Use of spatially refined satellite remote sensing fire detection data to initialize and evaluate coupled weather-wildfire growth model simulations

Availability of spatially refined satellite active fire detection data is gradually increasing. For example, the new 375 m Visible Infrared Imaging Radiometer Suite (VIIRS) data show improved active fire detection performance for both small and large size fires. The VIIRS data have proved superior to MODIS for mapping of wildfires events spanning several days to weeks of either continued or intermittent activity, delivering 12-h active fire data of improved spatial fidelity. The VIIRS active fire data are complemented by other satellite active fire data sets of similar or higher spatial resolution, including the new 30 m Landsat-8. Additional assets should include the upcoming 20 m Sentinel-2 Landsat-class satellite program by the European Space Agency to be launched in 2014-15. These improved active fire data sets are fostering new applications that rely on higher resolution input fire data. In this study, we describe the characteristics of the new VIIRS and Landsat-8 data and demonstrate one such new application of satellite active fire data in support of fire behavior modeling. We present results for a wildfire observed in June 2012 in New Mexico using an innovative approach to improving the simulation of large, long-duration wildfires, either for retrospective studies or forecasting in a number of geophysical applications. The approach uses (1) the Coupled Atmosphere-Wildland Fire Environment (CAWFE) Model, a numerical weather prediction model two-way coupled with a module representing the rate of spread of a wildfire's flaming front, its rate of consumption of different wildland fuels, and the feedback of this heat release upon the atmosphere - i.e. 'how a fire creates its own weather', combined with (2) spatially refined 375 m VIIRS active fire data, which is used for initialization of a wildfire already in progress in the model and evaluation of its simulated progression at the time of the next pass. Results show that initializing a fire that is 'in progress' with VIIRS data and a weather simulation based on more recent atmospheric analyses can overcome several issues and improve the simulation of late-developing fires and of later periods (particularly those with growth periods separated by lulls) in a long-lived fire.