The wildfire prediction model is crucial for accurate rescue and rapid evacuation. Existing models mainly adopt regular grids or fire perimeters to describe the wildfire landscape. However, these models have difficulty in explicitly demonstrating the local spread details, especially in a complex landscape. In this paper, we propose a wildfire spread model with an irregular graph network (IGN). This model implemented an IGN generation algorithm to characterize the wildland landscape with a variable scale, adaptively encoding complex regions with dense nodes and simple regions with sparse nodes. Then, a deep learning-based spread model is designed to calculate the spread duration of each graph edge under variable environmental conditions. Comparative experiments between the IGN model and widely used fire simulation models were conducted on a real wildfire in Getty, California, USA. The results show that the IGN model can accurately and explicitly describe the spatiotemporal characteristics of the wildfire spread in a novel graph form while maintaining competitive simulation refinement and computational efficiency (Jaccard: 0.587, SM: 0.740, OA: 0.800).

Sensor Web has been widely promoted and its application has evolved from original military usages to current ubiquitous civil and commercial applications. Its emergence has become a great strength to facilitate the Digital Earth concept. Although many Sensor Web applications and methods have been proposed to assist disaster management, they are not well suited to the unique features and application requirements of disaster management. Most researches focus on how to use the Sensor Web to monitor a certain phenomenon before a disaster happens and to provide early warning. However, there is a lack of study on the negative effects that a disaster may bring to the Sensor Web. For example, severe weather conditions, damaged infrastructure, and spatial isolation may directly make the Sensor Web out-of-service. Besides, disaster management is a complex subject and its domain knowledge needs to be clarified. In this paper, the domain of the disaster management is explored and its unique features are analyzed. Then the Sensor Web concept and its role in disaster management are explained. Afterwards, the challenges of the Sensor Web for disaster management and the possible solutions are discussed. Based on the above, we introduce a high-mobility emergency system to demonstrate a good solution of the Sensor Web for multi-purpose disaster management.

The construction method for chains of disasters or events is still one of the core scientific questions in studying the common rules of disaster’s evolution. Especially when dealing with the complexity and diversity of disasters, it is critical to make a further investigation on reducing the dependency of prior knowledge and supporting the comprehensive chains of disasters. This paper tries to propose a novel approach, through collecting the big scholar and social news data with disaster-related keywords, analysing the strength of their relationships with the co-word analysis method, and constructing a complex network of all defined disaster types, in order to finally intelligently extract the unique disaster chain of a specific disaster type. Google Scholar, Baidu Scholar and Sina News search engines are employed to acquire the needed data, and the respectively obtained disaster chains are compared with each other to show the robustness of our proposed approach. The achieved disaster chains are also compared with the ones concluded from existing research methods, and the very reasonable result is demonstrated. There is a great potential to apply this novel method in disaster management domain to find more secrets about disasters.

In contemporary cities, road collapse is one of the most common disasters. This study proposed a framework for assessing the risk of urban road collapse. The framework first established a risk indicator system that combined environmental and anthropogenic factors, such as soil type, pipeline, and construction, as well as other indicators. Second, an oversampling technique was used to create the dataset. The framework then constructed and trained a convolutional neural network (CNN)-based model for risk assessment. The experimental results show that the CNN model (accuracy: 0.97, average recall: 0.91) outperformed other models. The indicator contribution analysis revealed that the distance between the road and the construction site (contribution: 0.132) and the size of the construction (contribution: 0.144) are the most significant factors contributing to road collapse. According to the natural breaks, a road collapse risk map of Foshan City, Guangdong Province, was created, and the risk level was divided into five categories. Nearly 3% of the roads in the study area are at very high risk, and 6% are at high risk levels, with the high risk roads concentrated in the east and southeast. The risk map produced by this study can be utilized by local authorities and policymakers to help maintain road safety.

Background: An effective identification model is crucial to realise the real-time monitoring and early warning of forest fires from surveillance cameras. However, existing models are prone to generate numerous false alarms under the interference of artificial smoke such as industrial smoke and villager cooking smoke, therefore a superior identification model is urgently needed. Aims: In this study, we tested the Transformer-based model FireFormer to predict the risk probability of forest fire from the surveillance images. Methods: FireFormer uses a shifted window self-attention module to extract similarities of divided patches in the image. The similarity in characteristics indicated the probability of forest fires. The GradCAM algorithm was then applied to analyse the interest area of FireFormer model and visualise the contribution of different image patches by calculating gradient reversely. To verify our model, the monitoring data from the high-point camera in Nandan Mountain, Foshan City, was collected and further constructed as a forest fire alarm dataset. Key results: Our results showed that FireFormer achieved a competitive performance (OA: 82.21%, Recall: 86.635% and F1-score: 74.68%). Conclusions: FireFormer proves to be superior to traditional methods. Implications: FireFormer provides an efficient way to reduce false alarms and avoid heavy manual re-checking work.