Coal plays an essential role in the global energy system. The rapid determination of coal quality indices, including ash content (Ash), volatile matter (VM), moisture content (MC), fixed carbon (FC), and calorific value (CV), is significant for its clean and efficient utilization. This study investigates the feasibility and accuracy of using hyperspectral imagery combined with chemometric analysis to predict coal quality indices. The spectra of coal samples were characterized using hyperspectral images within the 400–2500 nm wavelength. Characteristic wavelengths relevant to coal quality were effectively extracted through spectral preprocessing and feature selection techniques. Subsequently, predictive models were developed by applying machine learning algorithms to these selected wavelengths. Evaluation using multiple performance metrics demonstrated that the optimal models achieved determination coefficients (Rp2) of 0.96, 0.97, 0.95, 0.93, and 0.94 for Ash, VM, MC, FC, and CV, respectively. The best-performing models were applied to visualize the spatial distribution of these quality indicators across the hyperspectral images. These results highlight the potential of hyperspectral imaging for accurate prediction and spatial mapping of coal quality parameters, providing a valuable technical reference for clean and efficient coal utilization. © 2025 Elsevier Ltd

Desertification remains a critical issue, particularly in desert fringe areas that are ecologically fragile and prone to desertification. This paper innovatively develops a more accurate desertification severity grading model for these areas and analyzes the desertification situation across different land cover types to improve existing control measures. An improved grading model based on feature space was constructed for Liaoning Province, and a random forest model was used to classify desertified land cover types. Combining these results with desertification severity provided fine-scale information. The results of the dynamic monitoring of desertification severity based on the projection distance model in MSAVI-DI (Modified Soil Adjusted Vegetation Index-Drought Index) feature space showed that the total area of desertification in the 2019 is 4930.28 km2. Over the past decade, the total desertified area has decreased by 373.71 km2. The fine-scale information analysis indicates that desertification management and ecological restoration should prioritize areas with more fragmented landscapes. Poorly managed and abandoned croplands are prone to further desertification, underscoring the importance of targeted management efforts. Overall, the study constructed a grading model of desertification severity for desert margins. At the same time, fine-scale information on desertification containing the severity of desertification and land cover type was obtained. This provides a more accurate basis for future desertification restoration.

Accurate mapping of large-scale forest aboveground carbon (AGC) stock is essential for understanding the role of forests in the global carbon cycle. Traditional forest resource inventory methods pose limitations due to their sparse spatial coverage and time-consuming data collection. While satellite sensors offer extensive spatial and temporal coverage, their low spatial resolution often introduces significant errors when ground measurements are directly matched with satellite image pixels. This study proposes a large-scale research framework for precise forest AGC stock mapping, which can effectively address these challenges by integrating unmanned aerial vehicle (UAV) imagery with spaceborne LiDAR data. Using Liaoning Province as the study area and based on field sample collection, UAV high-resolution images were used to generate expanded samples at the individual tree scale, and the forest AGC stock results at the spot scale were obtained through accumulation. Considering the number of expansion samples and the bias of the spaceborne LiDAR position caused by the terrain, we expanded the samples via GANs, and tested the effects of different spot radii on the model's fitting accuracy via a random forest regression model. Finally, selecting 12.5 m as the optimal fitting radius of the model, we obtained the forest AGC stock results at the spot scale of the GEDI and ICESat-2 in the province. Owing to the difficulty of one-to-one matching between spaceborne LiDAR spots and satellite image pixels, geographical correlation was performed to extract the average pixel value of multiple pixels covered by the light spot based on an area-weighting method as the model's input features. Combining this step with an ensemble machine learning algorithm, the final estimate of forest AGC stock in Liaoning Province was calculated to be 101.35 Tg, with an uncertainty of +/- 37.31 Tg. Our approach outperformed publicly available products, namely, AGBCY2021, ESA_CCI, and GEDI_L4B_V2, achieving an RMSE (%) of 19.86 %, and demonstrating the efficacy of the proposed method for quantifying uncertainty propagation in a multiscale analysis framework.

【目的】针对北方森林协同无人机激光雷达和无人机RGB影像进行单木地上生物量估测，以彰武县樟子松和杨树为研究对象，探究使用组合数据和单一数据对针、阔叶林单木地上生物量估测的影响，为彰武县防风固沙人工林单木地上生物量的精准预测提供技术参考。【方法】从LiDAR点云和基于RGB光学影像获取的数字正射影像图（DOM）中提取单木尺度高度、强度、密度、冠层结构、光谱、纹理和植被指数多特征，采用置换重要性（PI）和Boruta优选特征子集，结合地面实测单木地上生物量数据，使用随机森林（RF）、极端梯度提升树（XGBoost）和分类提升算法（CatBoost）3种典型机器学习方法构建樟子松和杨树地上生物量估测模型，对仅用LiDAR数据，仅用DOM数据以及联合二者的建模结果进行比较。【结果】1)点云高度和冠层结构是估测两个树种单木地上生物量的关键特征；而纹理特征仅对樟子松地上生物量估测产生积极影响；2)对于樟子松，基于组合数据的单木地上生物量估测精度最高，优于单一LiDAR和单一RGB影像。3种数据集的最优模型分别为ALL-PI-XGBoost、LiDAR-PI-XGBoost和DOM-PI-RF，测试...

The chronological classification of Jingdezhen blue and white porcelain in past dynasties has high academic and socioeconomic research value. In this study, RGB, non-imaging spectral, and imaging spectral datasets of Jingdezhen blue and white porcelain were constructed with small samples, and the spectral range was between 350-950 nm. The transfer learning classification and identification model of blue and white porcelain was established based on the VGG16 and ResNet50 models. On this basis, Dempster-Shafer (DS) evidence theory was used to construct a chronological classification and identification model of blue and white porcelain imaging spectral data that combined spatial and spectral information. The experimental comparison shows that in the case of small samples, the method incorporating the ResNet50 network model and stepwise discriminant analysis combined with the Long Short-Term Memory (LSTM) algorithm had the best classification effect, and the classification accuracy reached 90.47%. This study provides a new method and idea for the nondestructive identification and classification of cultural relics.

The reflective characteristics of remote sensing image information depend on the scale of the observed area, with high-resolution images providing more detailed feature information. Currently, monitoring refined industries and extracting regional information necessitate higher-resolution remote sensing images. Super-resolution reconstruction of remote sensing multispectral images not only enhances the spatial resolution of these images but also preserves and improves the spectral information of multispectral data, thereby providing richer ground object information and more accurate environmental monitoring data. To improve the effectiveness of feature extraction in the generator network while maintaining model efficiency, this article proposes the vision transformer improved super-resolution generative adversarial network (ViT-ISRGAN) model. This model is an improvement upon the original SRGAN super-resolution image reconstruction method, incorporating lightweight network modules, channel attention modules, spatial-spectral residual attention, and the vision transformer structure. The ViT-ISRGAN model focuses on reconstructing four types of typical ground objects based on Sentinel-2 images: urban, water, farmland, and forest. Results indicate that the ViT-ISRGAN model excels in capturing texture details and color restoration, effectively extracting spectral and texture information from multispectral remote sensing images across various scenes. Compared to other super-resolution (SR) models, this approach demonstrates superior effectiveness and performance in the SR tasks of remote sensing multispectral images.

矿区开采造成的土壤重金属污染严重影响作物产量、引发人体疾病；有效预防土壤重金属污染对健康的损害非常重要。高光谱快速、动态获取地物连续光谱信号的特点，为发展基于遥感的土壤重金属含量监测提供了新的思路。针对河北省涞源县典型铅锌矿区，实地采集矿区及周边土壤样本，基于SVC HR-1024i地物光谱仪(350～2 500 nm)获取土壤光谱反射率，通过对光谱数据进行平滑、一阶导数、多元散射校正、标准正态变换、多元散射校正后一阶导数、标准正态变换后一阶导数六种光谱数据组合变换，使用差值指数、比值和归一化方法从六种预处理数据中提取光谱指数，通过实验室化学测试分析得到土壤重金属镉、铅、锌含量，对不同重金属元素使用不同光谱变换方式进行预处理，得到不同类型重金属元素的最优光谱变换方式。采用差值指数、比值指数和归一化植被指数，提取不同光谱指数下的最优波段组合，从而得到用于不同重金属元素建模使用的最优自变量。基于随机森林和偏最小二乘回归法分别构建重金属元素反演模型。研究表明，通过对光谱数据预处理，可以有效地降低噪声，增强光谱特征。从结果来看，经过预处理后光谱数据与重金属含量相关性有所提高。对不同重金属元素建模选择对其最优的光谱指数自变量，增加了反演建模的有效特征。对三种重金属镉、铅、锌利用随机森林算法和偏最小二乘回归法建立预测模型，最优模型的R2分别达到了0.90、 0.91、 0.84,证实了该方法的有效性。该研究可为铅锌矿区土壤重金属含量反演建模提供依据，为矿区土壤重金属含量检测提供方法参考。

为满足煤炭矿区植被叶绿素含量高精度动态监测需求,该文以陕北大柳塔矿区为研究区,首先分析PROSAIL模型对矿区典型植被欧李、野樱桃的适用性,然后根据PROSAIL辐射传输模型建立查找表,结合基于正则化的代价函数对欧李、野樱桃叶绿素含量进行反演,并利用SNAP软件反演结果与地面实测数据对PROSAIL模型反演结果进行验证,最后利用所构建模型反演得到2016—2019年大柳塔矿区植被叶绿素含量空间分布。结果表明:PROSAIL模型模拟光谱与地面实测光谱的绝对偏差平均值最大为0.016,该精度满足植被参数反演;PROSAIL模型反演得到的欧李、野樱桃叶绿素含量与地面实测数据的决定系数、均方根误差和相对均方根误差分别为0.679、1.926和4.625%,优于SNAP软件反演结果,反演得到的大柳塔矿区叶绿素含量时空变化与实际植被生态修复情况和土地利用覆盖类型一致。研究结果可为矿区植被叶绿素反演和生态修复效果评估提供技术参考。

Mapping the aboveground biomass (AGB) of forests in a spatially continuous manner is essential to comprehend the carbon sequestration capacity of forest ecosystems. Automatic extraction of tree crown and height is an effective approach to estimating biomass at the tree level. In this study, we employed high-resolution unmanned aerial vehicle (UAV) images and a mask region-based convolutional neural network (Mask R-CNN) to automatically identify the tree crown and height of Pinus sylvestris, we added tree height information to the model training by setting special tree height label samples and determined the optimal model by comparing the tree crown and height detection accuracy of four different combinations (RGB-CHM-DSM, RGB-CHM, RGB-DSM, RGB) of UAV images. The findings demonstrate that the model with RGB-CHM-DSM combination has the highest crown extraction accuracy (Precision = 0.896, Recall = 0.916, F1 Score = 0.906, IoU = 0.822), and the tree height extraction results with RGB-CHM combination has the highest correlation with the tree height of UAV images (R2 = 0.93, RMSE = 0.25, rRMSE = 3.10). Higher extraction accuracy is achieved due to the inclusion of CHM features in the model, and the results can be used to estimate forest AGB at different scales, improve the accuracy and efficiency of forest carbon sink research, minimize the workload of field investigations, and reduce the cost of manual methods.

Chronological classification studies of successive ages of Jingdezhen blue and white porcelain have high research value both academically and socioeconomically. Compared with other chemical analysis meth-ods, hyperspectral remote sensing techniques have advantages such as non-contact and non-destructive nature. In this paper, Jingdezhen blue and white porcelain of various ages is taken as the research ob-ject, linear discriminant analysis is used to build a model of typical extracted spectral features that are sensitive to blue and white porcelain material type and age information, stepwise discriminant analy-sis and competitive adaptive reweighted sampling are used for feature selection, and continuous wavelet transform and spectral feature parametric-based methods are used for feature extraction. Then, a random forest algorithm and long short-term memory (LSTM) are combined to categorize Jingdezhen blue and white porcelain of various ages. The stepwise discriminant analysis paired with LSTM is the most accu-rate combination amongst all classification techniques with the same amount of control preferred char-acteristics. This research shows that the categorization of Jingdezhen porcelain from various ages may be accomplished using hyperspectral remote sensing technique in conjunction with the random forest algorithm and long short-term memory.& COPY; 2023 Consiglio Nazionale delle Ricerche (CNR). Published by Elsevier Masson SAS. All rights reserved.