Calibrating a System of Satellite Instruments - Satellite-based Applications on Climate Change - page 21

chamber prelaunch at different target and instrument temperatures, but it was found
that test artifacts often obscure the true nonlinearity of the system. Also, since there is
no readily available method for verifying the nonlinearity postlaunch, uncertainty
exists in using prelaunch determined nonlinearity in postlaunch calibration.
The new generation of instrument has significant improvements in both scan-cavity
and blackbody designs. The high-quality blackbodies in some cases are made directly
traceable to NIST radiometric standards. However, even with perfect traceability, it is
still desirable to verify the calibration between satellites. Otherwise, the links between
satellites are not known and may not be established.

2.4 Inter-satellite Instrument Calibration

Given the limited life span of a satellite for a typical mission, constructing a long-term
time series for climate change detection requires accurate and consistent data from a
series of satellites. It is well known that despite the best calibration efforts, biases and
inconsistencies still exist for the same series of radiometers on different satellites.
Unlike instrument noise which can be quantified precisely with on-orbit calibration
targets, biases are very difficult to characterize due to the lack of commonly traceable
on-orbit absolute calibration standards and the variable nature of biases both short
term and long term in response to the spacecraft and instrument thermal dynamics.
Several methods have been developed to address the inter-satellite calibration issue,
and each has its advantages and limitations.
The SNO (Simultaneous Nadir Overpass) method (Cao et al. 2004 , 2005a )was
developed for quantifying inter-satellite biases initially for instrument performance
monitoring and has been used by scientists in constructing time series for climate
change detection. This method is relatively simple and robust and is based on the fact
that any pair of polar-orbiting satellites with different altitudes can regularly observe
the Earth at orbital intersections at nearly the same time and that these events are
predictable with orbital perturbation models such as SGP4. The frequency of occur-
rence is a function of the altitude difference between the two satellites (typically once
every 2-10 days). Observations from the two satellites at the SNOs can then be
collocated pixel by pixel and the biases between them quantified. The uncertainties in
the SNO analysis are further reduced in a SNO time series where the inter-satellite
biases at the SNOs are studied as a function of time.
Using the SNO method, an on-orbit calibration reference network can be
established. The network can keep track of the long-term time series of inter-satellite
biases at the SNOs, GEO (Geostationary)/LEO (Low Earth Orbiting) satellites, and
selected vicarious sites for all operational satellites. Even without an absolute scale,
this will tie the calibration of all the satellites together to provide traceability of
individual satellites to each other. It is difficult to know which radiometer produces
the absolutely correct radiance in this scenario, but truth is likely to emerge from the
long-termmeasurements by a multitude of satellites. In addition, airborne radiometers
can be used as checkpoints for the long-term time series to provide calibration links to