Using cellular automata to assess the role played by wind direction in two large fire episodes in Portugal

Portugal is recurrently affected by severe wildfires, the fire season of 2017 representing the most tragic year with half a million of hectares burned and 115 deaths. The events that took place on October 15, 2017 deserve special attention, not only because the area burned on that day represents more than 50% of that burned during the entire year, but also because it resulted from the combination of very strong winds steered by the passage of hurricane Ophelia, very dry vegetation because of a prolonged drought affecting the country, very low atmospheric relative humidity and a record number of ignitions. Meteorological fire danger is usually rated using the Fire Weather Index (FWI) that is part of the Canadian Forest Fire Weather Index System. However, wind direction is not taken into account when defining FWI, and therefore it is worth investigating how this factor may affect the evolution of a given fire keeping all the remaining factor unaltered. The role played by wind direction is assessed using a cellular automata (CA) model to simulate two wind-driven wildfires that took place at Pataias-Burinhosa and Quiais on October 15, 2017. The CA model is first calibrated using winds derived from a regional weather forecasting model and sensitivity studies are then performed by systematically rotating the forecasted winds keeping all the other parameters constant. Results indicate a a progressive decrease in probability of burning from a 45º to a 90º counterclockwise rotation. These results suggest improving FWI by defining an FWI vector definition of an FWI vector, whose direction is that of the wind and magnitude is that of FWI. This vector should then be compared against the prevailing orientation of the vegetated area, and the closer the alignment between the two directions, the greater the meteorological fire danger.