Calibrating a System of Satellite Instruments - Satellite-based Applications on Climate Change - page 22

the absolute standard. Also, further research onmoon calibrationmay allow us to use it
as an absolute calibration standard in the visible/near-infrared for the long-term time
series according to NIST ( http://www.nist.gov/physlab/div844/grp06/lusi.cfm ).
Applications of the SNO method to microwave instruments have shown very
promising results for climate change detection analysis. Several factors contributed
to this success. First, the inter-satellite biases for microwave instruments appear to
change little over the short term and slowly over the long term. Second, the
microwave channel center frequencies are made to be stable and well known,
which significantly reduces uncertainties related to spectral calibration. Third,
each microwave instrument has its own onboard blackbody calibration, which
keeps track of the instrument degradation independently. It is found that the SNO
method works very well for microwave instruments sensing the mid-troposphere to
upper stratosphere, where the uncertainty in the bias is much smaller than the
instrument noise. For example, studies have demonstrates the excellent agreement
on the order of 0.1 K for the 53.6 GHz channel of AMSU on NOAA-16 and -17
(Cao and Tobin 2008 ).
The application of the SNO method to the visible/near-infrared and infrared
radiometers has great potential. Studies have shown that the SNO method is very
effective in quantifying the inter-satellite biases for these channels. Since the biases
are short-term invariant for the visible/near-infrared instruments, they can be used
for intercalibrating the satellites for global data. The dry atmosphere and highly
reflective surface for a broad range of solar zenith angles at the SNO sites in the
polar regions are advantageous for calibrating these channels (Jaross et al. 1998 ;
Masonis and Warren 2001 ). However, since the SNO method only provides a
relative calibration between two satellites and none of the NOAA satellites has
onboard calibration for the visible/near-infrared channels, the SNO calibration
alone is not sufficient to produce a recalibrated long-term time series for these
channels. This method would be more useful if one satellite can be relied on as a
stable standard, such as in the intercalibration of MODIS and NOAA radiometers
(Cao et al. 2008a ; Heidinger et al. 2002 ), but the difference in the spectral response
functions between them introduces uncertainties and makes the intercalibration
difficult. Therefore, this spectral bias must be resolved in intercomparisons such
as through hyperspectral analysis (Cao et al. 2010 ).
For infrared radiometers, studies have shown that the SNO method can quantify
inter-satellite biases with uncertainties smaller than the instrument noise (Cao and
Heidinger 2002 ). However, additional uncertainties exist when compared to that of
the microwave and visible instruments. First, the calibration accuracy may vary
over an orbit, as found for AVHRR (Wang and Cao 2008 ). Biases found at the
SNOs may not be the same in other parts of the orbit, and the bias may be orbital
and seasonal dependent. The inter-satellite bias can also depend on the scene
radiance (Shi et al. 2008 ). The calibration accuracy may also change long term in
response to a number of factors such as degradation and orbital drift. Second, for
infrared sounders, small differences in spectral response functions may mean that a
different layer of the atmosphere is observed, thus producing seasonal biases as the
atmosphere changes over time.