A near real-time web-system for predicting fire spread across the Cerrado biome

Abstract

Wildfires are aggravating due to climate change. Public policies need territorial intelligence to prevent and promptly fight fires, especially in vast regions like Brazil. To this end, we have developed a fire-spread prediction system for the Brazilian Cerrado, the biome most affected by wildfires in South America. The system automatically uploads hot pixels and satellite data to calculate maps of fuels loads, vegetation moisture, and probability of burning for simulating fire spread thrice a day for the entire Cerrado at 25 ha and for nine conservation units at 0.04 ha spatial resolution. In both versions, the model attains 65–89% of spatial match. Model results together with ancillary data, e.g., historical burned areas and annual CO2 emissions from fires, are available on an interactive web-platform that serves as a tool for fire prevention and fight, particularly in the selected conservation units where the platform is being used for daily operations.