Infrared Measurements of Pristine and Disturbed Soils 2. Environmental Effects and Field Data Reduction

We have expanded on previous methods for correcting sense, the requirement of thermodynamic equilibrium in thermal infrared field measurements for atmospheric Kirchhoff ’s law is violated (Badenas, 1997). downwelling radiance (Salisbury and D’Aria, 1992; Korb Nerry et al. (1990a,b) and Labed and Stoll (1991) et al., 1996; Hook and Kahle, 1996) by comparing the used the ‘‘emissivity box’’ approach (in which a large box sensitivity of various input parameters (sample temperasurrounds a sample to allow control of the environmental ture, diffuse gold plate emissivity, and temperature) on radiation) to determine broad-band (8–14 lm) emissivithis correction. We find that the sample temperature has ties in the field and laboratory. Salisbury and D’Aria the greatest influence on the applied atmospheric correc(1992), Korb et al. (1996), and Hook and Kahle (1996) tion (cf. Nerry et al., 1990a,b; Labed and Stoll, 1991), presented field and computational methods for deriving and we present a method in which the sample temperaspectral emissivity from field radiance spectra with supture is varied to minimize the residual atmospheric emisporting downwelling radiance measurements and radiosion lines in the measured field emissivity spectra. Direct metrically measured temperatures. These emissivity meacomparison of laboratory hemispherical reflectance measurements, once corrected for environmental effects, can surements of wet-sieved and dry-sieved samples with be directly compared to laboratory hemispherical reflecthese appropriately corrected field observations of undistance measurements. turbed and disturbed soils, respectively, provides a means We have found that these methods are effective in to predict the expected infrared spectral contrast differreducing field measurements, but that the results are ences between such soils (see preceding article). Elsehighly sensitive to the values of some of the input paramvier Science Inc., 1998 eters, values which may not be well known, but also for many purposes not crucial in themselves. In this article, we summarize the general data reduction methodology INTRODUCTION used with the Designs & Prototypes FTIR field spectroAn essential step in compositional remote sensing in the meter (cf. Korb et al., 1996; Hook and Kahle, 1996) and region of thermal emission is comparison of field and labevaluate how the individual input parameters (sample oratory spectra. Laboratory spectra are often obtained ustemperature, diffuse gold plate emissivity, and temperaing an integrating sphere to provide directional hemiture) affect the correction of field data of soils to absolute spherical reflectance, which Salisbury et al. (1994) showed emissivity. We find that the sample (target) temperature could be used to derive emissivity via Kirchhoff’s law has the greatest influence on these corrections, consistent within about 1%, despite the fact that, in the strictest with the results of Nerry et al. (1990a,b) and Labed and Stoll (1991), who used different field measurement techniques and broad-band radiometers, and Rivard et al. * Hawaii Institute of Geophysics and Planetology, University of (1995), who restricted their analyses to rocks. Hawaii at Manoa, Honolulu † Pacific Island Technology, Inc., Honolulu Two soil examples, one from Ft. A.P. Hill, Virginia, ‡ Current address: U.S. Geological Survey, Flagstaff, AZ 86004 and the other from Ft. Devens, Massachusetts, are preAddress correspondence to K. A. Horton, Hawaii Inst. of Geosented to illustrate the results obtained. These particular physics and Planetology, 2525 Correa Rd., Honolulu, HI 96822. Received 16 April 1997; revised 28 October 1997. sites were chosen from among those data sets in which