Zhijun Zhen

Leaf radiative transfer models provide a physical basis for leaf trait retrieval, but compensating parameter effects and measurement discrepancies constrain estimation accuracy and uncertainty characterization. This study proposes Training-Informed Differentiable Bayesian Inference (TI-DBI). This method integrates training-derived joint lognormal priors, wavelength-specific spectral biases, and residual scales into a differentiable PROSPECT-D model, enabling joint parameter inference via the No-U-Turn Sampler (NUTS). Evaluated on multi-source datasets, TI-DBI yields relative root mean square errors (rRMSEs) of 27.31%, 33.72%, 16.22%, and 37.66% for chlorophyll (C_ab), carotenoids (C_ar), equivalent water thickness (C_w), and leaf dry mass per area (C_m). Outperforming an uncalibrated Markov chain Monte Carlo (MCMC) baseline for all traits and partial least squares regression (PLSR) for C_ab, C_ar, and C_w, these results confirm the specific value of the data-informed probabilistic framework. Beyond point-estimation, TI-DBI substantially improves uncertainty characterization. It yields narrower interval widths and lower interval scores than the uncalibrated baseline, with prediction interval coverage probability (PICP) of the 95% highest density intervals (HDIs) reaching 93.14%, 96.46%, 89.62%, and 86.22% for Cab, Car, Cw, and Cm, respectively. Ablation and sensitivity studies confirm that both the learned priors and spectral error calibration drive these performance gains across varying training fractions and spectral configurations. TI-DBI strictly preserves the forward model’s physical structure while providing a robust, practical framework for jointly evaluating trait estimation accuracy and posterior uncertainty.

1. Soil moisture (SM) is a critical boundary condition for land–atmosphere interactions, directly influencing plant growth, development and surface runoff, and it is closely linked to global food a...

Subsurface combustion in coal mines poses a significant threat to ecosystem integrity, geological stability, and public safety. Effective risk mitigation requires continuous monitoring and accurate detection of combustion dynamics. In this study, an improved subsurface combustion index (SCI) was developed based on multisource remote sensing indicators, and long-term time series observations (2010–2025) were used to characterize its spatiotemporal evolution. Results show that dREGI achieved the highest anomaly discrimination among all evaluated vegetation indices, with an M-statistic of 1.4186, outperforming NDVI (1.1073) and EVI (0.8226). Adaptive principal component analysis identified dREGI and H as the dominant contributors to SCI construction. Separability analysis further demonstrated that integrating dREGI with LST and H improved the performance of the composite SCI by 16.3%, increasing its M-statistic from 0.959 to 1.115 relative to the dREGI-only baseline. Temporally, subsurface combustion exhibits a multi-stage evolution, with initial anomalies emerging around 2013, followed by a transitional phase during 2014–2018. Activity intensifies during 2019–2023, peaks in 2023, and declines in 2024, indicating residual combustion. Spatially, high-risk areas are concentrated in the eastern region, while moderate and low-risk zones occur in the central and western regions, respectively. These results demonstrate that the proposed indices provide a more robust and sensitive framework for early warning and spatial delineation of subsurface combustion zones.

Simulation de la réflectance de la végétation et inversion des propriétés bio-optiques basées sur un modèle de transfert radiatif tridimensionnel: Les données de télédétection sont de plus en plus utilisées, en particulier via les méthodes d'inversion, pour la surveillance de la végétation en raison de l'amélioration de la précision des mesures et de la résolution spatiale/spectrale/temporelle, ainsi que des progrès des méthodes d'interprétation des données de télédétection. Les modèles de transfert radiatif unidimensionnels simulent souvent avec une mauvaise précision la réflectance de la végétation, si bien que leur utilisation dans des procédures d'inversion des mesures de télédétection en termes de paramètres bio-optiques conduit à des paramètres estimés qui peuvent être très imprécis. Les modèles de transfert radiatif tridimensionnels sont beaucoup plus précis, car ils peuvent prendre en compte de manière réaliste l'architecture des couverts végétaux (FC), si bien qu'ils peuvent modéliser avec précision les mécanismes radiatifs complexes qui surviennent au sein des FC. Cependant, l'emploi des modèles 3D, et en particulier leur inversion, se heurte à trois problèmes majeurs. (1) Les modèles de simulation actuels ne permettent pas de simuler de manière continue l'évolution de la végétation du fait de la méconnaissance de la variation spatiale et temporelle des paramètres clés du sol. Cette difficulté est une contrainte importante pour la mise en œuvre de l'inversion de séries temporelles d'images de télédétection. (2) Les méthodes d'inversion spectrale foliaire existantes sont principalement applicables aux couverts homogènes tels que chaque unité spatiale d'inversion comprend plusieurs éléments de paysages (e.g., arbres). Cependant, l'observation de scènes qui comportent de nombreuses composantes dont la dimension est du même ordre de grandeur ou inférieure à la résolution spatiale du capteur, conduit à l'apparition de nombreux pixels mixtes. C'est en particulier le cas des villes, où la précision de l'inversion des propriétés optiques foliaires des arbres tend à être imprécise, voire très imprécise. (3) Les indices de végétation existants sont facilement saturés dans les zones à forte couverture végétale, ce qui limite fortement leur capacité d'inversion. Pour résoudre les problèmes mentionnés ci-dessus concernant l'inversion des images de télédétection, nous avons couplé le modèle DART (Discrete Anisotropic Radiative Transfer) de transfert radiatif 3D avec un modèle de croissance. Cette approche permet de simuler avec une meilleure précision l'évolution de la réflectance des couverts en fonction de leur développement. L'inversion de la végétation en milieu urbain est particulièrement complexe du fait de sa distribution éparse et de l'impact de l'architecture urbaine. Nous avons résolu ce problème via le développement de la méthode appelée "Etalonnage de DART". Nous avons aussi analysé les comportements des isolignes de végétation et proposé un décalage vers la droite du point d'intersection des isolignes calculé à partir de simulations DART pour atténuer l'effet de saturation de l'indice de végétation ajusté au sol (SAVI) dans les zones à fort FC.

Soybean is a key global crop for food and oil production, playing a vital role in ensuring food security and supplying plant-based proteins and oils. Accurate information on soybean distribution is essential for yield forecasting, agricultural management, and policymaking. In this study, we developed an Enhanced Chlorophyll Index (NRLI) to improve the separability between soybean and maize—two spectrally similar crops that often confound traditional vegetation indices. The proposed NRLI integrates red-edge, near-infrared, and green spectral information, effectively capturing variations in chlorophyll and canopy water content during key phenological stages, particularly from flowering to pod setting and maturity. Building upon this foundation, we further introduce a pixel-wise compositing strategy based on the peak phase of NRLI to enhance the temporal adaptability and spectral discriminability in crop classification. Unlike conventional approaches that rely on imagery from fixed dates, this strategy dynamically analyzes annual time-series data, enabling phenology-adaptive alignment at the pixel level. Comparative analysis reveals that NRLI consistently outperforms existing vegetation indices, such as the Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), and Greenness and Water Content Composite Index (GWCCI), across representative soybean-producing regions in multiple countries. It improves overall accuracy (OA) by approximately 10–20 percentage points, achieving accuracy rates exceeding 90% in large, contiguous cultivation areas. To further validate the robustness of the proposed index, benchmark comparisons were conducted against the Random Forest (RF) machine learning algorithm. The results demonstrated that the single-index NRLI approach achieved competitive performance, comparable to the multi-feature RF model, with accuracy differences generally within 1–2%. In some regions, NRLI even outperformed RF. This finding highlights NRLI as a computationally efficient alternative to complex machine learning models without compromising mapping precision. This study provides a robust, scalable, and transferable single-index approach for large-scale soybean mapping and monitoring using remote sensing.

The convergence of Artificial Intelligence (AI) with advanced remote sensing is revo-lutionizing the way we monitor and manage agricultural landscapes and vegetation. As we face unprecedented challenges from climate change and a growing global popula-tion, AI-driven analysis of data from satellite, aerial, and in situ sensors offers a trans-formative pathway to enhance food security and promote sustainable ecosystem man-agement. These intelligent systems provide critical, real-time insights into crop health, stress, and productivity, enabling a paradigm shift towards more precise, efficient, and re-silient agricultural practices. This Special Issue aims to explore the latest advancements in AI and machine learn-ing and their wide range of applications in agriculture and vegetation studies. We invite submissions that showcase innovative methodologies, cutting-edge sensor applications, and novel modeling approaches that contribute to the quantitative assessment and moni-toring of agricultural and vegetative systems in a new era of data-driven discovery. Articles may address, but are not limited to, the following topics: • Novel AI/ML Algorithm Development: Creation and optimization of new deep learning, reinforcement learning, and other machine learning architec-tures for processing remote sensing data. • Large-Scale Vegetation Mapping and Classification: Application of AI for land cover classification, species identification, and mapping invasive spe-cies at regional, national, and global scales. • Biophysical and Biochemical Parameter Retrieval: Using machine learning to estimate key vegetation variables such as Leaf Area Index (LAI), chlorophyll content, biomass, and canopy water content. • Precision Agriculture and Smart Farming: AI-driven applications for crop monitoring, yield prediction, soil health assessment, and the detection of dis-ease, pests, and nutrient deficiencies. • Forestry and Ecosystem Management: Monitoring deforestation, forest deg-radation, and regrowth; estimating forest carbon stocks; and applications for sustainable forest management. • Urban Ecology and Green Infrastructure: Mapping and assessing the health of urban vegetation, quantifying ecosystem services, and monitoring the im-pact of urbanization on plant life. • Time-Series Analysis and Change Detection: Advanced methods for analyz-ing long-term satellite data records to detect phenological shifts, land use change, and abrupt disturbances like fires or droughts. • Multi-Sensor Data Fusion and Integration: Innovative techniques for combin-ing data from diverse sources (e.g., hyperspectral, LiDAR, SAR, thermal, UAV, and satellite imagery) to create enhanced data products. • Vegetation Stress and Health Monitoring: Early detection and monitoring of vegetation stress from climatic events, pollution, and other anthropogenic pressures. • Big Data Processing and Cloud Computing: Development and use of scalable processing workflows on platforms like Google Earth Engine for continent- or planet-scale vegetation analysis. • Explainable AI (XAI) and Model Interpretability: Research focused on making complex AI models more transparent, understandable, and trustworthy for scientific and policy applications. • Generative AI and Data Augmentation: Using generative models to simulate realistic remote sensing data for training more robust and accurate AI sys-tems, especially in data-scarce environments.

Effectively leveraging multi-source unmanned aerial vehicle (UAV) observations for reliable object recognition is often compromised by environmental extremes (e.g., occlusion and low illumination) and the inherent physical discrepancies between modalities. To overcome these limitations, we propose DLANet, a lightweight, rotation-aware multimodal object detection framework that introduces a dual-level attention relearning strategy to maximize complementary information from visible (RGB) and thermal infrared (TIR) imagery. DLANet integrates two novel components: the Implicit Fine-Grained Fusion Module (IF²M), which facilitates deep cross-modal interaction by jointly modeling channel and spatial dependencies at intermediate stages, and the Adaptive Branch Feature Weighting (ABFW) module, which dynamically recalibrates modality contributions at higher levels to suppress noise and pseudo-targets. This synergistic approach allows the network to relearn feature importance based on real-time scene conditions. To support industrial applications, we construct the OilLeak dataset, a dedicated benchmark for onshore oil-spill detection. The experimental results demonstrate that DLANet achieves state-of-the-art performance, recording an mAP_0.5 of 0.858 on the public DroneVehicle dataset while maintaining high efficiency, with 39.04 M parameters and 72.69 GFLOPs, making it suitable for real-time edge deployment.

The accurate mapping of Leaf Area Index (LAI) and Canopy Chlorophyll Content (CCC) is crucial for studying physiological and ecological processes. Leaf Area Index (LAI) is essential for understanding processes such as photosynthesis, transpiration, and energy exchange between the land surface and the atmosphere. CCC serves as an indicator of plant health and productivity, and it is used to assess nitrogen content, which is vital for crop management. We conducted an inversion study for the PROSAIL model using the auto-differentiation (AD) method to retrieve LAI and CCC. The accuracy is assessed on two multispectral/hyperspectral simulated datasets and three measured datasets, covering various types of satellite/airborne image data and in-situ measurements. The RMSE of LAI and CCC is 1.0640 and 74.5518 ug/cm² for multi-spectral simulated datasets and 0.5537 and 39.5267 ug/cm² for hyperspectral simulated datasets under two observation directions. For Barrax agricultural crop datasets, the AD inversion method demonstrates a comparable LAI and an enhanced CCC inversion accuracy when juxtaposed with well-established Simplified Level 2 Product Prototype Processor (SL2P). The feasibility of the AD inversion method is demonstrated in multi-spectral, hyperspectral, and multi-sensor spectral application scenarios. This method has the potential to enhance the retrieval of vegetation biophysical parameters and improve the ecosystem monitoring in the future.

As a powerful tool for simulating bi-directional reflectance and radiative transfer in complex canopies, the Radiosity Graphics Model (RGM) suffers from a reduced runtime speed or even crashes when facing considerable computation load of the view factor for fine-grained simulation of heterogeneous canopy. In this work, the RGM Model at pixel scale (RGM _PS model) is proposed with the OpenACC acceleration techniques and two improved algorithms, which solve the overloaded view factor calculation and enhance scene availability without sacrificing accuracy. Two heterogeneous canopy scenario experiments were used for validation, including realistic single tree experiment and large-scale synthetic heterogeneous canopy experiment. The RGM _PS model has increased by nearly 70 speed times than the original RGM model, demonstrating its capability to model large-scale scenes spanning ten thousand square meters. The R ² between RGM and RGM _PS is over 0.94 and RMSE is below 0.0038. The cross-model validation between RGM _PS model and the Discrete Anisotropic Radiative Transfer (DART) model, achieved a high agreement with R ² as high as 0.98 in the near-infrared (NIR) band. An assessment conducted using airborne multi-angle measurements also demonstrated that the accuracy of the proposed solution was deemed satisfactory for BRF simulation, with RMSEs of 0.0031 and 0.0340 for the Red and NIR bands, respectively. Our research contributes to the development of more efficient and accurate bidirectional reflectance factor simulations for large heterogeneous canopy scenes using the RGM model.