RADIOMETRIC CHARACTERIZATION OF DIODE-ARRAY FIELD SPECTRORADIOMETERS

A study ii as conducted to evaluated the radiometric performance of tw o commercially available diode-array field spectroradiometers: the Spe ctron Engineering (SE) Model 590 that has been on she market for some 10 years and the several-year-old Analytical Spectral Devices (ASD) Pe rsonal Spectrometer (PS) 2. Both of these instruments provide rapid ac quisition (similar to 1 s) of a spectrum in the visible to near-infrar ed region (similar to 0.4-1.1 mu m) and are field-portable. These eval uations included: spectral bandwidth, second-order effects suppression , signal to noise ratios, noise equivalent radiances, linearity of res ponse with radiance and with integration time, temperature sensitivity , and polarization sensitivity In the configurations tested the SE590 instruments had a spectral bandwidth of about 16 nm full width at half -maximum the PS2 instruments, 3-4 nm. Neither of the instruments produ ced measurable second-order leakage. When viewing a barium-sulfate-coa ted integrating source internally illuminated with tungsten lamps, bot h instruments were able to obtain signal to noise ratios of 1000:1 or better between 500 m and 900 nm over most of the dynamic range provide d by the sources when the integration time was optimized for the light level. Signal to noise dropped at shorter wavelengths in part due to the low energy provided by the source and at longer wavelengths due to the decrease in the detector response. In linearity tests all of the instruments generally agreed within 5% of the reference radiometer bet ween 400 nm and 1000 nm; however, results were poorer at lower radianc e levels and at wavelengths longer than 1000 nm or shorter than 400 nm . Both of the instruments rely on a shutter to measure their dark curr ent, which is normally subtracted from the signal. However, closing th e shutter on all of the instruments altered the dark current readings to some extent such that the dark current that teas subtracted was not correct. When the error in dark current is compensated for, she appar ent linearity of the system improves considerably with weak signals. T he SE590 systems showed significant polarization sensitivity of up to 25%, whereas the PS2 instruments did not. Movement of the fiber optic cable on the PS2 instruments did result in measurable signal changes. Both instruments also showed the temperature sensitivity typical of si licon detectors; i.e., increasing sensitivity at the longer wavelength s with higher temperature.