Being able to predict where, when and how a fire is most likely to strike is vital during wildfire seasons across the globe. Fire risk management and fire behavior modeling require large volumes of data that change continuously over time and space, creating both the need and the opportunity to automate the tasks. These large volumes of data required for spatiotemporal calculations often rely on significant computer resources (i.e. processing power and storage). In the Geography of Natural Disasters Laboratory, University of the Aegean, Mytilene, Greece, an emphasis has been given in applying fire science algorithms in parallel computing environments.
Within this scheme, the AEGIS research project developed a Web Geographic Information System (GIS) as a cost effective, easy-to-use forest fire management platform for the pertinent end users. The research was implemented and applied in seven different study locations around the country of Greece with high-hazard, high-value and high-use forest and other multi-purpose areas: i) Rhodes Island, SE Aegean Sea; ii) Lesvos Island, NE Aegean Sea; iii) Halkidiki (incl. Mount Athos); iv) West Attica, Athens; v) Chania, Crete; vi) Messenia, Peloponnesus; and vii) Kastoria, Macedonia.
Field data were collected from all the study areas to create spatial and non-spatial databases. Data include road networks, vegetation cover types, fuel types, water sources, topography, dispatching resources, social structures, urban areas, wildfire history, etc. To create a reliable fire risk and fire behavior database, field inventories were conducted to collect information about vegetation cover types of the study areas; and then, to create spatial datasets through the use of remote sensing (by acquiring recent satellite images), spatial statistics and GIS techniques. Another major feature of the system is the sharing of real-time weather data streams from local Remote Automatic Weather Stations (RAWS). We integrated the available RAWS into the AEGIS system to achieve the proper information flow of weather data to the end users; weather is one of the most important parameters in fire confrontation activities. AEGIS also provides weather forecasts in a map format. Weather prediction maps were prepared with the operational use of the SKIRON state-of-the-art weather forecasting system.
One of our goals was to develop a pilot fire danger mapping system for Greece including not only the ignition risks but also the expected burned area. Artificial neural networks were used for the mathematical modeling of these complex phenomena. In addition, today’s firefighting needs require a system that will be able to conduct on a timely manner and without devastating delays fire behavior predictions based on topography, vegetation and weather conditions of the wildfire affected area. A giant lead compared to current worldwide fire behavior predictions is the use of a web-based system that allows users of the system to overcome the difficulties that arise from the lack of knowledge or the complexity in usage of current fire behavior systems (e.g. BehavePlus, FARSITE, FlamMap). Thus, we used the Minimum Travel Time (MTT) algorithm as a powerful fire behavior prediction system that runs behind the proposed web platform. The implementation of the fire propagation research was achieved with the collaboration of one of the top research teams worldwide in the field of fire behavior algorithms and software engineering, i.e. Dr. Mark Finney, Dr. Robert Keane and Dr. Alan Ager (from the world renowned USDA Missoula Fire Sciences Laboratory in Montana and the USDA Pacific Northwest Research Station in Oregon of USA), who were invited as external investigators during the project’s period. The algorithms of fire risk and fire behavior procedures were automated by using parallel processing (High Performance Computing and Cloud Computing) to speed up the calculations and promptly provide the outcomes through the web platform to end-users.
We designed and implemented the AEGIS application as a web-based platform that provides access to fire prediction data (risk and behavior), as well as additional information such as socioeconomic activities, roads, land uses, water tanks locations, patrol routes, fleet tracking, satellite images, vegetation types, terrain and weather data. All functionalities provided by AEGIS are accessible to local fire agencies and authorities of civil protection through an appropriate graphical user interface. Without the requirement of knowing the handling of complicated applications, end users utilize a wide range of maps directly in the web browser of their personal and laptop computers, tables or smartphones through a special application (app). All available output results are visualized by utilizing the Web-GIS design tools that provide powerful mapping and geo-processing functionalities free-of-charge.