detection and ranging (LiDAR) supplies new, massive and high-density data for estimating CBH. Many methods hadbeen generated to compute CBH indirectly using regression-based ways or directly using geometric/statistical LiDAR-based ways. However, there werefew methods to deal with the problem of understory, trunkand noise points caused by high-density UAV data. A robust method was firstlyproposed in this studyto directly estimate CBH from LiDAR data, which containedtwo significant skills: 1)understory vegetation removal for each tree using a polynomial curve,and 2) computing CBH by kernel densification ofthe elevation frequency histogram of LiDAR data. It could tolerate the understoryand trunk points better through kernel convolution. The method proposed in this study and a previous simple model were applied in a crabapple plot in Huailai Remote Sensing Comprehensive Experimental Station, Hebei Province, China and verified byfield-measured data. Itwasinspiringthat our method is slightly better, 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, R2 was 0.73, RMSE was 8.29 cm and bias was 2.68%for these trees. Generally, this method showedstrong usability for high-density UAV LiDAR data and high precision for measuringCBH of low trees.

Central Yunnan, constantly in a state of water shortage, is a region of fragile ecological environment whose regional agricultural production and economic development are seriously restricted by drought, despite its relatively developed social economy and dense population. In order to study the drought situation in this area, Landsat TM\OLI remote sensing image data of the region in 2008, 2011, 2014, and 2017 was selected for the construction of the NDVI-LST characteristic space temperature vegetation drought index (TVDI) using the normalized difference vegetation index (NDVI) and the land surface temperature (LST). The validity of TVDI was verified by Standardized Precipitation Index (SPI), Palmer Drought Severity Index (PDSI) and Potential Evapotranspiration (PET), and the TVDI calculation results were divided into 5 drought levels: wet, normal, moderately dry, dry and severe dry. It is shown that: (1) The TVDI values of different land use types have obvious changes and good applicability; (2) TVDI can effectively reflect the characteristics of drought and spatial-temporal changes in central Yunnan, and the severe drought is mostly distributed in the west and southwest; (3) In the past 10 years, the dynamics of drought in this area have changed significantly, mainly from drought to slight drought, and the overall drought has decreased with time. It is concluded that TVDI model accurately reflected the drought situation in central Yunnan and thus is an instructive tool for regional drought monitoring.

Airborne laser scanning (ALS) point cloud has been widely used in various fields, for it can acquire three-dimensional data with a high accuracy on a large scale. However, due to the fact that ALS data are discretely, irregularly distributed and contain noise, it is still a challenge to accurately identify various typical surface objects from 3D point cloud. In recent years, many researchers proved better results in classifying 3D point cloud by using different deep learning methods. However, most of these methods require a large number of training samples and cannot be widely used in complex scenarios. In this paper, we propose an ALS point cloud classification method to integrate an improved fully convolutional network into transfer learning with multi-scale and multi-view deep features. First, the shallow features of the airborne laser scanning point cloud such as height, intensity and change of curvature are extracted to generate feature maps by multi-scale voxel and multi-view projection. Second, these feature maps are fed into the pre-trained DenseNet201 model to derive deep features, which are used as input for a fully convolutional neural network with convolutional and pooling layers. By using this network, the local and global features are integrated to classify the ALS point cloud. Finally, a graph-cuts algorithm considering context information is used to refine the classification results. We tested our method on the semantic 3D labeling dataset of the International Society for Photogrammetry and Remote Sensing (ISPRS). Experimental results show that overall accuracy and the average F1 score obtained by the proposed method is 89.84% and 83.62%, respectively, when only 16,000 points of the original data are used for training.

The global digital elevation measurement (DEM) products such as SRTM DEM and GDEM have been widely used for terrain slope retrieval in forests. However, the slope estimation accuracy is generally limited due to the DEMs’ low vertical accuracy over complex forest environments. The Ice, Cloud, and land Elevation Satellite-2 (ICESat-2) mission shows excellent potential for slope estimation because of the high elevation accuracy and unique design of beam pairs. This study aimed to explore the possibility of ICESat-2 data for terrain slope retrieval in the United States forests. First, raw ICESat-2 data were processed to obtain accurate ground surfaces. Second, two different methods based on beam pairs were proposed to derive terrain slopes from the ground surfaces. Third, the estimated slopes were validated by airborne LiDAR-derived slopes and compared with SRTM-derived slopes and GDEM-derived slopes. Finally, we further explored the influence of surface topography and ground elevation error on slope estimation from ICESat-2 data. The results show that the ground surface can be accurately extracted from all scenarios of ICESat-2 data, even weak beams in the daytime, which provides the basis for terrain slope retrieval from ICESat-2 beam pairs. The estimated slope has a strong correlation with airborne LiDAR-derived slopes regardless of slope estimation methods, which demonstrates that the ICESat-2 data are appropriate for terrain slope estimation in complex forest environments. Compared with the method based on along- and across-track analysis (method 1), the method based on plane fitting of beam pairs (method 2) has a high estimation accuracy of terrain slopes, which indicates that method 2 is more suitable for slope estimation because it takes full advantage of more ground surface information. Additionally, the results also indicate that ICESat-2 performs much better than SRTM DEMs and GDEMs in estimating terrain slopes. Both ground elevation error and surface topography have a significant impact on terrain slope retrieval from ICESat-2 data, and ground surface extraction should be improved to ensure the accuracy of terrain slope retrieval over extremely complex environments. This study demonstrates for the first time that ICESat-2 has a strong capability in terrain slope retrieval. Additionally, this paper also provides effective solutions to accurately estimate terrain slopes from ICESat-2 data. The ICESat-2 slopes have many potential applications, including the generation of global slope products, the improvement of terrain slopes derived from the existing global DEM products, and the correction of vegetation biophysical parameters retrieved from space-borne LiDAR waveform data.