|
Titel |
Evaluating the performance and added value of full physics retrievals of XCO2 from GOSAT. |
VerfasserIn |
S. Guerlet, A. Butz, D. Schepers, O. Hasekamp, A. Galli, I. Aben |
Konferenz |
EGU General Assembly 2012
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 14 (2012) |
Datensatznummer |
250060670
|
|
|
|
Zusammenfassung |
The JAXA Greenhouse gas Observing Satellite (GOSAT), launched in 2009, is the first
operating mission dedicated to measuring CO2 and CH4 column-averaged dry air mole
fractions (XCO2 and XCH4) from space. Onboard GOSAT, the TANSO-FTS instrument
acquires near-infrared spectra of sunlight backscattered by the Earth’s surface and
atmosphere. We have developed a full physics algorithm, called RemoTeC, that
simultaneously retrieves XCO2, XCH4 as well as three effective aerosol parameters
representing particle amount, size and height. This method reduces the uncertainty caused by
the modification of lightpath by scatterers [Butz et al., 2009].
In this presentation we evaluate the CO2 retrieval performance of our algorithm. First, we
extend the validation study of Butz et al., 2011. From the comparison with 12 ground-based
stations of the Total Carbon Column Observing Network (TCCON) on almost two years of
data (April 2009 to April 2011), we find a single-sounding precision of our GOSAT retrievals
of 2.8 ppm and an interstation bias of 0.25%. If we consider daily means, the precision is 2.3
ppm on average. Furthermore, we investigate the importance of filtering for (thin) cirrus and
evaluate correlation of the GOSAT-TCCON differences with geophysical or instrumental
parameters. We do not find any significant correlation of errors on XCO2 with airmass,
scattering optical thickness, blended albedo or signal in the O2 A-band. There is,
however, a small dependence between errors and a combination of retrieved aerosol
properties.
We also compare RemoTeC retrievals with non-scattering retrievals, both around
TCCON and at the global scale. We find that when strict cloud and cirrus filters are
applied to the data, the performance of non-scattering retrievals around TCCON
stations is surprisingly good, with a precision of 3 to 3.5 ppm for five of the stations
and an interstation bias similar to that of full physics retrievals. However, on the
global scale, the errors in non-scattering retrievals appear to be larger and they
increase with albedo and scattering optical thickness. As a consequence, satellite
validation studies might lead to a too optimistic assessment of retrieval performance, if
all validation stations are located in regions of similar ground albedo and aerosol
load. We conclude that with respect to its validation purpose, the TCCON can be
improved if future site locations are chosen to extend the current range of ground
albedo. |
|
|
|
|
|