Experimental and model-based derivation of atmospheric point spread functions

Radiometric error exists for most remote sensing applications in the visible and near-infrared regions due to atmospheric effects which are often characterized by absorption and scattering mechanisms that modify the direct radiance and path radiance components of the total radiance observed by the sensor. However, another source of error exists due to scattering, and often turbulence, that causes loss of contrast between bright and dark targets. The phenomenon, sometimes called the adjacency effect, occurs when light from a bright surface is scattered into the field of view of a darker target, thus causing the brighter target to appear darker while the darker target appears brighter. At a given point in time and space, this effect can be modeled as an atmospheric point spread function (PSF). Much of the work to characterize atmospheric PSF has been done in coastal regions where there is a high aerosol concentration and a well-defined edge (shoreline). The work described attempts to extend these observations to an inland, mid-continent atmosphere. A unique spectral/spatial target was developed and deployed in a grassy area near the Brookings, South Dakota, airport. It consisted of a large tarp, 30 by 60 meters in size, against a uniform background. The tarp was made of nylon and dyed so that it was very bright in the blue region of the spectrum with a reflectance peak of 0.6 at approximately 510 nm. Over the rest of the visible spectrum the reflectance was approximately 0.05, and in the near infrared it again rose to roughly 0.5. The uniform background was recently mowed grass and, therefore, exhibited a typical vegetative spectrum that was quite dark in the blue. Since aerosol scattering is typically greatest in the blue portion of the visible spectrum, this target with its well defined spatial edge and unique spectral signature was assembled in an attempt to optimize detection of atmospheric PSF near 500 nm wavelengths