An automated procedure for reducing atmospheric features and emphasizing surface emissivity in hyperspectral longwave infrared (Lwir) images

The at-sensor radiance of a pixel in a longwave infrared (LWIR) image represents the surface temperature, the surface emissivity, which is similar during the day and at night, and the atmospheric contribution, which is expressed differently during the day and at night. Based on this, an automated procedure, which locates pixels for which each absorption feature in their radiance during the day appears as an emission feature at night, indicative of atmospheric contribution, was developed. The average day or night spectrum of these indicative pixels was applied as a gain factor spectrum to the entire day or night image, respectively, reducing the atmospheric contribution and emphasizing the surface spectral features, represented by emissivity, of each pixel in the image. The procedure was examined on LWIR hyperspectral data cubes acquired over two different areas, and enabled effective reduction of the atmospheric features in both area data sets.