Laboratory characterization of a CCD camera system for retrieval of bi-directional reflectance distribution function

The Remote Sensing Group of the Optical Science Center at the University of Arizona has developed a four-band, multi-spectral, wide-angle, imaging radiometer for the retrieval of the bi-directional reflectance distribution function;(BRDF) for vicarious calibration applications. The system consists of a fisheye lens with four interference filters centered at 470 nm, 575 nm 660 nm, and 835 nm for spectral selection and an astronomical grade 1024 x 1024-pixel, silicon CCD array. Data taken by the system fit in the array as a nominally 0.2 degree per pixel image. This imaging radiometer system has been used in support of the calibration of Landsat-5 and SPOT-satellite sensors. This paper presents the results of laboratory characterization of the system to determine linearity of the detector, point spread function (PSF) and polarization effects. The linearity study was done on detector array without the lens, using a spherical-integrating source with a 1.5-mm aperture. This aperture simulates a point source for distances larger than 60 cm; Data were collected as both a function of exposure time and distance from the source. The results of these measurements indicate that each detector of the array is linear to better than 0.5%. Assuming a quadratic response improves this fit to better than 0.1% over 88% of the upper end of the detector's dynamic range. The point spread function (PSF) of the lens system was measured using the sphere source and aperture with the full camera system operated at a distance of 700 mm from the source, thus the aperture subtends less than the field of view of one pixel. The PSF was measured for several field angles and the signal level was found to fall to less than 1% of the peak signal within 1.5-degrees (10 pixels) for the on-axis case. The effect of this PSF on the retrieval of modeled BRDFs is shown to be less than 0.2% out to view angles of 70 degrees. The final test presented is one to assess the polarization effects of the lens system by illuminating the camera system with the same spherical-integrating source with a 50-mm aperture with a linear polarizing filter. The degree of polarization of the system is shown to be negligible for on-axis imaging but to have up to a 20% effect for field angles of 70 degrees. The effect of the system polarization on the retrieval of modelled BRDFs is shown to be up to 3% for field angles of 70 degrees off nadir and a solar zenith angle of 70 degrees. Polarization response is therefore found to be the greatest source of error in the system. A method to account for polarization effects in digital camera imagery is proposed.