Ground-based Solar Absorption Studies for the Carbon Cycle Science by Fourier Transform Spectroscopy

(CC-FTS) Mission

Dejian Fu, K. Sung, C.D. Boone,

K. Walker and P.F. BernathJune 19th, 2007

1950 1960 1970 1980 1990 2000 2010310

320

330

340

350

360

370

380

390

Year

CO

2 C

on

cen

trat

ion

(p

pm

v)

Atmospheric Carbon Dioxide

Measured at Mauna Loa, Hawaii

Monthly Mean Values

Detrended Values

Introduction - Previous Observations

Jan06 May06 Sep06 Jan07375

380

385

390Annual Cycle

Keeling Curve

About the Keeling Curve The concentrations of CO2 were obtained by in situ surface measurements. It is the first evidence of atmospheric CO2 increase (Keeling C.D., Tellus, 1960). It shows a persistent year-to-year increase.

About the annual cycle of CO2

Biospheric respiration in winter produces CO2

Photosynthesis in summer consumes CO2

a Two plots were drawn using the data downloaded from the link: ftp://ftp.cmdl.noaa.gov/ccg/co2/in-situ/

Introduction - Previous Observations Washenfelder et al. GRL 2003 Dufour et al. JGR 2003

a VMR: Volume Mixing Ratio b FTS: Fourier Transform Spectrometer

Location: Kitt Peak, Arizona, USA Instrument: 1 meter McMath-Pierce FTSb

Time: 1978 - 1996

CO2

CH4 VMRa time series

CH4 VMRa annual cycle

CO2

Satellites are carrying out pioneering studies on global carbon budgets.

Orbiting Carbon Observatory, OCO (NASA, 2008) Three near infrared grating spectrometers Spectral resolution ~ 0.3 cm-1

Measuring CO2 only at high (1 km) spatial resolution

Greenhouse Gases Observing Satellite, GOSAT (JAXA, Japan, 2008) Fourier Transform Spectrometer Low spatial resolution ~ 10 km Single pixel detector

OCO

Satellite Missions for Greenhouse Gas Monitoring

Satellite images were downloaded from the homepages of each mission.

GOSAT

SCIAMACHYon ENVISAT

AIRS on AQUA

TES on AURA

IASI on METOP-A

MOPITT on AURA

CC-FTS Mission Next generation satellite mission for

carbon cycle science Heritage from the successful SCISAT-1

mission Orbits the earth at the tail of ‘A train’ Monitors many species using a satellite-

borne FTS: CO2, CH4, CO, N2O, (four gases listed in IPCCa 2007 report) and O2

High spatial resolution ~ 1 km Precision goals under most favorable

situation: ~ 0.3% for CO2 ; ~ 1% for other species

Global coverage To characterize distribution and variability of

sources and sinks on a regional scale Advanced study for the mission on the

Spectral resolution? Spectral range? Spectroscopic parameters?…

A: CCFTS orbiting the earth and primary modesB: Coverage strategy using a detector array with

8x8 elements and cross-track scanningC: An area 8 km square per FOV covers the area

of Kitchener-Waterloo, Canadian twin city with a population of 300,000

D: Footprint on University of Waterloo campus

Nadir

Glint

a Intergovernmental Panel on Climate Change (IPCC)

Ground-based Atmospheric Absorption Spectra

In order to determine optimum values for a greenhouse gas mission, advanced studies were accomplished for the mission on

Spectral resolution? Spectral range?

Spectroscopic parameters? …

Measurement sites Preliminary comparisons of spectral resolution effects using spectra from

Kiruna (67.84ºN, 20.41ºE, and 419 m above sea level), Sweden on 1st Apr. 1998 a.

Spectral region: 3950 cm-1 to 7140 cm-1;

Instrumentation: Bruker IFS 120 HR FTS. Further investigations using spectra from Kitt Peak, Arizona (31.9ºN,

111.6ºW and 2.1 km above sea level) and Waterloo, Ontario (43.5ºN, 80.6ºW and 0.3 km above sea level)

a Meier A, et al., Spectroscopic atlas of atmospheric microwindows in the middle infrared (2nd edition). IRF Technical Report No.48, ISSN 0284-1738, Kiruna, Apr 2004.

Atmospheric Absorption Spectra at Kiruna, Sweden

Spectral resolution of a FTS = 1/ (2 x Maximum Optical Path Difference) 0.0041 cm-1 (MOPD 120 cm), the typical resolution setting used by NDACCa FTSs 0.0100 cm-1 (MOPD 50 cm), the resolution setting close to that used by the FTSs in TCCONb

0.3000 cm-1 (MOPD 5/3 cm), the resolution setting used in OCO mission 0.1000 cm-1 (MOPD 5 cm), the resolution setting candidate for the CC-FTS mission

CO2

a The Network for the Detection of Atmospheric Composition Change: http://www.ndsc.ws/b Total Carbon Column Observing Network, Washenfelder R.A., et al. J. Geophys. Res. 2006

CO

Ground-based Solar Absorption Studies Using Fourier Transform Spectrometers at Two Observatories

Further comparisons of spectral resolution of 0.01 cm-1 and 0.1 cm-1 settings using spectra from two sites.

The McMath-Pierce Fourier Transform Spectrometer Located National Solar Observatory at Kitt Peak, Arizona A folded cat’s-eye Michelson interferometer housed in a vacuum vessel

(Maximum Optical Path Difference: 100 cm; Spectral Range: 550 to 45,000 cm-1) and more detail available in the link: http://nsokp.nso.edu/

The WAO DA8 Fourier Transform Spectrometer Located at Waterloo Atmospheric Observatory, Waterloo, Ontario A Michelson interferometer using ABB Bomem’s dynamical alignment

techniques (Maximum Optical Path Difference: 25 cm; Spectral Range: 400 to 55,000 cm-1)

Observations at National Solar Observatory at Kitt Peak, Arizona (31.9ºN, 111.6ºW and 2.1 km above sea level)

Date: July 25th, 2005

Region: 1700–15000 cm-1

Resolution: 0.01 cm-1

Detector: InSb

Scan Time: ~30 minutes

Filters: RG715

Number of Scans: 2

Observations at Waterloo Atmospheric Observatory at Waterloo, Ontario (43.5ºN, 80.6ºW and 0.3 km above sea level)

Sun TrackerWAO DA8 Spectrometer

Date: November 22nd, 2006

Region: 2000–15000 cm-1

Resolution: 0.1 cm-1

Detector: InSb and Si

Scan Time: ~16 minutes

Filters: RG715

Number of Scans: 20

Spectral Overview of the Atmospheric Absorption Spectra Recorded on 22nd November 2006 Using WAO DA8 Spectrometer

2000 4000 6000 8000 10000 12000 14000

0

0.2

0.4

0.6

0.8

1

Wavenumber (cm-1)

Inte

nsi

ty (

arb

itra

ry u

nit

)

InSb DetectorSi Detector

CH4

CH4

CO2

CO2

O2

O2

CO

N2O

Spectral resolution0.1 cm-1

InSb and Si detectors InSb: 2000 to 15000 cm-1

Si: 8500 to 15000 cm-1

Solar Zenith Angles (SZA) 66.61 InSb band 67.33 Si band

Number of Scans20

Scan Time~16 minutes

WAO

6100 6150 6200 6250 6300 6350 6400

0.2

0.6

1.0

4800 4825 4850 4875 4900 4925 4950

0.2

0.6

1.0

12900 12950 13000 13050 13100 13150

0.2

0.6

1.0

12900 12950 13000 13050 13100 13150

0.2

0.6

1.0

Inte

nsi

ty (

arb

itra

ry u

nit

)

Wavenumber (cm-1)

CO2 1.57 m from InSb detector

CO2 2.06 m from InSb detector

O2 0.76 m from InSb detector

O2 0.76 m from Si detector

Data Analysis – from spectra to retrieved columns Retrieval Code: SFIT2 (version 3.91)

Widely used for the analysis of ground-based infrared solar spectra Jointly developed at the NASA-Langley Research Center and the National Institute of Water

and Atmospheric Research at Lauder, New Zealand. Employs the optimal estimation method of Rodgers (1976, 1996 and 2000). Includes the a priori constituent profiles in the retrievals in a statistically sound manner Allows the simultaneous retrieval of the vertical profile of the target molecule together with

the total columns of interfering species

Model atmosphere 1) Pressure and Temperature profiles: combining the National Centers for Environmental

Prediction (NCEP) data (obtained from the Goddard Auto mailer science@hyperion.gsfc.nasa.gov) and the Mass-Spectrometer-Incoherent-Scatter model (MSIS-2000) output.

2) In the retrievals of spectra recorded at WAO: A priori VMRsa from the HALogen Occultation Experiment (HALOE) v.19 satellite data and mid-latitude daytime 2001 Michelson Interferometer for Passive Atmospheric Sounding (MIPAS). In the retrievals of spectra recorded at Kitt Peak: A priori VMRsa from Mark IV balloon FT-IR spectra obtained in northern mid-latitudes by G. C. Toon et al. (1995).

Spectroscopic parameters HITRAN 2004 data base [Rothman L.S. et al. 2004]a VMR: Volume Mixing Ratio

Fitting residuals for CO2 at 1.57 m

6190 6200 6210 6220 6230 6240 6250-5

0

5

Res

idu

als(

%)

6180 6190 6200 6210 6220 6230 6240 6250 6260

0

0.2

0.4

0.6

0.8

1

Wavenumber (cm-1)

Am

plit

ud

e (a

rbit

ary

un

it)

Meas. Spec.

Calc. Spec.

6180 6190 6200 6210 6220 6230 6240 6250 6260-5

0

5

Res

idua

ls(%

)

6180 6190 6200 6210 6220 6230 6240 6250 6260

0

0.2

0.4

0.6

0.8

1

Wavenumber (cm-1)

Am

plit

ude

(arb

itar

y un

it)

Meas. Spec.

Calc. Spec.

Date: 25th July 2005 Spectral Resolution: 0.01 cm-1

Solar Zenith Angle: 49.1

The fitting residuals are in the same order of Yang et al. 2002, Geophys. Res. Lett. 29(9).

Date: 22nd November 2006 Spectral Resolution: 0.1 cm-1

Solar Zenith Angle: 66.6

NSO WAO

30013-00001 (1 +42+ 3, 0 = 6228 cm-1) transition Numerous absorption features from CO2 with weak absorption by H2O, HDO and additionally CH4 Many lines with a wide range of intensities, which provides good retrieval sensitivities in both the

stratosphere and troposphere Thermal emission from the atmosphere and instrument are also negligible at these short wavelengths

Averaging Kernela for CO2 at 1.57 m

-0.2 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.00

10

20

30

40

50

60

Averaging kernel values

Alt

itu

de

(km

)

15.0 - 38.0 km

8.0 - 15.0 km

31.0 - 53.0 km

2.1 - 8.0 km

2.1 - 100.0 km

-0.2 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.00

10

20

30

40

50

60

Averaging Kernel Values

Alt

itu

de

(km

)

8.0 - 15.0 km

31.0 - 53.0 km

15.0 - 38.0 km

0.3 - 8.0 km

0.3 - 100.0 km

NSO WAO

Typical averaging kernels for CO2 retrievals in the thermal infrared at 14 microns and the near infrared at 1.6 microns (Crisp et al., 2004, Advances in Space Research)

a The averaging kernel is the derivative of a derived parameter with respect to its a priori state value, i.e., when this derivative is small (nearly 0) all of the information comes from the a priori and when it is large (near 1) then the information in the retrieval comes mainly from the measured data.

The required near surface to upper troposphere sensitivity (much better sampling than those observations thermal infrared region used in the previous missions) is achieved.

Only minor differences are seen between spectra with resolutions of 0.01 cm-1 and 0.1 cm-1

Fitting Residuals for CH4 at 1.68 m

5890 5900 5910 5920 5930 5940 5950 5960 5970 5980-5

0

5

Res

idua

ls(%

)

5890 5900 5910 5920 5930 5940 5950 5960 5970 5980

0

0.2

0.4

0.6

0.8

1

Wavenumber (cm-1)

Am

plit

ude

(arb

itar

y un

it)

Meas. Spec.

Calc. Spec.

5890 5900 5910 5920 5930 5940 5950 5960 5970 5980-5

0

5

Res

idua

ls(%

)5890 5900 5910 5920 5930 5940 5950 5960 5970 5980

0

0.2

0.4

0.6

0.8

1

Wavenumber (cm-1)

Am

plit

ude

(arb

itar

y un

it)

Meas. Spec.

Calc. Spec.

The fitting residuals are of the same order as in Washenfelder et al. 2003, Geophys. Res. Lett.

WAONSO

Date: 25th July 2005 Spectral Resolution: 0.01 cm-1

Solar Zenith Angle: 49.1

Date: 22nd November 2006 Spectral Resolution: 0.1 cm-1

Solar Zenith Angle: 66.6

The spectral region from 5880 to 5996 cm-1 was investigated for the CH4 retrieval.

-0.2 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.00

10

20

30

40

50

60

Averaging kernel values

Alt

itu

de

(km

)

15.0 - 38.0 km

8.0 - 15.0 km

2.1 - 100.0 km

2.1 - 8.0 km

Averaging Kernela for CH4

-0.2 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.00

10

20

30

40

50

60

Averaging Kernel ValuesA

ltit

ud

e (k

m)

15.0 - 38.0 km 8.0 - 15.0 km 0.3 - 8.0 km 0.3 - 100.0 km

Degrees of Freedom for signal (DOFS) = 2.58 Degrees of Freedom for signal (DOFS) = 2.12

WAONSO

The required near surface to upper troposphere sensitivity (much better sampling than those observations thermal infrared region used in the previous missions) is achieved.

Only minor differences are seen between spectra with resolutions of 0.01 cm-1 and 0.1 cm-1

a The averaging kernel is the derivative of a derived parameter with respect to its a priori state value, i.e., when this derivative is small (nearly 0) all of the information comes from the a priori and when it is large (near 1) then the information in the retrieval comes mainly from the measured data.

Fitting residuals for O2 at 0.76 m

The fitting residuals are in the same order as in Yang et al. 2005, JQSRT.

13075 13080 13085 13090 13095 13100 13105 13110-5

0

5

Res

idu

als(

%)

13075 13080 13085 13090 13095 13100 13105 13110

0

0.2

0.4

0.6

0.8

1

Wavenumber (cm-1)

Am

plit

ud

e (a

rbit

ary

un

it)

Meas. Spec.

Calc. Spec.

13075 13080 13085 13090 13095 13100 13105 13110-5

0

5

Res

idu

als(

%)

13075 13080 13085 13090 13095 13100 13105 13110

0

0.2

0.4

0.6

0.8

1

Wavenumber (cm-1)

Am

plit

ud

e (a

rbit

ary

un

it)

Meas. Spec.Clac. Spec. WAONSO

Date: 25th July 2005 Spectral Resolution: 0.01 cm-1

Solar Zenith Angle: 49.1

Date: 22nd November 2006 Spectral Resolution: 0.1 cm-1

Solar Zenith Angle: 66.6

b1Σ+g X3Σ-

g near infrared band (0 = 13121 cm-1) Provide constraints on both the surface pressure and optical path length variations associated with

scattering by aerosols in the atmosphere By taking the ratio of columns of CH4 and CO2 to O2 columns, systematic errors will be reduced as

long as they are measured under the same conditions Also used for cloud detection

Features in the Fitting Residuals

The largest discrepancies between the calculated and the measured transmittances are on the order of a few percent and are observed in the

vicinity of the absorption line centers.

Similar systematic fitting residual patterns in terms of positions and amplitudes also appeared in the results of previous work. They mainly

arise from the spectroscopic parameters including line intensity, self- and air-broadening coefficients, and self- and air-shift coefficients.

Away from the absorption lines, the fitting residuals from spectra recorded at NSO are generally larger than those obtained using spectra recorded at

WAO. This is because the WAO spectra have a higher Signal-to-Noise Ratio (SNR) mainly because of their lower spectral resolution.

The absolute accuracy of the CO2 retrievals obtained using spectroscopic parameters from HITRAN 2004 is expected to be limited to ~2% [Devi et al.,

. JMS In Press]. Recent studies show the improvements in the CO2 spectroscopic parameters

in the spectral region of 4550 to 7000 cm-1 with a precision of 1% or better. Very recently, Devi et al. [Devi et al., JMS In Press] made further

improvements in the CO2 spectroscopic parameters for the 6348 cm-1 band by considering line mixing and using speed-dependent Voigt line shape

functions. This work by Devi et al. provides the possibility of remote sensing CO2 with ~ 0.3% precision.

As demonstrated by Boone et al. JQSRT 2007, the use of speed-dependent Voigt line shape functions improves tropospheric remote sensing.

Deficiencies in spectroscopic parameters were also found for the CH4 and O2 retrievals. For example, the fitting residuals show obvious difficulty in fitting

the O2 continuum (not included in our forward model) for both 1.27 μm and 0.76 μm bands. However, no recent published work has presented

improvements to the spectroscopic parameters for CH4 and O2 over those in HITRAN 2004.

Line Parameters for Greenhouse Gases

CO2 Column Averaged Volume Mixing Ratio (VMR)

CO2 column averaged Volume Mixing Ratio = CO2 column / total column.

However, humidity can increase the total column by 0.5%, but does not change the CO2 column. Essentially the CO2 gets ‘diluted’ by the H2O.

The dry-air CO2 VMR is more directly related to source sinks and is a better tracer since it is not being influence by evaporation or condensation of H2O.

Fortunately, the O2 is diluted by the same amount as the CO2 since we measure all of species at the same time (i.e. same air mass). The mole fraction of O2 in dry air is fairly constant. O2 column / dry air total column = 0.2095.

Hence, we can get the CO2 VMR using the following formula CO2 column averaged VMR in dry air = CO2 column / O2 column / 0.2095Similarly, CH4 column averaged VMR in dry air = CH4 column / O2 column / 0.2095

CO2 Column Averaged VMR in Dry Air at WAO

a The simultaneously observed total columns of O2 at 0.76 m were used.b The precision of the observations can be estimated from one sigma standard deviation of the results of repeated

measurements under similar conditions, e.g., precision = 1 of CO2 VMR / mean value of CO2 VMR.

The precisions of the column-averaged VMRs of CO2 in dry airb

1.00 % for the CO2 6228 cm-1 band at 1.57 m; 1.07% for the CO2 6348 cm-1 band at 1.57 m; 0.60% for the CO2 band at 2.06 m;

CO2

6348 cm-1 band at 1.57 mO 6228 cm-1 band at 1.57 m 2.06 m bandO2

O 1.27 m band with HITRAN 2004 1.27 m band with Goldman * 0.76 m band with HITRAN 2004

[CO2]/[O2]/0.2095a

6348 cm-1 band at 1.57 m O 6228 cm-1 band at 1.57 m 2.06 m band

64 66 68 70 72 74 76 78

6

7

8

9x 10

21

Ver

tica

l Co

lum

n (

#/cm

2)

CO2

64 66 68 70 72 74 76 78

3.5

4.0

4.5

5.0 O2

66 68 70 72 74 76 78300

350

400

450

Solar Zenith Angle (degree)

Mix

ing

Rat

io (

pp

mv)

[CO2]/[O

2]/0.2095

x 1024

WAO

WAO

WAO

CH4 Column Averaged VMR in Dry Air at Two Sites

30 35 40 45 50 55 60 65 70 75 802

3

4x 10

19

Ver

tica

l C

olu

mn

(#/

cm2)

CH4 1.68 m NSO

64 66 68 70 72 74 76 78 802

3

4x 10

19

CH4 1.68 m WAO

30 35 40 45 50 55 60 65 70 75 801200

1500

1800

2100

Mix

ing

Rat

io (

in p

pb

v)[CH

4]/[O

2]/0.2095 1.68 m NSO

64 66 68 70 72 74 76 78 801200

1500

1700

2000

Solar Zenith Angle (degree)

[CH4]/[O

2]/0.2095 1.68 m WAO

a The simultaneously observed total columns of O2 at 0.76 m were used.b The precision of the observations can be estimated from one sigma standard deviation of the results of repeated

measurements under similar conditions, e.g., precision = 1 of CH4 VMR / mean value of CH4 VMR.

CH4 1.68 m band from NSO* 1.68 m band from WAO

[CH4]/[O2]/0.2095a ∆ 1.68 m band from NSO * 1.68 m band from WAO

The precisions of the column-averaged VMRs of CH4 in dry airb

1.07 % for the CH4 1.68 m band from NSO;

1.13% for the CH4 1.68 m band from WAO;

Summary An advanced study was carried out for a next generation satellite mission named

CCFTS, which will investigate greenhouse gas budgets, using absorption spectra recorded at three ground-based observatories.

In order to obtain the absorption features of CH4, CO2, CO, and N2O together with the

O2 A-band in a single spectrum and investigate the desired spectral resolution (0.01 cm-1 and 0.1 cm-1), further observations over a broad spectral region from 2000 to 15000 cm-1 were taken at Kitt Peak and Waterloo at a resolution of 0.01 cm-1 and 0.1 cm-1, respectively. The vertical sampling of these observations is quantified by computing averaging kernels as defined in the Rodgers optimal estimation method. The vertical sampling of observations with a spectral resolution of 0.1 cm -1 is similar to those with a spectral resolution of 0.01 cm-1. Precision of column averaged CO2 and CH4 VMR are about 1% for the observations at NSO and WAO. Considering overall performances and costs of the mission, a spectral resolution of 0.1 cm-1 (MOPD = 5 cm) is recommended for the CC-FTS mission.

Systematic fitting residuals are obvious in all of our retrievals and have been noted previously. These residuals are due to the deficiencies in the spectroscopic line parameters in the HITRAN 2004 database. To improve the precision of atmospheric observations, new laboratory measurements on the spectroscopic parameters are required.

Thanks to Meier A. and IRF, Sweden providing spectra recorded at Kiruna, Sweden.

Funding:

Canadian Space Agency, Natural Sciences and Engineering Research Council of Canada (NSERC), and other sources

Acknowledgements