Minimum travel time algorithm for fire behavior and burn probability in a parallel computing environment

Abstract

Fire management systems materialize the integration of fire science models and decision support planning modules. Their operational usage often requires the concurrent execution of a large number of fire growth simulations by multiple users. Intensive computations such as the creation of burn probability maps demand not only high expertise but also high computing power and data storage capacity. The purpose of this paper is to present some of the initial results of the AEGIS platform, which is a Web-GIS wildfire prevention and management information system currently under development. More specifically, the paper focuses on the utilization of the Minimum Travel Time (MTT) algorithm as a powerful fire behavior prediction system. MTT in AEGIS will be applied in a transparent way through its graphical user interface. Several end users will be able to conduct on-demand fire behavior simulations. To achieve this, end users must provide a minimum amount of inputs, such as fire duration, ignition point and weather information. Weather inputs can be either inserted directly or derived from selected remote automatic weather stations or forecasted weather data maps based on the SKIRON system (Eta/NCEP model). Seasonal burn probability maps will be also prepared and provided to the end users. Socioeconomic data, weather predictions, topographic and vegetation data will be combined with artificial neural networks to produce an ignition probability map. Based on the ignition probability map, thousands of potential ignition points located in areas of anticipated high risk will be generated. These ignitions will be further used as inputs on MTT simulations, running FConstMTT as a command line-based executable. FConstMTT calculations will be conducted on a parallel mode in Microsoft Azure infrastructure using a different subset of ignition points in each simulation. The current deployment of the AEGIS platform consists of a number of machines resided on premises and a scalable Cloud Computing environment based on the Microsoft Azure infrastructure. This parallel computing environment ensures high processing power availability and high data storage capacity. During a fire emergency, the scalability of the Cloud can also provide extra processing power and storage, if needed. It is anticipated that by integrating MTT into the AEGIS platform, the firefighting and civil protection agencies will gain great assistance to organize better and more reliable plans for fire confrontation.