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.

Forest fire size distribution (FSD) is one of the suite of indicators of forest fire regimes. It is applied in forest fire management, particularly for planning and evaluating suppression efforts. It is also used in forest management in the context of emulating natural fire disturbances. Given the recent growth in research and applied interest in this topic, we review and synthesise the state of knowledge on FSD, and identify sources of knowledge uncertainties and future research directions. Based on literature, it is common for forest fires to follow the power law probability distribution, particularly the truncated subtype, under a variety of forest types and forest and fire management practices. Other types of FSD are also observed, but under specific circumstances. Although there is evidence that observed FSDs vary both over space and time, the knowledge is too fragmented to generalise the cause–effect relationships for such variation. As well, it is not clear how the various methods of studying FSD and their spatio-temporal scales influence derivations of FSDs. We suggest that a hypothetico-deductive research approach, combining empirical studies with process-based simulations is an effective means to advance the knowledge of FSD. We suggest caution in the use of FSD in forest management because applying different distributions or even different parameters for the same distribution may result in great fire size class differences and thus different implications for forest 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.

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 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.

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.