Background In contrast to physical and quasi-empirical models, empirical models constructed for local fuel types provide significantly greater accuracy and applicability in predicting rate of spread (ROS).Aims To present an empirical modelling methodology for specific regions and fuel types.Methods We constructed experimental fuel beds with four levels of fuel loads, three levels of fuel moisture content (FMC) and eight slope gradients in a laboratory burning bed, resulting in 288 combustion experiments. We used data collected from these experiments to develop an empirical model specifically designed to predict the ROS of surface fires in Pinus koraiensis plantations.Key results Model evaluation revealed that direct application of Rothermel's model yielded substantial errors, with a pronounced tendency to overpredict ROS under most conditions. In contrast, the newly developed empirical model, which explicitly incorporates the effects of fuel bed structure, FMC and slope, demonstrated markedly improved accuracy while substantially reducing overprediction bias.Conclusion Overall, prediction accuracy increased with higher ROS. Using local fuel laboratory combustion data to construct the empirical model resulted in excellent predictive performance.Implications This study presents a robust methodology for developing predictive models of ROS for surface fires across various fuel types, providing valuable support for forest fire management and suppression efforts in northeastern China.

Dead Fine Fuel Moisture Content (DFFMC) is a critical factor influencing wildfire risk and fire spread behavior in forest fire management. DFFMC field-measurement relies on manual sampling, suffering from slow response, high labor costs, and limited spatial coverage. Moreover, existing predictive models of DFFMC are mostly based on single machine learning algorithms, which struggle to balance spatial generalization and local fitting capabilities, thereby limiting overall model performance. This study proposes a DFFMC prediction approach that integrates a stacking ensemble learning model with a hybrid dataset from different regions and Internet of Things (IoT) technology, offering the advantages of high accuracy, high spatial generalization, and rapid responsiveness. A stacking ensemble learning model was trained using publicly available international datasets covering diverse ecological and climatic zones. To evaluate the model's spatial generalization capability, field data collected from Bajia Country Park in Beijing, China, were used exclusively as an independent validation set. The model demonstrated strong predictive performance on the domestic dataset, achieving a correlation coefficient of 0.91 and a mean absolute error below 2. Key drivers analysis revealed that humidity and precipitation are the key drivers of DFFMC. Partial dependence plots indicate nonlinear DFFMC responses when humidity exceeds 60% and precipitation surpasses 3 mm. Bivariate dependence analysis further highlights complex interactions among meteorological factors, underscoring the value of multi-factor modeling for accurate DFFMC prediction and wildfire risk management.

Forest fires have become a significant global threat, with many negative impacts on human habitats and forest ecosystems. This study proposed a forest fire identification method by fusing visual and infrared images, addressing the high false alarm and missed alarm rates of forest fire monitoring using single spectral imagery. A dataset suitable for image fusion was created using UAV aerial photography. An improved image fusion network model, the FF-Net, incorporating an attention mechanism, was proposed. The YOLOv5 network was used for target detection, and the results showed that using fused images achieved a higher accuracy, with a false alarm rate of 0.49% and a missed alarm rate of 0.21%. As such, using fused images has greater significance for the early warning of forest fires.

Forest fires are one of the most devastating natural disasters, and technologies based on remote sensing satellite data for fire prevention and control have developed rapidly in recent years. Early forest fire smoke in remote sensing images, on the other hand, is thin and tiny in area, making it difficult to detect. Satellites with high spatial resolution sensors can collect high-resolution photographs of smoke, however the impact of the satellite’s repeat access time to the same area means that forest fire smoke cannot be detected in time. Because of their low spatial resolution, photos taken by satellites with shorter return durations cannot capture small regions of smoke. This paper presents an early smoke detection method for forest fires that combines a super-resolution reconstruction network and a smoke segmentation network to address these issues. First, a high-resolution remote sensing multispectral picture dataset of forest fire smoke was created, which included diverse years, seasons, areas, and land coverings. The rebuilt high-resolution images were then obtained using a super-resolution reconstruction network. To eliminate data redundancy and enhance recognition accuracy, it was determined experimentally that the M11 band (2225–2275 nm) is more sensitive to perform smoke segmentation in VIIRS images. Furthermore, it has been demonstrated experimentally that improving the accuracy of reconstructed images is more effective than improving perceptual quality for smoke recognition. The final results of the super-resolution image segmentation experiment conducted in this paper show that the smoke segmentation results have a similarity coefficient of 0.742 to the segmentation results obtained using high-resolution satellite images, indicating that our method can effectively segment smoke pixels in low-resolution remote sensing images and provide early warning of forest fires.

Forest fire is a ubiquitous disaster which has a long-term impact on the local climate as well as the ecological balance and fire products based on remote sensing satellite data have developed rapidly. However, the early forest fire smoke in remote sensing images is small in area and easily confused by clouds and fog, which makes it difficult to be identified. Too many redundant frequency bands and remote sensing index for remote sensing satellite data will have an interference on wildfire smoke detection, resulting in a decline in detection accuracy and detection efficiency for wildfire smoke. To solve these problems, this study analyzed the sensitivity of remote sensing satellite data and remote sensing index used for wildfire detection. First, a high-resolution remote sensing multispectral image dataset of forest fire smoke, containing different years, seasons, regions and land cover, was established. Then Smoke-Unet, a smoke segmentation network model based on an improved Unet combined with the attention mechanism and residual block, was proposed. Furthermore, in order to reduce data redundancy and improve the recognition accuracy of the algorithm, the conclusion was made by experiments that the RGB, SWIR2 and AOD bands are sensitive to smoke recognition in Landsat-8 images. The experimental results show that the smoke pixel accuracy rate using the proposed Smoke-Unet is 3.1% higher than that of Unet, which could effectively segment the smoke pixels in remote sensing images. This proposed method under the RGB, SWIR2 and AOD bands can help to segment smoke by using high-sensitivity band and remote sensing index and makes an early alarm of forest fire smoke.

Training a deep learning-based classification model for early wildfire smoke images requires a large amount of rich data. However, due to the episodic nature of fire events, it is difficult to obtain wildfire smoke image data, and most of the samples in public datasets suffer from a lack of diversity. To address these issues, a method using synthetic images to train a deep learning classification model for real wildfire smoke was proposed in this paper. Firstly, we constructed a synthetic dataset by simulating a large amount of morphologically rich smoke in 3D modeling software and rendering the virtual smoke against many virtual wildland background images with rich environmental diversity. Secondly, to better use the synthetic data to train a wildfire smoke image classifier, we applied both pixel-level domain adaptation and feature-level domain adaptation. The CycleGAN-based pixel-level domain adaptation method for image translation was employed. On top of this, the feature-level domain adaptation method incorporated ADDA with DeepCORAL was adopted to further reduce the domain shift between the synthetic and real data. The proposed method was evaluated and compared on a test set of real wildfire smoke and achieved an accuracy of 97.39%. The method is applicable to wildfire smoke classification tasks based on RGB single-frame images and would also contribute to training image classification models without sufficient data.

Forest fire smoke detection based on deep learning has been widely studied. Labeling the smoke image is a necessity when building datasets of target detection and semantic segmentation. The uncertainty in labeling the forest fire smoke pixels caused by the non-uniform diffusion of smoke particles will affect the recognition accuracy of the deep learning model. To overcome the labeling ambiguity, the weighted idea was proposed in this paper for the first time. First, the pixel-concentration relationship between the gray value and the concentration of forest fire smoke pixels in the image was established. Second, the loss function of the semantic segmentation method based on concentration weighting was built and improved; thus, the network could pay attention to the smoke pixels differently, an effort to better segment smoke by weighting the loss calculation of smoke pixels. Finally, based on the established forest fire smoke dataset, selection of the optimum weighted factors was made through experiments. mIoU based on the weighted method increased by 1.52% than the unweighted method. The weighted method cannot only be applied to the semantic segmentation and target detection of forest fire smoke, but also has a certain significance to other dispersive target recognition.

Fire is an important disturbance agent in the boreal forests of China. The aggressive fire suppression policy of China since 1988 has resulted in a large financial investment in support of fire brigade capabilities and the maintenance of fire management infrastructure. We developed a spatially explicit burn probability (BP) model to evaluate the effectiveness of improved fire management in Daxing’anling, China. The BP model can emulate the burn probability of the forest landscape by simulating daily wildfire occurrences, spread, and suppression for simulated years. Two scenarios were used for fire simulations in this study. The base scenario used the infrastructure data and parameters of fire suppression capability from the 1968–1987 period, and the intensive scenario used the data and parameters from the 1988–2012 period. The simulated annual burned areas for 1968–2012 showed a fluctuating trend similar to the historical fire records. Compared with the base scenario, the burn probability decreased by 73.6% under the intensive scenario, which suggests that improved fire management could significantly reduce the burn probability. This study shows that the BP model can model the effects of fire management activities on the forest landscape level and evaluate the effectiveness of fire management strategies or management measures.

Projecting the burn probability (BP) under future climate scenarios would provide a scientific basis for the implementation of forest fire adaptation technology. This study compared the changes in the climate, fire weather, and burn probability during the fire season in Daxing’anling, China. A burn probability model was established and used to simulate the daily fire occurrence and spread at baseline (1971–2000) and into the 2030s (2021–2050) based on the outputs from five global climate models (GCMs) (GFDL-ESM2M, Had GEM2-ES, IPSL-CM5A-LR, MIROC-ESM-CHEM, and Nor ESM1-M) under four climate scenarios (RCP2.6, RCP4.5, RCP6.0, and RCP8.5). The results showed that the average daily maximum temperature in the fire season will be increased by 2.1 °C (+16.6%) in the 2030s compared with the baseline and precipitation in the fire season will be increased by 7.1%. The average fire weather index (FWI) of the fire season in the 2030s will be increased by 4.2%, but this change is not significant. There will be 39 fires per year in the 2030s, representing an increase of 11.4%. The accuracy of simulated burned areas was 71.2% for the 1991–2010 period. The simulated and observed burned areas showed similar interannual fluctuations during period 1971–2010. The potential burned areas in the 2030s will increase by 18.8% over those in the baseline period and the BP will increase by 19.4%. The implementation of proactive fire management in areas with high predicted BP values will be key for an effective mitigation of future wildfire impacts.

Forest fire growth models (FGMs) are widely used in both research and operations. FGMs involve modelling complex physical-chemical dynamic processes over large spatially heterogeneous forest landscapes and long periods under changing weather conditions. Because of their complexity, it is difficult to validate these models. A typical approach is to graphically compare predicted boundaries to the corresponding boundaries of actual fires, which provides is a visual rather than quantitative evaluation of modelling errors of forest fire spread. In this paper, we propose a method to quantify two-dimensional spread process modelling errors, in this case for forest fire spread modelling. We introduce several indices that can be used to quantify spatio-temporal modelling errors of two-dimensional spread processes explicitly and to evaluate overall modelling errors. We demonstrate the effectiveness of the indices through a case study in which the modelling errors of a forest fire simulated by a FGM are compared with those of a reference fire. The case study illustrates that the spatio-temporally explicit indices do work to quantify modelling errors of forest fire spread compared to a reference model and that this error analysis is not only useful for validating FGMs but also provides a basis for improving them. Because of the similarity of other two-dimensional spread processes to forest fire spread, we suggest potential applications of the method in other spatial spread processes, such as the spread of forest insect and contagious disease. The limitations of the method are presented.

Forest management planners usually treat potential fire loss estimates as exogenous parameters in their timber production planning processes. When they do so, they do not account for the fact that forest access road construction, timber harvesting, and silvicultural activities can alter a landscape’s vegetation or fuel composition, and they ignore the possibility that such activities may influence future fire losses. We develop an integrated fire and forest management planning methodology that accounts for and exploits such interactions. Our methodology is based on fire occurrence, suppression, and spread models, a fire protection value model that identifies crucial stands, the harvesting of which can have a significant influence on the spread of fires across the landscape, and a spatially explicit timber harvest scheduling model. We illustrate its use by applying it to a forest management unit in the boreal forest region of the province of Alberta in western Canada. We found that for our study area, integrated fire – forest management planning based on our methodology could result in an 8.1% increase in net present value when compared with traditional planning in which fire loss is treated as an exogenous factor.

Emulating natural forest disturbance is an increasingly popular forest management paradigm that is considered a means of achieving forest sustainability. Adopting this goal requires a sound understanding of natural disturbances at scales that correspond to management policies and strategies. In boreal forest landscapes driven by periodic stand-replacing fires this requires knowledge of fire regime characteristics, especially their spatial and temporal variability as well as stochasticity. The major goal of this study was to demonstrate the utility of fire regime simulation modeling to explore the variability of fire regime characteristics, with respect to formulating and assessing forest management strategies. We conducted a modeling experiment in a boreal forest landscape of northwestern Ontario, Canada, to examine its long-term fire regime in relation to forest policies on harvest size distribution. We used BFOLDS, a spatially explicit fire regime model that simulates individual fire events mechanistically in response to fire weather, fuel patterns, and terrain. The fire regimes in four large eco-regions were modeled for a 200-year period under three fire-weather (cold, normal, and warm) scenarios, with replications. We found that fire size distribution in all eco-regions followed power law under all weather scenarios, but their slopes and intercepts varied among eco-regions and fire weather scenarios. Warming fire weather increased burn rates and fire numbers in all eco-regions, albeit to different degrees. Overall, the variability among eco-regions was higher than the variability among fire weather scenarios, and among replicates. Comparisons of simulated fire size classes with those from an 86-year long fire history showed that empirical data cannot capture the variability that could be revealed by simulation modeling. We also show that fire size distribution is spatially heterogeneous within eco-regions, and provide several suggestions for forest policy directions with respect to forest harvest size distributions and harvest rates, based on the variability of fire regime characteristics. An assessment of present forest policies of emulating natural disturbances that guide forest harvest sizes showed that these are incongruent with simulated fire size distributions under all scenarios with one exception. Overall, this study illustrates the value of scenario simulation modeling to explore and quantify the variability of forest fire regime, for use in forest policies and strategies that attempt to emulate natural disturbance.

The harvesting of forest stands can reduce landscape flammability by fragmenting fuel continuity in ways that make it difficult for fires to spread and by providing firefighters with fuel discontinuities they can use as anchor points for suppression operations. We describe a methodology for assessing the impact of harvesting designated forest stands on landscape flammability and expected losses. We combine assessments of the probability that fires will be ignited at any point on the landscape with probabilistic predictions concerning how long escaped fires will burn and how they will spread. Shortest path methods are used to identify critical paths that link potential ignition points with values at risk. We then rank stands with respect to their ability to disrupt those critical paths and thereby reduce landscape flammability and fire losses. We describe how we applied our methodology to a 12 964-ha forested area of boreal forest in the province of Alberta, Canada. Our results indicate that the crucial stands in our study area, those that have the most significant impact on landscape flammability and fire loss, tend to be those that are flammable and located on or close to critical paths that link areas where fires are most likely to occur with values at risk.

Wildfire poses threats to public safety and property in many Wildland-Urban Interface (WUI) areas. We describe a spatial Decision Support System (DSS) that can be used to help evaluate fuel treatment measures for WUI areas in terms of their impact on burn probabilities across the landscape. Our DSS is embedded in a computer-based geographic information system (GIS) platform that describes the structure of the landscape, its fuel mosaic, where structures and other values at risk are located and where fires might ignite and how they might spread across the landscape. It has a graphic interface that includes a digital map of the landscape on which fire managers and planners can delineate areas where proposed fuel treatment activities might take place. Mathematical models of fire ignition, suppression and spread processes are used to predict the potential impact of such measures on the landscape and display those impacts on a burn probability map which shows the predicted probability that any point on the landscape will burn given the current landscape and the proposed fuel treatment measures that are to be evaluated.