Ichoku, C., R. Levy, Y. J. Kaufman, L. A. Remer, R.-R. Li, V. J. Martins, B. N. Holben, N. Abuhassan, I. Slutsker, T. F. Eck, and C. Pietras, 2002: Analysis of the performance characteristics of the five-channel Microtops II sun photometer for measuring aerosol optical thickness and precipitable water vapor. J. Geophys. Res., 107, NO. D13, doi: 10.1029/2001JD001302.
Five Microtops II Sun photometers were studied in detail at the NASA Goddard
Space Flight Center (GSFC) to determine their performance in measuring aerosol optical
thickness (AOT or tal) and precipitable column water vapor (W). Each derives tal from
measured signals at four wavelengths l (340, 440, 675, and 870 nm), and W from the 936
nm signal measurements. Accuracy of tal and W determination depends on the reliability
of the relevant channel calibration coefficient (V0). Relative calibration by transfer of
parameters from a more accurate Sun photometer (such as the Mauna-Loa-calibrated
AERONET master Sun photometer at GSFC) is more reliable than Langley calibration
performed at GSFC. It was found that the factory-determined value of the instrument
constant for the 936 nm filter (k = 0.7847) used in the MicrotopsÕ internal algorithm is
unrealistic, causing large errors in V0(936), ta936, and W. Thus, when applied for transfer
calibration at GSFC, whereas the random variation of V0 at 340 to 870 nm is quite small,
with coefficients of variation (CV) in the range of 0 to 2.4%, at 936 nm the CV goes up to
19%. Also, the systematic temporal variation of V0 at 340 to 870 nm is very slow, while at
936 nm it is large and exhibits a very high dependence on W. The algorithm also computes
ta936 as 0.91 ta870, which is highly simplistic. Therefore, it is recommended to determine
ta936 by logarithmic extrapolation from ta675 and ta870. From the operational standpoint
of the Microtops, apart from errors that may result from unperceived cloud contamination,
the main sources of error include inaccurate pointing to the Sun, neglecting to clean the
front quartz window, and neglecting to calibrate correctly. If these three issues are
adequately taken care of, the Microtops can be quite accurate and stable, with root-meansquare
(rms) differences between corresponding retrievals from clean calibrated Microtops
and the AERONET Sun photometer being about ±0.02 at 340 nm, decreasing down to
about ±0.01 at 870 nm.
