Environmental Engineering Reference
In-Depth Information
bias correction patterns for these satellites. Differencing of the bias corrections of two
satellites gives inter-satellite difference information. With a differencing of the bias
corrections for two SNO-calibrated satellites, one can see that the bias corrections are
consistent with Fig. 8.6 where inter-satellite biases are on the magnitude of
0.05-0.1 K for NOAA-10 through NOAA-14 (Zou et al. 2006 ). In summary, bias
correction pattern analyses indicate that the inter-calibrated MSU data have reached
their performance expectation in the reanalysis data assimilation.
One remaining problem in the inter-satellite calibration is that the absolute value
of the inter-calibrated MSU/AMSU brightness temperature has not been adjusted to
an absolute truth, although inter-satellite biases have been removed. This is part of
the reason for the relative biases between the climate reanalysis and inter-calibrated
MSU observations as shown in Fig. 8.7 after 1987. As described earlier, the
calibration offset of the reference satellite,
δR N10 , was arbitrarily assumed to be
zero. This assumption does not affect the variability and trend analysis of the TCDR
products developed from the FCDR. However, it influences the relative biases
between the reanalysis and FCDR. This offset needs to be determined in FCDR
validation processes against certain reference observations. Plans for such a valida-
tion are being developed at NOAA/NESDIS as part of the effort of FCDR and
TCDR development. Once this is done, the bias removed, identical multi-satellite
MSU/AMSU FCDRs may be used as an anchor or reference dataset for bias
corrections of other observations in the reanalysis data assimilation. This is
expected to help the climate reanalysis to be more consistent with the satellite
observations.
8.3 Atmospheric Temperature TCDR from Merged MSU/
AMSU-A Data
Development of atmospheric temperature TCDR involves proper treatment of
errors from several different sources. These include, but are not limited to, antenna
pattern effect, incident angle effect, diurnal drift errors, short overlaps between
certain satellite pairs, Earth-location dependency in biases, residual biases left from
non-perfect instrument calibration, orbital decay, and frequency differences
between MSU and AMSU-A channels. Correction algorithms for these effects
have been developed by different investigators for TCDR development. In the
following, the algorithms implemented in the NOAA MSU/AMSU atmospheric
temperature TCDR are briefly described.
8.3.1 Antenna Pattern Correction
When the satellite antenna main beam looks at the Earth to make an observation,
the antenna side lobes receive a small amount of radiation from cold space, the
Search WWH ::




Custom Search