co 2 lineshapes near 2060nm thinh q. bui california institute of technology isms 2014
TRANSCRIPT
CO2 Lineshapes Near 2060nm
Thinh Q. BuiCalifornia Institute of Technology
ISMS 2014
Orbiting Carbon Observatory 2
• OCO-2 aims to determine CO2 concentrations to a precision of 1 ppm (0.3%)
• Satellite measurements will provide answers to important carbon cycle questions, such as:
-What are the present CO2 sinks?
-Why are they so variable?
-How will they respond to climate change?
• These satellite measurements will require the most accurate laboratory measurements of spectroscopic parameters.
Spectral Bands Recorded by OCO-2 Spectrometer "weak” "strong" O2 A-BAND CO2 CO2
B-X (0,0) Band 30013-00001 20013-00001
765 nm
Provide constraints on surface pressure, optical path length, and cloud/aerosol distribution
Most linear with respect to CO2 column
Provide constraints on cloud/aerosol distribution, water vapor column, and temperature profiles
Previous Works
Non-Voigt line profiles are necessary to properly model CO2 lineshapes (“weak” & “strong” regions)
I. FTS measurements (Devi et al. (2007) J. Mol. Spec., Predoi-Cross et al. (2007, 2009) J. Mol. Spec.)
II. Laser based measurements (Hikida et al. (2005, 2006) J. Mol. Spec., Joly et al. (2008) JQSRT)
A. Casa et al. (2009) J. Chem. Phys measured R(12) in the (20012)(00001) band. Retrieved an unphysical narrowing parameter
B. Long et al. (2011) J. Chem. Phys. measured R(16) in the (30012)(00001). Treating simultaneous Dicke narrowing and speed dependence gave the best fit residuals
What line profile model is optimal for CO2 strong band?
2-4m2060nm Laser
OI
Servo
Frequency Stabilized HeNe
AOM
l-meter
DDG
Ringdown Cavity PZT
R=99.99%
Instrumentation
Leff=23.5 km
0 20 40 60 80tim e (s)
-0.006
0
0.006-0.006
0
0.006
0
0.4
0.8
1.2
Rin
g-d
own
sign
al (
V) 300 shot average:
= 3 .7097(53) s
= 6 .3783(82) s
Re
sid
ua
ls (
V)
Empty-cavity
Absorbing medium
Long et al. 2012 Chem. Phys. Lett., 536, 1-8
Experimental Conditions:
• ~90ppm CO2 in synthetic air (no Ar)
• 7 kPa – 27 kPa (50-200 Torr) total pressure
• Room temperature
Measured Spectra
Air-broadened CO2
• Isolated line regime within impact approximation (avoid line-mixing)
Line Profiles for Modeling Spectra
• Doppler Broadening (Gaussian)• Collisional Broadening (Lorentzian)
• Dicke Narrowing- narrowing of Doppler profile- collisions change velocity, not quantum states- Soft (diffusional) and hard collisions
• Speed Dependence - narrowing of Lorentz profile, asymmetry
(speed-dependent shifts)
VOIGT PROFILE (VP)
RAUTIAN-SOBELMAN(GALATRY (GP)/NELKIN-GHATAK (NGP))
SD VOIGT (SDVP)SD NELKIN-GHATAK(SDNGP)
Correlated effectsCorrelated effects – simultaneous velocity & phase changing collisions
PARTIALLY-CORRELATED SD NELKIN-GHATAK(pcSDNGP)
Single Spectrum Fits
1
2
3
4
5
6
(c)
-1 [1
0-6 c
m-1
]
-40
0
40 VP
0 10 20 30 40 50 60 70 80 90
(G Hz)
-8
0
8 SD N G P
-8
0
8 G P
-8
0
8 N G P
-8
0
8 SD VP
Re
sidu
als
(10
-9 c
m-1
)
210 Torr 155.8 Torr 95.4 Torr 48.1 Torr
S ingle-spectrum fits - R (24) line
Q F m ulti = 854
Q F m ulti = 3374 (3040)
Q F m ulti = 3007
Q F m ulti = 3079
Q F m ulti = 3554 (3030)
1
2
3
4
(c)
-1 [1
0-6 c
m-1
]
-40
0
40 VP
0 10 20 30 40 50 60 70 80 90
(G Hz)
-8
0
8 SD N G P
-8
0
8 G P
-8
0
8 N G P
-8
0
8 SD VPR
esi
dua
ls (
10-9
cm
-1)
180.3 Torr 127.8 Torr 95.4 Torr 48.1 Torr
S ingle-spectrum fits - R (30) line
Q F m ulti = 730
Q F m ulti = 2841 (2667)
Q F m ulti = 2668
Q F m ulti = 2689
Q F m ulti = 2864 (2654)
- All parameters are fitted independently, ignoring correlations
Black = without speed-dependent collisional shifts, Red = with speed-dependent collisional shiftsObservation of simultaneous Dicke narrowing and speed-dependence effects!
opt = diff - ( - i) = Re(opt)+iIm(opt)
Correlation between Dicke and speed-dependent narrowing mechanisms
• If Dicke narrowing was the only source of line narrowing, Re(nopt) would be linear with pressure
• Observed quadratic behavior
• To a good approximation, aw is independent of pressure
• Fitted aw has negative slope
0 5 10 15 20 25 30
Pressure (kPa)
0
0.1
0.2
0.3
Re
(op
t) (G
Hz)
R (24) N G P sing le-spectrum fit
R (24) G P sing le-spectrum fit
d iffp
0 1 2 3 4
L / D
5 10 15 20 25 30
Pressure (kPa)
0.06
0.07
0.08
0.09
0.1
0.11
aW
Linear fit
R (24) SD VP single -spectrum fit(a) (b)diffusional value
observed
ndiff(p)
Multi-spectrum fits
- Since speed-dependence and Dicke narrowing parameters are competitive effects, multi-spectrum fits are necessary to remove correlation
• Fix Doppler widths calculated from measured laboratory temperature
• Lorentz halfwidths were constrained to be linear with pressure G(p) = gairp
• Collisional narrowing frequencies were constrained to be linear with pressure ndiff(p) = *b p
1
2
3
4
5
6
(c)
-1 [1
0-6 c
m-1
]-40
0
40
0 10 20 30 40 50 60 70 80 90
(G H z)
-8
0
8
-8
0
8
-8
0
8
-8
0
8
Re
sid
uals
(10
-9 c
m-1
)
210 Torr 155.8 Torr 95.4 Torr 48.1 Torr
M ulti-spectrum fits - R (24) line
-8
0
8
VP
SD N G P
G P
N G P
SD VP
pC SD N G P
Q F m ulti = 802
Q F m ulti = 2828 (2695)
Q F m ulti = 2293
Q F m ulti = 2364
Q F m ulti = 2864 (2702)
Q F m ulti = 2893 (2798)
1
2
3
4
(c)
-1 [1
0-6 c
m-1
]
-40
0
40
0 10 20 30 40 50 60 70 80 90
(G H z)
-8
0
8
-8
0
8
-8
0
8
-8
0
8
Re
sid
uals
(10
-9 c
m-1
)
180.3 Torr 127.8 Torr 95.4 Torr 48.1 Torr
M ulti-spectrum fits - R (30) line
-8
0
8
VP
SD N G P
G P
N G P
SD VP
pC SDN G P
Q F m ulti = 684
Q F m ulti = 2669 (2535)
Q F m ulti = 2252
Q F m ulti = 2322
Q F m ulti = 2689 (2582)
Q F m ulti = 2688 (2675)
Residuals look reasonable, what about the fitted parameters?
0 5 10 15 20 25 30
Pressure (kPa)
-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
Re(
op
t) (G
Hz)
single-spectrum R (24) SDNG P fits
m ulti-spectrum R(24) SDN G P fit
d iffp
0 5 10 15 20 25 30
Pressure (kPa)
-0.05
0
0.05
0.1
0.15
0.2
a W
single-spectrum R (24) SDNG P fits
m ulti-spectrum R(24) SDN G P fit
fitted aW for Re(vopt) constra ined from SD NG P m ulti-spectrum fit
(a) (b)
Removing correlation between Dicke narrowing & speed-dependent by multi-spectrum fits
Multi-spectrum fits provided physical results, but Re(nopt) << ndiff.
Recall: opt = diff - ( - i) = Re(opt)+iIm(opt)
First evidence for the correlations between velocity-changing and phase-changing collisions (v-p correlations) in CO2 spectra!
ndiff(p)
Recall: opt = diff - ( - i) = Re(opt)+iIm(opt)
Fitted v-p correlation parameters (Im(nopt) and as) to determine value for the correlation parameter h
0 5 10 15 20 25 30
Pressure (kPa)
-0.1
-0.05
0
0.05
0.1
0.15
Im(
op
t) (G
Hz)
single-spectrum R (24) pCSDN G P fits
m ulti-spectrum R(24) pCSD NG P fit
0 5 10 15 20 25 30
Pressure (kPa)
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
a S
single-spectrum R (24) pCSDN G P fits
m ulti-spectrum R(24) pCSD NG P fit
fitted aS for Im (vopt) constra inedfrom pC SDNG P m ulti-spectrum fit
(a) (b)
hG hD
R(24) 0.116(4) 0.22(6)R(30) 0.1101(12) 0.020(17)
Implications for OCO-2 and remote sensing
Large systematic uncertainties (up to 5%) from spectroscopy are present in retrieved CO2 concentration if improper line profiles are used!
0.98
0.99
1
1.01
1.02
A/A
pCS
DN
GP
R e ference leve l (pC S D N G P)
R (24) line
R (30) line
0.92
0.93
0.94
0.95
0.96
0.97
0.98
0.99
1
1.01
/ p
CS
DN
GP
R e ference leve l (pC S D N G P)
R (24) line
R (3 0 ) lin e
(a) (b)V
P
SD
VP
GP
NG
P
SD
NG
P
pC
SD
NG
P
HIT
RA
N'1
2
VP
SD
VP
GP
NG
P
SD
NG
P
pC
SD
NG
P
Conclusions
• CO2 strong band lineshapes display simultaneous Dicke narrowing and speed-dependence (Long et al. 2011), but additional velocity-phase changing correlations & asymmetry are observed at 2.06mm CO2
• pCSDNGP profile (partially correlated SDNGP) was ideal at this isolated line (low pressure) regime
• Neglecting narrowing effects introduces large deviations (5%) in the measured air-broadening parameter and 1% deviations in the integrated areas for CO2
• Multi-spectrum fitting is necessary to quantify and remove correlations
Milinda Rapusinghe, Daniel Hogan, Mitchio OkumuraCaltechDavid Long, Vincent SironneauNIST, GaithersburgAgata Cygan, Roman Ciuryło, Daniel LisakNicolaus Copernicus University
$$$ NASA & NESSF Fellowship
Acknowledgements