CO2 Vertical Profile Constraints from OCO and Thermal IR Measurements
Le Kuai
Ge 152 Term paper
1
Abstract
Understanding changes in atmospheric carbon dioxide (CO2) concentrations, their global sources and
sinks, and the carbon cycle dynamics and processes that control their variability has emerged as one of the
principal challenges of 21st century Earth system science. Satellite observations of atmospheric CO 2 are
poised to revolutionize our understanding of global carbon cycle science by providing unprecedented
spatiotemporal resolution and coverage. Major advances are expected with the launch of the Orbiting
Carbon Observatory (OCO) in 2009.
Here, we carry out an information analysis for a combined retrieval of OCO and thermal IR data (e.g.
TES; the principle is same for AIRS data). Preliminary results show that there is considerable increase in the
information content and degrees of freedom in the combined retrieval compared to retrieval using only OCO
or only TES data. This approach will significantly improve the estimation of atmospheric carbon sources
and sinks by providing observational constraints on vertical as well as horizontal and temporal distributions
of atmospheric CO2 in data assimilation and data fusion approaches.
2
Introduction
Carbon dioxide (CO2) is the second important green house. Its atmosphere concentration rapidly increases
from 280 to 37 parts per million (ppm) within last 40 years. However, the geographic distribution of sourcesink for these CO2 and their variations are not well understood.
The Orbiting Carbon Observatory (OCO) will for the first time provide global, space-based observations
with high spatial resolution and high accuracy to identify CO2 sources and sinks and quantify their
variability over the seasonal cycle [Crisp, et al., 2004]. The measurement of CO2 from space has
dramatically improved spatiotemporal coverage. It will measure reflected sunlight in three near-infrared
spectral regions (the 0.76 µm O2 A-band and in the CO2 bands at 1.61 and 2.06µm). The 3-band
spectrometric approach using NIR reflected sunlight has the capability for highly accurate Xco2
measurements [Kuang, et al., 2002].
Tropospheric Emission Spectrometer (TES), launched aboard the Aura spacecraft in 2004, is a Fourier
Transform spectrometer measuring infrared spectral radiances from 3.2 to 15.4 microns.
TES observes thermally active atmospheric or surface quantity. TES data routinely provides vertical profiles
of the following: Ozone, Carbon monoxide, Temperature, Water vapor, HDO, Emissivity, Effective cloud
parameters, Cloud optical depth, and Cloud height. The principle is same for AIRS data.
3
(a)
(b)
Fig. 1 (a) OCO weak CO2 band in NIR region; (b) TES strong CO2 band in Thermal
(a)
(b)
(c)
Fig. 2 Averaging kernels for OCO retrieval (a), TES retrieval (b) and simultaneous retrieval (c)
4
Using the OCO weak CO2 band (1.62 µm, Fig. 1(a)) in the NIR region for the CO2 retrieval, the averaging
kernels have peaks near surface (below 5 Km) (Fig. 2(a)). For TES retrieval, we use strong CO2 band (15 µm,
Fig. 1(b)) in thermal infrared spectra. Thus, the peaks of TES averaging kernel are in middle and upper
troposphere (between 5 to 15 Km, Fig, 2(b)), which is very similar to AIRS retrieval. A simultaneous
retrieval of OCO and TES (or AIRS) will combine these pieces of complementary information. Fig. 2(c)
shows that the joint retrieval averaging kernel for levels below 5 Km retain the structure of OCO averaging
kernels. While the averaging kernels at levels of 7, 9 and 11 Km have peaks at middle and upper troposphere
but are not zero at surface.
Whether there will be benefit from the simultaneous retrievals of OCO and AIRS satellite observations can
be investigated by the information analysis.
Information Analysis
In information theory, the Shannon entropy or information entropy is a measure of the uncertainty associated
with a random variable. Self information is a measure of the information content (H) associated with the
outcome of a random variable. It is in a unit of information, for example bits, nats or hartleys. In the
measurement space, the information content is the difference between the entropy of the prior estimate of y
and the posterior estimate Rodgers, 2000]. The degrees of freedom (ds) for signal describe the number of
useful independent quantities [Rodgers, 2000]. It is a measure of information.
The relations between information content, degrees of freedom for signal, the singular values of K and the
averaging kernel matrix are showed below:
5
ˆ 1
I n  A  (K T S1K  S a1 )S a1  SS
a
1
1
2
H   i ln(1  i )   ln I n  A
2
2
d s   i i2 /(1  i2 )  tr ( A)

1
S a : the a priori covariance matrix;
S : the measurement error covariance matrix;
K : the Jacobian; A: the averaging kernel;
1
i : the eigenvalues of K
K  S 2 KS a2
We did the information analysis for 3 retrievals. Table 1 shows the information content, degrees of freedom
and first three eigenvalues of K . Here, about 3 ppm variation of CO2 was used in the three cases for the a
priori covariance matrix. The SNR for both OCO and TES is ~ 100. The analysis shows that for individual
OCO/TES retrieval, the degrees of freedom for signal are both less than 1. However, the simultaneous
retrieval of OCO and TES increase es the degrees of freedom to 1.2. The advantages of joint retrieval are also
displayed in the enhanced information content. In Fig. 3 the first three eigenfunction for each case are
showed.
OCO and TES (AIRS) provide the complementary components of the global remote sensing system that can
be used to quantify atmospheric CO2. The information analysis suggested that there are benefits from the
simultaneous retrieval. With these observations, we can better characterize geographic distributions of CO2
sinks and their variability.
H
d
λ
1
λ
2
λ
3
OCO
2.78
0.99
6.76
0.99
0.01
TES
1.22
0.87
2.02
0.26
0.04
OCO&TES 3.01
1.20
7.04
0.52
0.09
s
Table 1. Information analysis
6
Fig. 3 First three eigenfunctions for (a) OCO retrieval, (b) TES retrieval and (c) simultaneous retrieval
Conclusions
The information analysis did show us the benefit of the simultaneous retrieval of OCO and TES due to the
complementary pieces of information provided by OCO and TES/AIRS respectively. There is considerable
increase in the information content and degrees of freedom in the simultaneous retrieval compared to
retrieval using only OCO or only TES data. Therefore, the simultaneous retrieval with high-resolution,
spectrometric approach has the potential for highly accurate measurement for the CO2 vertical profile.
7
Future work
In this work, we only use one band for OCO/TES retrieval. For the future work, more TES (and AIRS)
bands and including the OCO strong CO2 band can be used for the information analysis and retrieval. In
addition, the simultaneous retrievals of OCO and TES (and AIRS) will be performed for a single clear sky.
A calibration and validation program of the retrieval CO2 vertical profile is also needed to identify and
eliminate the biases.
Reference
Crisp, D., R. M. Atlas, F. M. Breon, L. R. Brown, J. P. Burrows, P. Ciais, B. J. Connor, S. C. Doney,
I. Y. Fung, D. J. Jacob, C. E. Miller, D. M. O'Brien, S. Pawson, J. T. Randerson, P. J. Rayner, R. J.
Salawitch, S. P. Sander, B. Sen, G. L. Stephens, P. Tans, G. Toon, P. O. Wennberg, S. C. Wofsy, Y. Yung,
Z. M. Kuang, B. Chudasama, G. Sprague, B. Weiss, R. Pollock, D. Kenyon, and S. Schroll, 2004: The
Orbiting Carbon Observatory (OCO) mission. Advances in Space Research, 34, 700-709.
Kuang, Z. M., J. Margolis, G. Toon, D. Crisp, and Y.L. Yung, 2002: Spaceborne Measurements of
Atmospheric CO2 by High-resolution NIR Spectrometry of Reflected Sunlight: An Introductory Study.
Geophys. Res. Lett., 29, 1716-1719.
Rodgers, C. D., 2000: Inverse Methods for Atmospheric Sounding: Theory and Practice. World
Scientific, 256 pp.
8