AIS radiometry and the problem of contamination from mixed spectral orders

Airborne Imaging Spectrometer data from Mono Lake, CA, are studied in order to establish the spectral radiance of test areas under solar illumination. The objective is to provide a method of atmospheric correction for major absorbers from the spectrometer data themselves. Crucial to the analysis is radiometric calibration of the instrument. Good agreement is found between calculated and measured radiances for uniform surface targets (beaches), but simulations of atmospheric properties with LOWTRAN 6 lead to unreasonably low values of atmospheric precipitable water. Absorptions from carbon dioxide are not detected in the AIS data, but are strongly present in the LOWTRAN 6 model. The apparent low contrast of all atmospheric absorption bands leads to a study of contamination from overlapping spectral orders in the AIS data. The suspected contamination is shown unambiguously to be present beyond approximately 1500 nm and consists of an extra radiance term including atmospheric bands from the λ/2 wavelength interval. The magnitude of the contamination is a factor of 1.5–2 greater than the expected uncontaminated signal alone. The spectral band filling at 1400 nm, i.e., an apparent finite transmittance in a band expected to be saturated, cannot be accounted for by order mixing because of the 800-nm blocking filter used and must arise from some other cause. A rigorous removal of the unwanted spectral contamination does not seem possible for any data taken in the spectral region 1200–2500 nm. Rough estimates for observations in the interval 900–2100 nm might be pieced together if a suitable after-the-fact radiometric calibration of the instrument can be formulated.