Atmospheric Temperature Climate Data Records from Satellite Microwave Sounders - Satellite-based Applications on Climate Change - page 106

Satellite-based CDRs are further segmented into fundamental CDRs (FCDRs),
which are calibrated and quality controlled sensor data that have been improved
over time, and thematic CDRs (TCDRs), which are geophysical variables derived
from the FCDRs (NRC 2004 ). For MSU/AMSU, the FCDR is the swath Level-1c
radiance data that are produced immediately after the instrument calibration. Since
FCDR is used for subsequent satellite retrievals and data assimilations in numerical
weather predictions and modeling reanalyses, its quality has a significant impact on
the accuracies of these subsequent applications. Consistent, high-quality FCDR
should have instrument calibration errors removed as much as possible.
Although fundamentally important, however, the FCDR is usually not directly
used for climate change analysis due to data irregularity in location and time and
also because of different interpretations of the radiance data from the retrieved
geophysical variables that are used to describe the climate change. As a result, a
TCDR is developed on top of the FCDR for the purpose of direct climate analysis
and trend calculations. A TCDR (thematic climate data record) is generally a
gridded dataset with fixed time interval that is easy to use. However, developing
TCDR requires more bias corrections and adjustments of different error sources
than applied for FCDR development. These include satellite sampling errors such
as those related to satellite drift. Retrieval algorithms are also needed for many
geophysical variables when there is a nonlinear relationship between the radiances
and the physical variables to be retrieved. These bias correction and retrieval
processes cause the TCDRs to have more error sources than the FCDRs.
Recently, significant progress has been made at NOAA/NESDIS in the develop-
ment of both the FCDR and atmospheric temperature TCDR from the MSU/AMSU
observations using the simultaneous nadir overpass (SNO) intercalibration meth-
odology (Zou et al. 2006 , 2009 ; Zou and Wang 2010 , 2011 ). The method largely
removed instrument calibration errors related to inaccurate calibration nonlinearity
and solar heating variability on the instrument and thus resulted in a consistent,
high-quality radiance FCDR. Such a feature in the FCDR has the advantage that it
prevents instrument calibration errors from transferring to the gridded analysis level
where they, if not corrected, could become mixed up with the diurnal drift errors.
This error coupling may cause difficulties in bias correction in the TCDR develop-
ment and ultimately cause uncertainties in the trend determination. In addition, with
biases removed, identical multi-satellite FCDRs can be an ideal candidate as an
anchor or reference dataset for bias corrections of other observations in reanalysis
data assimilation, which may help the climate reanalysis to be consistent with the
satellite observations as much as possible.
In this chapter, the SNO methodology for MSU/AMSU inter-satellite calibration
is reviewed, and the resulting FCDR performance in terms of inter-satellite bias
reduction is described. Impact of the inter-calibrated FCDR on the reanalysis bias
correction improvement is demonstrated. Bias correction approaches of various
error sources for MSU/AMSU atmospheric temperature TCDR are described.
Finally, updated 34-year atmospheric temperature trends derived from the SNO-
calibrated TCDR are provided.