Advances in Remote Sensing and GIS applications in Forest Fire Management From local to global assessments

Abstract

The conceptual definition of a fire risk assessment system should include the most relevant components associated to the fire process. Fire occurrence factors have been traditionally classified in three groups: Fuels, Heat source and Oxygen. In the case of wildland fires, heat source is mostly related to the starting of the fire (fire ignition), and can be further divided in natural-caused (lightning, volcano eruption) and human-caused. Oxygen is needed for combustion, and it is mostly related to fire propagation, being the critical parameters wind speed, direction and slopes. Fuels are both related to fire ignition (the drier, the more likely to ignite) and propagation (the more fuel load available, the more energy will be released and favour further ignition). Fuel moisture is associated to short-term weather factors, topographic conditions and soil characteristics, while fuel load and geometrical properties are related to climate, soil and land use patterns. In addition to these factors, fire risk assessment should also consider the potential damages caused by fires, which are very much dependent on fire characteristics (energy released, residence time, flame length, etc.). Following these ideas, a comprehensive fire risk assessment system has been developed within the Fireglobe project (www.fireglobe.es), funded under the Spanish Program for Science and Research. To implement the proposed risk framework, the first phase focused on the generation of risk factors for the whole Spanish Iberian territory (both the Canary and the Balearic islands were not considered at this stage). Human factors, lightning probability, fuel moisture content of both dead and live fuels, and propagation have been considered. Additionally, fire vulnerability has been assessed by analyzing values at risk and landscape resilience. Once the variables were generated, the fire risk model was defined by integrating the input variables using statistical and physical approaches. Finally, the validation has been performed using fire statistics derived from the fire seasons of 2010 and 2011.