7% when tied to the fixed-time absolute calibrations, which
have an accuracy of about 5%.Clearly, for future systems such as the EOS, one must make careful provision for in-operation calibration of the visible channels. Such provision is, however, quite difficult because internal calibration sources are subject to degradation. While ARM should keep abreast of, and even provide input to, decisions about this calibration capability, these new systems will not be operable until late in the ARM program. Therefore, in this report, we focus our attention on indirect calibration of existing systems. Such absolute indirect calibrations fall into two general classes: reflectance methods and radiance methods. Reflectance methods involve measuring the reflectance of a target surface from the ground or a low-flying aircraft, then using a radiation transfer code to calculate the radiance at the aperture of the satellite. Radiance methods involve measuring the radiance of a target from a high-flying aircraft whose viewing geometry and time are arranged to coincide with the satellite overpass, thereby measuring most of the atmospheric effects on the radiation transport directly. The residual atmospheric effects, significantly smal-ler than in the reflectance method, are then calculated using a radiation transport code.
Representative best case accuracies for each of the two methods are summarized in Table S.2. The two methods currently yield about the same accuracy, i.e., 5-7% for AVHRR, but are dominated by different uncertainties. Atmospheric corrections, most notably aerosols, dominate the uncertainties in the reflectance method, while radiometer calibration and coalignment of the aircraft and satellite fields-of-view dominate the uncertainty of the radiance method.
The absolute accuracies of both these methods is expected to improve in the next 3-5 years. Improvements in radiometer calibration and navigation algorithms should enable radiance methods to attain 3-4% absolute accuracy for AVHRR. Similarly, improvements in aerosol characterization and in ground reflectance measurements should enable reflectance methods to attain about 5% accuracy for AVHRR.
Relative calibrations, such as those developed under ISCCP (International Satellite Cloud Climatology Program), can then be used to interpolate between absolute calibration periods. Relative calibrations are made by grouping scenes into several classes of targets of similar albedo and assuming that this albedo remains essentially constant. While the total uncertainty in this method is estimated at about 5%, drifts in calibration on the order of 2% of the absolute calibration can be detected.
Figure S.1 shows the results from combining fixed-time, absolute calibrations with relative calibration to "fill in" the inter-calibration intervals for both
AVHRR and VISSR/VAS. The relative calibrations are found to have an absolute accuracy of about 7% when tied to the fixed-time absolute calibrations, which
have an accuracy of about 5%.
We recommend that ARM use both reflectance and radiance methods to detect and eliminate systematic errors in absolute indirect cali- brations, and that these methods be supplemented with relative calibrations through the ISCCP.
