Estimating radiated flux density from wildland fires using the raw output of limited bandpass detectors

Abstract

We have simulated the radiant emission spectra from wildland fires such as would be observed at a scale encompassing the pre-frontal fuel bed, the flaming front and the zone of post-frontal combustion and cooling. For these simulations, we developed a ‘mixed-pixel’ model where the fire infrared spectrum is estimated as the linear superposition of spectra of many (n ~ 30) greybody emitters of randomly selected areal fraction, emissivity and temperature. Our model neglects contributions from atomic and molecular line emission from combustion gasses. The purpose of these simulations was to allow unambiguous use of limited bandwidth detectors to estimate the total power emitted from a wildland fire. From the simulations we observed a well-defined relationship between ground-leaving radiance (W m–2 sr–1) and limited bandpass sensor-reaching radiance for many different detector spectral responses. Error in the relationship is least when the detector sampled in the mid-wave portion of the infrared spectrum (~3–5 μm) where flaming combustion emits most strongly. We validate our approach to estimating total power using data from experimental burns. The ability to estimate total power from limited bandpass measurements has great utility in the observation of wildland fires from ground-based instruments and aircraft and satellite platforms.