Haihui Wang

Smoke detection by remote sensing plays a vital role in forest fire monitoring due to its wide coverage and fast update frequency. At present, two main challenges remain: 1) the spatio-temporal variability of smoke's shape and color, and 2) the drastic scale differences of smoke instances across images, both of which hinder the generalization of existing approaches. Traditional CNNs are effective in capturing local details but struggle with large-scale context modeling. Transformers offer global self-attention and long-range dependency modeling but often miss fine-grained information, especially for small targets. To address these issues, we propose SmokeNeXt, a dual-stream, multiscale adaptive model tailored for smoke classification. By fusing the Conv-branch's local detail extraction with the Trans-branch's global context modeling, a semantic merge module with a self-designed multi-axis convolutional cross-attention enables cross-stream feature complementation. In addition, A gated redundancy suppression module is introduced to dynamically filter out irrelevant background noise, enhancing robustness to appearance variability. Experiments on the FASDD_RS and USTC_SmokeRS datasets demonstrate that SmokeNeXt effectively mitigates both challenges. On FASDD_RS, it achieves 97.8% accuracy and 99.1% AR@$\bf {\phi _{0.04\%\rm{--}0.8\%}}$, with high recall even under extreme small-scale smoke conditions, and consistently focusing on smoke regions across diverse appearances and backgrounds. On USTC_SmokeRS, SmokeNeXt achieves an accuracy of 96.2%. Compared with ResNet-50, ConvNeXt-B, and MobileNetV3, SmokeNeXt improves AR by 8.0%, 1.5%, and 6.6%, respectively, indicating strong generalization under extreme scale variation and morphological diversity.

This study investigated mass flux patterns of layered moist leaf samples to illustrate the wide measurable response across a representative set of vegetation types. A series of experiments with 24 plant species was conducted using a cone calorimeter. The samples' moisture content (FMC) varied between 3.2% and 78.6%, and the radiant heat flux was set at 35, 55, 70, and 85 kW m ⁻² . The initial mass loss of the samples exhibited quasilinear trend, allowing for the extraction of pre‐ignition mass flux , which represents the combined release of moisture and pyrolysis products. Under the same radiation heat flux, was essentially identical for the samples with the same properties but varying mass loading, consistent with the observed ignition modes and times. Measurements indicated that fluctuates widely in a vast range between 3.7 and 21.5 g m ⁻² s ⁻¹ for the samples undergoing smoldering and in the presence of a flame; no systematic linkage between ignition mode and the measured was observed. As rises, ignition time shortens and peak HRR increases, mirroring the field‐observed cessation of fire spread when FMC exceeds ~40%. Live leaves exhibited an additional mass loss flux that lengthened ignition time by one order of magnitude, especially for high FMCs. While data for FMC > 40% provide a quantitative moisture target for green barrier irrigation and prescribed burning buffer zones, the identified variation patterns among FMC, and may supply input bounds for future fire spread model parameterization.

Accurately reproducing the measured scattering matrix of black carbon (BC) through numerical simulations remains a challenge. Researchers have developed various morphological models of BC and computed their scattering matrices in attempts to replicate experimental measurements. However, prior simulation endeavors frequently encountered issues such as significant discrepancies with observational data, implausible particle shapes, or unsuitable computational parameters. In this study, we developed a fractal‐based overlapping and necking model to represent the morphology of bare BC particles. We computed the scattering matrices for both individual particles and particle population using these models and compared the results with the previously reported measurements. Our findings revealed that the overlapping model reproduces the measured scattering matrix elements more accurately, whereas the necking model fails to achieve similar consistency. At a wavelength of 532 nm, the overlapping model yields F ₂₂ (π)/ F ₁₁ (π) ranging from 0.80 to 0.99 for single particles and from 0.86 to 0.99 for particle population, both of which are much closer to the experimental observations than those of the necking model. In contrast, only a small subset of results from the necking model falls within the measured range. The overlapping model outperforms the necking model in reproducing the scattering matrix and should be preferred for representing bare BC particles. The established understanding provides useful guidance for retrieving microphysical parameters of BC from polarization features and for diminishing the uncertainties associated with its radiative forcing estimates.

Forest and grassland fire behavior prediction is increasingly critical under climate change, as rising fire frequency and intensity threaten ecosystems and human societies worldwide. This paper reviews the status and future development trends of wildfire behavior modeling and prediction technologies. It provides a comprehensive overview of the evolution of models from empirical to physical and then to data-driven approaches, emphasizing the integration of multidisciplinary techniques such as machine learning and deep learning. While conventional physical models offer mechanistic insights, recent advancements in data-driven models have enabled the analysis of big data to uncover intricate nonlinear relationships. We underscore the necessity of integrating multiple models via complementary, weighted fusion and hybrid methods to bolster robustness across diverse situations. Ultimately, we advocate for the creation of intelligent forecast systems that leverage data from space, air and ground sources to provide multifaceted fire behavior predictions in regions and globally. Such systems would more effectively transform fire management from a reactive approach to a proactive strategy, thereby safeguarding global forest carbon sinks and promoting sustainable development in the years to come. By offering forward-looking insights and highlighting the importance of multidisciplinary approaches, this review serves as a valuable resource for researchers, practitioners, and policymakers, supporting informed decision-making and fostering interdisciplinary collaboration.

In this study, an ecological framework was developed to sort out the existing forest fire prevention techniques. The subsequent analysis involved comparing the ecological values and application prospects of these techniques developed in different time periods. As ecological applications, fire regimes reflect vegetation response to wildfires, providing valuable insights for shaping the fire risk and behaviors in forests through fuel treatment and vegetation modification. Fuel treatment and the construction of green fire barriers are both rooted in existing ecosystems and possess ecological characteristics. While fuel thinning focuses on reducing the potential fire intensity and severity, green fire barriers have been more targeted for fire prevention purposes. Among these techniques, green fire barriers demonstrate unique sustainability and have the potential to generate long-term ecological and environmental benefits. Through the comprehensive utilization of several fuel management formulas, we can effectively combine the fire prevention demands with ecological maintenance and environment protection. This integrated approach promotes the development of fire-resilient ecosystems and desirable living environments in a more realistic and sustainable manner.

Prediction of future fire probability and severity lays the foundation for formulating ecological fire prevention and global environmental protection policies. The present work focuses on overview of the fire regimes and their constitution basis to promote the understanding on the fire regime parameters and the related mechanism operated in the ecosystem. Through the application of remote sensing technology and fire models in fire regime research and predictions, research progress and achievements were made in systematically quantifying the fire characteristic parameters, with the emphasis on the influence of forest stand characteristics on fire regimes under climate change. It has been well understood that the fire regimes have obvious spatiotemporal characteristics from regional to global scale, and the fire regimes are closely related to forest stands and climate. The future fire regimes are likely to be dominated by climate changes. For this reason, the future research on fire regimes needs to further study the spatiotemporal fire process and quantify the parameters of fire regimes through constantly exploring the mechanism of the fire regimes driven by climate changes and the control measures; meanwhile, the research on fire regimes will continue to serve as a basis for forest fire management.

Forest fire ecology is a discipline to study the ecological consequences of forest fires and the interrelationships among them. With the intensification of climate change and the deterioration of forest fire status, forest fire ecology has become a very active branch of ecology. In this paper we reviewed the development history of forest fire ecology and expounded the extension and connotation of forest fire ecology, fire factors and regimes as well as their interaction with the ecological systems. Previous ecologists have conducted a lot of research on the impact of forest fire on the ecological environment and the adaptability of ecosystems to forest fire. The development of forest fire ecology can be divided into three periods: in the initial period the destruction of forest fire was recognized with the study objective being focused on forest fire control; in the expansion period, efforts were made to the application of forest fire in the fire management and the third period was symbolized as the founding of the entire discipline. With the application of the advanced technology and mathematical models, forest fire ecology provides successful examples in carrying out long-term prediction on the forest community dynamics after fire intrusion. Planned burning have been developed as effective means of forest fire management, whereas ecological fire mitigation is considered as a new concept in forest fire management.