Reasoning on multi-sensor geographic smoke spread data for fire development and risk analysis

This paper presents the general architecture of a multi-sensor geographic information system which allows for the effective use of sensor data and geographic information in fire incident management and fire development analysis. The proposed platform allows the generation of real-time heatmaps that show the space-time distribution of fire/smoke risk levels across an area of concern based on multi-view sensing. Such levels are there to assist the decision makers in taking actions and aims at facilitating quick fire emergency response. Results of several real fire experiments in a large-scale road tunnel and in a multi-compartment set-up show the feasibility of the approach. Furthermore, by temporal analysis of the within- and between-variance of the sensor estimations, an indication on the accuracy/certainty of the measurements at each moment in time can be given. In this way, problems that arise in one sensor can be compensated by the other surrounding sensors. Finally, user experience studies have shown that the platform improves the current visualization of fire characteristics.