Estimating fire properties by remote sensing

Contemporary knowledge of the role of fire in the global environment is limited by inadequate measurements of the extent and impact of individual fires. Observations by operational polar-orbiting and geostationary satellites provide an indication of fire occurrence but are ill-suited for estimating the temperature, area, or radiant emissions of active wildland and agricultural fires. Simulations of synthetic remote sensing pixels comprised of observed high-resolution fire data together with ash or vegetation background demonstrate that fire properties including flame temperature, fractional area, and radiant-energy flux can best be estimated from concurrent radiance measurements at wavelengths near 1.6, 3.9, and 12 /spl mu/m. Successful observations at night may be made at scales to at least 1 km for the cluster of fire data simulated. During the daytime, uncertainty in the composition of the background and its reflection of solar radiation would limit successful observations to a scale of approximately 100 m or less. Measurements at three wavelengths in the long-wave infrared would be unaffected by reflected solar radiation and could be applied to separate flame properties in a binary system of flame and background. However, likely variation in the composition of the background and its temperature limit the approach to measurements that are of high resolution in relation to the scale of the flaming front. Alternative approaches using radiances at wavelengths near 4 and 12 /spl mu/m alone must fail absent a correction for the background, yet the correction is made imprecise by uncertainty in composition of the background where it comprises more than one-third of a pixel.